i; il N 1si N ! i__1 1 ^~~~~~~~~~~~~~~~~~~~~l
120 2 S g g , , ,, ,,, = ..,.. ,___,,_ . , -_ or,
|E | fi | | u u a~~~~~~~~~~~~~~~~~~~~~~~
u E | g � E3 '_ ~~~~~~~~~~~~~~~~~~~~~~~~
� g | | | 11 z ~~~~~~~~I
E~~~~~~~~~~~~~~~~~~~~~~~~~ G ||11
E E 0 E B *1 ,_s -__~~~~~~~~~~~~~~~~~~~l:I
g | i | | 111 [~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-,
g E | S | | 1 11 ER _~~~~~-
g~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | 1I I
B g g B i B U * , ,_* __ I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~
| g g | 0 E I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 11
THE WORLD BANK
ECONOMIC REVIEW
EDITOR
Moshe Syrquin
CONSULTING EDITOR
Sandra Gain
EDITORIAL BOARD
Kaushik Basu, Cornell University and University of Delhi David Dollar
Francois Bourguignon, DELTA, Paris Gregory K. Ingram
Willem H. Buiter, University of Cambridge John Page
Gerald K. Helleiner, University of Toronto Lant H. Pritchett
Mark R. Rosenzweig, University of Pennsylvania Jacques van der Gaag
L. Alan Winters
The World Bank Economic Review is a professional journal for the dissemination of World Bank-
sponsored research that informs policy analyses and choices. It is directed to an international readership
among economists and social scientists in government, business, and international agencies, as well as in
universities and development research institutions. The Review emphasizes policy relevance and opera-
tional aspects of economics, rather than primarily theoretical and methodological issues. It is intended for
readers familiar with economic theory and analysis but not necessarily proficient in advanced mathematical
or econometric techniques. Articles will illustrate how professional research can shed light on policy
choices. Inconsistency with Bank policy will not be grounds for rejection of an article.
Articles will be drawn primarily from work conducted by World Bank staff and consultants. Before being
accepted for publication by the Editorial Board, all articles are reviewed by two referees who are not mem-
bers of the Bank's staff and one World Bank staff member; articles must also be recommended by at least one
external member of the Editorial Board.
The Review may on occasion publish articles on specified topics by non-Bank contributors. Any reader
interested in preparing such an article is invited to submit a proposal of not more than two pages in length
to the Editor.
The views and interpretations expressed in this journal are those of the authors and do not necessarily
represent the views and policies of the World Bank or of its Executive Directors or the countries they rep-
resent. The World Bank does not guarantee the accuracy of the data included in this publication and
accepts no responsibility whatsoever for any consequences of their use. When maps are used, the bound-
aries, denominations, and other information do not imply on the part of the World Bank Group any judg-
ment on the legal status of any territory or the endorsement or acceptance of such boundaries.
Comments or brief notes responding to Review articles are welcome and will be considered for publica-
tion to the extent that space permits. Please direct all editorial correspondence to the Editor, The World
Bank Economic Review, The World Bank, Washington, D.C. 20433, U.S.A.
The World Bank Economic Review is published three times a year (January, May, and September) by the
World Bank. Single copies may be purchased at $12.95. Subscription rates are as follows:
Individuals Institutions
1-year subscription US$30 US$50
2-year subscription US$55 US$95
3-year subscription US$70 US$130
Orders should be sent to: World Bank Publications, Box 7247-7956, Philadelphia, PA 19170-7956,
U.S.A. Subscriptions are available without charge to readers with mailing addresses in developing countries
and in socialist economies in transition. Written request is required every two years to renew such
subscriptions.
C) 1997 The International Bank for Reconstruction and Development / THE WORLD BANK
1818 H Street, N.W., Washington, D.C. 20433, U.S.A.
All rights reserved
Manufactured in the United States of America
ISBN 0-8213-3782-3; ISSN 0258-6770
Material in this journal is copyrighted. Requests to reproduce portions of it should be sent to the
Office of the Publisher at the address in the copyright notice above. The World Bank encourages dis-
semination of its work and will normally give permission promptly and, when the reproduction is for
noncommercial purposes, without asking a fee. Permission to make photocopies is granted through the
Copyright Clearance Center, Suite 910, 222 Rosewood Drive, Danvers, MA 01923, U.S.A.
This journal is indexed regularly in Current Contents/Social & Behavioral Sciences, Index to Interna-
tional Statistics, journal of Economic Literature, Public Affairs Information Service, and Social Sciences
Citation Index". It is available in microform through University Microfilms, Inc., 300 North Zeeb Road,
Ann Arbor, MI 48106, U.S.A.
THE WORLD BANK ECONOMIC REVIEW
Volume 11 May 1997 Number 2
Financial Market Fragmentation and Reforms in Ghana, 195
Malawi, Nigeria, and Tanzania
Ernest Aryeetey, Hemamala Hettige, Machiko Nissanke,
and William Steel
Civil Liberties, Democracy, and the Performance 219
of Government Projects
Jonathan Isham, Daniel Kaufmann, and Lant H. Pritchett
The Relative Efficiency and Implementation Costs 243
of Alternative Methods for Pricing Irrigation Water
Yacov Tsur and Ariel Dinar
Managing Price Risk in the Pakistan Wheat Market 263
Rashid Faruqee, Jonathan R. Coleman, and Tom Scott
Explaining Industrial Growth in Coastal China: 293
Economic Reforms ... and What Else?
Ashoka Mody and Fang-Yi Wang
Organized Labor and the Political Economy 327
of Product Market Distortions
Martin Rama
What Can New Survey Data Tell Us about Recent Changes 357
in Distribution and Poverty?
Martin Ravallion and Shaohua Chen
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2 195-218
Financial Market Fragmentation and Reforms
in Ghana, Malawi, Nigeria, and Tanzania
Ernest Aryeetey, Hemamala Hettige, Machiko Nissanke, and William Steel
This article reports the findings from surveys of formal and informal institutions and
their clients in Ghana, Malawi, Nigeria, and Tanzania. It investigates the hypothesis
that reforming financially repressive policies would not be sufficient to overcome frag-
mentation of financial markets because of structural and institutional barriers to inter-
actions across different market segments. The four countries have substantially frag-
mented financial markets, with weak linzkages between formal and informal segments
and interest rate differentials that cannot be adequately explained by differences in
costs and risks. Nevertheless, the relatively low transaction costs and loan losses of
informal institutions indicate that they provide a reasonably efficient solution to infor-
mation, transaction cost, and enforcement problems that exclude their clients from
access to formal banking services. The findings imply that financial liberalization and
bank restructuring in the African context should be accompanied by complementary
measures to address institutional and structural problems, such as contract enforce-
ment and information availability, and to improve the integration of informal and
formal financial markets.
Expecting to hasten financial deepening and reduce fragmentation of financial
markets, governments in many Sub-Saharan African countries initiated finan-
cial policy reforms in the 1980s. This article examines the experience in four
countries and raises the issue of whether policy reform programs need to be
accompanied by measures to address the institutional and structural problems
of financial systems in Africa. We use survey findings to compare the behavior
of informal and formal financial markets in handling risks and transaction costs.
We evaluate indicators of financial deepening and lending to the private sector.
Ernest Aryeetey is with the Institute for Statistical, Social, and Economic Research at the University
of Ghana, Hemamala Hettige is with the Policy Research Department at the World Bank, Machiko
Nissanke is with the School of Oriental and African Studies at the University of London, and William
Steel is with the Private Sector Finance Group of the Africa Region at the World Bank. This study was
supported by the World Bank Research Committee, the Swedish International Development Association,
the Overseas Development Institute, the School of Oriental and African Studies (University of London),
and the Leverhulme Trust. The fieldwork was conducted by Ernest Aryeetey (Ghana), Mboya Bagachwa
(Tanzania), Chinyamata Chipeta (Malawi), M. L. C. Mkandawire (Malawi), and Adedoyin Soyibo
(Nigeria), with assistance from Martin Wall on the flow of funds and Deborah Johnston on editing. The
authors are grateful for comments from the referees and from Gerald Caprio, Carlos Cuevas, Jean-
Jacques Deschamps, Marcel Fafchamps, Sergio Pereira Leite, Kazi Matin, Richard Meyer, Ademola
Oyejide, and Hennie van Greuning.
� 1997 The International Bank for Reconstruction and Development/THE WORLD BANK
195
196 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
The countries studied-Ghana, Malawi, Nigeria, and Tanzania-have similar
types of financial systems but different degrees of financial development and
liberalization, permitting cross-country comparisons.
The analysis distinguishes between efficient specialization for market niches
by different segments of informal and formal finance and fragmentation with
impediments to efficient intermediation. Under efficient specialization for dif-
ferentiated risk and cost characteristics, interest rate differentials reflect differ-
ences in cost of funds, transaction costs, and risk. In fragmented markets, wide
differences in risk-adjusted returns occur because funds and information do not
flow between segments, and clients have limited access to different financial
instruments, resulting in low substitutability. Where poor information and con-
tract enforcement make it too costly for formal financial institutions to serve
small businesses and households, informal sector techniques may have an im-
portant role to play in serving these financial market segments.
Section I provides some background on initial conditions and policy reforms.
Section II presents the analytical framework used to examine market responses
and performance. Section III presents the evidence on segmentation, and section
IV analyzes the responses of different segments to policy reforms. Section V
concludes with policy implications.
1. BACKGROUND
The review in "Adjustment in Africa" (World Bank 1994) acknowledges the
limited progress in financial sector reform in Africa and calls for some rethink-
ing of strategy. Financial liberalization may need to be accompanied by mea-
sures to address institutional weaknesses and structural obstacles that inhibit
financial market efficiency and integration.
In most African countries, the indigenous private sector consists largely of
households and small-scale enterprises that operate outside the formal financial
system. Analysts refer to the informal sector by many terms, such as unorga-
nized, noninstitutional, and curb markets. Conforming to recent trends in the
literature, we use the term "informal finance" to refer to all transactions, loans,
and deposits occurring outside the regulation of a central monetary or financial
market authority (Adams and Fitchett 1992). The semiformal sector has charac-
teristics of both the formal and informal sectors-for example, legally registered
institutions that are not directly regulated by the financial authorities.
Informal savings activities in Africa are widespread but generally self-
contained and isolated from those of formal institutions (Adams and Fitchett
1992 and Bouman 1995). There is evidence of demand for external finance by
enterprises that want to expand beyond the limits of self-finance but that have
historically lacked access to bank credit (Aryeetey and others 1994; Levy 1992;
Liedholm 1991; Parker, Riopelle, and Steel 1995; and Steel and Webster 1992).
Better integration among different segments of the financial system-formal,
semiformal, and informal-could facilitate economic development by mobiliz-
Aryeetey, Hettige, Nissanke, and Steel 197
ing household resources more effectively and improving the flow of financial
resources to enterprises with high potential (Seibel and Marx 1987).
We selected Ghana, Malawi, Nigeria, and Tanzania for this study on the ba-
sis of their reasonably comparable financial systems, financially repressive poli-
cies prior to reform in the late 1980s, well-documented financial systems, and
experienced local researchers. Financial policies pursued in the four sample coun-
tries in the prereform period shared certain financially repressive characteristics,
such as restriction on market entry, often coupled with public ownership; high
reserve requirements; interest rate ceilings; quantitative control on credit alloca-
tion; and restrictions on capital transactions with the rest of world (Johnston
and Brekk 1991; Montiel 1996).
Financial repression discouraged investment in information capital. Savings
mobilization was not actively pursued. Financial systems lacked active liquidity
and liability management and incentives to increase efficiency, resulting in high
costs of financial intermediation. Although the nature of particular measures
varied by country, in general the allocation of investible funds shifted from the
market to the government. The degree of government control over banking in-
stitutions was higher in socialist-oriented Tanzania and Ghana than in Malawi
and Nigeria, which encouraged indigenous private agents following indepen-
dence. Governments often used banking institutions as a source of implicit taxa-
tion, for example, by imposing high reserve requirements in the range of 20-25
percent of assets (more than 80 percent in Ghana in the early 1980s) and by
financing operating losses of parastatals (Collier and Gunning 1991). In the
period before adjustment, the share of government and public enterprises in
total domestic credit was 86 and 95 percent in Ghana and Tanzania, respec-
tively, and well over 50 percent in Malawi and Nigeria.
Governments implemented financial sector reforms to address these condi-
tions through liberalization and balance-sheet restructuring. The reforms decon-
trolled interest rates and credit allocation and included efforts to strengthen regu-
latory and supervisory frameworks. Although the general thrust of these measures
was similar for all four countries, the initial conditions differed, including banks'
and borrowers' net worth and the scale of fiscal imbalances preceding financial
sector reform. Policy sequences and the pace of reforms also differed across coun-
tries. All of the countries initiated policy reforms during the period 1985-87
(although implementation in Tanzania was very slow before 1991).
Analysts frequently mention the partial nature of reforms and inadequate in-
stitution building as explanations for the disappointing outcomes of financial
liberalization in Sub-Saharan Africa (World Bank 1994). The experience of the
Southern Cone countries in South America shows that important conditions for
successful liberalization include macroeconomic stability, prudential supervision,
and an adequate regulatory framework. The financial reform programs intro-
duced in Ghana and Malawi addressed these conditions, at least to some extent.
Ghana reduced fiscal imbalances before decontrolling the interest rate and credit
allocation over a two-year period and restructured banks and their balance sheets.
198 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
The country paid early attention to strengthening the regulatory and supervi-
sory environment and to developing money and capital markets. In Malawi,
too, major fiscal and public enterprise reforms prior to financial liberalization
reduced the cost of bank restructuring. The reforms gradually decontrolled in-
terest rates and implemented institution-building measures. Neither country ex-
perienced major financial crisis.
In Tanzania problems arose from delays in restructuring parastatals, which
were the banks' main borrowers. Banks' net worth deteriorated significantly as
they continued to extend credit to poorly performing parastatals. Nonperforming
loans accumulated, greatly increasing the cost of balance-sheet restructuring.
Thus, weaknesses on the institutional side impeded progress in policy reforms.
In Nigeria financial sector reforms were thrown into crisis by the sequencing of
reform measures and the lack of the necessary prerequisites for liberalization. In
particular, wholesale deregulation of interest rates and market-entry require-
ments in the early years aggravated the instability of the financial system.
A series of corrective measures had to be adopted, raising questions of policy
credibility.
Our fieldwork shows that, in comparison with the disappointing response of
formal institutions to reform measures, informal financial agents responded dy-
namically in the adjustment period in all four countries. In particular, we ob-
serve signs of innovation in the semiformal financial sector. However, with weak
linkages between segments of the financial market, these new developments have
as yet had little measurable impact on market fragmentation, resource mobiliza-
tion, and financial intermediation.
II. ANALYTICAL FRAMEWORK
Two leading theoretical paradigms in contemporary financial economics pro-
vide analytical frameworks for examining the impact of policy reforms on finan-
cial market fragmentation. These paradigms complement each other but focus
on different policy-based or structural and institutional explanations.
A Policy-Based Explanation of Financial Market Fragmentation
The financial repression hypothesis (McKinnon 1973; Shaw 1973; and Fry
1982, 1988) attributes underdeveloped and inefficient financial systems to gov-
ernment policy failures, which result from excessive intervention. The hypoth-
esis sees repressive policies as the prime cause of fragmentation (Roe 1991).
Ceilings on deposit and loan rates tend to raise the demand for and depress the
supply of funds. Unsatisfied demand for investible funds then forces financial
intermediaries to ration credit by means other than the interest rate, while an
informal market develops at uncontrolled rates. A fragmented credit market
emerges in which favored borrowers obtain funds at subsidized, often highly
negative, real interest rates, while others must seek credit in inefficient, expen-
sive informal markets.
Aryeetey, Hettige, Nissanke, and Steel 199
In this view, removing restrictive policies should enable the formal sector to
expand and thereby eliminate the need for informal finance. Financial liberaliza-
tion would lead to financial deepening; improved efficiency, resulting in lower
spreads between borrowing and lending rates; and increased flow of funds be-
tween segments, including better access to formal finance for previously
marginalized savers and borrowers.
Structural and Institutional Explanations of Financial Market Fragmentation
Other authors have concentrated on structural and institutional features of
the financial markets of developing countries to explain fragmentation. Hoff
and Stiglitz (1990) advance an explanation based on imperfect information on
creditworthiness and differences in the costs of screening, monitoring, and con-
tract enforcement across lenders. In the presence of imperfect information and
costly contract enforcement, market failures result from adverse selection and
moral hazard, which undermine the operation of financial markets. Adverse
selection occurs as interest rates increase and borrowers with worthwhile invest-
ments become discouraged from seeking loans. The quality of the mix of loan
applications changes adversely as interest rates increase. Further, borrowers have
an incentive to adopt projects that promise higher returns but have greater risks
attached. This increases the risk of default. Moral hazard occurs when some
applicants borrow to pay high interest on existing loans to avoid bankruptcy or
borrow without the intention or the capacity to pay back loans. Thus the level
of interest rates affects the risk composition of financial portfolios (Stiglitz and
Weiss 1981 and Stiglitz 1989). Concerned about greater risk, lenders may resort
to nonprice rationing rather than raise interest rates when faced with excess
demand for credit. As a result, credit rationing may characterize market equilib-
rium even in the absence of interest rate ceilings and direct allocation. Liberal-
ized markets do not necessarily ensure Pareto-efficient allocation (Stiglitz 1994).
Problems arising from imperfect information are likely to be most pronounced
in low-income countries, where information flows are limited by poor commu-
nications, and gathering information is often costly. Poor information systems
encourage segmentation by raising the cost to formal institutions of acquiring
reliable information on both systemic and idiosyncratic risks for all but the larg-
est clients. In contrast, informal agents rely on localized, personal information
that gives them local monopoly power but constrains their ability to scale up.
Segmentation may also result from weaknesses in the infrastructure that sup-
ports the financial system. For example, the adequacy of the legal infrastructure
affects the costs and risks of contract enforcement, which in turn influence both
the willingness of lenders to enter into financial agreements and the type of
security they will accept. The ability to offset the risk of default may be limited
by the absence of a well-functioning insurance market and of markets for the
sale of confiscated collateral (Binswanger and Rosenzweig 1986). In low-income
countries, reliance on collateral excludes many otherwise creditworthy small-
scale borrowers, especially where land tenure is not legally explicit. Market seg-
200 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
ments that formal banks avoid for these institutional reasons may nevertheless
be served by informal agents who use personal relationships, social sanctions,
and collateral substitutes such as reputation and group responsibility to ensure
payment.
Synthesizing Alternative Explanations of Financial Market Fragmentation
The explanations for segmentation discussed above are not necessarily mutu-
ally exclusive. Ghate (1988) suggests that the informal sector consists of two
parts. The autonomous part, represented by indigenous bankers, rotating sav-
ings and credit associations (ROSCAS; see Bouman 1995), and pawnbrokers, his-
torically antedates the formal sector. The reactive part developed in response to
controls over the formal sector. In this respect, informal sector credit can be
viewed as residual finance, satisfying spillover demand by those excluded from
the formal market (Bell 1990).
Roemer and Jones (1991) also make a useful distinction between a parallel
market and a fragmented market. Parallel markets arise principally to evade
government controls and regulations, but markets can become fragmented in
the absence of government controls due to inherent operational characteristics.
Roemer and Jones suggest that "Credit markets in developing countries display
characteristics of both parallelism and fragmentation" (p. 8). Evaluated in this
light, the financial repression hypothesis is concerned with parallelism, while
the imperfect information paradigm implies that fragmentation may persist de-
spite liberalization.
Structural and institutional barriers across segments provide the opportunity
to exploit monopoly power, thus perpetuating fragmentation. A pronounced
feature of financial markets in Sub-Saharan Africa is the separation of formal
and informal sectors into almost discrete enclaves (Seibel and Marx 1987). A
critical policy-related question is whether segment-specific advantages can be
translated into market efficiency; measures to promote integration of segments
may be necessary (Seibel 1989). As financial sector reforms address policy-
induced bottlenecks, the extent to which structural and institutional deficiencies
constrain efficient specialization becomes more observable.
Hypotheses about the Effects of Liberalization on Access to Formal Finance
Under the financial repression hypothesis, liberalization of restrictive poli-
cies on interest rates and entry leads to greater access to formal finance for
previously marginalized borrowers, lower spreads between borrowing and
lending rates, increased financial flows between segments of the financial
market, and a diminished role for informal finance. Lack of a well-defined
time period in which the results should occur limits our ability to test this
hypothesis. However, we anticipate that some perverse effects will occur ini-
tially, with the removal of interest rate ceilings and the restructuring of bank
portfolios. We conducted our study more than three years after the initiation
of reforms in each of the countries, a period sufficient to observe the initial
Aryeetey, Hettige, Nissanke, and Steel 201
effects on informal finance, although reform of the formal financial sector
was not necessarily complete.
If informal finance represents an efficiency-improving solution to structural
problems of imperfect information and contract enforcement, we would expect
to observe specialized techniques designed to minimize transaction costs and
risks in dealing with narrow market segments. If structural and institutional
constraints are important, reforms in the formal financial sector would have
little impact on informal activities, which would respond more to changes in
financial demand and supply in the real economy than to changes in financial
policies.
Methodology for Constructing the Sample
We collected data on 283 informal financial institutions and 174 bank branches
in the four countries during 1992 and 1993 (see tables 1 and 2). Altogether the
sample has 160 observations for Ghana, 104 for Malawi, 104 for Nigeria, and
89 for Tanzania. We attempted to survey bank branches representing all major
commercial and development banks in each country and a representative sample
Table 1. Survey Sample of Informal Nonbank Financial Institutions, 1992-93
(number of observations)
Rotating
savings Savings
and credit and credit
Savings Money Traders, associations cooperatives Credit
Country collectors lenders landlords (ROSCAS) (sccs) unions Othera Total
Ghana 28 12 - 18 12 18 2 90
Malawi - 23 29 9 9 - - 70
Nigeria 15 20 - 12 10 4 3 64
Tanzania - - 30 10 19 - - 59
Total 43 55 59 49 50 22 5 283
Percent 15.2 20.8 20.8 17.3 17.7 7.8 1.8 100.0
- Not available.
a. Savings and loan companies, finance houses.
Source: Authors' calculations based on survey data.
Table 2. Survey Sample of Formal Banking Institutions, 1992-93
(number of observations, including branches)
Commercial and Development Total in
Country merchant banks banks Othera Total rural areas
Ghana 38 14 18 70 35
Malawi 14 3 17 34 15
Nigeria 34 0 6 40 8
Tanzania 6 15 9 30 5
Total 92 32 50 174 63
Percent 52.9 18.4 28.7 100.0 36.2
a. Rural banks (Ghana), community and people's banks (Nigeria), building society and union of
savings and credit cooperatives (Malawi), postal bank (Tanzania).
Source: Authors' calculations based on survey data.
202 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
of specialized banking institutions (such as rural banks, community banks, build-
ing societies, and postal banks). For informal financial institutions, no system-
atic enumeration was available that could serve as a sampling frame. Further-
more, differences in the nature of informal institutions are found across countries.
Hence we selected representative respondents from three broad categories of
informal institutions (see section III), based on interviews with borrowers and
prior knowledge by the local research teams. Absent a basis for determining the
sample's representativeness in terms of the numbers and assets of different types
of institutions, the analysis focuses on differences in institutional characteristics,
behavior, and performance between categories and between informal and for-
mal financial institutions.
The questionnaires sought data on the agents themselves, portfolio character-
istics, interest rates, risk management, transaction costs, delinquency rates, and
linkages to other institutions. We used the surveys to obtain retrospective infor-
mation on changes over the preceding two years. Retrospective data are subject
to bias because less successful institutions that failed are excluded from the sample.
However, the observation of the researchers based on previous research in the
sector was that dropout rates were relatively low for most informal financial
agents. The data can be considered representative of agents who stayed in busi-
ness during the period under review, although they cannot be generalized to
estimate changes at the national level, given the absence of census and panel
data. The results are presented in more detail in Nissanke and Aryeetey (forth-
coming), Aryeetey and others (1997), Aryeetey (1994 and 1996), Bagachwa (1995
and 1996), Chipeta and Mkandawire (1996a and 1996b), and Soyibo (1996a
and 1996b).
We gathered data on formal financial flows and indicators from published
sources and central bank authorities. We compared data for 1987-92 with data
for 1981-86 to analyze changes that occurred after reforms were under way.
III. FINANCIAL MARKET SEGMENTATION
This section summarizes the specialized techniques of informal financial insti-
tutions and compares them with those of banks. The evidence indicates the ex-
tent to which informal markets are more efficient than formal financial markets.
Types of Informal Financial Institutions
Financial transactions involve the exchange of money in the present for a
promise to pay in the future. The ability to enforce these contracts is critical for
the survival of a financial intermediary. Unlike financial transactions in the for-
mal sector, transactions in the informal sector rarely involve legal documenta-
tion. We identify three basic approaches to risk and contract enforcement prob-
lems. One category specializes in either the credit or the savings side of the market.
Another category bases the financial transaction on a personal or business rela-
Aryeetey, Hettige, Nissanke, and Steel 203
tionship. A third category provides full financial intermediation between savers
and borrowers. We draw roughly a third of the sample from each of these three
categories.
SPECIALISTS IN ONE SIDE OF THE MARKET. Moneylending covers a wide range of
credit arrangements that differ across countries, with interest rates ranging from
0 to as much as 100 percent a month. (In general, the most common source of
informal finance consists of relatives and friends; this type was not covered in
the survey because of its noncommercial character.) All the informal moneylenders
surveyed base their lending decisions on firsthand knowledge of the borrower.
In our sample, there are few professional moneylenders. More commonly, part-
time moneylenders use surplus funds from other sources such as a commercial
business. Professional and part-time moneylenders account for 55 (of the total
457) observations in the combined survey sample.
Individuals who operate primarily on the savings side are found only in West
Africa (43 observations in Ghana and Nigeria). Savings collectors take regular
deposits (usually on a daily basis) of an amount determined by each client and
return the accumulated sum (typically at the end of each month) minus one
day's deposit as commission. These mobile bankers form a symbiotic relation-
ship with market traders, protecting daily earnings from competing claims and
ensuring working capital to restock supplies at the end of the month (Miracle,
Miracle, and Cohen 1980 and Aryeetey and Steel 1995). Savings collectors place
most of their deposits in banks for safekeeping, but they sometimes extend ad-
vances to their best clients before the end of the month.
RELATIONSHIP-BASED LENDERS. Rotating savings and credit associations are
pervasive in all the countries studied. ROSCAs are known as (among other names)
susu in Ghana, esusu in Nigeria, upatu or mchezo in Tanzania, and chilemba or
chiperegani in Malawi. ROSCAs are membership groups in which all members
pay in set amounts at regular intervals to a common pool, which goes to each
member in turn (usually randomly, but some variations allow bidding).
Intermediation occurs between members whose turn comes earlier and later within
a small, closed group over a fixed period of time. Mutual trust offsets the risk
that early recipients will drop out. In another type of savings and credit
association, members save jointly toward common objectives such as school
fees, annual festivals, or community development, sometimes making loans at
high rates to increase the accumulated amount. Rotating savings and credit
associations represent 49 observations in the four countries in our sample.
Traders are an important source of informal credit in all the countries stud-
ied. They supply either inputs or cash advances to farmers, linked to purchase of
produce at a highly discounted price. In Malawi and Tanzania, landlords and
estate owners often lend to their tenants. Individual lenders who have long-term
business relationships with their clients account for 59 observations in our sample.
204 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
FINANCIAL INTERMEDIARIES. Savings and credit cooperatives (sccs), or societies,
raise savings from and make loans to members. Although they are membership
organizations, sometimes raising money from shares as well as voluntary deposits,
they are relatively large and open to new members, unlike ROSCAS. Credit unions
are registered as such and represent a more formal form of scc based on share
capital. sccs and credit unions (with 50 and 22 observations, respectively, in the
combined sample) use repeat transactions to screen borrowers. Other semiformal
institutions (5 observations), such as finance houses, have emerged to both
mobilize and lend funds to the general public (see section IV).
Management of Information and Risk
The banks surveyed in the sample countries view small borrowers as riskier
than large ones for reasons often related to the difficulty of obtaining accurate
information about them: geographical remoteness, illiteracy, and unreliable in-
comes. Through heavy emphasis on stringent collateral requirements for loans
and high minimum deposit requirements, banks effectively screen out the vast
majority of small clients. We rarely observe foreclosure on collateral or legal
actions in our survey, reflecting weak and uncertain legal systems.
Informal lenders draw heavily on information obtained through personal,
social, and business relationships in order to preselect clients. ROSCAs, SCCs, and
credit unions operate with group membership selection criteria. Traders and
landlords lend only to their customers and tenants. Most informal lenders do
not use interest rates to discriminate among clients. Through prescreening, all of
the borrowers of each lender fall into a similar risk category.
Informal lenders generally require security but are much more flexible than
banks in accepting personal guarantees, arrangements with employers, and mov-
able property. About 60 percent of moneylenders in Nigeria, 63 percent in Tan-
zania, and 83 percent in Ghana require such security, as do 76 percent of credit
unions in Ghana (but smaller sccs and ROSCAs generally do not). In interlinked
transactions, the crops pledged, the equipment provided, or the land involved
serve as collateral. Informal enforcement is easier than going through the legal
system. For example, a landlord-lender could make productive use of pledged
farmland, whereas a bank would face a long, expensive legal process to seize it.
Personal relationships and social pressure, either within membership groups or
through family members, are often instrumental in ensuring repayment without
aggressive enforcement measures.
Our survey finds that linking loans with real sector transactions is a common
informal technique in all countries. Traders may provide materials and equip-
ment on credit or make a cash loan contingent on purchasing such inputs or
selling the crop to the trader. In Malawi all estate owners making loans linked
them in this way. The lower implicit rate (6 percent a month) for linked loans
than for unlinked cash loans (9 percent) from trader-lenders in Tanzania and
loan sizes that are five times the size of unlinked loans show that interlinked
transactions reduce uncertainty.
Aryeetey, Hettige, Nissanke, and Steel 205
Evidence of Financial Market Fragmentation
Weak linkages between segments of the financial market and differences in
returns that cannot be explained by costs and risks indicate fragmentation. To
study the costs and risks, we gathered data on interest rates, default risk, and
transaction costs, although we find it difficult to make precise comparisons for
loan instruments of widely different terms and conditions.
Financial flows from formal to informal markets are negligible. Informal fi-
nancial agents generally have a limited capital base and little access to borrowed
funds. Even those moneylenders who can access bank credit through their other
business activities rarely do so for the purpose of on-lending. The main sources
of the expanding supply of loanable funds by informal agents are mobilized
savings and reinvested profits (including from other activities).
By contrast, informal deposit mobilizers (except for ROSCAS) frequently main-
tain bank accounts, especially in urban areas. In Ghana 89 percent of informal
operators report having a bank account, in Nigeria 82 percent, and in Tanzania
97 percent in urban areas and 67 percent in rural areas. We find no direct link-
ages between informal agents, although some clients use savings collectors to
accumulate funds for contributions to their ROSCA or credit union. Informal cli-
ents generally have neither a savings nor a credit relationship with formal banks,
and few can obtain credit from more than one source.
Interest rates vary widely across informal institutions, as well as between for-
mal and informal markets. Moneylenders' rates are generally at least 50 percent
above formal rates, with average monthly interest ranging from under 10 per-
cent in Tanzania to 48 percent in Malawi, reaching as high as 100 percent a
month in individual cases. The average monthly interest rate of sccs in Malawi
is also relatively high at 13 percent, well above rates in the formal sector, whereas
the average of 2.6 percent a month in Tanzania is comparable to the 31 percent
a year charged by the state-owned commercial bank.
Delinquency and default rates of informal lenders are generally low relative
to banks in the sample countries. In Ghana 70 to 80 percent of informal lenders
have no delinquent borrowers compared with 80 to 86 percent in Nigeria. In all
cases, lenders are confident that delinquent borrowers will repay within three
months of the loan maturing. Eventual default rates in Tanzania are as low as
0.1 percent for sccs, 2.5 percent for ROSCAS, and 4 percent for traders and land-
lords. In contrast, commercial banks report very high rates of nonperforming
loans, averaging from 45 percent in Nigeria to more than 80 percent in Tanza-
nia (in part a problem of state banks and parastatal borrowers), with only lim-
ited recovery through portfolio restructuring exercises.
Loan administration costs (screening, monitoring, and contract enforcement)
are generally lower as a percentage of loan amounts for informal lenders than
for banks. Most of the informal lenders' costs are in prescreening the client's
ability to repay, not the particular use of the funds, whereas banks devote con-
siderable resources to project evaluation. The value of the time that moneylend-
206 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
ers allocate to administering loans is equivalent to only 0.6 to 3.2 percent of
loan amounts across the four countries, compared with 1.7 to 12.9 percent for
bank loans to small-scale enterprises (as high as 18.9 percent for large-scale
enterprises). Credit unions fall within the same range as moneylenders, while
their less formal counterparts, sccs, are consistently lower at 1 percent or less.
The part-time nature of much informal lending and the lack of overhead help
explain the relatively low costs.
The cost of lending also depends on the cost of funds. Banks' cost of funds in
1992, as indicated by deposit rates, were 16 percent in Ghana, 17 percent in
Malawi, 18 percent in Nigeria, and 22 percent in Tanzania. (The inflation rate,
measured by the gross domestic product (GDP) deflator, in 1992 was 13 percent
in Ghana, 18 percent in Malawi, 65 percent in Nigeria, and 19 percent in Tan-
zania.) Most of the informal financial units surveyed mobilize their own funds,
usually at very low cost; they have no access to bank loans for on-lending. Sav-
ings collectors have a negative cost of funds, because they receive payment for
taking deposits. ROSCA members evidently have a low opportunity cost of funds,
because they persist despite the absence of interest payments. The opportunity
cost of funds for moneylenders who are also traders is low because they gener-
ally lend out temporarily idle funds.
The evidence indicates that financial markets in the sample countries are highly
fragmented. Formal and informal lenders are polarized at extreme ends of the
market, with relatively little overlap of clientele. Each informal and formal insti-
tution selects a narrow range of clients and products. Although some informal
agents link households and small businesses to the formal financial system through
their deposit mobilization activities, this is a one-way link with virtually no link-
age on the credit side.
Furthermore, risk-adjusted returns do not appear comparable across segments.
Informal interest rates are generally much higher than formal rates, yet informal
lenders have both lower transaction costs and lower default rates. Among infor-
mal lenders, the variation in rates is much wider than the variation in transac-
tion costs and default rates. Informal lenders appear better able to enforce col-
lateral than banks and to have a relatively low opportunity cost of funds. Hence
the relatively high rates charged by informal moneylenders are likely to repre-
sent substantial monopoly power vis-a-vis borrowers who lack access either to
formal credit or to membership-based informal finance.
Nevertheless, the relatively low transaction costs and loan losses of informal
agents in serving clients who lack access to the formal banking system indicate
that they provide a reasonably low-cost solution to the information and enforce-
ment problems that characterize African economies.
IV. FINANCIAL SECTOR RESPONSES FOLLOWING REFORMS
This section investigates the financial repression hypothesis. It looks at how
different segments of the financial market responded to the introduction of lib-
Aryeetey, Hettige, Nissanke, and Steel 207
eralization measures in terms of deposit mobilization, financial deepening, lend-
ing, and interest rate spreads. Reformers expected that bank depositors would
switch to interest-bearing, longer-term deposits and that banks would increase
their lending to the private sector, including small enterprises. They also ex-
pected interest rate spreads to diminish and the importance of informal financial
markets to dwindle.
It should be stressed that financial sector reforms are ongoing, encompassing
wide-ranging measures rather than just liberalization of interest rates and credit
allocation. Because the breadth and depth of financial reforms vary consider-
ably across the case-study countries, it is inappropriate to make a final conclu-
sion on the outcome of reforms per se. It is now widely accepted that financial
reform is a lengthy process, requiring progress in institution building, as well as
policy liberalization.
Financial Deepening and Deposit Mobilization
The countries in our sample made little progress in savings mobilization, with
fluctuating growth in the number of depositors and the size of deposits follow-
ing the introduction of reforms. Figures 1-4 show little net change in the mobi-
lization of deposits by banking institutions, as measured by the ratios of cur-
rency in circulation (Ml and M2) to GDP. The M2-to-GDP ratios of the sample
countries-31 percent for Tanzania over 1987-92, 15 to 21 percent for the oth-
ers-lie below those for countries of comparable income per capita from other
regions, such as Honduras (31 percent), Bangladesh (33 percent), Pakistan (43
percent), and India (48 percent).
Ghana's financial depth was the lowest among the countries studied, de-
spite some improvements after the mid-1980s, and remained far below the
levels attained in the 1970s. Though Malawi achieved greater financial depth
than Ghana, the pattern was likewise one of recovery after an initial decline
following liberalization, with no definite trend between 1975 and 1992. Since
1980 Malawi's currency-to-GDP ratio has remained under 5 percent, although
nontransaction demand for money (the difference between M2 and Ml) was
higher than in the other three countries, accounting for 10-14 percent of
GDP.
Nigeria's indicators of financial deepening were affected by the difficulties
experienced after liberalization attempts. Figure 3 shows that both M2-to-
GDP and (M2-M1)-to-GDP ratios declined sharply in the late 1980s, partly
because the government abruptly withdrew public sector deposits from the
banking system. Although these ratios have since recovered, Nigeria's pro-
cess of financial deepening appears to have stalled in the 1990s. Tanzania
had a higher M2-to-GDP ratio than the other study countries in the late 1970s,
but the banking system lost ground in savings mobilization in the initial years
of economic reform (1984-88). Recently currency accounted for more than
a third of M2, and the nontransaction demand for money had declined
noticeably.
208 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Figure 1. Indicators of Financial Deepening in Ghana, 1975-92
Percent
30.
25 -
20 - ~~ _
15.
0 -- - --- -- - --- --- --- --- --- ---
1975 1980 1985 1990 1992
-- Currency/GDP Mi/GDP M2/GDP
Note: Ml measures money supply as demand deposits plus currency in circulation; M2 is MI plus time
and savings deposits.
Source: IMF (various years); Bank of Ghana (various years).
Figure 2. Indicators of Financial Deepening in Malawi, 1975-92
Percent
30
25
20
15
10
1975 1980 1985 1990 1992
----Currency/GDP ----Mi/GDP M2/GDP
Note: Ml measures money supply as demand deposits plus currency in circulation; M2 is Ml plus time
and savings deposits.
Source: IMP (various years); Reserve Bank of Malawi (various years).
Liquid short-term instruments continued to dominate the liabilities of banking
institutions during the reform period, although the general trend was toward a
smaller share of demand deposits (table 3). In Nigeria the share of time deposits
actually fell 15 percentage points between 1980 and 1992. Only in Tanzania was
demand for time deposits clearly both strong and rising. In sum, the deposit base
of banking institutions remained volatile, with only limited change in the struc-
ture of liability.
RESOURCE MOBILIZATION BY INFORMAL FINANCIAL INSTITUTIONS. In contrast to
the disappointing performance of the formal financial sector following reforms,
the survey results show that informal financial institutions in all four countries
responded dynamically to increased demand for their services in the liberalized
environment. The capital base of moneylenders in Nigeria grew 264 percent
over two years (1990 to 1992) and that of savings and loan companies grew 148
percent; in Malawi the combined average increase was 73 percent over two years.
Aryeetey, Hettige, Nissanke, and Steel 209
Figure 3. Indicators of Financial Deepening in Nigeria, 1975-92
Percent
35
30
25
20
15
1975 1980 1985 1990 1992
----- Currency/GDP M1/GDP M2/GDP
Note. Ml measures money supply as demand deposits plus currency in circulation; M2 is Ml plus time
and savings deposits.
Source IMF (various years); Central Bank of Nigeria (various years).
Figure 4. Indicators of Financial Deepening in Tanzania, 19 75-92
Percent
50
40
30 -- -- -_ - - - -
30
20
1975 1980 1985 1990 1992
----- Currency/GDP - M1/GDP M2/GDP
Note: Ml measures money supply as demand deposits plus currency in circulation; M2 is Ml plus time
and savings deposits.
Source: IMF (various years); Bank of Tanzania (various years).
Survey results indicate that deposits increased in informal sector institutions.
Determining national trends in aggregate deposits is difficult because of sea-
sonal and annual fluctuations in the amounts reported and the absence of na-
tionwide data. Indications are that the number of informal institutions was in-
creasing in all countries, thus multiplying the increases per institution reported
from the survey data. In Tanzania the total volume of deposits rose 67 percent in
the sccs surveyed and 113 percent in ROSCAs from 1990 to 1992, due to in-
creases in both the number of members and the average size of deposits. In
Nigeria deposits rose 100 percent in credit unions, 56 percent in sccs, and 77
percent in ROSCAS over the same period. In Malawi deposits grew 44 percent in
community funds from 1989 to 1991 and 45 percent in ROSCAs (both faster in
urban areas), mainly as a result of deposits from additional members. The aver-
age number of clients per savings collector surveyed rose from 250 in 1990 to
438 in 1992 in Nigeria and from 155 to 290 in Ghana, and average monthly
deposits rose 51 percent in Nigeria and 64 percent in Ghana.
210 THE WORLD RANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 3. Changes in Composition of Bank Deposit Liabilities, 1980 and 1992
(percentage of total deposits)
Demand deposits Savings deposits Time deposits
Country 1980 1992 1980 1992 1980 1992
Ghana 71a 57b 28a 34b la 8b
Malawi 40 38 - - 60' 62c
Nigeria 48 44 15 34 37 22
Tanzania 67 45 10 17 23 30
- Not available.
a. 1983.
b. 1991.
c. Figures are for both time and savings deposits.
Source: Aryeetey 1994; Bagachwa 1996; Soyibo 1996a; the figures for Malawi are calculated from
Reserve Bank of Malawi, various years.
TRENDS IN LENDING. Reformers tried to restructure bank loan portfolios to
expunge nonperforming loans, largely to public enterprises, and to enable banks
to resume lending to private enterprises. Although the share of lending by
commercial banks to the private sector generally increased in Ghana and
Tanzania, lending to the public sector remained high, reflecting past development
strategy (table 4). The share of the private sector in bank lending was higher in
Malawi and Nigeria (compared with Ghana and Tanzania), but contracted during
1987-92. Despite these limited signs of improvement in private sector shares of
credit, slow growth of credit overall meant that the ratio of private credit to GDP
actually fell in Malawi and Nigeria and remained relatively low in Ghana at 4
percent and in Tanzania at 2 percent. Indeed, in all four countries, the ratio of
private sector lending to GDP was remarkably lower (2 to 11 percent) than in
many countries with comparable income per capita, for example, 50 percent in
Indonesia and 20 percent in Kenya (International Monetary Fund 1995).
There was also little change in banks' lending profile within the private sector
portfolio. Banks continued to concentrate on their traditional large, established
customers and to avoid small-scale enterprises and small farmers. In Ghana large
enterprises (30 or more workers) took as much as 74 and 50 percent of loans
extended to the private sector by commercial banks and development banks,
Table 4. Credit Allocation between the Private and Public Sectors, 1981-93
(percent)
Share of credit Private sector Ratio of
to the public lending in total private sector
sector in total credit commercial bank lending credit to GDP
Country 1981-86 1987-92 1986 1990 1993 1981-86 1987-92
Ghana 86.3 74.5 13.6 27.6 35.8 2.4 4.1
Malawi 63.3 53.6 39.4 52.5 40.4 14.6 9.1
Nigeria 55.3 50.3a 47.2 63.5 44.9 17.4 11.3
Tanzania 94.8 61.6b 7.2 14.6 27.1 2.3 2.2
a. 1987-91.
b. 1987-88.
Source: Authors' calculations based on various annual reports from relevant central banks.
Aryeetey, Hettige, Nissanke, and Steel 211
respectively, although they represented less than 10 percent of firms and employ-
ment. In Malawi the small enterprise sector (fewer than 30 workers) received
only 15 percent of total loan volumes in 1992, while large enterprises received 63
percent of total loans disbursed. Only in Nigeria was there little disparity in the
number of loans to the small and large enterprise sectors, although large enter-
prises continued to receive a greater share of total loans disbursed.
In sum, despite liberalization and attempts to introduce greater competition,
formal finance did not become more accessible to a broad section of the real
economy. Sectoral credit distribution remained dominated by short-term credit
for trade, as in Ghana and Nigeria, or by financing for the processing and mar-
keting of agricultural exports, as in Malawi and Tanzania. Usually only large
manufacturing firms received credit from banks. Other characteristics of formal
sector loans, such as maturities and real average loan sizes, hardly changed after
reforms were initiated.
INFORMAL CREDIT DEMAND AND SUPPLY. Strong increases in the number of loan
applications received and approved are observed for almost all informal lenders
in the sample countries. The only notable exceptions are the relatively small
credit unions in Nigeria and community funds and savings and credit cooperatives
in Malawi. The activities of moneylenders increased sharply in all four countries;
the number of loans rose 20 to 130 percent (60 to 73 percent for traders in
Tanzania and Malawi). In many cases, loan approval rates rose along with the
number of applications, implying an increase in the supply of as well as the
demand for funds. Nevertheless, substantial excess demand was reported; for
example, 42 percent of moneylenders and 40 percent of savings collectors in
Ghana were unable to satisfy all the loans demanded by clients they considered
creditworthy.
The survey results suggest that the growth in operations of informal agents
was related more to growth of the real economy than to financial sector devel-
opments. In Tanzania, for example, trader-lenders and landlords obtained about
35-40 percent of their loan capital from other economic activities, and 85 per-
cent reported that their capital base was growing. Liberalization of grain mar-
kets during the 1980s in particular fostered private traders who provided short-
term financing for crops. In Ghana, increased financial market activity from
liberalization of product markets and increased imports associated with struc-
tural adjustment expanded the savings mobilized by savings collectors and the
profits of larger trader-moneylenders.
INTEREST RATES AND SPREADS. Under liberalized policies, formal sector lending
and deposit rates were expected to settle at a market-clearing level. An initial
increase in the spread between lending and deposit rates was expected, because
banks needed time to reshape their cost structures. The spread was then expected
to narrow as more efficient business practices were adopted under increasing
competition.
212 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
However, lending rates and spreads remained persistently high during the
reform years (table 5). In most cases, high spreads persisted more than seven
years after reforms began. The ratio of average spread to lending rate remained
the same between 1987 and 1992 in Malawi (0.3) and Tanzania (0.5) and rose
in Ghana (from 0.4 to 0.5). The ratios were high relative to Indonesia (0.2),
Bangladesh (0.3), and the Philippines (0.3). (See International Monetary Fund
1995.) This trend in spreads indicates low competition in financial markets and
high cost of funds and transaction costs in bank lending. On top of nominal
lending rates, many banks impose servicing fees, equivalent to an extra 2-5 per-
cent. High reserve requirements intended to absorb excess liquidity in Ghana
could also explain high spreads; however, banks voluntarily held reserve instru-
ments well in excess of requirements, indicating that excess liquidity persisted
despite high spreads. Thus, there is little evidence of improved efficiency of in-
termediation in the banking sector.
In some cases moneylenders' interest rates declined following financial
sector reforms. The survey respondents did not associate these changes with
prevailing interest rates or competition, but rather with the increased supply
of funds from liberalized trade (for moneylenders whose primary activity
was trading) and with the inability of many clients to pay high traditional
rates.
Portfolio Management
Reform measures had limited impact on banks' portfolio management at the
time of the study. Even in Ghana and Malawi, where reforms were relatively
orderly, most banking institutions continued to operate in an extremely con-
strained environment, with underdeveloped market-supporting infrastructure and
a poor base of information.
Lending remained constrained by external factors such as policy uncertainty,
resulting in a low-lending trap despite latent excess demand for credit-particu-
larly by small-scale enterprises with good opportunities but insufficient collat-
eral. Furthermore, de facto crowding out of the private sector persisted in many
countries because of the presence of high-yielding government securities. The
Table 5. Interest Rates and Spreads in Sample Countries, 1987 and 1992
(percent per year)
Average fixed deposit rate Average lending rate Spread
Country 1987 1992 1987 1992 1987 1992
Ghana 19.0a 15.0' 30.0b 29.0b 11.0 14.0
Malawi 14.3 16.5 19.5 22.5 5.3 6.0
Nigeria 13.1 18.0 - 31.2 - 13.2
Tanzania 14.5 16.0 29.0 31.0 14.5 15.0
- Not available.
a. Minimum on six-month deposit rate.
b. Maximum secured lending rate.
Source: Authors' calculations based on annual reports from relevant central banks.
Aryeetey, Hettige, Nissanke, and Steel 213
portfolios of banking institutions remained dominated by a high incidence of
nonperforming loans and excess liquidity.
Although many informal agents grew along with demand for their services,
they had difficulty moving beyond their particular sphere of specialization. In
general, informal lenders' liability base was narrow, limited to deposits from a
specific group of people or surplus income earned by the lender from other eco-
nomic activities. For example, the lending base of each scc was limited by the
incomes of members, the frequency of deposits, and the size of membership
within which it could retain cohesion in its operation. There is little evidence of
sccs or informal agents borrowing externally to lend to their clientele. Their
average size of loans remained far smaller than that of banks, while the maturi-
ties of their loans were shorter. The demand for medium-size, medium-term
loans remained largely unsatisfied as both formal and informal segments of the
financial system continued to serve their narrow market niches.
New Institutional Developments
Some signs indicate that emerging nonbank, semiformal institutions in West
Africa increased competition (reflected in lower interest rates charged by mon-
eylenders) and began to fill underserved market niches. In Ghana, for example,
a nongovernmental organization (NGO) adopted the methods of savings collec-
tors, and a new savings and loan company targeted market women and small
businesses. Private finance houses in Nigeria provided services such as loans,
hire-purchase, equipment leasing, factoring, project financing, and debt admin-
istration. They could not take regular deposits, but they could borrow amounts
not below 4100,000 from investors. Their willingness to lend short- and me-
dium-term funds to clients who often could not satisfy the collateral require-
ments of conventional banks helped to improve the access to finance of sscs in
Nigeria (unlike the other countries studied).
Nevertheless, efforts to fill gaps in the financial system often had difficulties.
Nigeria's finance houses were poorly regulated, and many collapsed. Attempts
to establish unit banks in rural areas in both Ghana (rural banks) and Nigeria
(community banks) had only limited success, with high rates of distress resulting
from high costs and management problems. Malawi's Investment and Develop-
ment Fund and Small Enterprise Development Organization incurred substan-
tial losses in trying to serve indigenous small-scale enterprises, in part because
they also provided costly training, technical assistance, and advisory services.
Ghana's 1993 Non-Banking Financial Institutions Law helped foster new in-
stitutions for leasing, factoring, venture capital schemes, and discounting, as
well as savings and loan companies. However, savings and loan companies tend
to compete with savings collectors for the smallest depositors and borrowers,
while leasing companies and venture capital schemes are mainly interested in the
upper end of the market, where banks have always operated. Thus a gap re-
mains in meeting demand from growing small enterprises and other underserved
groups.
214 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
V. CONCLUSIONS AND POLICY IMPLICATIONS
Fragmentation of financial markets in Ghana, Malawi, Nigeria, and Tanza-
nia has persisted more than seven years after the initiation of financial policy
reforms. Fragmentation persists both because implementation of reform pro-
grams has been incomplete and because the reforms have not been accompanied
by adequate complementary measures to address underlying institutional and
structural constraints. Reforms have focused on the formal financial sector. But
our study shows that simply removing financially repressive policies is not suffi-
cient to increase financial depth or to induce banks to reach a wider clientele.
In contrast, informal financial agents have responded positively to demand
from clients who continue to lack access to formal finance. Expansion of de-
mand and supply in informal markets appears related more to growth of real
sector activities than to changes in financial policies.
In the prevailing situation of imperfect information and uncertainty, informal
financial agents in the four countries demonstrated a comparative advantage in
serving the large share of the population with little access to formal intermediar-
ies. Informal financial institutions used a variety of specialized methodologies to
mitigate the problems caused by information asymmetries and to contain risks
and transaction costs. In dealing with small clients, informal institutions used
methods that enabled them to achieve relatively low transaction costs and de-
fault rates (compared with what banks reported for both large and small clients).
Formal financial deepening is a long-term process that also requires a sound
macroeconomic environment, stronger regulatory and supervisory frameworks,
improved information flows, and legal and judicial reforms to facilitate contract
enforcement. Until costs to formal institutions of acquiring information and
enforcing contracts are significantly reduced, informal financial institutions will
retain a comparative advantage in their market niches. For some time to come,
the efficiency of the financial system as a whole can be improved by enabling
informal and emerging semiformal financial institutions to function and better
integrate with the rest of the system.
Extensive institution-building measures clearly must be part of effective fi-
nancial reform programs in Africa. Given the observed difficulties in improving
the regulatory and supervisory framework and the soundness of bank portfo-
lios, as well as in sustaining macroeconomic stability, a sensible approach is
"cautious gradualism on deregulation of interest rates and portfolio restrictions,
but prompt moves on institution building" (Caprio, Atiyas, and Hanson 1994,
pp. 436-37). Reform of the regulatory and supervisory system should not only
address formal institutions. It should also treat different tiers of the financial
system according to their distinct characteristics, the likely benefits of regula-
tion, and the ability of governments to regulate them effectively. Reform pro-
grams should balance encouraging innovative institutions, such as those emerg-
ing in some countries, with regulating institutions that are sufficiently large to
come under the purview of formal financial authorities.
Aryeetey, Hettige, Nissanke, and Steel 215
The study findings support the view that incentives and support for linkages
among segments of the financial market may be needed to accelerate integration
of formal and informal financial institutions. Greater flows between segments
would help equalize risk-adjusted returns by drawing on the comparative ad-
vantages of each. To expand financial market segments viewed as risky by banks,
it is likely to be more effective to induce banks to link up with institutions that
use appropriate methods than to expect banks to lend directly. For example,
partial guarantee of a line of credit to an NGO or association of informal agents
for on-lending in small amounts would make more sense than guaranteeing di-
rect small loans by banks. Savings collectors could expand credit to their cli-
ents-largely women traders-if they had recourse to a commercial bank line of
credit, and the resulting increase in business would allow them to mobilize sav-
ings for deposit in commercial banks. Technical assistance to (and prudential
regulation of) semiformal intermediaries would help give formal institutions
greater confidence in lending to them. Banks could provide a deposit instrument
adapted for savings and credit societies.
Improving contract enforcement through reform of the legal system is a fun-
damental long-term institutional measure that would encourage formal finan-
cial institutions to serve dynamic small-scale enterprises. This may require intro-
ducing special commercial laws and courts. Measures to facilitate taking collateral
in forms other than landed property, such as laws and courts that facilitate the
seizure of equipment and stock in case of default, would encourage leasing and
working capital loans to smaller businesses.
Difficulties in obtaining reliable information and in managing risks cause frag-
mentation by raising the costs to formal institutions of entering household and
small-scale enterprise market segments. These difficulties enable informal agents,
who have developed individualized information and social networks, to form
local monopolies. Measures to improve the flow of information about borrow-
ers include establishing credit bureaus, creating registries for recording secured
debt, and making audits available to small businesses at reasonable cost.
The study findings indicate that liberalization of financially repressive poli-
cies has limited impact on financial deepening without complementary measures
to address problems of information, risk management, and contract enforce-
ment. Innovative methodologies of informal and semiformal institutions, how-
ever, are overcoming these barriers to small financial transactions. Including
these methodologies in financial development strategies offers important poten-
tial to improve financial intermediation and widen access to financial services in
low-income countries.
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Adams, Dale W., and Delbert A. Fitchett. 1992. Informal Finance in Low-Income Coun-
tries. Boulder, Colo.: Westview Press.
216 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Aryeetey, Ernest. 1994. "Financial Integration and Development in Sub-Saharan Africa:
A Study of Informal Finance in Ghana." Working Paper 78. Overseas Development
Institute, London. Processed.
. 1996. "The Formal Financial Sector in Ghana after the Reforms." Working
Paper 86. Overseas Development Institute, London. Processed.
Aryeetey, Ernest, Amoah Baah-Nuakoh, Tamara Duggleby, Hemamala Hettige, and
William F. Steel. 1994. Supply and Demand for Finance of Small Enterprises in Ghana.
World Bank Discussion Paper 251. Washington, D.C.: World Bank.
Aryeetey, Ernest, Hemamala Hettige, Machiko Nissanke, and William F. Steel. 1997.
Financial Market Fragmentation and Reforms in Sub-Saharan Africa. World Bank
Technical Paper 356. Washington, D.C.: World Bank.
Aryeetey, Ernest, and William F. Steel. 1995. "Savings Collectors and Financial Inter-
mediation in Ghana." Savings and Development 19(2):191-211.
Bagachwa, Mboya S. D. 1995. "Financial Integration and Development in Sub-Saharan
Africa: A Study of Informal Finance in Tanzania." Working Paper 79. Overseas De-
velopment Institute, London. Processed.
.1996. "Financial Linkage and Development in Sub-Saharan Africa: The Role of
Formal Financial Institutions in Tanzania." Working Paper 87. Overseas Develop-
ment Institute, London. Processed.
Bell, Clive. 1990. "Interaction between Institutional and Informal Credit Agencies in
Rural India." The World Bank Economic Review 4(3):297-327.
Binswanger, Hans, and Marc Rosenzweig. 1986. "Behavioral and Material Determi-
nants of Production Relations in Agriculture." Journal of Development Studies
22(3):503-39.
Bouman, F. J. A. 1995. "ROSCA: On the Origin of the Species." Savings and Develop-
ment 19(2):117-46.
Caprio, Gerald, Jr., Izak Atiyas, and James Hanson, eds. 1994. Financial Reform: Theory
and Experience. Cambridge, U.K.: Cambridge University Press.
Chipeta, Chinyamata, and M. L. C. Mkandawire. 1996a. "Financial Integration and
Development in Sub-Saharan Africa: The Formal and Semi-Formal Financial Sectors
in Malawi." Working Paper 89. Overseas Development Institute, London. Processed.
.1996b. "Financial Integration and Development in Sub-Saharan Africa: A Study
of the Informal Financial Sector in Malawi." Working Paper 85. Overseas Develop-
ment Institute, London. Processed.
Collier, Paul, and J. W. Gunning. 1991. "Money Creation and Financial Liberalization in
a Socialist Banking System: Tanzania 1983-88." World Development 19(5):533-38.
Fry, M. J. 1982. "Models of Financially Repressed Developing Economies." World De-
velopment 10(9):731-50.
.1988. Money, Interest, and Banking in Economic Development. Baltimore, Md.:
Johns Hopkins University Press.
Ghana, Bank of. Various years. Quarterly Economic Bulletin. Accra.
Ghate, P. B. 1988. "Informal Credit Markets in Asian Developing Countries." Asian
Development Review 6(1):64-85.
Hoff, Karla, and Joseph E. Stiglitz. 1990. "Imperfect Information and Rural Credit Mar-
kets: Puzzles and Policy Perspectives." The World Bank Economic Review 4(3):235-50.
International Monetary Fund. Various years. International Financial Statistics. Wash-
ington, D.C.
Aryeetey, Hettige, Nissanke, and Steel 217
Johnston, R. Barry, and Odd Per Brekk. 1991. "Monetary Control Procedures and Fi-
nancial Reform: Approaches, Issues, and Recent Experiences in Developing Coun-
tries." In Philippe Callier, ed., Financial Systems and Development in Africa. EDI
Seminar Series. Washington, D.C.: World Bank.
Levy, Brian. 1992. "Obstacles to Developing Small and Medium-Sized Enterprises: An
Empirical Assessment." The World Bank Economic Review 7(1):65-83.
Liedholm, Carl. 1991. "Dynamics of Small- and Micro-Scale Enterprise and the Evolv-
ing Role of Finance." GEMINI Working Paper 26. U.S. Agency for International Devel-
opment, Washington, D.C. Processed.
Malawi, Reserve Bank of. Various years. Financial and Economic Review [quarterly].
Lilongwe.
McKinnon, R. I. 1973. Money and Capital in Economic Development. Washington,
D.C.: Brookings Institution.
Miracle, Marvin P., Diane S. Miracle, and Laurie Cohen. 1980. "Informal Savings Mo-
bilization in Africa." Economic Development and Cultural Change 28(4, July):701-
24.
Montiel, Peter. 1996. "Financial Policies and Economic Growth: Theory, Evidence, and
Country-Specific Experience from Sub-Saharan Africa." Journal of African Econo-
mies. Supplement to Vol. 5, No. 3, Part I.
Nigeria, Central Bank of. Various years. Economic and Financial Review [annual]. Lagos.
Nissanke, Machiko, and Ernest Aryeetey. Forthcoming. Financial Integration and De-
velopment: Financial Gaps under Liberalization in Four African Countries. London:
Routledge.
Parker, Ronald L., Randall Riopelle, and William F. Steel. 1995. Small Enterprises Ad-
justing to Liberalization in Five African Countries. World Bank Discussion Paper
271. Washington, D.C.: World Bank.
Roe, A. R. 1991. "Financial Systems and Development in Africa." A conference report
of an Economic Development Institute policy seminar, Nairobi, 1990. World Bank,
Washington, D.C. Processed.
Roemer, Michael, and Christine Jones. 1991. Markets in Developing Countries: Paral-
lel, Fragmented, and Black. San Francisco: ICS Press for International Centre for Eco-
nomic Growth and Harvard Institute for International Development.
Seibel, Hans Dieter. 1989. "Linking Informal and Formal Financial Institutions in Af-
rica and Asia." In Jacob Levitsky, ed., Microenterprises in Developing Countries.
London: Intermediate Technology Publications.
Seibel, Hans Dieter, and Michael T. Marx. 1987. Dual Financial Markets in Africa.
Saarbrucken: Breitenbach Publishers.
Shaw, E. S. 1973. Financial Deepening in Economic Development. New York: Oxford
University Press.
Soyibo, Adedoyin. 1996a. "Financial Linkage and Development in Sub-Saharan Africa:
The Role of Formal Financial Institutions in Nigeria." Working Paper 88. Overseas
Development Institute, London. Processed.
-. 1996b. "Financial Linkage and Development in Sub-Saharan Africa: A Study of
Informal Finance in Nigeria." Working Paper 90. Overseas Development Institute,
London. Processed.
Steel, William F., and Leila M. Webster. 1992. "How Small Enterprises in Ghana Have
Responded to Adjustment." The World Bank Economic Review 6(3):423-38.
218 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Stiglitz, Joseph E. 1989. "Financial Markets and Development." Oxford Review of Eco-
nomic Policy 5(4):55-66.
.1994. "The Role of the State in Financial Markets." In Michael Bruno and Boris
Pleskovic, eds., Proceedings of the World Bank Annual Conference on Development
Economics 1993. Washington, D.C.: World Bank.
Stiglitz, Joseph E., and Andrew Weiss. 1981. "Credit Rationing in Markets with Incom-
plete Information." American Economic Review 71(3):393-410.
Tanzania, Bank of. Various years. Economic and Operations Report. Dar es Salaam.
World Bank. 1994. Adjustment in Africa: Reforms, Results, and the Road Ahead. A
World Bank Policy Research Report. New York: Oxford University Press.
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 219-42
Civil Liberties, Democracy, and the Performance
of Government Projects
Jonathan Isham, Daniel Kaufmann, and Lant H. Pritchett
This article uses a cross-national data set on the performance of government invest-
ment projects financed by the World Bank to examine the link between government
efficacy and governance. It demonstrates a strong empirical link between civil liberties
and the performance of government projects. Even after controlling for other determi-
nants of performance, countries with the strongest civil liberties have projects with an
economic rate of return 8-22 percentage points higher than countries with the weakest
civil liberties. The strong effect of civil liberties holds true even when controlling for
the level of democracy.
The interrelationship among civil liberties, civil strife, and project performance sug-
gests that the possible mechanism of causation is from more civil liberties to increased
citizen voice to better projects. This result adds to the evidence for the view that in-
creasing citizen voice and public accountability-through both participation and bet-
ter governance-can lead to greater efficacy in government action.
Discussions of governance often generate more rhetorical heat than empirical
light. Governance, like religion, is a broad topic that inspires strong beliefs and
is difficult to measure reliably. Even a consensus on definitions is elusive: what
do we mean by governance? A World Bank policy paper defines governance as
"the manner in which power is exercised in the management of a country's eco-
nomic and social resources for development," which does not easily lend itself
to quantification (World Bank 1992, p. 1). We hope to shed some empirical
light on one dimension of governance by demonstrating a positive link between
a country's civil liberties and the performance of the government's investment
projects.
Governance involves actions of publicly vested authorities. We label three
interrelated dimensions of government action as what, how, and how well. What
public decisions are taken-including the enactment of laws, policies, and regu-
lations-affects the allocation of public expenditures and investments and de-
Jonathan Isham is with the Institutional Reform and the Informal Sector (IRIS) Center at the University
of Maryland, Daniel Kaufmann is with the Harvard Institute for International Development (on leave
from the World Bank), and Lant H. Pritchett is with the Policy Research Department at the World
Bank. The authors thank Deon Filmer and Phil Keefer for helpful comments, as well as seminar
participants at Columbia University, the University of Maryland, the Northeast Universities Development
Conference, and the World Bank. This article is a reworking of "Governance and the Returns on
Investment: An Empirical Investigation" (World Bank Policy Research Department Working Paper 1550).
C) 1997 The International Bank for Reconstruction and Development / THE WORLD BANK
219
220 THE WORLD BANK ECONOMIC REVIEW, VOL. I1, NO. 2
termines incentives for all other actors. How public decisions and authority are
exercised depends on underlying social structures, political structures, and offi-
cial and unofficial institutions. How well public decisions and authority are ex-
ercised determines the efficacy of government in accomplishing its objectives.
Although researchers have written a lot on what and how, they have written
less on how well. In this article, we analyze the impact of one element of how-
the degree of civil liberties-on one element of how well-the returns on gov-
ernment investments. Section I reviews recent empirical, cross-national litera-
ture linking economic outcomes with government action as well as recent work
on the efficacy of government action. Section II discusses data on economic rates
of return of public investment projects financed by the World Bank. These data
provide a unique quantitative measure of government performance that is com-
parable across countries. Section III presents evidence of a strong relationship
between civil liberties and these rates of return; this relationship is robust to a
wide variety of controls, including measures of democracy. Section IV explores
the links among civil liberties, citizen voice, and project performance.
I. WHAT, How, AND How WELL
Economic and social outcomes so depend on governance-for good and for
ill-that the what, how, and how well of government action underlie the richest
social science traditions. We cannot begin to do justice (even in outline) to this
literature. Therefore, in setting the context for our new results, we limit the
scope of our review to recent empirical, cross-national research focused princi-
pally on economic outcomes (and hence written mostly by economists).
Much recent literature concerns the impact on economic growth of what gov-
ernments do. A small share of these studies examines the effects of directly mea-
surable government actions on growth, such as levels and patterns of public
investment expenditures (Easterly and Rebelo 1993 and Devarajan, Swaroop,
and Zou 1996). A larger share examines the effects of a specific outcome associ-
ated with government actions, including school enrollment rates (Barro 1991),
outcome-based measures of outward orientation (Harrison 1995 and Dollar
1992), financial depth (King and Levine 1993), macroeconomic instability (Fischer
1993), and investment in machinery (de Long and Summers 1993). Most such
studies, however, offer no explicit link between specific government policies and
actions (for example, building more schools) and the growth-promoting out-
come (for example, higher enrollment rates).
Another strand of literature analyzes the effects of underlying social struc-
tures, political structures, and institutions that determine how governments
exercise public decisions and authority. Much of this work focuses on the
effects of civil and political liberties (Dasgupta 1993). Lipset (1960) demon-
strated the association between higher levels of income and higher levels of
civil liberties and of popular political participation. However, whether de-
mocracy promotes or hinders economic growth remains ambiguous. In the
Isham, Kaufmann, and Pritchett 221
1960s and 1970s scholars debated whether democracy was an insuperable
obstacle to development. Many argued that a premature move to democracy
hindered growth by increasing the influence of special interest groups, fo-
menting the competition for policy-induced rents, lowering savings rates,
reducing the stability of policy (especially macroeconomic policy), and fos-
tering political instability. This position seemed reasonable at the time. The
top 20 fastest growing major economies in 1960-74 included only three con-
sistent democracies (and only one of those was a developing country), four
decidedly authoritarian Asian economies, and four socialist countries. Cur-
rent research, although deeply divided, tends to find no causal link at all
between democracy and growth. Researchers have revisited the issue as part
of the resurgence of empirical work on economic growth (Weede 1983;
Kormendi and Meguire 1985; Scully 1988; Grier and Tullock 1989; Helliwell
1992; Barro 1994; and Bhalla 1994). Przeworski and Limongi (1993) and
Alesina and Perotti (1994) provide excellent reviews.
The what and how of government action are, of course, critically linked. Poli-
cies and actions matter, and underlying conditions partially determine the choice
of good or bad policies. The current studies on growth rarely document this
link. However, the literature on central bank independence has established the
connection between specific political institutional mechanisms that promote cen-
tral bank independence and economic outcomes of inflation and growth
(Cukierman, Webb, and Neyapti 1992). And recent work by Alesina (1996)
shows how institutional arrangements affect budgetary outcomes.
Much of the ambiguity about the impact of democracy on growth revolves
around whether more or less popular political participation leads to better or
worse policy outcomes. Two contrasting arguments seem to be well documented.
One argument is that more democratic arrangements may lead to greater public
investments in infrastructure, greater (and more equitable) investments in human
capital, more open trade policies (Tavares and Wacziarg 1996), and better provi-
sion of a secure legal system and property rights (Clague and others 1997). The
other argument is that more democratic arrangements may have negative effects
on government policies and actions when vested interests lobby for preferential
treatment and against efficiency-enhancing reforms (Olson 1965). Negative ef-
fects might occur when local pressures block needed investments because of "not
in my backyard" attitudes or when interest groups engage in wars of attrition in
order to avoid the costs of stabilization and promote populist macroeconomic
policy (Alesina and Drazen 1991). Recent work on economic reform is not en-
tirely sanguine about the ability of democratic politics as usual to bring about
economic reforms because the magnitudes of redistribution relative to efficiency
gains are often large (Rodrik 1996). According to one view of the success of
some East Asian governments in pursuing sensible macroeconomic policies, au-
thoritarian leaders effectively insulated meritocratically selected civil servants from
direct popular pressures (World Bank 1993b). This view also recognizes that
authoritarian leaders in other contexts have pursued disastrous policies.
222 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
A relatively small amount of literature addresses how well government ac-
complishes its objectives. A recent set of papers uses private-service ratings for
foreign investors to analyze government efficacy. Mauro (1995) examines the
impact of various measures of institutional quality and suggests that corruption
is associated with lower economic growth, primarily by reducing investment.
Knack and Keefer (1995) find significant negative effects of the overall quality
of government on economic outcomes. Chong and Calder6n (1996) explore the
connection between these same institutional quality indexes and economic
inequality.
We focus not on the impact of the how or how well of government policies on
aggregate economic outcomes, but on the connection between how and how
well. In particular, we examine the link between civil liberties and democracy-
critical determinants of how governments exercise public decisions and author-
ity-and the efficacy of public investments.
Why might economists expect such a link? In competitive markets, share-
holders and consumers goad managers of private corporate firms-with sepa-
rate management and ownership-into efficacy. In competitive markets, profit-
maximizing shareholders can choose alternative investments, and discriminating
buyers can choose alternative suppliers. By contrast, shareholders and consum-
ers do not pressure publicly vested authorities through these channels of choice.
Citizens cannot freely choose to own shares of another country. Citizen prefer-
ences are not linked to revenues for government services, because taxation is
ultimately coercive. Accordingly, other channels induce government performance,
including accountability, openness, transparency, predictability, and the rule of
law (Brautigam 1992). In Hirschman's evocative phrase, while markets create
managerial discipline and induce efficacy through the exercise of choice, govern-
ments are principally disciplined through the exercise of voice (Hirschman 1970).
However, very few empirical studies have documented the link between
citizen voice-facilitated by openness-and accountability and performance
(Paul 1992, 1994, and 1996). Comparing the performance of public irriga-
tion systems in India and Korea, Wade (1994) finds that when irrigation
officials face more local connections and accountability, the systems per-
form better than traditional arrangements that insulate civil servants from
performance pressures. Dreze and Sen (1989) argue that no country with a
free press has ever had a major famine. They postulate that a free flow of
information pressures (even nondemocratic) governments into public action.
Literature on the involvement of potential beneficiaries in government-
financed investment projects also suggests the importance of citizen voice
(World Bank 1995 and Korten and Siy 1988). For instance, Isham, Narayan,
and Pritchett (1995) show that aid-financed rural water supply projects per-
formed much better with greater participation of the beneficiaries. Overall,
these results suggest that citizen voice is an important determinant of gov-
ernment accountability and efficacy but do not identify the underlying social
and political conditions conducive to citizen voice.
Isham, Kaufmann, and Pritchett 223
This unexplored chain of reasoning-from social and political conditions to
citizen voice to government efficacy-frames the key hypothesis explored here.
We hypothesize that basic civil liberties-such as the freedom of individual ex-
pression, a pluralistic and free media, the ability of groups to organize, and
freedom of dissent and criticism-facilitate greater citizen voice and hence more
effective government action. We also consider whether citizen voice requires (or
is enhanced by) democracy. For example the country that Wade (1994) argues
had less public sector accountability (India) was clearly more democratic.
II. PROJECT PERFORMANCE AS AN INDICATOR OF GOVERNMENT EFFICACY
Conceptual and practical difficulties explain most of the lack of cross-
national research on determinants of the efficacy of government action. Deep
conceptual disagreements about what governments ought to do, including the
objectives that governments ought to pursue and the appropriate means to achieve
those objectives, plague the efforts to measure efficacy. These differences imply
that efficacy cannot be inferred from the success or failure in achieving mea-
sured aggregate outcomes like economic growth. Mistaken beliefs may cause
government to pursue policies that are inefficient, or even counterproductive,
relative to its ultimate objectives. For instance, many governments have actively
and deliberately discouraged many types of foreign investment. Whether that
policy has been effectively implemented is a distinct question from whether it
has promoted the desired outcomes.
In addition, a practical difficulty hinders the analysis. Nearly all data con-
cerning government actions concern public resources spent on inputs, not com-
parable outcomes. The data document finances allocated for roads, but not roads
built, and spending on health clinics, but not health outcomes. Nearly every
government supports education in a roughly similar way and collects a fair amount
of data on education spending. But analysts cannot compare cross-country effi-
cacy without comparable measures of student learning that are extremely rare in
developing countries. Overall, because governments do not spend money equally
effectively, we can learn very little from input data alone, and certainly nothing
about government efficacy (Pritchett 1996). For example, Putnam (1993) recog-
nizes this problem and devises his own measures of government efficacy for
assessing the performance of regional governments in Italy, where the scope of
regional government responsibility is assigned.
Our data provide an opportunity to overcome these conceptual and practical
obstacles. The data rate on a comparable quantitative scale the success of invest-
ment projects that governments have chosen to undertake. We use the economic
rate of return (ERR) as an indicator of outcomes (not just expenditures) calcu-
lated similarly for all countries. Moreover, we do not compare the amounts
different governments chose to invest, either in total or in distribution across
sectors. Rather we compare returns on government investments. The data also
have the advantage of being microeconomic and hence much less susceptible to
224 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
argument about reverse causation. Although the level of economic growth could
affect the level of civil liberties, we find it unlikely that the returns on individual
projects would affect the level of civil liberties.
The Data
The World Bank's Operations Evaluation Department (OED) constructed our
data on the performance of government investment projects financed by the World
Bank, including both loans from the International Bank for Reconstruction and
Development (IBRD) and credits from the International Development Association
(IDA). We exclude adjustment (or program) lending from our analysis, because it
raises a large set of problems with evaluation, which have been addressed on sev-
eral occasions both by the World Bank and its staff (Pritchett and Summers 1993)
as well as by other less sympathetic analysts. After full disbursement of each World
Bank loan-typically five to eight years after the opening of the loan-staff from
the World Bank and borrower country jointly write a project completion report
assessing project performance. The project completion report, or implementation
completion report, is usually written by a staff member in the World Bank division
that supervised the loan, but typically not by anyone with major project approval
responsibilities. This practice minimizes the incentives to dissemble about project
performance. As part of project assessment, OED staff judge each project as satis-
factory or unsatisfactory in achieving its development objectives.
In addition, for projects in eight economic subsectors with readily quantified
and valued project benefits-infrastructure, agriculture, industry, energy, water,
urban development, transport, and tourism-project staff, sometimes in collabo-
ration with OED, calculate an ERR. The ERR iS the discounted stream of project costs
and benefits over the life of the project, evaluated at economic (as opposed to
financial) prices and calculated following (roughly) the methodology of Squire and
van der Tak (1975). (See Little and Mirrlees 1991 for a discussion of economic
pricing in World Bank appraisals and the quality of cost-benefit analysis overall.)
The OED staff calculate the ERRS after project completion (ex post) in contrast to the
ex ante ERRS computed as part of the internal World Bank procedures for project
approval. Ex ante and ex post calculations of the ERRS differ by an enormous gap
(6-10 percentage points on average). The gap has a huge variability: regressing ex
post on ex ante ERRs results in an R2 of only about 0.2. Pohl and Mihaljek (1992)
study the determinants of this gap. Follow-up studies tend to find that even the ex
post ERRS tend to overstate the true economic rate of return because in many cases
projects do not sustain the benefit flows as long as anticipated in the ex post ERR
calculations. For ex post ERRS evaluators know actual implementation costs and
have somewhat more information about actual operating costs and demand, but
must still estimate most of the future stream of benefits.
Government Efficacy
Are the rates of return on government investment projects a reasonable proxy
for government efficacy? To find the answer, we address two issues. First we
Isham, Kaufmann, and Pritchett 225
evaluate the reliability and representativeness of the sample of World Bank-
financed projects. Second, we distinguish the impact of civil liberties on govern-
ment efficacy from other country- and project-level determinants of project
performance.
PROJECT ERRS AS AN EFFICACY INDICATOR. Suppose we know the ERR on every
government project j undertaken in country i in period t, ERRi'. Then we could
calculate the average ERR simply by averaging over all projects. But we do not
know the ERRS for all projects in any country, much less for all projects in many
countries. We can, however, observe the ERR on the subset of projects financed
by the World Bank.
Statistical inference based on this sample is difficult for three reasons. First, al-
though our sample contains an absolutely large number of projects, the median
number of projects per country is only 9 (average 13.5). Therefore, the average of
these few projects is at best a very noisy indicator of a country average. Second,
projects financed by the World Bank represent only a small fraction of most govern-
ments' investments. In our sample the average ratio of World Bank disbursements
to government investment is just 6 percent. Third, there is a great deal of within-
country heterogeneity in project returns (between-country variation in ERRS accounts
for only 13 percent of the total ERR variance), while there is very little variance over
time in country conditions like civil liberties. This combination implies that the gov-
ernance variables (many of which are for a single point in time in any case) do not
vary sufficiently to allow country fixed-effects estimation. If we have a representa-
tive sample, however, these problems merely stack the deck against us; these prob-
lems create low explanatory power and large standard errors. Thus, the results will
reveal whether we can overcome these problems.
The present empirical exercise does not focus on the representativeness of
the sample, because World Bank involvement in the project may raise the
ERR (compared with other government projects) through increased attention
and resources. Instead, we investigate the potential relationship between ERRS
and civil liberties that is specific to World Bank-financed projects. A simple
growth accounting relationship allows us to estimate the relationship be-
tween overall returns to capital and our sample of ERRS. The regression re-
sults suggest that ERRS are representative of economywide (not just govern-
ment) returns.
If the difference in performance of World Bank-financed projects com-
pared with the government portfolio depends on a country's civil liberties,
then a sample selection bias exists. This bias could happen for two reasons.
Countries can choose which of their possible projects to finance through the
World Bank. Thus the first reason for potential bias is that this choice may
involve cream skimming, in which governments seek World Bank financing
for projects with very high expected ERRS. Or, second, it may involve laggard
dumping, in which governments offer the World Bank the most problematic
projects and finance the best projects out of their own budget. In addition to
226 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
these potential causes for bias, the country's civil liberties could affect the
World Bank's selection of projects. The World Bank as a development insti-
tution invests in a wide variety of investment climates subject to the projects'
meeting some minimum criteria. We return to this selection problem below
in discussing the empirical results.
OTHER DETERMINANTS OF PROJECT PERFORMANCE. Even if the sample is
representative of the returns on the projects in the government's investment
portfolio, many factors influence the realized return other than government
efficacy. We can think of a schedule of projects as a frontier of potential or
achievable project returns from which the government chooses a subset.
Economywide and project-specific factors determine the location of this schedule
of returns (Isham and Kaufmann 1992, 1995 and Kaufmann and Wang 1995).
We identify the possible returns so that we can identify the deviations from this
potential as an indicator of government performance.
The lack of a strong correlation between the ERRS and other possible mea-
sures of project performance augurs against an interpretation of ERRS as an
indicator of government efficacy. For instance, the "Business Environmental
Risk Intelligence" and "International Country Risk Guide" rank countries by
various characteristics that indicate their attractiveness for foreign investment.
These various measures are not significantly correlated with the ERRS in our
data set (although they do show a reasonable correlation with the civil liber-
ties variables). In part the lack of correlation might occur because these pri-
vate sector ratings are flawed indicators of government effectiveness, as they
are designed for foreign investors. Governments that are not attractive to for-
eign investors on these criteria might still be reasonably effective in imple-
menting their own projects.
The basic unit of observation in the data is the project, implemented in a
specific country over a specific period. Prior to adding any indicator of civil
liberties to our analysis, we specify an equation that relates the ERRit" to three sets
of nongovernance variables: sectoral dummies, country characteristics, and re-
gional dummies. We report the results on these control variables in table A-1.
An annual publication by OED on evaluation results uses these data to examine
project performance by a number of characteristics (see World Bank 1993a).
Background papers for the most recent publication also examine the effect of
various country aggregate variables, including inflation and World Bank-
specific inputs, like supervision activity (Kilby 1995).
Three econometric issues deserve mention. First, the time-varying variables,
such as the black market premium, must be matched to the period relevant to
project performance. While the arguments can be made in favor of various weights
(such as disbursement profile weights), we use a three-year weighted average of
the time-varying variable, going back from the year in which the project evalua-
tion was done. Second, although the projects vary tremendously in total cost,
from $1.7 million to more than $1 billion, the standard tests do not indicate any
Isham, Kaufmann, and Pritchett 227
conditional heteroscedasticity as a function of project size, nor does weighting
ordinary least squares (OLS) estimates by project size affect the results. Third, by
OED convention the lowest ERR reported is negative 5 percent, which implies that
the data are truncated from below; hence, the reported regression results use
Tobit estimation unless otherwise noted. However, because only 8.4 percent of
the sample is at the truncation point (-5), the Tobit estimates are quite similar to
simple OLS estimates (Greene 1981). It is hard to believe that much is gained (or
lost) by using Tobit estimates.
We include a set of sectoral dummy variables because the sectors differ sub-
stantially in their ability and in their techniques for assessing the ERR. By includ-
ing the sector dummies, the differing patterns of sectoral investment across coun-
tries do not affect the estimates of the other parameters. We also include a dummy
variable for project complexity, which accounts for a subset of agricultural
projects, including all integrated rural development, irrigation and drainage, and
livestock projects, which presented some particular difficulties (World Bank
1988). Our estimates reconfirm that the ERRS for these projects are about 4 per-
centage points lower on average.
We include a set of time-varying country characteristics that potentially de-
termine returns. We use the economywide capital-labor ratio because a higher
capital-labor ratio lowers the potential return on capital. Our estimates confirm
this relation: a unit increase in the natural log of the capital-labor ratio reduces
the ERR by between 1 and 1.6 percentage points (table A-1). We use the terms of
trade because many analysts suspect that terms of trade shocks determine project
returns, both in the affected sector and in the economy as a whole. We do not
find a particularly large or significant effect. Policy and outcome variables also
potentially influence returns. We consider the black market premium to be an
omnibus indicator of distorted policies because it is associated with overvalued
exchange rates, trade distortions, and macroeconomic instability, all of which
have a strong negative impact on ERRS. Even accounting for the black market
premium, projects do better in countries with a larger fiscal surplus. We ex-
pected that gross domestic product (GDP) growth would also have a large impact
on returns, but the effect is modest.
We also include a set of regional dummies based on the World Bank group-
ings for Latin America and the Caribbean, Sub-Saharan Africa, South Asia,
East Asia, and Europe, the Middle East, and North Africa. We find as ex-
pected that projects in Sub-Saharan Africa do much worse (10 percentage
points), projects in Latin America and the Caribbean and in Europe, the
Middle East, and North Africa do about 5 percentage points worse, and coun-
tries in East Asia (which includes in addition to the high-performing East
Asian countries, the underperforming Southeast Asian and Pacific countries)
do about 3 percentage points worse (table A-1). The inclusion of the regional
controls does have a significant impact on the estimates of other variables,
so in all subsequent tables we report regressions with and without regional
controls.
228 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
III. CIVIL LIBERTIES, PROJECT PERFORMANCE, AND DEMOCRACY
Our results here are similar to those reported by Isham and Kaufmann (1995),
who argue that many variables, such as policy distortions, affect both public
and private sector projects. We ask whether civil liberties have an additional
effect on project performance if we control for the set of project and country
factors. We describe four measures of basic civil liberties that are relevant to the
ability of citizens to exercise voice and present the results of including these
measures as determinants of ERRS. We then look at the robustness of the relation
between civil liberties and ERRS using a wide variety of controls, including mea-
sures of democracy.
Measuring Civil Liberties
Freedom House (1994) publishes a ranking of civil liberties on a scale of 1 to
7 for 165 countries from 1972 to 1994 based on a checklist of 14 civil liberties.
The checklist includes media free of censorship, open public discussion, freedom
of assembly and demonstration, freedom of political organization, nondiscrimi-
natory rule of law in politically relevant cases, freedom from unjustified political
terror, free trade unions and peasant organizations, free businesses and coop-
eratives, free professional and other private organizations, free religious institu-
tions, personal social rights (for example, the right to own property and to travel
internally and externally), socioeconomic rights, freedom from gross socioeco-
nomic inequality, and freedom from gross government indifference or corrup-
tion. Humana (1986) ranks human rights achievement in 89 countries for 1985
on a scale of 0 to 100 (the actual range for our sample is 13 to 91) based on the
definition of human rights adopted by the General Assembly of the United Na-
tions in 1966 under the International Covenant on Civil and Political Rights.
The Humana index includes such items as the right of peaceful assembly, free-
dom of opinion and expression, the right and opportunity to take part in the
conduct of public affairs, the right to freedom of opinion and expression, and
the right to form trade unions. Coppedge and Reinicke (1990) rank 170 coun-
tries on two dimensions-media pluralism and freedom to organize-on a scale
of 1 to 3 for the year 1985.
Creating a reliable empirical cross-country indicator of civil liberties is obvi-
ously difficult, and any measure will be subjective and hence debatable. But the
actual differences across countries in liberties are so large that, in spite of the
complexity and subtleties, any reasonable assessment will produce the same ba-
sic pattern across countries. This result is indicated by the high correlations among
these measures of civil liberties. The correlation of the Freedom House index
(averaged over 1979-86) with the Humana index is 0.83, with freedom to orga-
nize, 0.78, and with media pluralism, 0.81. The correlation of the Humana in-
dex with freedom to organize is 0.68, and with media pluralism, 0.79. The cor-
relation of freedom to organize with media pluralism is 0.82. (Coppedge and
Reinicke's use of the information in the Freedom House and Humana studies in
Isham, Kaufmann, and Pritcbett 229
their own ranking procedure may account for at least part of the high correla-
tion between the latter two and former two series.)
Civil Liberties and Project Performance
Each of the four measures of civil liberties shows a statistically significant and
empirically large association with the return to projects (table 1). The estimates
that include regional dummy variables suggest that if the Freedom House civil
liberties index improved from that for the worst country (1) to that for the best
(7, as in Costa Rica), the ERR would be predicted to increase 8 percentage points,
50 percent of the mean ERR of 16. Similarly, improving from the least civil liber-
ties by the Humana index (13) to one of the best (91, again, Costa Rica) would
improve the ERR by an amazing 20 percentage points.
Each of the civil liberties indexes and other determinants of project perfor-
mance differs in scale. Therefore, to compare the different effects, we calculate
the predicted increase in the ERR if each index were improved by 1 standard
deviation (column 3 of table 1). A standard deviation improvement in civil liber-
ties would raise the predicted ERR 1.9 points using the Freedom House index,
4.5 points using the Humana index, and 2.6 points using the media pluralism
Table 1. The Impact of Civil Liberties Indicators on the Economic Rate
of Return of Government Projects, Controlling for Economic
and Project Variables
Effect of a 1 standard deviation
Without increase in civil liberties on
regional With regional the economic rate of returna
Index variables variables (percentage points)
Freedom House civil liberties, 1.95 1.32 1.9
1978-87 (0.0Oo)* (0.047)-
Humana, 1982-85 0.251 0.256 4.5
(0.009)* (0.025)-
Media pluralism, 1983-87 2.89 2.85 2.6
(0.013)* (0.062)"^'
Freedom to organize, 1983-87 2.45 -0.057 2.7b
(0.006)* (0.969)
p-level less than 0.05.
* p-level less than 0.10.
Note: The base specification includes capital-labor ratio, black market premium, GDP growth, fiscal
surplus, terms of trade changes, sectoral dummies, and a dummy for complex projects (see table A-1).
The estimation is based on annual values for 1978-87 for the Freedom House civil liberties index. For
the other three indexes, single values were extrapolated to cover the sample period. We report p-levels
of the test for whether the coefficient is 0 rather than test statistics themselves. The p-level is the significance
level at which the null hypothesis can be rejected, hence a p-level less than 0.05 indicates a rejection of
the null hypothesis at (at least) the 5 percent level. The p-levels are in parentheses. Sample sizes are 649
for the Freedom House civil liberties index, 236 for the Humana index, 389 for media pluralism, and
389 for freedom to organize.
a. The standard deviations-for the entire sample for which each variable is available-are 1.47 for
the Freedom House civil liberties index, 17.8 for the Humana index, 0.91 for media pluralism, and
1.12 for freedom to organize.
b. Using the estimate without regional dumrnies.
Source: Authors' calculations.
230 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
index. These effects of civil liberties on project returns are empirically large com-
pared with those of macroeconomic policy, an effect that has received a great
deal of attention (World Bank 1991). The average of the standardized effect of
the four civil liberties indicators on project returns (2.9 percentage points) is
much larger than equivalent changes in terms of trade shocks, fiscal deficits, or
GDP growth (column 4 of table A-1). Improving civil liberties by a standard
deviation would improve project performance by about as much as a standard
deviation fall in the black market premium (3.31, table A-1). Although the total
effect of good macroeconomic policies is larger (as the effects are additive), clearly
civil liberties are as important as any other single determinant of project success.
The relationship between civil liberties and ERRS is the central positive finding
of this article. We show that this result is robust to outliers, to the measure of
project performance, to possible financing selection effects, and to the inclusion
of other variables in the base specification. Of special interest, the inclusion of
indicators of political liberties or democracy does not shift the estimates of the
importance of civil liberties.
OUTLIERS. A concern with any econometric result is its sensitivity to a few
observations. Although the civil liberties indicators we use are bounded, some
projects have extreme values for the ERR (the maximum is 155, table A-2). We
address the robustness of the estimates to extreme observations and influential
data points in two ways. First, we estimate a Tobit specification with lower and
upper truncation. Censoring the ERRS above at 50 percent (roughly 2 standard
deviations above the mean) does not affect the results. Second, we estimate the
same specifications using quantile (median) regression, a technique that is much
more robust to extreme observations than Tobit estimates. All the civil liberties
variables that are significant in table 1 are also significant using median
regressions.
PROJECT PERFORMANCE INDICATOR. The results are not unique to the ERR. If we
use the binary "satisfactory or unsatisfactory" rating created by OED, we obtain
qualitatively similar results. We have a larger sample of projects using only this
rating as the measure of project performance because we include social sector
projects that normally do not receive an ERR. (See Kaufmann and Wang 1995
for a discussion of the performance of social sector projects as a function of
macroeconomic policies.) Table 2 reports the estimates of a Probit regression.
Naturally, because the binary indicator discards a great deal of statistical
information, we obtain less precise results: the p-levels are generally higher, and
the estimates for the Humana ranking are even insignificant.1 Nevertheless, the
other variables show large increases in the likelihood of a successful project
when implemented in countries with higher civil liberties.
1. In the tables we report p-levels of the test whether the coefficient is 0 rather than test statistics
themselves. The p-level is the significance level at which the null hypothesis can be rejected, hence a p-
level less than 0.05 indicates a rejection of the null hypothesis at (at least) the 5 percent level.
Isham, Kaufmann, and Pritchett 231
Table 2. The Impact of Civil Liberties on the Probability of a Project Being
Rated Satisfactory Using a Probit Regression, Controlling for Economic
and Project Variables
Effect of a 1 standard deviation
Without increase on the probability
regional With regional of project successa
Index variables variables (percentage points)
Freedom House civil liberties, 0.018 0.022 3.2
1978-90 (0.056)* (0.060)-
Humana, 1982-86 -0.00067 0.0012 2.1
(0.589) (0.388)
Media pluralism, 1983-90 0.022 0.054 4.9
(0.296) (0.045)1
Freedom to organize, 1983-90 0.042 0.040 4.5
(0.009)1 (0.085)1
* p-level less than 0.05.
*- p-level less than 0.10.
Note: The value reported is not the coefficient in the Probit regression, but the marginal change in
the probability of a successful project as the variable changes, evaluated at the means of all independent
variables. See table A-1 for the complete specification. The estimation is based on annual values for
1978-87 for the Freedom House civil liberties index. For the other three indexes, single values are
extrapolated to cover the sample period. Sample sizes are 1,155 for the Freedom House civil liberties
index, 604 for the Humana index, 740 for media pluralism, and 740 for freedom to organize. The
p-levels of the test for whether the Probit coefficient is 0 are in parentheses; note that this is not the same
as the p-level of the statistic reported.
a. The standard deviations-for the entire sample for which each variable is available-are 1.47 for
the Freedom House civil liberties index, 17.8 for the Humana index, 0.91 for media pluralism, and 1.12
for freedom to organize.
Source: Authors' calculations.
For instance, from table 2 using the mean of the Freedom House variable, an
increase of 1 standard deviation in civil liberties lowers the probability of a
failed project 3.2 percentage points, which reduces the predicted failure rate 16
percent (from the mean failure of 20 percent). Similarly, an increase of 1 stan-
dard deviation in media pluralism reduces the failure rate almost 5 percentage
points, or 25 percent (table 2).2
SELECTION EFFECTS. Do selection effects create the relationship between civil
liberties and the performance of World Bank-financed projects? We consider
two perspectives on World Bank project selection decisions. One interpretation
says that the World Bank's Articles of Agreement preclude explicit consideration
of noneconomic factors, particularly civil liberties or political factors, in the
selection of World Bank projects. This view suggests that World Bank project
selection should be uncorrelated with civil liberties. We create an indicator to
measure World Bank involvement in a country's investment as the ratio of World
Bank loan or credit disbursements to total government investment. We find a
negative bivariate correlation between World Bank involvement and civil liberties,
principally because the World Bank has greater involvement in poorer countries,
2. It takes some calculations to compare, but the magnitudes of the effects are roughly similar to
those with ERRS.
232 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
which on average have fewer civil liberties. Controlling for per capita income
and population, we find no correlation between World Bank involvement and
civil liberties using the Freedom House index or the media pluralism index. We
find only mildly positive correlation between World Bank involvement and civil
liberties using the Humana index (p-level, 0.07). If we introduce World Bank
involvement as an independent explanatory variable, we obtain reasonably robust
results, in that the sign and magnitude of the effects are roughly similar in all
regressions and the statistical significance is maintained except in some of the
regressions with regional dummies.
The second perspective says that decisions concerning World Bank project
selection are based on the ex ante ERRS, not the ex post ERRS (which are obvi-
ously available only after project completion). If World Bank project selection
causes the partial association of civil liberties and ERRS, the association should
appear in the ex ante ERRS. However, when we use the ex ante ERR as the depen-
dent variable, we find no relationship with the Humana index (p-level, 0.98), no
relationship with the media pluralism index (p-level, 0.59), and a modest nega-
tive relationship with the Freedom House index (p-level, 0.10). These results
suggest that differences in the implementation of the projects, not differences in
project selection for financing, cause the relation between project performance
and civil liberties.
OMITTED VARIABLES. The partial association between civil liberties and ERRS is
robust. We explore the possibility, however, that some other variable is associated
with both project performance and civil liberties and hence that the partial
association of civil liberties is an artifact of bias from an omitted variable. This
possibility has two versions: an incidental association between the omitted variable
and civil liberties, or, much worse, civil liberties as a proxy for the true omitted
variable. We address these concerns in turn, with a separate section devoted to
the impact of civil liberties and democracy on project performance.
The specification of the variables included in the general specification for
project performance is not tightly theoretically constrained. Our analysis con-
cerns the robustness of the project performance result. Therefore, we experi-
ment with "data undermining" by searching for variables whose inclusion changes
the civil liberties results. Besides those variables reported in our base specifica-
tion in table A-1, we experimented with the inclusion of other variables. We
tried the stock of education because greater human capital perhaps led to higher
returns. We tried an indicator for trade policy because results by L6pez (1995)
suggest an interaction between trade and returns to capital. We tried
ethnolinguistic fractionalization, which Easterly and Levine (1996) show is as-
sociated with good economic outcomes and good government policies. And we
tried dummy variables for whether the country gained independence from France,
Spain, or the United Kingdom and for the year the country gained indepen-
dence; Chong and Calder6n (1996) argue that these factors have a lasting effect
on government institutional arrangements. We also added a dummy variable for
Isham, Kaufmann, and Pritchett 233
IDA credits (as separate from IBRD loans) and found no difference. Although each
of these variables is plausibly correlated with both civil liberties and government
efficacy, their inclusion in the project performance equation did not substan-
tially alter the magnitude or significance of the civil liberties coefficient.3
Civil Liberties and Democracy
By far the most important question on robustness is whether the results re-
flect civil liberties or capture some more directly political element. Civil and
political liberties and more democratic political regimes are closely associated
with each other, both of necessity (a certain degree of civil liberties is a precon-
dition for democracy) and in practice. Yet there are clear analytical and practi-
cal distinctions between civil liberties and more strictly political rights and prac-
tices. In particular, the degree of civil and political liberties varies widely among
nondemocracies. At the extremes, totalitarian regimes clearly differ from au-
thoritarian regimes in the degree to which the regime attempts to control nonpo-
litical dimensions of society and in the degree to which it tolerates opposition,
criticism, and dissent. Therefore, finding an association between more civil lib-
erties and better ERRS does not imply an association between different types of
political regimes and better performance. Here we explore the association be-
tween ERRS and political liberties and type of political regime.
As with civil liberties, measuring and classifying political regimes raises sub-
stantial difficulties. The most widely used measure of democracy in the eco-
nomic literature is the Freedom House index of political liberties, a subjective
ranking from 1 to 7 based on 11 indicators of political rights: chief authority
recently elected by a meaningful process; legislature recently elected by a mean-
ingful process; fair election laws; fair reflection of voter preference in distribu-
tion of power; multiple political parties; recent shifts in power through elec-
tions; significant opposition vote; freedom from domination by the military,
foreign powers, and other powerful groups; no major group or groups denied
reasonable self-determination; decentralized political power; and informal con-
sensus (de facto opposition power). Alesina and others (1992) construct another
index of type of political regime that provides an annual ranking for 1982-94
for 43 countries by democratic status on a three-point scale. The complexity of
the classification of political systems does not impede a reasonably reliable cross-
national ranking of countries. The correlation of the Freedom House political
liberties index with Alesina's democracy index is 0.69.
When we include the indicators of civil liberties in the equation for project
performance together with indicators of democracy, the civil liberties indicators
retain all of their importance, while the democracy indicators do not have any
additional explanatory power (see table 3). The Freedom House political liber-
ties variable shows a weak association alone, and when combined with a civil
liberties variable it is consistently negative (sometimes significantly so). Simi-
3. Although the results on some of these variables might be of independent interest, we do not report
the results because we do not want to appear to be mining the data.
234 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 3. The Impact of Civil Liberties and Democracy Variables
on the Economic Rate of Return of Government Projects
Civil liberties indicator
No civil Freedom House
liberties civil liberties Humana Media
Variable indicator index index pluralism
Freedom House political 0.138 -2.08 -0.798 -0.594
liberties index (0.805) (0.025)* (0.526) (0.175)
Civil liberties indicator 3.39 0.297 3.41
(0.003)* (0.024)* (0.062)'*
Alesina democracy index 3.61 3.96 5.77 6.03
= 2 (medium democratic) (0.163) (0.135) (0.159) (0.088);*
Alesina democracy index 0.651 0.989 -0.376 2.51
= 1 (most democratic) (0.757) (0.638) (0.921) (0.434)
Civil liberties indicator 1.25 0.271 2.67
(0.081)** (0.023)* (0.082)-'
Sample size
Using the Freedom House
political liberties index 649 649 236 448
Using the Alesina
democracy index 372 372 236 448
* p-level less than 0.05.
* p-level less than 0.10.
Note: The base case regression is as in table 1 including the regional dummies (see table A-1). We
report p-levels of the test whether the coefficient is 0 rather than test statistics themselves. The p-level is
the significance level at which the null hypothesis can be rejected, hence a p-level less than 0.05 indicates
a rejection of the null hypothesis at (at least) the 5 percent level. The p-levels are in parentheses.
Source: Authors' calculations.
larly, including Alesina's democracy index has no impact on the estimates of the
impact of civil liberties.4 We do not place much importance on these negative
results on democracy because the two variables, civil liberties and political re-
gime, move closely together (the correlation of the civil and political Freedom
House variables is 0.89). Their closeness creates both statistical and interpreta-
tional problems, but the civil liberties variable is not a proxy for democracy.
More important than the statistical concerns is the problem of practical inter-
pretation. Because the civil and political liberties variables typically move in
tandem, the question of the impact of changing civil liberties without changing
democracy may not be practically relevant. Nearly every policy change that
changes civil liberties is likely to have as its natural counterpart a political change
as well. Hence the usual ceteris paribus assumption-that all else (particularly
political liberties) remains the same-in assessing shifts in civil liberties is inap-
propriate and should be replaced with an assumption that the two variables
move together. The results from column 2 of table 3 (in which the civil and
4. This finding is robust to the use of other indicators of democracy. We also used an indicator of
type of political regime created by the IRIS center at the University of Maryland and an indicator of fair
elections from Coppedge and Reinicke (1990). Using these indicators gave similar results of no partial
impact of democracy and unchanged estimates on civil liberties.
Isham, Kaufmann, and Pritchett 235
political Freedom House variables have the same scale) show that if we increase
both civil and political variables by 1 (on the common scale of 1 to 7), the ERR
would increase about 1.31 percentage points. The regression suggests that this
net effect is due to a large positive effect of civil liberties (3.39) offset by a large
negative political effect (-2.08). Most important, the joint shift (1.31) is of the
same magnitude of the shift in civil liberties alone (1.32) estimated from table 1.
This result suggests that the total effect of an improvement in civil liberties is
positive, even accounting for the induced political changes.
IV. CIVIL LIBERTIES, CIVIL STRIFE, AND PROJECT PERFORMANCE
In the data, an interesting interrelationship among civil liberties, civil strife,
and project performance suggests that the possible mechanism of causation is
from more civil liberties to increased citizen voice to better projects. After con-
trolling for population, higher indicators of some types of civil strife, such as an
increased number of riots, protest demonstrations, and strikes, are strongly posi-
tively correlated with project performance (table 4). High ERR countries have
average rates of return twice as high (22.2) as low ERR countries (11.2). High
ERR countries have many more riots, demonstrations, and political strikes per
capita (adjusted for population) than countries with poor project performance.
The civil unrest variables (riots, protest demonstrations, and strikes) come as
the number of incidents per country per year (Banks 1979, updates). This means
that countries with larger populations have a greater absolute number of inci-
dents. However, it does not seem right simply to normalize to per capita, as
there are plausibly some increasing returns to scale in civil unrest. Consequently,
for each of the three variables we regress the absolute number of incidents on
population*ln(population), which is equivalent to adjusting the per capita level
for the total population in semilog form. We report the residual of this regres-
sion as excess civil unrest over the amount expected for a given level of popula-
tion. The population adjustment is also very significant, and the R2 varies from
0.02 (strikes) to 0.18 (riots). The results reported below were unchanged by
using other concave functional forms in place of this semilog form.
That greater civil tension is associated with better projects might appear puz-
zling. Typically, analysts associate all forms of political and social instability
with worse investment climate. They base this reasoning on associating civil
strife with risks to private projects and with political instability. In our analysis,
governments finance all the projects. We tried including as separate regressors
indicators of political instability, such as the Taylor and Jodice (1983 and supple-
ments) series on irregular government transfers and an index by Alesina and
Perotti (1993) on sociopolitical instability, but neither had any impact on project
success or the civil liberties variables.
Some degree of civil tension reflects a citizen's ability to agitate and influence
government's behavior without negative repercussions, a mechanism that plau-
sibly leads to greater accountability and hence better choice and implementation
236 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 4. The Impact of Civil Strife Variables on the Economic Rate of Return
of Government Projects
Civil strife indicatora
Political Protest
Impact measure Riots strikes demonstrations
Deviation of civil strife from population-adjusted levelb
High-ERR countriesc 2.48 3.19 0.30
Medium-ERR countriesd 0 -0.02 0.16
LOW-ERR countriese -0.19 -0.23 -0.04
Correlation of population-adjusted level of civil strife 0.27 0.34 0.17
with Freedom House civil liberties index (0.000)* (0.000)* (0.000)"
Estimates of the project performance regressionf
Without the Freedom House civil liberties index 0.42 1.67 0.81
(0.040)* (0.097)'* (0.003)'
With the Freedom House civil liberties index 0.21 0.45 0.68
(0.34) (0.683) (0.013)"
* p-level less than 0.05.
** p-level less than 0.10.
Note: We report p-levels of the test whether the coefficient is 0 rather than test statistics themselves.
The p-level is the significance level at which the null hypothesis can be rejected, hence a p-level less than
0.05 indicates a rejection of the null hypothesis at (at least) the 5 percent level. The p-levels are in
parentheses. The sample size is 649.
a. Values for the civil strife indicators are per capita, adjusted for total population size. We regress
the absolute number of incidents on populatiostln(population), which is equivalent to adjusting the per
capita level for the total population in semilog form, and report the residual of this regression as excess
civil unrest over the amount expected for a given level of population. The population adjustment is also
very significant, and the R-squared varies from 0.02 (strikes) to 0.18 (riots).
b. ERR categories are determined by average rates of return classified by country for all countries
with at least 10 projects for 1974-87.
c. There are six high-ERR countries, three in South Asia and three in East Asia. The average ERR for
the high-ERR countries is 22.2.
d. There are 11 medium-ERR countries, five in Latin America and the Caribbean; two in Sub-Saharan
Africa; three in Europe, the Middle East, and North Africa; and one in South Asia. The average ERR for
the medium-ERR countries is 17.
e. There are 12 IOW-ERR countries, nine in Sub-Saharan Africa, two in Latin America and the Caribbean,
and one in South Asia. The average ERR for the lOw-ERR countries is 11.2
f. This is the base regression (see table A-1) without sectoral or regional dummies.
Source: Authors' calculations.
of projects. Indeed, table 4 shows that higher civil liberties are strongly associ-
ated with higher levels of riots, demonstrations, and political strikes (although
regional dummy variables sharply attenuate this effect). Table 4 also shows,
even controlling for our set of exogenous and policy variables, a positive and
significant relation between the ERR and the number of riots, protest demonstra-
tions, and political strikes. However, adding the degree of civil liberties sharply
reduces the estimated impact of political manifestations: the coefficient on riots
falls from 0.42 to 0.21, and the coefficient on strikes falls from 1.67 to 0.45.
For a given level of civil liberties, neither riots nor political strikes are associ-
ated with better performance (although the protest demonstrations variable does
retain some effect). The results support a chain of causation that runs from
greater civil liberties to higher levels of citizen involvement and political partici-
Isham, Kaufmann, and Pritchett 237
pation-including as one dimension civil manifestations-to better projects.
Environments that allow civil strife or unrest to occur also allow other mecha-
nisms for expression of popular (dis)content with government performance. The
availability and effectiveness of those mechanisms improve government efficacy.
V. CONCLUSIONS
The extent of a country's civil liberties has a substantial impact on the suc-
cessful implementation of government investment projects financed by the World
Bank. This impact of civil liberties is as empirically large as the more celebrated
impact of economic distortions on project returns. Given that citizen voice is an
important precondition for government accountability and, not coincidentally,
that voice is suppressed in the absence of civil rights, this result is perhaps not
surprising. This result adds to the evidence for the view that increasing citizen
voice and public accountability-through both participation and better gover-
nance-can lead to greater efficacy in government action. Some analysts argue
that there is a trade-off between liberties and development. We find the opposite
evidence, that suppressing liberties is likely to be inimical to government perfor-
mance. This has obvious implications not just for governments but also for de-
velopment assistance (Picciotto 1995 and OECD 1995).
The most important aspects of civil liberties and political regimes go beyond
whether they promote or discourage economic outcomes. Here we have exam-
ined the instrumental value of civil liberties and political structure in producing
greater efficacy of government. Although we have focused on the instrumental
value, we want to emphasize that we believe government respect for civil liber-
ties is valuable regardless of its instrumental economic value.
(Appendix tables begin on the following page.)
238 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table A-1. Base Specification for the Nongovernance Determinants
of the Economic Rate of Return of Government Projects, 1974-87
With regional dummies
Estimate Effect of a 1 standard
Number of without deviation increase
dummy regional on the economic
Variable Meana variables dummies Estimate rate of return
Exogenous variable
ln(capital/labor) 8.22 -1.09 -1.66 -1.67
[1.01] (0.067)** (0.060)**
Dummy for project 319 -4.29 -4.23
complexity (0.017)* (0.016)*
Terms of trade shock -3.29 0.0015 0.001 0.0035
[3.3S] (0.889) (0.922)
Policy variable
Black market premia 46.6 -0.046 -0.037 -3.31
[89.5] (0.000)* (0.000)*
Fiscal surplus -5.21 0.197 0.266 0.925
[3.48] (0.149) (0.063)--
GDP growth 3.71 0.193 0.013 0.646
[3.35] (0.357) (0.949)
Regional dummy variableb
East Asia 278 -3.33
(0.154)
Latin America and the 314 -4.74
Caribbean (0.072)-
Europe, the Middle East, and 283 -4.93
North Africa (0.100*)
Sub-Saharan Africa 430 -10.8
(0.000)#
Sectoral dummy variablec
Agriculture 604 0.027 1.39
(0.992) (0.602)
Energy and public utilities 339 -3.92 -3.18
(0.136) (0.220)
Transport and tourism 413 3.85 6.24
(0.137) (0.016)*
Urban 48 10.1 11.9
(0.011)* (0.003)*
p-level less than 0.05.
* p-level less than 0.10.
Note: We report p-levels of the test whether the coefficient is 0 rather than test statistics themselves.
The p-level is the significance level at which the null hypothesis can be rejected, hence a p-level less than
0.05 indicates a rejection of the null hypothesis at (at least) the 5 percent level. The p-levels are in
parentheses. The sample size is 761.
a. Standard deviations are in square brackets. Standard deviations are calculated for the entire sample.
b. South Asia (184 observations) is excluded. Regions are based on World Bank regional classifications.
c. Industry (84 observations) is excluded.
Source: Authors' calculations. For exogenous and policy variables, World Bank data.
Isham, Kaufmann, and Pritchett 239
Table A-2. Summary Statistics
Standard Number of
Variable Mean deviation Range countries Years
Economic rate of return 16.01 15.16 -5-155 56a 1974-90
Civil liberties index
Freedom House 4.68 1.47 1-7 56 1974-90
Humana 55.13 17.80 13-91 38 1986
Media pluralism 2.50 0.91 1-4 56 1985
Freedom to organize 2.45 1.12 1-4 56 1985
Political liberties index
Freedom House 4.73 1.85 1-7 55 1974-90
Alesina 2.52 0.79 1-3 55 1974-82
Civil unrest indicator
Riots 0.14 1.61 -3.83-17.50 56 1974-89
Protest demonstrations 0.29 1.63 -0.79-14.54 56 1974-89
Strikes 0.07 0.50 -0.43-3.50 56 1974-89
a. 1,488 projects.
Source: Authors' calculations; Freedom House (1994); Humana (1986); Alesina and others (1992);
Banks (1979 and updates).
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Alesina, Alberto. 1996. "Budget Deficits and Budget Institutions." NBER Working Paper
5556. National Bureau of Economic Research, Cambridge, Mass. Processed.
Alesina, Alberto, and Allan Drazen. 1991. "Why Are Stabilizations Delayed?" Ameri-
can Economic Review 81 (December): 1170-88.
Alesina, Alberto, Sule Ozler, Nouriel Roubini, and Philip Swagel. 1992. "Political Insta-
bility and Economic Growth." NBER Working Paper 4173. National Bureau of Eco-
nomic Research, Cambridge, Mass. Processed.
Alesina, Alberto, and Roberto Perotti. 1993. "Income Distribution, Political Instability,
and Investment." NBER Working Paper 4486. National Bureau of Economic Research,
Cambridge, Mass. Processed.
1. 994. "The Political Economy of Growth: A Critical Survey of the Recent Lit-
erature." The World Bank Economic Review 8(3):351-73.
Banks, Arthur S. 1979 (and subsequent updates). "Cross-National Time Series Data
Archive." Center for Social Analysis, State University of New York at Binghamton.
Processed.
Barro, Robert. 1991. "Economic Growth in a Cross Section of Countries." Quarterly
Journal of Economics 106(2):407-43.
. 1994. "Democracy and Growth." NBER Working Paper 4909. National Bureau
of Economic Research, Cambridge, Mass. Processed.
Bhalla, Surjit. 1994. "Freedom and Economic Growth: A Virtuous Cycle." In Axel
Hadeneus, ed., Democracy's Victory and Crisis: Nobel Symposium 1994. Cambridge,
U.K.: Cambridge University Press.
240 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Brautigam, Deborah. 1992. "Governance, Economy, and Foreign Aid." Studies in Com-
parative International Development 27(3, fall):3-25.
Chong, Alberto, and Cesar Calder6n. 1996. "Institutions and Income Inequality." De-
partment of Economics, University of Rochester, New York. Processed.
Clague, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson. 1997. "Prop-
erty Rights under Democracy and Dictatorship." The Journal of Economic Growth.
1(2):243-76.
Coppedge, Michael, and Wolfgang H. Reinicke. 1990. "Measuring Polyarchy." Studies
in Comparative International Development 25(1, spring):51-72.
Cukierman, Alex, Steven Webb, and Bilin Neyapti. 1992. "Measuring the Independence
of Central Banks and Its Effect on Policy Outcomes." The World Bank Economic
Review 6(September):353-98.
Dasgupta, Partha. 1993. An Inquiry into Well-Being and Destitution. Oxford: Clarendon
Press.
De Long, Bradford, and Lawrence H. Summers. 1993. "How Strongly Do Developing
Economies Benefit from Equipment Investment." Journal of Monetary Economics
32(3):395-415.
Devarajan, Shanta, Vinaya Swaroop, and Hong-Fu Zou. 1996. "The Composition of
Public Expenditures and Economic Growth." Journal of Monetary Economics
37(2):313-44.
Dollar, David. 1992. "Outward-Oriented Developing Economies Really Do Grow More
Rapidly: Evidence from 95 LDCs, 1976-1985." Economic Development and Cultural
Change 40(April):523-44.
Dreze, Jean, and Amartya Sen. 1989. Hunger and Public Action. New York: Oxford
University Press.
Easterly, William, and Ross Levine. 1996. "Africa's Growth Tragedy." Policy Research
Department, World Bank, Washington, D.C. Processed.
Easterly, William, and Sergio Rebelo. 1993. "Fiscal Policy and Economic Growth: An
Empirical Investigation." Journal of Monetary Economics 32(3):417-58.
Fischer, Stanley. 1993. "Role of Macroeconomic Factors in Growth." Journal of Mon-
etary Economics 32(3):485-512.
Freedom House. 1994. Freedom in the World: Political Rights and Civil Liberties. New
York.
Greene, William H. 1981. "On the Asymptotic Bias of the Ordinary Least Squares Esti-
mator of the Tobit Model." Econometrica 49(3):505-13.
Grier, Kevin B., and Gordon Tullock. 1989. "An Empirical Analysis of Cross-National
Economic Growth, 1951-1980." Journal of Monetary Economics 24(September) :259-
76.
Harrison, Ann. 1995. "Openness and Growth: A Time-Series, Cross-Country Analysis
for Developing Countries." NBER Working Paper 5221. National Bureau of Economic
Research, Cambridge, Mass. Processed.
Helliwell, John. 1992. "Empirical Linkages between Democracy and Growth." NBER
Working Paper 4066. National Bureau of Economic Research, Cambridge, Mass.
Processed.
Hirschman, Albert 0. 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms,
Organizations, and States. Cambridge, Mass.: Harvard University Press.
Isham, Kaufmann, and Pritchett 241
Humana, Charles. 1986. World Human Rights Guide. London: Hodder and Stoughton.
Isham, Jonathan, and Daniel Kaufmann. 1992. "Does Policy-Based Lending Affect the
Productivity of Investment Projects?" Background paper for Adjustment Lending and
Mobilization of Private and Public Resources for Growth. Policy Research Depart-
ment, World Bank, Washington, D.C. Processed.
. 1995. "The Forgotten Rationale for Policy Reform: The Productivity of Invest-
ment Projects." Policy Research Working Paper 1549. Policy Research Department,
World Bank, Washington, D.C. Processed.
Isham, Jonathan, Deepa Narayan, and Lant Pritchett. 1995. "Does Participation Im-
prove Performance? Establishing Causality with Subjective Data." The World Bank
Economic Review 9(2):175-200.
Kaufmann, Daniel, and Yan Wang. 1995. "Macroeconomic Policies and Project Perfor-
mance in the Social Sectors." World Development 23(5):751-65.
Kilby, Christopher. 1995. "Risk Management: An Econometric Investigation of Project-
Level Factors." Background paper for The Annual Review of Evaluation Results 1994.
Operations Evaluation Department, World Bank, Washington, D.C. Processed.
King, Robert G., and Ross Levine. 1993. "Finance, Entrepreneurship, and Growth: Theory
and Evidence." Journal of Monetary Economics 32(December):513-42.
Knack, Stephen, and Philip Keefer. 1995. "Institutions and Economic Performance: Cross-
Country Tests Using Alternative Institutional Measures." Economics and Politics
7(3):207-28.
Kormendi, Roger C., and Philip G. Meguire. 1985. "Macroeconomic Determinants of
Growth." Journal of Monetary Economics 16(2):141-63.
Korten, Frances F., and Robert Siy, Jr., eds. 1988. Transforming Bureaucracy: The Ex-
perience of the Philippine National Irrigation Administration. West Hartford,
Conn.: Kumarian Press.
Lipset, Seymour. 1960. Political Man. Garden City, N.J.: Doubleday.
Little, I. M. D., and J. A. Mirrlees. 1991. "Project Appraisal and Planning Twenty Years
On." In Proceedings of the World Bank Annual Conference on Development Eco-
nomics 1990, pp. 351-82. Washington, D.C.: World Bank.
L6pez, Ram6n. 1995. "Synergy and Investment Efficiency Effects of Trade and Labor
Market Distortions." European Economic Review 39(7, August):1321-44.
Mauro, Paulo. 1995. "Corruption, Country Risk, and Growth." Quarterly Journal of
Economics 110(3):681-712.
OECD (Organization for Economic Cooperation and Development). 1995. Participatory
Development and Good Governance. Development Cooperation Guideline Series. Paris.
Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of
Groups. Cambridge, Mass.: Harvard University Press.
Paul, Samuel. 1992. "Accountability in Public Services: Exit, Voice, and Control." World
Development 20(July):1047-60.
. 1994. "Does Voice Matter?" Policy Research Working Paper 1388. Policy Re-
search Department, World Bank, Washington, D.C. Processed.
. 1996. "A Citizen Report Card on Public Services: Mixing Barks and Bites?"
Processed.
Picciotto, Robert. 1995. Putting Institutional Economics to Work: From Participation
to Governance. Discussion Paper 304. Washington, D.C.: World Bank.
242 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Pohl, Gerhard, and Dubravko Mihaljek. 1992. "Project Evaluation and Uncertainty in
Practice: A Statistical Analysis of Rate-of-Return Divergences in 1,015 World Bank
Projects." The World Bank Economic Review 6(2):255-77.
Pritchett, Lant. 1996. "Mind Your p's and q's: The Value of Public Capital Is Not the
Cost of Public Investment." Policy Research Working Paper 1660. Policy Research
Department, World Bank, Washington, D.C. Processed.
Pritchett, Lant H., and Lawrence H. Summers. 1993. "The Structural Adjustment De-
bate." American Economic Review 83(May):383-89.
Przeworski, Adam, and Fernando Limongi. 1993. "Political Regimes and Economic
Growth." Journal of Economic Perspectives 7(3):51-71.
Putnam, Robert. 1993. Making Democracy Work. Princeton, N.J.: Princeton University
Press.
Rodrik, Dani. 1996. "Understanding Economic Reform." Journal of Economic Litera-
ture. 39(March):9-41.
Scully, Gerald W. 1988. "The Institutional Framework and Economic Development."
Journal of Political Economy 96(3):652-62.
Squire, Lyn, and Herman van der Tak. 1975. Economic Analysis of Projects. Baltimore,
Md.: Johns Hopkins Press.
Tavares, Jose, and Romain Wacziarg. 1996. "How Democracy Fosters Growth." De-
partment of Economics, Harvard University, Cambridge, Mass. Processed.
Taylor, Charles Lewis, and David A. Jodice. 1983. World Handbook of Political and
Social Indicators. New Haven, Conn.: Yale University Press.
Wade, Robert. 1994. "Public Bureaucracy and the Incentive Problem." Background paper
for World Development Report 1994: Infrastructure for Development. World Bank,
Washington, D.C. Processed.
Weede, Erich. 1983. "The Impact of Democracy on Economic Growth: Some Evidence
from Cross-National Analysis." Kyklos 36(1):21-39.
World Bank. 1988. Rural Development: World Bank Experience, 1965-86. Operations
Evaluation Study. Washington, D.C.
-. 1991. World Development Report 1991: The Challenge of Development. New
York: Oxford University Press.
1992. Governance and Development. Washington, D.C.
1993a. Annual Evaluation Results for 1991. Washington, D.C.
. 1993b. The East Asian Miracle. New York: Oxford University Press.
.1994. World Development Report 1994: Infrastructure for Development. New
York: Oxford University Press.
. 1995. "The World Bank and Participation." Operations Policy Department,
Washington, D.C. Processed.
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 243-62
The Relative Efficiency and Implementation Costs
of Alternative Methods for Pricing Irrigation Water
Yacov Tsur and Ariel Dinar
A useful means for achieving efficient allocation of irrigation water is to put the right
price tag on it. This article discusses some of the more pervasive pricing methods and
compares their efficiency performance, paying special attention to the impact of the
cost of implementing each method on its efficiency. The article uses an empirical ex-
ample to demonstrate numerically the relative efficiency of the different pricing meth-
ods and the important role of implementation costs. The volumetric, output, input,
tiered, and two-part tariff methods all can achieve efficiency, although the type of
efficiency varies from one method to another. These methods also differ in the amount
and type of information, and the administrative cost, needed in their implementation.
The example indicates that water pricing methods are most pronounced through their
effect on the cropping pattern-more so than through their effect on water demand for
a given crop. Implementation costs have a large effect on water tariffs and on welfare
and hence should have an important role in determining the desirable method to use in
any given water situation.
Water is an essential input in various economic sectors. Growing populations,
improved lifestyle, and dwindling water supplies (both in terms of quantity and
quality) exacerbate the competition for scarce water resources. It is thus of great
importance that the existing water resources be allocated efficiently. In an eco-
nomically efficient allocation, the marginal benefit of water use should be equal
across all users; otherwise, society benefits by reallocating water to the sector
with the highest marginal benefit. A useful means for achieving efficient water
allocation is to put the right price tag on it. Consequently, a variety of methods
for pricing water have been developed. They differ in their implementation, the
institutions they require, and the information on which they are based. In this
article we discuss some of the more pervasive pricing methods and compare
their efficiency performance, paying special attention to the cost associated with
implementing each method.
For reasons such as economies of scale in supply, presence of externalities,
small number of participants, uncertainty, and strong temporal interdependen-
Yacov Tsur is with the Department of Agricultural Economics and Management at The Hebrew
University of Jerusalem and the Department of Agricultural and Applied Economics at The University
of Minnesota. Ariel Dinar is with the Agriculture and Natural Resources Department at the World
Bank. The authors gratefully acknowledge comments from K. William Easter, Eithan Hochman, Herve
Plusquellec, James Roumasset, David Steeds, and three anonymous referees as well as data collection on
pricing methods in some countries by R. M. Chellappan.
C) 1997 The International Bank for Reconstruction and Development I THE WORLD BANK
243
244 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
cies, the management of irrigation water systems is often regulated by some sort
of public intervention. Consequently, a plethora of management methods has
evolved in the thousands of years since people first practiced irrigation (Young
and Haveman 1985).
A large volume of literature deals with irrigation water management in gen-
eral and water pricing in particular (see, for example, Rhodes and Sampath 1988;
O'Mara 1988; Cummings and Nercissiantz 1992; Le Moigne and others 1992;
Sampath 1992; Small and Carruthers 1991; Shah 1993; Plusquellec, Burt, and
Wolter 1994; and Tsur and Dinar 1995). Yet, to the best of our knowledge, the
efficiency performance of the different management practices has not yet been
compared comprehensively. We attempt to fill some of this gap by evaluating
how several of the commonly used methods for pricing irrigation water fare on
the efficiency scale. We consider a sustainable (or steady-state) water situation
and thus avoid intertemporal considerations such as those that emerge when
water is mined from a nonreplenishable aquifer (see Tsur, Park, and Issar
1989).
Section I describes the methods for pricing water and looks at information
and institutional aspects regarding their implementation, an issue that has re-
ceived very little attention (see Roumasset 1987 and Easter and Tsur 1995).
Section II provides details of pricing methods that are applied in several coun-
tries. Section III defines efficiency concepts in the context of water pricing and
evaluates the performance of the different pricing methods in this regard. Sec-
tion IV presents an empirical example to demonstrate numerically the relative
efficiency of the different pricing methods and the important role of implemen-
tation (transaction) costs. Section V concludes.
I. PRICING METHODS FOR IRRIGATION WATER
AND THEIR IMPLEMENTATION COSTS
The costs of supplying irrigation water consist of the variable costs of pro-
cessing and delivering water to end users and the fixed costs of capital operation
and maintenance (o&M). Variable costs depend on the amount of water deliv-
ered, while fixed costs do not. In most countries, fixed costs are heavily subsi-
dized (United Nations 1980). The method by which irrigation water is delivered
affects the variable cost as well as the irrigation technology applied and the
feasible pricing methods. Water may flow continuously or in certain time peri-
ods (in which case it may or may not be delivered on demand); the conveyance
system may consist of open channels or closed pipes. Often the irrigation water
in a region is delivered by more than one method, depending on tradition, physi-
cal conditions, and water facilities and institutions (United Nations 1980).
Bos and Wolters (1990) investigated farmers representing 12.2 million hect-
ares (1 hectare = 10 dunams _ 2.5 acres) of irrigated farms worldwide and found
that in more than 60 percent of the cases water authorities charge on a per unit
area basis. Less than 15 percent of the irrigation projects charge for water using
Tsur and Dinar 245
a combination of area and volumetric methods. About 25 percent of the projects
charge using the volumetric method.
The descriptions of methods for pricing irrigation water in this section draw
on Rhodes and Sampath (1988) and Sampath (1992). Implementation of these
pricing methods requires appropriate institutions, such as a central (national,
regional, district, village) water agency, and entails the costs of administration,
monitoring and collection of information, and enforcement. We briefly discuss
the costs of collecting information and the type of institutions needed to imple-
ment each pricing method.
Volumetric Pricing
Volumetric pricing methods charge for water using a direct measurement of
the volume of water consumed. Variations of the volumetric approach include
(a) indirect calculations based on measurement of minutes of known flow (as
from a reservoir) or minutes of uncertain flow (proportion of the flow of a river)
and (b) charges for a given minimal volume even if it is not consumed.
Volumetric pricing requires information on the volume of water used by each
user, that is, it requires facilities to meter water. Once water meters are installed,
implementation is fairly straightforward, involving routine maintenance and pe-
riodic meter readings. In the absence of a water market, a central water author-
ity or water user organization is required to set the price, monitor use, and col-
lect fees. The implementation cost associated with volumetric pricing is relatively
high.
Output Pricing and Input Pricing
Output pricing methods charge irrigators a water fee for each unit of output
they produce. Thus, output pricing requires information on the output level of
each water user. Its advantage is that it does away with the need to measure
individual water consumption, which in many regions (particularly in develop-
ing countries) is an expensive, or even impossible, task. If the crop is used for
export (for example, cotton), output must be measured, and the cost associated
with imposing water fees is small. Otherwise, the measurement of output can be
as formidable as that of water, implying that output pricing is rather a poor
means for pricing water (indeed, examples of output pricing are rare).
Input pricing methods charge for water use by taxing inputs. Irrigators pay a
water fee for each unit of a certain input-for example, fertilizer-used.
Area Pricing
Area pricing charges for water used per irrigated area, depending on the kind and
extent of crop irrigated, the irrigation method, the season of the year, and other
factors. In many countries, the water rates are higher when the water is delivered
from storage (investment) than when it is diverted directly from streams. The rates
for pumped water are usually higher than those for water delivered by gravity. In
some cases, farmers also must pay the per acre charges for nonirrigated acres.
246 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Area pricing is easy to implement and administer and does not require water
conveyance facilities to be metered. This method needs only land-by-crop data
(if the per hectare water fees vary across crops) or only farm size data (if a
unified fee is used). Simplicity and low agency costs explain the popularity of
this method.
Tiered Pricing and Two-Part Tariff Pricing
With tiered pricing-a multirate volumetric method-water rates vary as the
amount of water consumed exceeds certain threshold values. Two-part tariff
pricing methods involve charging irrigators a constant marginal price per unit of
water purchased (volumetric marginal cost pricing) and a fixed annual (or ad-
mission) charge for the right to purchase water. All farmers pay the same admis-
sion charge. This pricing method has been advocated, and practiced, in situa-
tions where a public utility produces with marginal cost below average cost and
must cover total costs (variable and fixed).
Both tiered pricing and two-part tariff pricing elabQrate on the volumetric
method. Once water meters are installed, the extensions to multiple rates (tiered
pricing) are straightforward; the two-part tariff requires, in addition to the volu-
metric rate, a fixed admission fee per farmer.
Betterment Levy Pricing
Betterment levy pricing methods charge water fees per unit area, based on the
increase in land value accruing from the provision of irrigation.
Water Markets
Water markets exist in different forms throughout the world, in industrial
and developing countries alike. Water markets may be formal or informal, orga-
nized or spontaneous. Their participants may trade water rights (for example,
the right to purchase some quantities of water at a particular price during spe-
cific periods of time), or they may trade water at the spot or for future delivery.
In some countries, markets for water or for water rights have been formed and
determine water prices, usually measured on the basis of volume or flow of
water. They range from sanctioned markets for water rights, such as in Chile
(Hearne and Easter 1995), to spontaneous spot markets, such as in Brazil (Kemper
1996). Compared with an administratively imposed price, well-defined tradable
rights should formalize and secure the existing water rights held by users, econo-
mize the transaction costs, and increase the efficiency of water use by inducing
users to internalize the full opportunity cost of water, as determined by the market.
In a stylized water market, in any year, each irrigator is given a water endow-
ment (or entitlement) and is free to sell or buy shares of entitlements from other
farmers at the going rate. Water entitlements may be based on historical or legal
rights, or they may be set by an elected or assigned committee (or water agency).
Endowments may vary from year to year according to the availability of water.
This method requires no water meters for individual users below the diversion
Tsur and Dinar 247
point and is sure to internalize any private information farmers have. Water
markets are likely to provide incentives for water to flow from less productive to
more productive users.
To operate properly, water markets require a well-defined structure of water
rights, a clear and comprehensive set of rules for trading these rights, and a
judicial body for overseeing the trading activities and resolving disputes. In ad-
dition, water markets require a well-developed conveyance system for trans-
porting water to all participants.
II. EXAMPLES OF PRICING METHODS IN SEVERAL COUNTRIES
This section provides examples of the various pricing methods as they are
applied in several countries.
California (United States)
Multirate volumetric pricing of publicly supplied water is common in the
state of California in the United States. Depending on the irrigation district,
prices range between $2 per acre-foot to more than $200 per acre-foot (1 acre-
foot = 1,256 cubic meters). On average, farmers paid about $5 per acre-foot for
water from the federal Central Valley Project in 1988, compared with $48 per
acre-foot average capital depreciation cost and $325 per acre-foot average mar-
ginal cost of delivery (Rao 1988). Cummings and Nercissiantz (1992) estimate
the average water price at $19.32 per acre-foot, which they claim covers a mere
39 percent of the estimated scarcity value (the in situ value of groundwater). The
recent prolonged drought in California from 1986 to 1992 has led to the devel-
opment of innovative water banks and water markets through which water prices
are determined (see Easter and Tsur 1995).
India
Irrigation pricing methods vary throughout India, depending on geographic
location, the command area of the project (region, state, country), the system of
irrigation (storage, diversion, pumped), crops grown, seasons, the nature of agree-
ment (long lease, short lease), and the procedure used to extract penalties for
unauthorized use (Gole, Amble, and Chopra 1977). Some examples of pricing
methods are (United Nations 1980):
* Area charges that vary by crop or across seasons
* Area charges that vary according to the method of irrigation (flood, ridges,
or furrows)
* Area charges that are agreed on for one or more years (to be paid whether
or not water is used)
* Volumetric rate per estimated volume of water consumed, applied generally
in areas with pumped irrigation and tube- wells (estimates are based on
crop water requirements)
248 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
* Penalty rates per acre charged for using water in an unauthorized manner
or for wasting water
* Percolation rates charged for each cultivated acre within 200 yards of a
canal that receives percolation or leakage water from the canal
* A flat area charge covering all areas serviced by the project, whether or not
they are actually irrigated during a given season or year
* A betterment levy, applied per unit area served by the project. For example,
water charges for farmers in Tamil Nadu in 1993 were Rs200-Rs210 per
hectare (in 1993, 31.5 rupees = 1 U.S. dollar). This rate is considered to be
among the highest water charges in India (World Bank 1995).
Jordan
Most of the agricultural activity in Jordan is concentrated in the Jordan
Valley, while the majority of the population lives in the urban centers in
upland areas. So, in addition to competition over scarce water resources, a
conveyance cost is associated with transferring water to urban uses. Crop
water requirements vary substantially between regions because of soil and
climatic conditions. Upland irrigation is based mainly on the extraction of
groundwater. Private wells are not monitored. The cost of pumped water in
1993 was estimated at JDO.05 per cubic meter (in 1993 1 Jordanian dinar =
1.5 U.S. dollars, and in 1986 JD1 = $2.85). In the Jordan Valley, water is
provided through pipes to more than three-quarters of the irrigated land.
Water authorities use volumetric pricing, but water is greatly underpriced.
For example, in the East Ghor canal (the Jordan Valley Irrigation Project),
the water authority charged farmers JD0.003 per cubic meter for the first 1.5
meters of irrigation depth and JDO.006 per cubic meter for any additional
amount. o&M costs alone were estimated at JDO.02 to JDO.03 per cubic
meter (Arar 1987). In 1993 the water authority priced all irrigation water in
the Jordan Valley at JDO.006 per cubic meter irrespective of the volume used
(Hayward and Kumar 1994).
Although most of the water supply to agriculture is piped and easy to moni-
tor, the existing pricing method does not take advantage of it. The water author-
ity does not monitor or price irrigation water in the upland area. In the Jordan
Valley, the volume supplied to individual users is measured, but the price does
not influence efficient use of water. The water authority makes allocations based
on the crop grown and availability of water. The policy of favoring equity over
efficiency may encourage farmers to grow low-profit crops, most of which can-
not cover the real cost of water. Profitable crops such as citrus (irrigated with
10,000 cubic meters of water per hectare), bananas (20,000 cubic meters per
hectare), and grapes (8,000-11,000 cubic meters per hectare) are water-
intensive. Because water is scarce in Jordan, price signals may not be sufficient
to allocate the water; irrigators need additional guidance from the government
in the form of preferred cropping patterns. This may create an additional policy
dilemma for food security.
Tsur and Dinar 249
In contrast to water for irrigation, water for municipal and industrial uses,
which is also metered, is priced on the basis of block tariffs (tiered pricing) im-
posed every three months. The water component in the pricing method (in addi-
tion to a sewerage charge) varies by location, which reflects the marginal cost of
water supply. Water charges in the municipal and industrial sectors vary be-
tween JDO.06 and JDO.6 per cubic meter (in 1993), depending on the quantity
of water used and the location (Hayward and Kumar 1994).
Morocco
In Morocco, costs vary by location, based on specific conditions in each re-
gion. For example, in the Haouz irrigation district, water is supplied both for
irrigated agriculture (65,000 hectares) and for urban use in Marrakech (650,000
inhabitants). In 1993-94 the Haouz Office, which is the district's management
arm, supplied about 300 million cubic meters of water, of which Marrakech
received 35 million (50-60 percent of its annual consumption, with the rest
from groundwater sources). The Haouz Office provides water to Marrakech
free of service charge. The town treats and distributes the water to households
through a metered system that allows a tiered volumetric pricing method (aimed
at covering treatment and distribution costs only) and to industries for a flat fee.
For households in Marrakech, the bimonthly charge per cubic meter was
DHO.73 for the first 24 cubic meters, DH2.17 for 24-60 cubic meters, and
DH3.25 for any quantity beyond 60 cubic meters (8.4 Moroccan dirhams = 1
U.S. dollar in 1993). The industrial rate was DH2.01 per cubic meter (Morocco,
Direction de la Statistique, 1993). Irrigators in the Haouz irrigation district pay
an average price of DH0.16 per cubic meter of water. There are discounts re-
lated to geographical areas and to certain infrastructure setups. For example,
farmers can pay a discounted price by participating in maintenance of the irriga-
tion system. Water volumes are measured by several means. In perimeters with
surface irrigation, gates are used to measure flow, which is converted to volume.
In sprinkler-irrigated perimeters, volume is measured directly using meters in-
stalled at the farm-level outlet.
The operational cost of monitoring water use and enforcing payments in the
Haouz irrigation district provides a good example of the costs of pricing sys-
tems. These costs, not including investment in measuring equipment, can be
calculated from data available in Morocco; Ministry of Agriculture and Agricul-
tural Development (1994); and from the Haouz Office (El Hadj El Hallani, per-
sonal communication, 1995). In 1994 the collection rate for fees to cover these
costs was 76 percent.
The Haouz Office employs 175 irrigation water monitors for regulating wa-
ter distribution in the canals, 56 staff for invoicing, and 12 staff for collecting
payments. The average annual salary at the irrigation sector of the Haouz Office
in 1994 was DH7,700 (Agro-Concept 1995). The total annual cost of monitor-
ing, regulation, and enforcement is therefore DH1,871,100, or DHO.004 per
cubic meter delivered. This very conservative estimate does not take into ac-
250 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
count additional variable costs. For example, temporary staff are also hired in
the peak season to monitor water use. A rough estimate suggests that the Haouz
Office hires 5-10 temporary personnel per 400 hectares of irrigated land, which
adds about 875-1,750 additional temporary staff to the payroll.
Spain
There are important differences in water tariffs paid by farmers in Spain
(Maestu 1995). Two types of tariffs were designed to compensate the govern-
ment for its investment in, and operation and maintenance of, publicly financed
water projects.
First, the water basin authorities charge the beneficiaries of the waterworks
an annual regulation tariff and a water use tariff. The regulation tariff is calcu-
lated as 4 percent of the initial investment costs adjusted annually for inflation.
The amount of investment to which the 4 percent is applied is a political deci-
sion that can vary over time and by basin. The water basin authority sets the
water use tariff to cover average o&M costs, based on estimated future budget
costs. The authority charges farmers for o&M costs either on the basis of irri-
gated area (using information on crops grown and standard per-area water coef-
ficients) or on the basis of volume in new irrigation projects that are equipped
with metering devices. The final charge to the farmer also may include charges
to the local user associations (irrigation cooperatives).
Second, the water basin authorities charge a tariff for occupation of the pub-
lic domain for water. This charge is equivalent to 4 percent of the value of the
land used for the waterworks (dams, reservoirs,.canals, roads). In addition, the
authorities charge a pollution tax that is calculated individually. Farmers are
exempt from this tax at present.
In Almeria, the charge for water, including all relevant tariffs mentioned ear-
lier, is Ptasl6 per cubic meter (125 pesetas = 1 U.S. dollar in 1995), including
the cost of energy, which users pay to the irrigation cooperative. In Tajo-Segura,
the tariff for irrigation water is Ptal per cubic meter.
Turkey
Pricing and cost recovery policies vary among water use sectors in Turkey.
Water authorities charge domestic and industrial users by volume. They charge
farmers an annual area-based fee that varies by crop and region. In projects
operated by the State Hydraulic Works (DSI), that fee has two components, an
o&M component and a capital cost recovery surcharge component. The o&M
charge is supposed to recover costs born by DSI in the previous year, with no
adjustment for inflation. Because the government has the right to adjust the fee,
it is usually set at a level lower than that proposed by DSI. In 1993 the o&M
component for wheat in the Southern Anatolia Project was LT163,000 per hect-
are, compared with LT448,300 per hectare for gravity and pump irrigation. The
o&M fee for cotton was LT462,000, compared with LT1,086,800 per hectare
for gravity and pump irrigation (13,585 Turkish liras = 1 U.S. dollar in 1993).
Tsur and Dinar 251
The capital cost recovery surcharge is based on land area. DSI can charge users
for this component only 10 years after completion of the project, and for a
period not to exceed 50 years, with no inflation considerations. The capital cost
recovery surcharge varies by region, ranging between LT4,100 and LT8,500 per
hectare. The reported collection rate for the o&M fees and the capital cost re-
covery surcharge in 1992 was 33 percent (Kasnakoglu and Cakmak 1995).
Chile
Chile is one of the few countries where tradable water rights have been
officially established. Water rights are allocated to users in the form of shares
of the river flow (for example, there are 25,000 shares in the Rio Alqui, each
supposed to deliver 1 liter per second in a good year; see Hearne and Easter
1995). Economic analysis of water markets in four of Chile's river valleys-
the Maipo, Elqui, Limari, and Azapa valleys (Hearne and Easter 1995)-
demonstrates that the market transfer of water use rights produces substan-
tial economic gains from trade in both the Elqui and Limari valleys. These
economic gains produce rents for both buyers and sellers. In the Elqui valley,
net gains from trade per share were within the range of transaction values
observed in the 1990s in Chile of Ch$400,000. In the Limari valley, gains
from trade per share were three times the recent price of Ch$1,200,000 for a
share of water from the Cogoti reservoir (403 Chilean pesos = 1 U.S. dollar
in June 1993). When trading was active, especially in the Limari valley, trans-
action costs did not present an appreciable barrier to trading. Nonetheless,
in the large canal systems with fixed flow dividers in the Elqui and Maipo
valleys, there have been few transactions.
1II. EFFICIENCY OF WATER ALLOCATION
An efficient allocation of water-or any other scarce resource-is one that
maximizes the total net benefit that can be generated by the available quan-
tity of the resource given the available state of technology. If the net benefit
to be maximized involves variable (short-run) costs and abstracts from capi-
tal and other costs that are fixed in the short run, the allocation is quasi (or
short-run) efficient. When the fixed inputs are chosen optimally, the short-
and long-run outcomes are the same. In the absence of distortions (taxes) or
costs associated with implementing an allocation scheme (for example, col-
lecting water fees, monitoring, enforcing quotas), an efficient allocation is
first-best or Pareto efficient. In the presence of distortionary actions or imple-
mentation costs, an allocation that maximizes the total benefit net of all costs,
including implementation and distortionary costs, is second-best efficient
(Baumol and Bradford 1970). Such is the situation, for example, when taxes
distort decisions regarding input and output and collecting these taxes is
costly. In this section we discuss the performance of the various pricing meth-
ods vis-a-vis efficiency criteria.
252 THE WORLD BANK ECONOMIC REVIEW, VOL, I ], NO. 2
Volumetric Pricing
The optimal volumetric pricing rule requires that the water price be set equal
to the marginal cost of water supply. In the absence of implementation costs, the
(variable) cost of supply consists solely of the cost of delivery. In this case, the
marginal cost pricing rule, in which the water price is set at the level of marginal
delivery cost, is optimal. Water pricing, however, entails costly activities, such
as maintaining and reading water meters, administering the collection of water
fees, and resolving disputes with farmers. The costs incurred by these activities,
referred to as implementation costs, are augmented to the cost of delivery and
become an integral part of the cost of supply. The marginal cost of water supply
consists of the marginal delivery cost and the marginal implementation cost.
In the presence of implementation costs, the optimal volumetric pricing rule
is of the form:
water price = marginal delivery cost + marginal implementation cost
where all values are measured in dollars per cubic meter (detailed derivation can
be found in Tsur and Dinar 1996). Because the marginal cost pricing rule ig-
nores implementation, the presence of implementation costs requires departure
from marginal cost pricing. It also implies that volumetric pricing cannot achieve
a first-best (efficient) outcome. Volumetric pricing thus may not be superior to
other pricing methods, such as those based on output or area fees. The pricing
method that achieves the highest social benefit then becomes a practical matter
that depends crucially on the implementation costs associated with each method.
This point is illustrated in section IV.
Output Pricing
The output pricing method prices water by imposing a tax on output. The
optimal tax depends on the nature of the production technology and the imple-
mentation costs (see Tsur and Dinar 1996 for details). The allocation obtained
under output pricing is second best when implementation costs are nil. This is
because the output fee and the zero price of water will distort decisions regard-
ing input and output away from the first-best outcome achieved under the mar-
ginal pricing rule. The presence of implementation costs constitutes another source
of deviation from a first-best allocation.
Without implementation costs, output pricing is inferior to volumetric pric-
ing (that is, it achieves a lower social benefit): it achieves only a second-best
allocation, while volumetric pricing achieves a first-best allocation. With imple-
mentation costs, however, both methods are second best, and the method that
generates a higher benefit depends on the costs associated with implementing
each method.
Area Pricing
With area pricing, farmers pay a fixed fee per hectare or acre for the right to
receive irrigation water. The per hectare fee is a fixed cost that, once paid, can
Tsur and Dinar 253
no longer affect decisions regarding input and output. It can, however, affect the
choice of crop (if per hectare water fees vary across crops) or induce some farm-
ers to switch to unirrigated farming, thereby affecting the aggregate demand for
water. For farmers who pay the water fee, the demand for irrigation water is
larger than it would be under marginal cost pricing, and the resulting water
allocation is inefficient. However, the implementation costs associated with per
area pricing are smaller than those associated with volumetric or output pricing.
Therefore, area pricing may well generate a higher social benefit.
Tiered Pricing and Two-Part Tariff Pricing
Tiered pricing is common when water demand or supply have periodic (sea-
sonal, daily) variations. During periods of excess supply, setting the water price
equal to the marginal cost of supply achieves (short-run) efficiency. During peri-
ods of excess demand, the water price accounts also for water scarcity and is
increased by the scarcity rent. An alternative tiered pricing method increases the
water price each time demand exceeds one of a few prespecified levels.
The two-part tariff method consists of volumetric pricing plus a fixed admis-
sion charge per farmer. The admission charge can serve to balance the budget of
the water supply agency, thus extending short-run volumetric pricing to account
for long-run fixed costs. The implementation of the annual admission charge as
a Pigouvian poll tax avoids the distortionary effects of other tax schemes. Some
analysts therefore consider the two-part tariff method as capable of achieving
long-run efficiency (see Feldstein 1972a, 1972b; and Laffont and Tirole 1994,
pp. 19-34).
Water Markets
The basic premise of modern economics is that markets, under certain condi-
tions, achieve first-best efficiency when no implementation costs are present.
These "certain conditions" include a competitive environment (no single agent
can affect outcomes), fully informed agents operating under certainty, no exter-
nalities, and no increasing returns to scale in production. In the case of water,
these conditions are frequently violated. Water is expensive to transport; hence,
water markets tend to be localized, consisting of a limited number of partici-
pants, some of whom may be able to influence outcomes. Water supply is often
uncertain. Water resources (for example, rivers, aquifers) may be shared by many
users who inflict externalities on one another (for example, groundwater pump-
ing by one farmer reduces the water level and increases pumping costs to other
farmers). Finally, water supply systems, like other public utilities, may exhibit
increasing returns to scale. For these reasons, water markets are unlikely to at-
tain a first-best allocation in actual practice.
Yet, even when distorted, the suboptimal outcomes of water markets may
outperform the other pricing methods when implementation costs are taken into
consideration. Introducing water markets amounts to privatizing the water sec-
tor, an immediate result of which is that the cost associated with the collection
254 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
of information is internalized. This eliminates a major source of implementation
costs. In addition, water markets induce the transfer of water from less produc-
tive to more productive farmers and eliminate corruption incentives to which
centralized allocation mechanisms are more sensitive.
Which Method to Implement?
The preferred pricing method is the one that yields the highest social ben-
efit. In the absence of implementation costs, the volumetric method (or one
of its related methods-tiered or two-part tariff pricing) is optimal. With
implementation costs, other methods may perform better. Implementation
costs vary widely from region to region because of variations in climate,
demography, social structure, water rights, water facilities, history, and gen-
eral economic conditions. Therefore, the net benefit associated with each
method also varies from region to region. In the following section we per-
form some calculations to illustrate the effects of implementation costs and
inefficiencies on water prices and welfare.
IV. AN EMPIRICAL ILLUSTRATION
We present here a numerical example to evaluate the performance of some of
the pricing methods discussed above regarding efficiency of water allocation.
We consider the production of two crops, cotton and wheat, by means of two
inputs, water (q) and nitrogen (x). Nitrogen is purchased in the marketplace;
water is provided by a water agency. Other inputs are assumed to be fixed. We
use a quadratic approximation for the per hectare water-nitrogen production
functions, gj(q,x) = a; + fq + yjx + 5jq2 + OiX2 + rjqx, where j denotes cotton or
wheat. Table 1 lists the parameters estimated by Hexem and Heady (1978).
Output and nitrogen prices, using the state of Haryana in India as an example,
were taken from India, Directorate of Economics and Statistics (1993) and are
presented in table 2. Additional production costs that are not related to water or
nitrogen are $78.50 per acre ($196.30 per hectare) for cotton and $45.10 per
Table 1. Parameter Estimates of the Quadratic Production Functions
for Cotton and Wheat
Coefficient Cotton Wheat
Intercept, a 233.71 -10,414
Water (acre-inch), , 23.65 852.01
Nitrogen (pounds per acre), y 0.438 11.6
Water*water, o -0.182 -12.9
Nitrogen *nitrogen, 4 -0.0033 -0.032
Water*nitrogen, T3 0.0209 0.0925
Range of water input (acre-inches) 8-40 0-40
Experimental range of nitrogen input (pounds per acre) 0-120 0-200
Note: Yield is measured in pounds per acre. Metric conversion: 1 acre-inch = 102.8 cubic meters; 1
pound = 2.24 kilograms; 1 pound per acre = 1.102 kilograms per hectare.
Source: Hexem and Heady (1978).
Tsur and Dinar 255
Table 2. Output and Input Prices for Cotton and Wheat Production
(U.S. dollars)
Cotton Wheat
Indicator Per pound Per ton Per pound Per ton
Output 0.8 1,750 0.15 300
Nitrogen 0.089 0.199 0.089 0.199
Note: Prices are from the state of Haryana in India and are in constant 1993 U.S. dollars (Rs31.5 =
Si).
Source: India, Directorate of Economics and Statistics (1993).
acre ($112.70 per hectare) for wheat (India, Directorate of Economics and Sta-
tistics 1993).
Modifying Howitt and Vaux (1995), the marginal cost (MC) of water delivery
(not including implementation costs) is represented by C'(q) _ MC(q) = 11.5 +
0.000671q, where C' is dollars per acre-inch and q is measured in acre-inches.
The cost of supplying q acre-inches of water is thus 11.5q + 0.000671q2/ 2.
The farmer chooses the cropping pattern and allocates inputs, taking all
prices-including the price of water-parametrically. The water agency chooses
the pricing method and the water rates. We consider the efficiency effects of
volumetric, output, input, and area pricing, allowing prices to vary between
crops. The results are presented in table 3.
We assume that implementation costs are incurred as a fraction of water pro-
ceeds and that these costs vary among the pricing methods. The last column of
table 3 provides the implementation costs. For example, case 1 is free of imple-
mentation costs, while the 0.05 entry of case 2 implies that from each $1.00
raised as water proceeds, $0.05 are used up by water pricing activities. (Tsur
and Dinar 1996 provide a detailed analysis of such implementation costs.)
A few interesting observations emerge from the results in table 3. First, wel-
fare is affected most dramatically by the effect of water pricing on choice of
crop. Without water pricing (case 0), the farmer chooses to grow cotton, be-
cause it gives the highest profit. Cotton, however, consumes a lot of water (82.89
acre-inches per acre) and costs the society dearly to deliver the water. Thus,
despite the farmer's high profit of $1,011.84 per acre, the social benefit net of
the cost of water is only $56.35 per acre. Introducing a simple per acre pricing
scheme under which the water fee is $231 or more for each acre of cotton and
nothing for an acre of wheat (case 9) induces the farmer to switch to wheat
production. The result is that water consumption decreases to 33.82 acre-inches
per acre, the farmer's profit decreases to $781.05 per acre, and the social benefit
increases to $391.79 per acre. In neither case are water fees actually collected (in
the first case water is supplied free of charge, and in the second case farmers
avoid paying for water by choosing to grow wheat), but the mere effect on choice
of crop increases the social benefit almost sevenfold.
The second observation concerns the sensitivity of water prices to implemen-
tation costs. In cases 1-4 volumetric pricing is employed with escalating imple-
Table 3. Efficiency Effects of Alternative Methods for Pricing Irrigation Water
Water Water Nitrogen Farmer's Cost of Social Implementation
input proceeds input Output profit water benefit costs
Minimum water feel (acre-inches (U.S. dollars (pounds (pounds (U.S. dollars (U.S. dollars (U.S. dollars (percentage of
Case and pricing method Cotton Wheat Crop per acre) per acre) per acre) per acre) per acre) per acre-inch) per acre) water proceeds)
0. None 0.00 0.0000 Cotton 82.886 0.00 311.97 1,299.510 1,011.84 955.49 56.35 0.0
1. Volumetric 9.70 11.5210 Wheat 30.823 355.10 216.52 5,220.540 408.71 354.78 409.03 0.0
2. Volumetric 5.90 5.5690 Wheat 32.369 180.27 218.76 5,309.928 596.75 372.59 395.41 5.0
3. Volumetric 4.10 2.0690 Wheat 33.278 68.83 220.07 5,333.858 711.65 383.07 392.26 7.5
4. Volumetric 3.10 0.0000 Wheat 33.815 0.00 220.85 5,338.023 781.05 389.26 391.79 10.0
5. Volumetric with
balanced budget 9.70 11.5103 Wheat 30.825 354.81 216.53 5,220.750 409.03 354.81 391.29 5.0
6. Output 0.18 0.0000 Wheat 33.815 0.00 220.85 5,338.023 781.05 389.26 391.79 0.0
7. Input 2.50 0.0017 Wheat 33.814 9.14 220.67 5,337.916 780.67 389.25 391.79 0.0
8. Input 2.50 0.0000 Wheat 33.815 0.00 220.85 5,338.023 781.05 389.26 391.79 10.0
9. Area 231.00 0.0000 Wheat 33.815 0.00 220.85 5,338.023 781.05 389.26 391.79 0.0
10. Area with
balanced budget 621.00 389.2610 Wheat 33.815 389.26 220.85 5,338.023 391.79 389.26 391.79 0.0
a. For cases 1-5, the water fee is measured in dollars per acre-inch of water used, in cases 6-8, in dollars per pound of output or input, and in cases 9-10, in dollars per
acre of land.
Source: Authors' calculations.
Tsur and Dinar 257
mentation costs: case 1 entails no implementation costs, case 2 entails a cost of
5 percent (that is, $0.05 of each $1.00 of water proceeds are used to cover
expenses associated with implementation activities), case 3 entails a cost of 7.5
percent, and case 4 entails a cost of 10 percent. In all cases the water price for
cotton production is kept high enough to induce farmers to grow wheat. The
price of water for wheat production drops from $11.52 per acre-inch in case 1
to $5.57 per acre-inch in case 2. With 7.5 percent implementation costs, the
price of water is further reduced to $2.07 per acre-inch in case 3. When imple-
mentation costs are 10 percent or more (case 4), the pricing activities are costly
enough to render water pricing undesirable, except for the nominal charge on
water going to cotton production, whose only role is to switch production away
from this water-intensive crop.
The third observation that emerges from table 3 is that an inefficient but
simple method such as per acre pricing may outperform an efficient but compli-
cated method, taking implementation costs into account. Cases 4 and 9 yield the
same outcome using different methods: in case 4, volumetric pricing is employed,
and in case 9 area pricing is used. In both cases, only water in cotton production
is priced to discourage production of this crop. If, however, volumetric pricing
entails some fixed costs, for instance, because of the need to install water meters,
which have not yet been incurred, then it is better to use area pricing and avoid
the fixed costs and the ensuing implementation costs associated with volumetric
pricing.
From cases 1-4, higher implementation costs lead to lower water prices and thus
lower water proceeds that are insufficient to cover the costs of delivery. Often the
water agency is required to have a balanced budget. We look at the effect on welfare
of a balanced budget constraint imposed on volumetric pricing in case 5, which
imposes the constraint on case 2 (with 5 percent implementation costs). The result is
that the farmer's profit is reduced quite significantly from $596.75 to $409.03 per
acre, while social benefit decreases slightly from $395.41 to $391.29 per acre. Thus,
mandating a balanced budget on the water agency inflicts a heavy toll on farmers.
Without this constraint, the water agency's deficits would have to be financed by
taxpayers' money. Given its small effect on total welfare, the choice of whether to
impose the balanced budget constraint is mostly political, involving considerations
of income distribution and pressure groups.
Case 6 considers output pricing. It appears that because of the distortionary
effects of this method, water is priced to affect only the choice of crop, not the
choice of water input, given that the right crop (wheat) was chosen.
Cases 7 and 8 look at input pricing with 0 and 10 percent implementation
costs, respectively. As in the other cases, the main role of water pricing is to
affect the choice of crop by imposing a tax of $2.5 per pound or more on nitro-
gen applied in cotton production. Given this tax, wheat is the chosen crop, and
the impact on water input of taxing nitrogen is rather limited when implementa-
tion costs are nil and vanishes completely when implementing the input tax takes
up $0.10 or more from each $1.00 of taxes raised.
258 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Cases 9 and 10 consider area pricing without and with a balanced-budget
constraint, respectively. Imposing a fee of $231 or more for each acre of cotton
and $0 for each acre of wheat is sufficient to induce the profit-seeking farmer to
grow wheat (case 9). When the farmer is also required to cover the cost of water
delivery, this cost is imposed as a per acre fee on wheat, and a higher per acre fee
on cotton is needed to ensure that cotton will not be chosen (case 10). From
society's point of view, the balanced budget constraint makes no difference (the
social benefit is the same in both cases): with it the burden of paying for water
delivery falls on the user (the farmer), and without it the burden falls on the
taxpayers.
V. CONCLUSIONS
In this article we investigated the efficiency performance of several methods
of pricing irrigation water, paying special attention to the costs associated with
implementing them. Efficient use of irrigation water requires that the pricing
method affect demand. The volumetric, output, input, tiered, and two-part tar-
iff methods all satisfy this condition and can achieve efficiency, although the
type of efficiency (short or long run, first or second best) varies from one method
to the other. These methods also differ in the amount and type of information
and in the administrative cost needed to implement them. Pricing methods that
do not influence water input directly, such as area pricing, lead to inefficient
allocation. Such methods, however, are in general easier to implement and ad-
minister, and they require a modest amount of information.
We found that water pricing methods are most pronounced through their
effects on the cropping pattern-more so than through their effect on water
demand for a given crop. Implementation costs are found to have a large effect
on water prices and on welfare and hence should have an important role in
determining the desirable method to use in any given water situation. In the
conditions of the numerical example, for instance, moderate implementation
costs of 10 percent (that is, $0.10 of each $1.00 raised as water proceeds are
used to finance pricing-related activities) render the (second-best) efficient volu-
metric method equal in performance to an inefficient but simple per area pricing
method that entails no implementation costs. If the volumetric method also in-
volves fixed costs, such as the cost of installing water meters, then area pricing is
superior to volumetric pricing. With implementation costs of less than 10 per-
cent and a previously installed metered water conveyance facility, the volumet-
ric method is superior.
A volumetric method that uses the marginal cost pricing rule achieves first-
best (the maximum attainable total benefit) efficiency in the absence of imple-
mentation costs. But this method requires information on the water application
of each user (that is, metered water) and in general entails implementation costs.
In such cases the optimal departure from marginal cost pricing achieves second-
best efficiency.
Tsur and Dinar 259
The output (or input) pricing method cannot achieve first-best efficiency, be-
cause it distorts input-output decisions. Without implementation costs, the out-
come of an optimal output pricing can be considered as second best. The pres-
ence of implementation costs introduces another source of deviation from
first-best efficiency (in addition to the distortionary effect mentioned above),
hence the outcome may be considered as third best. However, whether output
pricing is inferior to volumetric pricing depends on the magnitude of implemen-
tation costs, because implementing these methods entails different activities and
requires different information and data. In the numerical example, input pricing
is better than output pricing, and both input and output pricing are inferior to
volumetric pricing in the absence of implementation costs. The introduction of
10 percent implementation costs makes all three methods equivalent, because
water prices are used to affect only the choice of crop, not the demand for water
once the right crop has been chosen.
Area pricing can affect water input through its effect on choice of crop but
cannot otherwise affect demand for water. Area pricing is, however, easy to
implement and administer and requires minimal information. Our numerical
example shows that with moderate 10 percent implementation costs on volu-
metric pricing, area pricing is as good as volumetric pricing. And when volumet-
ric pricing involves fixed costs (for example, the cost of installing water meters),
area pricing outperforms volumetric pricing.
Despite numerous imperfections (caused mainly by spatial and intertemporal
externalities, small number of participants, uncertainty, and economies of scale
in supply), the market mechanism is still an excellent means for securing the
transfer of water from low-value to higher-value activities. It puts the burden of
information collection on water users and avoids problems of asymmetric infor-
mation that are commonly found in principal-agent situations. The cost of in-
formation collection-a major component of implementation costs-is thus dras-
tically reduced. Water markets require well-developed water conveyance facilities,
a system of water rights and water endowment (or entitlement) for each user
contingent on the availability of water, a complete set of rules for trading in
water endowments and in water rights, and the appropriate institution to over-
see trading activities and resolve conflicts when they arise. Once the water insti-
tutions and conveyance facilities are in place, the implementation costs associ-
ated with water markets are small (or negligible), which is why this mechanism
is an attractive means for achieving efficiency.
Efficiency, an important objective, may not always warrant the social cost
associated with implementing pricing methods that are considered efficient. Imple-
mentation costs should always be considered, and the pricing method to be used
depends crucially on these costs. Other forces that work against efficient pricing
are either political or considerations of equity and fairness. If farmers are well
organized, they can effectively collude to exert political pressure on their own
behalf. In addition, politicians may find that it is in their interest to support
farmers, because it increases their chances for reelection (see de Gorter and Tsur
260 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
1991), and one manifestation of this support may be subsidized water. Equity
considerations in pricing irrigation water, which are discussed in Tsur and Dinar
(1995), imply that the pricing of water should not make farmers worse off.
Raising water prices, for instance, entails lowering farm income as well as land
values (Rosegrant and Binswanger 1994). This brings in the issue of whether
water is an appropriate means for achieving social ends such as income distribu-
tion. Results of the preliminary analysis of Tsur and Dinar (1995) suggest a
rather limited scope for water policies in achieving income distribution goals,
but further work is needed before definite conclusions can be reached.
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
"Agro-Concept, Financial Analysis of the Haouz Office." 1995. Report to the Ministry
of Agriculture, Rabat. Processed.
Arar, Ablullah. 1987. "Short Report on Cost of Irrigation Water and Irrigation Water
Charges in Some Arab Countries." In Technical Papers from the Expert Consultation
on Irrigation Water Charges, vol. 2. Rome: Food and Agriculture Organization of
the United Nations and U.S. Agency for International Development.
Baumol, W. J., and D. F. Bradford. 1970. "Optimal Departure from Marginal Cost
Pricing." American Economic Review 60(June):265-83.
Bos, M. G., and W. Wolters. 1990. "Water Charges and Irrigation Efficiencies." Irriga-
tion and Drainage Systems 4:267-78.
Cummings, R. G., and Vahram Nercissiantz. 1992. "The Use of Water Pricing as a
Means for Enhancing Water Use Efficiency in Irrigation: Case Studies in Mexico and
the United States." Natural Resources Journal 32(fall):731-55.
de Gorter, Harry, and Yacov Tsur. 1991. "Explaining Price Policy Bias in Agriculture:
The Calculus of Support-Maximizing Politicians." American Journal of Agricultural
Economics 73(November):1244-53.
Easter, K. W., and Yacov Tsur. 1995. "The Design of Institutional Arrangements for
Water Allocation." In Ariel Dinar and Edna Loehman, eds., Water Quantity/Quality
Management and Conflict Resolution, pp. 107-17. Westport, Conn.: Praeger.
Feldstein, M. S. 1972a. "Distributional Equity and the Optimal Structure of Public Prices."
American Economic Review 62:32-36.
1972b. "Equity and Efficiency in Public Sector Pricing: The Optimal Two-Part
Tariff." Quarterly Journal of Economics 86:175-87.
Gole, C. V., V. N. Amble, and M. M. L. Chopra. 1977. "Rationalization of Irrigation
Rates in a Developing Country." ICID [International Committee on Irrigation and
Drainage] Bulletin 26(1):31-49.
Hayward, J. A., and K. S. Kumar. 1994. "Jordan Agricultural Sector Adjustment Loan:
Water Component." Middle East Department, World Bank, Washington, D.C. Pro-
cessed.
Hearne, Robert R., and K. William Easter. 1995. Water Allocation and Water Markets:
An Analysis of Gains-from-Trade in Chile. World Bank Technical Paper 315. Wash-
ington, D.C.: World Bank.
Tsur and Dinar 261
Hexem, Roger R., and Earl 0. Heady. 1978. Water Production Functions for Irrigated
Agriculture. Ames: Iowa State University Press.
Howitt, Richard E., and Henry Vaux. 1995. "Competing Demands for California's Scarce
Water." In Ariel Dinar and Edna Loehman, eds., Water Quantity/Quality Manage-
ment and Conflict Resolution, pp. 271-87. Westport, Conn.: Praeger.
India, Directorate of Economics and Statistics. 1993. Indian Agriculture at a Glance.
23d ed. New Delhi.
Kasnakoglu, Haluk, and Erol Cakmak. 1995. "Economic Value and Pricing of Water in
Agriculture." Paper presented at the seminar Economic Aspects of Water Manage-
ment. CIHEAM-International Center for Advanced Mediterranean Agronomic Studies,
Marrakech. Processed.
Kemper, Karin E. 1996. The Cost of Free Water. Linkoping Studies in Arts and Science
137. Linkoping, Sweden: Institute of Tema Research.
Laffont, Jean-Jacques, and Jean Tirole. 1994. A Theory of Incentives in Procurement
and Regulation. Cambridge, Mass.: MIT Press
Le Moigne, Guy, Shawki Barghouti, Gershon Feder, Lisa Garbus, and Mei Xie, eds.
1992. Country Experiences with Water Resources Management: Economic, Institu-
tional, Technological, and Environmental Issues. World Bank Technical Paper 175.
Washington, D.C.: World Bank.
Maestu, Josefina. 1995. "Evolution of Water Tariffs in Spain and Present Debates."
Paper presented at the seminar Economic Aspects of Water Management. CIHEAM-
International Center for Advanced Mediterranean Agronomic Studies, Marrakech.
Processed.
Morocco, Direction de la Statistique. 1993. "Annuaire statistique du Maroc, 1992."
Rabat. Processed.
Morocco, Ministry of Agriculture and Agricultural Development. 1994. "Agricultural
Development Regional Office of the Haouz." Regional Agricultural Development
Office of the Haouz Office, Rabat. Processed.
O'Mara, Gerald T., ed. 1988. Efficiency in Irrigation: The Conjunctive Use of Surface and
Groundwater Resources. A World Bank Symposium. Washington, D.C.: World Bank.
Plusquellec, Herve, Charles Burt, and Hans W. Wolter. 1994. Modern Water Control in
Irrigation: Concepts, Issues, and Applications. World Bank Technical Paper 246.
Washington, D.C.: World Bank.
Rao, P. K. 1988. "Planning and Financing Water Resource Development in the United
States: A Review and Policy Perspective." American Journal of Economics and Soci-
ology 47(January):81-96.
Rhodes, G. F., and R. K. Sampath. 1988. "Efficiency, Equity, and Cost Recovery Impli-
cations of Water Pricing and Allocation Schemes in Developing Countries." Cana-
dian Journal of Agricultural Economics 36:103-17.
Rosegrant, Mark, and Hans Binswanger. 1994. "Markets in Tradable Water Rights:
Potential for Efficiency Gains in Developing-Country Water Resource Allocation."
World Development 22(November):1613-25.
Roumasset, James. 1987. "The Public Economics of Irrigation Management and Cost
Recovery." Agriculture and Natural Resources Department, World Bank, Washing-
ton, D.C. Processed.
Sampath, R. K. 1992. "Issues in Irrigation Pricing in Developing Countries." World
Development 20(7):967-77.
262 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Shah, Tushaar. 1993. Groundwater Markets and Irrigation Development: Political
Economy and Practical Policy. Bombay: Oxford University Press.
Small, Leslie E., and Ian Carruthers. 1991. Farmer Financed Irrigation: The Economics
of Reform. Cambridge, U.K.: Cambridge University Press.
Tsur, Yacov, and Ariel Dinar. 1995. "Efficiency and Equity Considerations in Pricing
and Allocating Irrigation Water." Policy Research Paper 1460. Policy Research De-
partment, World Bank, Washington, D.C. Processed.
. 1996. "On the Relative Efficiency of Pricing Irrigation Water and Their Imple-
mentation." Center for Agricultural Economics Research, Rehovot, Israel. Processed.
Tsur, Yacov, Hokyoun Park, and Arie Issar. 1989. "Fossil Groundwater Resources as a
Basis for Arid Zone Development? An Economic Inquiry." International Journal of
Water Resources Development 5:191-201.
United Nations. 1980. Efficiency and Distributional Equity in the Use and Treatment of
Water: Guidelines for Pricing and Regulations. New York.
World Bank. 1995. "Tamil Nadu Water Resources Consolidation Project." Staff Ap-
praisal Report 13506-IN. South Asia Department II, World Bank, Washington, D.C.
Processed.
Young, R. A., and R. H. Haveman. 1985. "Economics of Water Resources: A Survey."
In A. V. Kneese and J. L. Sweeny, eds., Handbook of Resource and Energy Econom-
ics, vol. 2. New York: Elsevier.
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 263-92
Managing Price Risk in the Pakistan Wheat Market
Rashid Faruqee, Jonathan R. Coleman, and Tom Scott
The government intervenes in the wheat market in Pakistan to ensure food security for
consumers and to provide adequate and stable incomes for producers. The cost of this
intervention is high, and its impact on the performance of agriculture has been signifi-
cantly negative. The World Bank is urging policy changes such as removing agricul-
tural trade restrictions, price supports, and subsidies. However, policymakers often
resist such reforms, fearing that they will expose the domestic market to fluctuating
international commodity prices.
This article assesses the risk management needs of the sector and evaluates whether
using financial instruments-such as commodity hedging using futures, options, or
swaps-would improve risk management. Simulations based on monthly data for 1994
show that market-based methods of risk management could reduce the impact of in-
ternational price volatility on the domestic market without incurring high government
cost or distorting price signals.
The Pakistan government has long intervened in the wheat sector because of its
importance as Pakistan's leading agricultural commodity. Interventions in the
sector seem to have two objectives-to protect the interests of consumers by
keeping the domestic price below the import parity price and to protect the
interests of producers by reducing price fluctuations and guaranteeing a support
price. Foremost among the mechanisms used to meet the first objective, the gov-
ernment has set an import subsidy to keep domestic prices below import parity
levels and has banned private sector trading on international markets. Although
the government has succeeded in stabilizing wheat prices, the policy has had a
significant economic cost in that it distorts the market signals facing farmers and
private traders throughout the sector. The direct link between these signals and
the volatile international price of wheat makes subsidy payments to wheat farm-
ers both large and highly unstable.
Policymakers generally recognize the need to end the highly distortionary policy
of keeping domestic wheat prices artificially low. However, they fear the pos-
sible short-run economic and political repercussions of agricultural price insta-
bility that would accompany the phasing out of public sector direct intervention
in wheat marketing. They hesitate to implement market liberalization policies
fully in the absence of alternative price stabilization mechanisms. (See Claessens
Rashid Faruqee and Jonathan R. Coleman are with the Agriculture and Natural Resources Division
of the South Asia Region at the World Bank. Tom Scott is with Sparks Companies, Memphis, Tennessee.
This research was funded by the World Bank's Research Support Budget (RPo 679-70).
D 1997 The International Bank for Reconstruction and Development / THE WORLD BANK
263
264 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
and Duncan 1993 and Gilbert 1993 on the stability of international commodity
prices.)
Developing countries' interest in market-based risk management techniques,
including the use of commodity futures, options, and swaps, has grown signifi-
cantly in recent years. The use of such financial instruments could provide Paki-
stan with an attractive method of managing its price risk, as long as the mecha-
nisms are understood well and used appropriately. Because they require less
government intervention and are more cost-effective than alternative approaches,
financial risk management instruments may be preferable to more intervention-
ist stabilization methods in Pakistan.
This study deals with price risk of the wheat crop in Pakistan. It has two
major objectives. First, it seeks to assess the risk management needs, if any, of
the wheat sector by identifying the market participants and institutions exposed
to risk and measuring the levels of those risks. Second, it seeks to evaluate whether
market-based financial instruments would provide a less distortionary method
of managing price risks than the stabilization methods currently used in the
wheat sector.
This study adds to the growing body of research on how developing countries
can hedge the risk associated with fluctuating agricultural commodity prices.
The recent World Bank book by Claessens and Duncan (1993), Managing Com-
modity Price Risk in Developing Countries, contains eleven case studies. These
case studies and others in the academic literature are fairly limited in their scope
and coverage. They focus largely on exporting countries in Latin America and
Africa. And they concentrate on a somewhat limited group of agricultural com-
modities, mainly cocoa, coffee, and cotton. See, for example, Myers (1993) and
Claessens and Varangis (1993) on Costa Rican coffee exports; Satyanarayan,
Thigpen, and Varangis (1993) on francophone African cotton exports; Varangis,
Thigpen, and Akiyama (1993) on Egyptian cotton exports; and Claessens and
Coleman (1993) on Papua New Guinea's gold, copper, coffee, cocoa, logs, and
palm oil exports.
Few studies have been undertaken on hedging in the grain market. Larson
(1993) examines the management of price risks for maize imports in Mexico.
Much of Larson's paper concerns domestic price stabilization using variable
border tariffs and subsidies to keep domestic prices within a price band. The
study includes a discussion of how the government could use options to man-
age the risk of international price movements that would require subsidy
payments to keep domestic prices within the price band. Sheales and Tomek
(1987) examine the effectiveness of hedging wheat prices in Australia using
U.S. futures markets. Faruqee and Coleman (1996) review these studies in
more detail.
To date, very little work has been done from the perspective of a developing-
country importer wishing to hedge price risk in the world grain markets, and
very few studies have looked at the prospects for commodity hedging in Asian
countries. So, although this article focuses on hedging wheat in Pakistan, it has
Faruqee, Coleman, and Scott 265
wider relevance to other Asian countries that rely heavily on grain imports to
meet their food consumption needs.
I. WHEAT PRICE SYSTEM AND INCIDENCE OF PRICE RISKS
Determining the risk management needs of the wheat sector requires analyz-
ing the various prices facing different market participants. Market participants
facing prices fixed by the government have no price risk, whereas those facing
highly unstable prices are likely to be the most interested in risk management.
Prices and Subsidies
The provincial food departments (PFDS) in Pakistan are the chief institutions
through which the government implements its price support policy. The law
requires PFDS to purchase any volume of wheat delivered to them as long as the
wheat meets certain quality standards. The predetermined price paid to farmers
is known as the procurement price. Because PFDs must accept all deliveries, the
procurement price becomes a floor below which the free market price cannot
fall. This price is fixed throughout the year and is constant across all centers
nationwide. Provincial food departments sell the majority of their wheat to pri-
vate flour mills at the release, or issue, price, which is set at the same level in all
areas of the country. This policy aims to control the price of wheat at the whole-
sale level (although small differences between the release and wholesale prices
do result from transportation margins and quality premiums), thereby reducing
the price of flour to consumers (because wheat represents a large share of the
total cost of producing ftour).
The policy of artificially depressing prices is costly to the economy and the
government. The price policy has a significant economic cost in that it distorts
the market signals facing farmers and private traders throughout the sector (World
Bank 1994). In particular, the system of pan-territorial pricing weakens private
sector incentives for wheat transportation, while pan-seasonal pricing provides
disincentives for private sector storage.
The system incurs subsidies, paid mainly by the provincial governments, be-
cause revenues received by the PFDS from the sale of wheat (at the release price)
are generally less than the cost of procuring wheat (the procurement price plus
transport, handling, and storage charges). Between 1984-85 and 1993-94, the
average annual subsidy amounted to 340 rupees (PRs) per ton, approximately
18 percent of the procurement price. The total subsidy payments ranged from
PRs533 million in 1984-85 to PRs3.3 billion in 1986-87 and averaged PRs1.3
billion over the 1984-93 period.1
The highest level of government decides the level of wheat imports. The Min-
istry of Agriculture implements the import of wheat and handles the financing
of imports, including foreign aid and ocean shipping. The level of wheat imports
depends on several factors, including the level of public sector stocks, the ex-
1. A billion is 1,000 million.
266 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
pected procurement of domestic output, the handling capacity of ports, the level
of reserve stocks, the conditions of the international wheat market, and foreign
exchange reserves.
The PFDS buy imported wheat from the federal government at the same release
price at which they sell domestically produced wheat to private millers. In addi-
tion, they pay the in-country transportation costs. In general, the government,
buying at the CIF (the value including the cost, insurance, and freight) import
price, pays more than it receives from selling to the PFDs at the release price. A
specific federal budget allocation for imported wheat subsidies finances this
margin between the CIF import price and the release price.
In 1993-94 this subsidy amounted to PRs590 per ton. Given total imports for
1993-94 of approximately 1.9 million tons, the government incurred a total
subsidy payment of about PRs1.1 billion. This subsidy payment per ton has
been quite variable over time, ranging between PRs341 per ton in 1987-88 and
PRs1,660 per ton in 1989-90. Total subsidy payments have also fluctuated dra-
matically, as a result of both variations in the subsidy per ton and the level of
imports. For example, the total subsidy payment was only PRs2O5 million in
1987-88 (when only 601,000 tons were imported), compared with PRs2.9 bil-
lion the next year (when imports exceeded 2 million tons).
Thus the government's intervention in the wheat market has two main objec-
tives. It achieves the first objective, to reduce the average price paid by consum-
ers vis-a-vis the import parity price, by setting the release price below the inter-
national price and making up the difference with a government subsidy. Over
the period May 1980 to April 1995, the government subsidy was about PRs1,000
per ton. It achieves the second objective, to protect producers against the fluc-
tuations of international commodity prices, by establishing a fixed procurement
price below which domestic prices do not fall. Over the period May 1980 to
April 1995, the coefficient of variation on the procurement price was about 6
percent, compared with 14 percent for the import parity price. This article fo-
cuses on the second objective of price stabilization. Although the government
could adopt several methods of stabilizing prices, such as the use of buffer stocks
or a buffer fund (see Faruqee and Coleman 1996, annex II), here we explore an
alternative that involves hedging risk using financial instruments. Thus we com-
pare the use of hedging only with the current method of stabilization, without
looking at alternative methods.
Incidence of Price Risk
We measure the level of price risk facing different market participants by the
standard deviation and coefficient of variation (the ratio of the standard devia-
tion to the mean) of prices paid and received over time (a 16-year period be-
tween May 1980 and April 1995). Table 1 reports the domestic prices and
results.
In terms of price exposure, the results show that domestic farmers and private
traders face relatively little price risk. The government-determined prices have
Faruqee, Coleman, and Scott 267
Table 1. Measures of Price Variability in Pakistan, May 1980-April 1995
Standard Coefficient of
Price per ton Mean deviation variation (percent)
Procurement price (rupees) 2,566 163.6 6.4
Release price (rupees) 2,645 154.2 5.8
Wholesale price (rupees)
Lahore 3,044 254.6 8.4
Multan 2,809 222.4 7.9
FOB Pacific Northwest (U.S. dollars) 158.5 27.3 17.2
CIF Karachi (U.S. dollars) 187.6 38.6 20.6
CIF Karachi (rupees) 3,778 532.7 14.1
FOB Pacific Northwest with EEP (U.S. dollars) 154.3 31.6 20.5
CIF Karachi with EEP (U.S. dollars) 183.4 42.4 23.1
CIF Karachi with EEP (rupees) 3,687 614.6 16.7
Government import subsidy (rupees) 1,042 617.6 59.3
Note: Prices are deflated by the producer price index. Lahore and Multan are cities in Pakistan. FOB
denotes free on board (used to value exports); CIF denotes cost, insurance, and freight (used to value
imports); and EEP denotes the U.S. government's Export Enhancement Program.
Source: Government of Pakistan (1995) and International Wheat Council (various issues).
coefficients of variation of 6.4 and 5.8 for the procurement price and release
price, respectively (table 1). Wholesale wheat prices in the cities of Lahore and
Multan have also varied very little since the early 1980s, with coefficients of
variation of about 8 percent.
By contrast, international prices have been substantially more volatile than
domestic prices. The U.S. dollar price of wheat at Pacific Northwest ports has a
coefficient of variation of 17.2 percent; when freight charges are included, the
figure rises to 20.6 percent (table 1). Interestingly, when we use the Karachi U.S.
dollar price to convert the price of wheat into rupees, the instability of the price
series declines considerably with a coefficient of variation of only 14.1 percent.
One explanation for this change is that although wheat prices and freight costs
have declined in real terms over time, the value of the rupee in terms of the U.S.
dollar has fallen. As a result, the decline in the commodity price has been offset
by the change in the exchange rate, with the net effect that the international
price in rupees has remained relatively stable.
The U.S. Export Enhancement Program (EEP) was introduced in 1985 to
boost exports of U.S. agricultural products following their precipitous de-
cline in the first half of the 1980s. The EEP pays subsidies to U.S. exporters to
allow them to sell agricultural products in targeted countries at competitive
prices (below U.S. market prices). The program helps U.S. products meet
subsidized competition, expands U.S. agricultural exports, and encourages
negotiations on agricultural trade problems. Wheat is the chief commodity
sold under the EEP (accounting for more than 85 percent of the sales value of
all EEP commodities), and EEP sales account for 50 percent of total U.S. wheat
exports since 1985.
268 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
The EEP subsidies have increased the instability of wheat prices considerably,
with the U.S. dollar prices at Pacific Northwest ports and at Karachi having
coefficients of variation in excess of 20 percent (table 1). It is important to note
that comparing the coefficients of variation of prices with and without EEP sub-
sidies overestimates the impact of these export subsidies on variability, because
the EEP subsidies lower the average price (the denominator in the coefficient of
variation calculation), thereby giving rise to an increase in the coefficient of
variation for a fixed standard deviation. Finally, we measure the government's
price risk exposure by the variability of subsidy payments per ton (simplified as
the CIF price measured in rupees less the release price). The coefficient of varia-
tion of this subsidy series is almost 60 percent, indicating that in a typical year
subsidy payments will be 60 percent above or below the average payment. This
indicates the high degree of instability and risk that the government faces each
year.
These findings have several important implications for risk management. They
indicate that the government has been successful in stabilizing prices through its
procurement and policy of fixed producer and miller prices. Because of its
distortionary impact on economic incentives and because there may be more
effective and less costly methods of price stabilization, the policy may neverthe-
less be inadvisable. Another important implication for risk management is that
given the current policy regime, farmers and millers have little incentive to man-
age risk on their own behalf. In effect, the government has crowded out private
sector risk management, and farmers and millers have little need to worry about
fluctuating prices when making production and investment decisions.
Price stabilization policies involving government procurement and pricing do
not remove price risk from the economy as a whole but merely transfer the risk
within the economy. The policies transfer the risk from wheat market partici-
pants in the form of unstable prices to the government (and ultimately to tax-
payers) in the form of unstable subsidy payments.
II. PRICE STABILIZATION, HEDGING, AND THE ROLE OF GOVERNMENT
Under the existing system, the government stabilizes wheat prices for farmers
and traders through its price and procurement policies. In the short term, such
government intervention may be justified because market failures, such as the
lack of available market information, unfamiliarity of farmers with risk man-
agement techniques, and absence of an effective system of brokerage, prevent
market participants from engaging in price risk management through the pri-
vate sector. In the long term, however, the government should stop its direct
intervention and instead focus on providing economic and institutional condi-
tions conducive to private sector risk management activities, including the es-
tablishment of futures exchanges in Pakistan.
There are several strong economic arguments against government interven-
tion. For example, government stabilization of commodity markets generally
Faruqee, Coleman, and Scott 269
constrains the active participation of the private sector, particularly in storage,
transportation, and general trading activities. This is of great relevance to Paki-
stan, which has a system of pan-seasonal and pan-territorial pricing. This sys-
tem has seriously weakened the economic incentives for private sector involve-
ment in storage and transportation. Price stabilization also leads to welfare losses
associated with the failure of producers and consumers to react to market sig-
nals (Massell 1969). For instance, if producers are insulated from the market
through a government stabilization scheme, they will tend to overproduce in
periods of lower international prices when domestic prices are artificially raised,
and to underproduce in periods of high international prices when domestic prices
are artificially lowered. Further, stabilizing prices does not stabilize income or
profit. Instead, stabilizing prices with year to year fluctuations in production
would result in greater income instability than if prices were allowed to adjust to
the level of supplies (Thomas 1985). Stabilization schemes are difficult to imple-
ment, often requiring huge bureaucracies. They are also expensive (stabilization
can be so expensive to operate that the costs of operating the program outweigh
any benefits that might accrue to producers and consumers) and highly prone to
political manipulation, as experience from many countries has shown (Knudsen
and Nash 1990).
Although economic arguments generally do not support price stabilization by
the government, considerable social and political pressures do. However, there
may be ways to provide stabilization that are less distorting of economic incen-
tives and that are more consistent with the market and trade liberalization re-
forms currently taking place in Pakistan.
Under the current system, the government pools the price risk of wheat farm-
ers and traders and assumes it in the form of unstable subsidy payments. Having
assumed the risk, however, the government can employ mechanisms with which
to transfer the risk to entities willing and able to take it on. One way of transfer-
ring this risk would be to hedge the price of wheat with futures markets. Hedg-
ing involves the buying and selling of financial assets whose values are linked to
the underlying commodity markets. Four major types of hedging instruments
can be used-forward contracts, futures contracts, options, and swaps. Manag-
ing price risks through these mechanisms could be highly beneficial to Pakistan
because doing so facilitates better financial management and planning and al-
lows buyers and sellers of commodities to protect themselves against the poten-
tially catastrophic consequences of sudden and unforeseen changes in market
conditions.
In the long term, however, the government should phase out direct interven-
tion (including public sector hedging activities) and confine itself to establishing
economic conditions supportive of private sector hedging activities. The govern-
ment should also provide the preconditions to set up commodity futures ex-
changes in Pakistan. (Several developing countries have established local futures
and options exchanges. Examples include Argentina for grains and livestock;
Brazil for livestock, coffee, cotton, and gold; China for various metals and agri-
270 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
cultural commodities; Hungary for grains and hogs; India for pepper; Malaysia
for palm oil, tin, and cocoa; the Philippines for copra, sugar, coffee, soybeans,
and dry coconut; and Zimbabwe for corn and beans.)
For the private sector to engage in hedging activities using U.S. futures ex-
changes, the government should remove the obstacles that deter their use. Varangis
(1994) identifies several such obstacles that are common in many developing
countries, including Pakistan. First, legal and regulatory barriers prevent market-
based hedging. Foreign exchange controls, for example, which are common in
many developing countries, can make hedging impossible. In terms of regula-
tory barriers, since the structural adjustment program in Pakistan was intro-
duced, financial markets have become increasingly liberalized, and most foreign
exchange controls have been lifted. In particular, there are no controls on trans-
actions on the current account, including goods, services, and transfers. There
are restrictions on the capital account; however, no current laws or statutes
automatically prohibit hedging in commodities futures markets by Pakistani
residents.
Second, the current system of fixing procurement and release prices for the
whole season and announcing the prices well before planting means that private
farmers and traders do not need to manage their own risk. Third, farmers and
traders lack familiarity with futures markets and expertise in how to use them.
In some developing countries the misconception that hedging is a form of specu-
lation presents a major obstacle. Fourth, misunderstanding the tradeoffs between
risks and returns can lead to the perception that hedging strategies that result in
higher total import bills are counterproductive. Fifth, some hedging instruments
require up-front costs that can represent obstacles for some potential market
participants. Option contracts, for example, require a premium, futures con-
tracts require a margin, and some forms of financial collateral may be required
for swaps and over-the-counter arrangements.
Also in the long term, and as an alternative to hedging on U.S. exchanges,
Pakistan could establish local exchanges. Varangis and Larson (1996) identify
several preconditions for establishing futures exchanges in developing countries.
These are highly relevant to setting up a wheat exchange in Pakistan, and an
important future role of the government would be to ensure that such precondi-
tions are met. These include the development of infrastructure in areas such as
communications, transportation, and information processing; strong commer-
cial and financial sectors; the absence of government intervention in the wheat
market; a strong legal and regulatory framework in establishing a futures mar-
ket; and sufficient capital among potential market participants to forestall
counterparty risk (that is, sufficient capital to form a viable clearing entity).
Although not insurmountable in the long term, such conditions are not likely to
emerge in the short term.
Establishment of a wheat futures market would benefit Pakistan by improv-
ing price discovery and reducing basis risk. (The basis risk is the difference be-
tween the futures price in the United States and the market price in Pakistan.) A
Faruqee, Coleman, and Scott 271
futures market would reduce the basis risk by specifying wheat contracts for the
varieties and qualities of local wheat and for delivery within the country (in
Lahore or Karachi, for example). Other benefits include more publicly available
information on wheat prices, improved transmission of price and other com-
modity information, improved credit systems, more responsive capital markets,
uniformity in repayment rules and market surveillance, reduced transactions
costs, and more accurate forward prices (Varangis and Larson 1996).
In the next section, we compare these factors with the benefits and costs of
using U.S. futures exchanges. U.S. exchanges have an advantage because they
have well-established rules and regulations and are very liquid. Higher levels of
liquidity mean reduced transactions costs that can outweigh the basis and ex-
change rate risks. The main disadvantage of using a U.S.-based exchange is that
the basis risk and exchange rate risk can be large. Overall, however, the advan-
tages outweigh the disadvantages, and Pakistan should use U.S. exchanges until
it can set up viable domestic exchanges.
III. EFFECTIVENESS OF HEDGING WHEAT PRICE RISK
Even with no legal and institutional barriers and no informational or aware-
ness constraints, hedging still might not provide Pakistan with an effective means
of managing its commodity price risk. The effectiveness of hedging depends on
the nature of the commodity traded, the timing of purchases, land and ocean
transportation charges, exchange rate movements, export subsidies, and other
policy variables-factors that disassociate the prices quoted on the commodity
exchanges with those actually paid by importers in Pakistan. All of these factors
create a difference between the prices quoted on commodity exchanges and the
prices actually paid by wheat importers. Greater unpredictability in the basis, or
the difference between the two prices, reduces the effectiveness of managing risk
by hedging.
By observing the basis over time, experienced hedgers are able to predict
the basis with a good degree of accuracy. Unforeseen differences between the
futures contracts and cash prices result in an unpredicted basis. This risk, the
basis risk, cannot be managed by hedging. Because hedging does not elimi-
nate all uncertainty, it can be viewed as merely substituting basis risk for
price risk. Overall risk is nevertheless reduced because basis risk is consider-
ably less than price risk (because cash and futures prices tend to be closely
correlated).
Analysts commonly test for the correlation between cash prices of govern-
ment wheat purchases on the international market and wheat futures prices by
regressing a time series of nearby futures contract prices on the corresponding
cash price series (nearby prices are the prices closest to the expiration date). The
higher the correlation, as measured by the R2 statistic, the greater the extent to
which movements in cash prices can be explained by movements in futures prices
and therefore the more effective the hedging operations. We can measure the
272 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
basis risk by the variability in cash prices not explained by futures price move-
ments, or by 1 - R2.2
To quantify the potential effectiveness of hedging Pakistani wheat on U.S.
futures exchanges, we test three futures contracts-no. 1 soft white wheat traded
on the Minneapolis Grain Exchange, no. 2 hard red winter wheat traded on the
Kansas City Board of Trade, and no. 2 soft red winter wheat traded on the
Chicago Board of Trade. We chose these contracts because they cover wheat
whose characteristics are closest to those of wheat commonly imported by Paki-
stan. We collected monthly data for the three contracts from February 1991
(when the Minneapolis contract started trading) to April 1995, providing a total
of 51 observations. We first test the correlations against four different import
(cash) prices-the U.S. dollar FOB (free on board) price of western white wheat
at Pacific Northwest ports, the price of wheat delivered at Karachi, the Karachi
price adjusted for EEP, and the Karachi price in Pakistani rupees (table 2). Be-
cause we find all the price series to be nonstationary based on the Durbin-Watson
test, we transform them by taking first differences. Subsequent tests of the trans-
formed series show them to be stationary in all cases.
We find a high degree of correlation between the wheat futures contract price
on the Minneapolis Grain Exchange and the U.S. dollar FOB price of western
white wheat quoted at Pacific Northwest ports. The results in table 2 show that
variations in the futures prices can explain 89 percent of the variation in the
cash prices, with a basis risk of only 11 percent (most likely reflecting variability
in transportation costs). This indicates that at least the U.S. dollar FOB price at
Pacific Northwest ports faced by Pakistan could be fairly well hedged by trading
wheat futures contracts on the Minneapolis Grain Exchange.
We then test correlations between the western white wheat price adjusted for
freight charges and the Minneapolis wheat futures contract. Because freight rates
between Pacific Northwest ports and Karachi have varied little since early 1991,
the correlation is the same (R2 of 89 percent), with a corresponding basis risk of
only 11 percent (table 2). This indicates that hedgers in Pakistan could manage
the risk of fluctuating U.S. dollar CIF prices, assuming the government does not
implement EEP subsidies.
2. Measuring the basis risk in this manner is sometimes complicated by the statistical properties of
time series data. In particular, the validity of testing the correlation between cash and futures prices
using regression requires that each price series be stationary. A stationary series is one in which the
underlying stochastic process generating the series is invariant with respect to time (that is, the stochastic
process is in equilibrium over time about a constant mean level, and the probability of any given fluctuation
around that mean level is the same at any point in time). Typically, time series price data are nonstationary
because they are influenced by seasonal factors. Fortunately, several straightforward tests for stationarity
such as the Durbin-Watson test of Sargan and Bhargava and the Dickey-Fuller test can be performed
(Palaskas and Varangis 1991). In most cases, cash and futures price series that are found to be
nonstationary can be transformed into stationary series simply by taking first differences (that is, the
price in period T minus the price in period T - 1). The differenced series can then be regressed against
one another, with the R2 coefficient from the regression providing a valid measure of hedging effectiveness
and basis risk.
Faruqee, Coleman, and Scott 273
Table 2. Hedging Effectiveness, Basis Risk, and Hedge Ratios in Pakistan,
1991-95
Contract and price R21 Basis riskb Hedge ratioc
Minneapolis soft white wheat no. 1
FOB U.S. dollar price at Pacific Northwest ports 0.89 0.11 0.92
CIF U.S. dollar price at Karachi 0.89 0.11 0.94
CIF U.S. dollar price at Karachi adjusted for EEP subsidiesd 0.59 0.41 0.91
CIF rupee price at Karachi adjusted for EEP subsidiesd 0.55 0.45 0.84
Kansas City hard red winter wheat no. 2
FOB U.S. dollar price at Pacific Northwest ports 0.62 0.38 0.75
CIF U.S. dollar price at Karachi 0.61 0.39 0.76
CIF U.S. dollar price at Karachi adjusted for EEP subsidiesd 0.42 0.58 0.76
CIF rupee price at Karachi adjusted for EEP subsidiesd 0.39 0.61 0.74
Chicago soft red winter wheat no. 2
FOB U.S. dollar price at Pacific Northwest ports 0.50 0.50 0.66
CIF U.S. dollar price at Karachi 0.49 0.51 0.66
CIF U.S. dollar price at Karachi adjusted for EEP subsidiesd 0.33 0.67 0.64
CIF rupee price at Karachi adjusted for EEP subsidiesd 0.30 0.70 0.65
Note: Calculations are based on monthly observations for February 1991-April 1995 (51
observations). FOB denotes free on board (used to value exports); CIF denotes cost, insurance, and freight
(used to value imports); and EEP denotes the U.S. government's Export Enhancement Program.
a. From ordinary least squares regression; cash price = a + b' nearby futures price. All price series
were transformed into first differences. Regression period from February 1991 to April 1995.
b. The differences between the futures price in the U.S. and the market price in Pakistan.
c. Slope coefficient regression between first differences of cash and nearby futures prices.
d. Regressions are run from September 1992, the first time Pakistan qualified for EEP subsidies.
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues),
and authors' calculations.
The picture changes dramatically when we test correlations between Minne-
apolis futures prices and the U.S. dollar Karachi price of western white wheat
adjusted for EEP subsidies. The R2 from the regression is only 0.59, implying a
basis risk of 41 percent (table 2). This decline in correlation reflects not only the
instability of EEP payments but also the fact that they represent a large percent-
age of the overall purchase price. The finding also suggests that continuation of
subsidies would significantly limit the effectiveness of hedging as a mechanism
for managing risk. Converting the CIF Karachi price into rupees yields a slightly
lower correlation with the Minneapolis futures prices (R2 of 0.55), a level also
well below that needed to make hedging effective (table 2).
We finally test the same set of correlations using the prices of the no. 2 hard
red winter wheat futures contract traded on the Kansas City Board of Trade and
the no. 2 soft red winter wheat futures contract traded on the Chicago Board of
Trade. Overall, changes in the prices of these futures contracts are less corre-
lated with changes in the relevant wheat prices for Pakistan. Even before adjust-
ing for transport costs, EEP subsidies, and exchange rates, changes in the futures
price on the Chicago Board of Trade can explain only half the changes in the
U.S. dollar western white wheat price at Pacific Northwest ports (table 2). This
274 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
finding indicates that the Chicago wheat futures contract would not be an effec-
tive hedging instrument for Pakistani importers.
If there is no basis risk and if changes in futures prices explain all changes in
cash prices, hedgers should cover all cash transactions with futures contracts.
When there is basis risk, however, hedgers should generally cover only a portion
of their cash position. Statistical analysis of cash and futures prices can deter-
mine the hedge ratio, an important policy variable.
In Pakistan, the instability of wheat import costs depends on the variability of
both the volume and price of imports. However, because the government con-
trols the volume of imports, fluctuating prices are the main source of risk expo-
sure, and controlling price fluctuations is the main objective of risk management
strategies. We can view the hedging decision as a portfolio selection problem in
which the hedger selects the optimal proportions of unhedged (cash) and hedged
(futures) wheat imports. In this case, risk management strategies aim to mini-
mize the variance in the value of the portfolio of hedged and unhedged imports.
Based on portfolio selection theory, we can demonstrate that the optimal hedge
ratio is equivalent to the slope coefficient in the ordinary least squares regression
between changes in the cash and futures prices (Ederington 1979). Calculating
the optimal hedge ratio in this manner, we assume that the hedger seeks to mini-
mize risk. Selecting the portfolio of hedged and unhedged imports that mini-
mizes risk may result in a higher import bill than would otherwise apply. Whether
the importers consider the higher import bill acceptable depends on their aver-
sion to risk. Infinitely risk-averse importers seek to minimize risk. Less risk-
averse importers are willing to bear some risk in order to reduce the cost of
imports. Given the government's concern over commodity price risks, it seems
reasonable to assume that the government is infinitely risk averse and to select
hedge ratios accordingly.
This assumption may not be justified given our argument that the govern-
ment is better able to pool and absorb risks than individual farmers and millers.
An alternative way to derive an optimal hedge ratio would be to equate the
marginal benefit from hedging (measured in terms of the value of risk reduction)
with the marginal cost of hedging (brokerage fees). Using the price series with
and without hedging, we derive estimates of the risk benefits from hedging based
on formulas developed by Newbery and Stiglitz (1981, pp. 93). We calculate the
value of risk reduction (risk benefit) for values of the hedge ratio ranging be-
tween 0 and 1 and compare it with the cost of hedging in each case. Then, using
numerical methods, we determine an optimal value of the hedge ratio at the
point where the marginal cost of hedging equals the marginal benefit of hedging.
The problem with this approach is that a value for the coefficient of relative risk
aversion has to be assumed, which requires specifying the decisionmaker's util-
ity function (in this case the government's). This problem is intractable, and
many researchers simply assume a coefficient of relative risk aversion equal to 1
and then measure the sensitivity of the risk benefits to different values of the
coefficient (Akiyama and Varangis 1991 and Coleman and Larson 1993). As-
Faruqee, Coleman, and Scott 275
suming a relative risk aversion coefficient equal to 1, we obtain an optimal hedge
ratio of about 0.85, which increases to 0.89 for a coefficient of 2. This indicates
that a hedge ratio close to 0.9 would be appropriate for a fairly wide range of
assumptions about the government's preferred level of risk.
Assuming risk minimization, hedging the U.S. dollar FOB wheat prices at Pa-
cific Northwest ports using the Minneapolis Grain Exchange yields a hedge ra-
tio of 0.92 (table 2). This means that if the government wishes to purchase, say,
2 million tons of wheat, it would need to cover 1.84 million metric tons with
futures contracts, or roughly 13,522 contracts (assuming about 136 tons per
contract).3 Hedge ratios range from 0.91 for the U.S. dollar CIF price adjusted
for EEP subsidies to 0.94 for the U.S. dollar CIF price without the subsidy (table
2). Hedge ratios decline using the Chicago wheat contract, ranging between
0.64 and 0.66. In general, the hedge ratios decline as the level of basis risk in-
creases because R2 measures the effectiveness of the hedging, and 1 - R2 mea-
sures the basis risk. Thus the greater the basis risk, the less effective the hedging,
and the lower the basis risk, the more effective the hedging.
The analysis indicates, absent export subsidies, the potential effectiveness of
hedging Pakistani wheat purchases using the soft white wheat contract that is
traded on the Minneapolis Grain Exchange. The existence of export subsidies
severely limits the effectiveness of hedging, however. Two developments suggest
that export subsidies may be reduced in the future. The Uruguay Round of the
General Agreement on Tariffs and Trade (GATT), which was signed in late 1993,
contains a key provision to reduce the overall level of export subsidies (a provi-
sion most affecting the United States and the European Union). The agreement
calls for a 21 percent decline in the volume of export subsidies and a 36 percent
drop in their value from a 1986-90 base period. In the United States, new farm
legislation (the Federal Agricultural Improvement and Reform Act, 1996) has
restructured agricultural programs, and budgetary pressures have limited agri-
cultural spending. These developments could result in a reduction of export sub-
sidies below the levels required under the GATT.
IV. ANALYSIS OF HEDGING STRATEGIES USING FINANCIAL INSTRUMENTS
The policy of maintaining wheat prices below an equilibrium (import parity
price) is not tenable and should be discontinued. Therefore, in this section we
outline the policy and institutional environment in which we assume hedging
takes place. Other strategies could be developed. For example, the government
could abstain from importing wheat when the parity price is below the fixed
price, in order to let the private sector import at the import parity price. This
would reduce the cost of flour and would also benefit consumers. This policy
would work as a call option from the government to wheat millers: the govern-
ment would subsidize imports as long as the local price is lower than the import
3. All quantities of wheat are measured in metric tons. Each contract is for 5,000 bushels, with 36.74
bushels per metric ton.
276 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
parity price, but when the parity price falls below the fixed price, the govern-
ment would allow millers to benefit from lower world prices. The government
could use hedging to manage its exposure to higher international prices.
We make the following assumptions. The government eliminates the wheat
import subsidy and sets a price (the release price) at which it sells imported
wheat to mills equal to the expected average import parity price for the coming
year. To provide the market with stability, the government announces the re-
lease price at the beginning of the year, and that price remains fixed throughout
the year. By buying at a variable international price and selling at a fixed domes-
tic price, the government effectively pools the risk of individual market partici-
pants and assumes the risk for itself. In particular, the government exposes itself
to the risk of international prices rising more than expected, thereby requiring a
subsidy to maintain the fixed price. Of course, if the international price falls
below the fixed domestic price, the government would impose a tax on wheat
imports, bringing the import price up to the domestic price level. Because the
international wheat price cannot be predicted accurately, we cannot expect the
subsidies required when the international price rises higher than the fixed do-
mestic price to be offset by the revenues received when the international price
falls below the fixed domestic price. To manage this risk, the government can
hedge using financial instruments. In the following sections, we outline and evalu-
ate the effectiveness of three hedging strategies using futures, options, and swaps.4
Strategy I: Hedging with Futures Contracts
One possible hedging strategy using futures contracts would enable the gov-
ernment to lock in an international price for its wheat purchases at the begin-
ning of the year. This price would equal the weighted sum of wheat futures
prices maturing at various months throughout the coming year, with weights
determined by the quantities of wheat imported in the months between each
contract expiration. Variations of this strategy concentrate or disperse hedging
among various contract months. Selection of the month in which to hedge in-
volves judgment and expertise. For the purpose of these examples, we use the
conventional hedge approach, matching calendar months with the correspond-
ing futures contract months.
Say, for example, that in December 1993 the government wishes to fix the
release price for the 1994 crop. Wheat futures contracts can expire in five differ-
ent months (March, May, July, September, and December); in December 1993
prices for delivery in each of these months in 1994 are established in the market.
On the basis of historical import trends, the government could predict fairly
well the proportions of the total import requirements before each of the delivery
months (for example, January, February, and March, 10 percent; April and May,
4. The examples do not superimpose hedging strategies on existing patterns and practices of wheat
purchases. Instead, the strategies show that hedging with futures can make purchasing much more
flexible, lock prices further into the future, and make alternative methods of purchasing wheat more
desirable.
Faruqee, Coleman, and Scott 277
20 percent; June and July, 30 percent; August and September, 25 percent; and
October, November, and December, 15 percent). It could use these proportions
to obtain a weighted import price for the coming year. The government would
guarantee this price and could use it to set the fixed release price assuming ex-
pected freight costs, export subsidies, and exchange rates.
We evaluate this strategy using actual cash and futures prices for 1993 and
1994. It should be noted that at the time of this example it is unlikely that the
large volume could have been effectively hedged on the Minneapolis Grain Ex-
change. Contracts from other exchanges (the Chicago Board of Trade and Kan-
sas City Board of Trade) could also have been used, because the Minneapolis
Grain Exchange white wheat contract did not have a great deal of liquidity in
1993 and 1994. For simplicity, however, we confine the futures operations as-
pect of this example to Minneapolis Grain Exchange contracts. This raises two
practical issues that need to be addressed with regard to hedging white wheat.
First, as already noted, liquidity on the white wheat contract is low and there-
fore represents a problem for hedging large quantities. Second, we find the basis
volatility between the FOB white wheat on the Chicago Board of Trade and Kan-
sas City Board of Trade wheat contracts to be substantially higher than for
white wheat futures contracts on the Minneapolis Grain Exchange. However,
contract volume on the Minneapolis Grain Exchange white wheat contract is
growing, and the prospects of this market providing an adequate hedge in the
future are improving. More important, as the role of governments in the export
wheat trade declines, the wheat contracts on the Chicago Board of Trade and
Kansas City Board of Trade should more closely reflect global export wheat
prices. Under these conditions, futures contracts will provide a better hedging
mechanism than has been the case in the past. The critical point is that the recent
and significant structural changes in the wheat market mean that past relation-
ships between cash and futures prices may not hold in the future, and it is pos-
sible that in the new trade environment prices on the Chicago Board of Trade
and the Kansas City Board of Trade may better reflect global supply and de-
mand conditions. This being the case, the type of techniques described here would
be even more effective for managing price risk than the tools currently available.
In this example, in mid-December 1993 the government decides that for cal-
endar year 1994 it needs about 1.2 million tons of imported white wheat (roughly
the average annual volume of imports over the past ten years) and wishes to
purchase 100,000 tons each month during the year. The government also wishes
to lock in the prevailing mid-December price of $133 per ton for the entire
purchase, on the basis of which it announces the fixed release price (assuming
expected freight costs, export subsidies from suppliers, and exchange rates). In
executing the strategy, the government buys 100,000 tons of wheat on the first
trading day of each month and buys and sells futures contracts with expiration
dates coinciding with future purchases.
Table 3 gives the monthly transactions and net positions for this strategy.
Here we discuss two months, January and December, to illustrate how the hedg-
Table 3. Analysis of Cash and Futures Transactions
(U.S. dollars per ton unless otherwise noted)
Futures transactions.
Contract Net positiond
Date of Cash price expiration date Date Futures price Futures price Gain with (effective price
transactions (1994) paid (1994) bought paid received Gainb hedge ratioc paid)
January 3 133.4 March 12/16/93 136.7 136.5 -0.18 -0.17 133.7
February 1 130.4 March 12/16/93 136.7 132.3 -4.40 -4.05 134.6
March 1 127.5 March 12/16/93 136.7 129.0 -7.71 -7.09 134.7
April 4 126.8 May 12/16/93 136.3 130.1 -6.25 -5.75 132.7
May 2 137.8 May 12/16/93 136.3 139.4 3.12 2.87 135.1
June 1 134.1 July 1/1/94 134.1 133.4 -0.73 -0.67 134.9
July 1 132.3 July 1/1/94 134.1 134.1 0.00 0.00 132.4
August 1 127.9 September 2/16/94 131.5 133.4 1.84 1.69 126.3
September 1 147.0 September 2/16/94 131.5 150.6 19.10 17.57 129.5
October 3 169.0 December 3/1/94 130.4 174.5 44.09 40.56 128.6
November 1 166.4 December 3/1/94 130.4 169.4 38.95 35.83 130.7
December 1 164.2 December 3/1/94 130.4 166.5 36.01 33.13 131.3
Note: In the hedging strategy represented here, the government purchases 100,000 tons of wheat on the first trading day of each month. It buys and sells futures
contracts with expiration dates coinciding with future purchases. See section IV of the text.
a. The quantity purchased is 92,000 tons, equivalent to 676 futures contracts. The brokerage fee is $0.15 per ton.
b. Price received minus price paid.
c. The gain times the hedge ratio of 0.92.
d. The cash price minus the gain with the hedge ratio plus the brokerage fee.
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues), and authors' calculations.
Faruqee, Coleman, and Scott 279
ing operates. On January 3, 1994, the government purchases 100,000 tons of
wheat on the international market at a price of $133.40 per ton. On the same
day, it sells 676 of the March futures contracts at a price of $136.50 per ton.
Buying the March futures at $136.70 per ton and selling them at $136.50 yields
a loss of $0.18 per ton, or $0.17 per ton with the 0.92 hedge ratio. Including a
brokerage fee of $0.15 per ton, the government pays an effective price of $133.70
per ton ($133.40 per ton cash price plus $0.17 per ton loss from the futures
transaction plus the $0.15 per ton brokerage fee). By comparison, on December
1, 1994, the government purchases 100,000 tons of wheat on the international
market at a price of $164.20 per ton. On the same day, it sells 676 of the Decem-
ber futures contracts at a price of $166.40 per ton. Buying the December futures
back in March at $130.40 per ton and selling them at $166.40 per ton in De-
cember yields a profit of $36.01 per ton, or $33.13 per ton with the 0.92 hedge
ratio. Including a brokerage fee of $0.15 per ton, the government pays an effec-
tive price of $131.30 per ton ($164.20 per ton cash price less $33.10 per ton
gain from the futures transaction plus the $0.15 per ton brokerage fee).
The pattern of cash wheat prices shows a decline into April of 1994 followed
by strong price increases in the succeeding months. It is important to remember
that the primary objective of the hedging strategy is to establish an import price
at or near the desired level of $133 per ton. Losing sight of this will lead to the
erroneous conclusion that it would be better not to have hedged purchases up to
April 1994, a period of declining market prices.
Figure 1 and the first two columns of table 4 present a comparison of the
gross FOB import price (what the government would pay if it had not hedged)
and the net FOB import price (the price that it would pay for white wheat FOB
Portland net of brokerage charges and including the gain or loss from futures
transactions). With hedging, the government would pay a lower price in six of
the twelve months; in two of the remaining six months the difference in the net
import price is less than $1 per metric ton.
It is also informative to look at the total import cost for wheat under each
scenario. Under the nonhedged scenario (gross FOB import price) total expendi-
ture is PRs6.1 billion (average monthly price times import volume of 1.2 million
tons); under the hedged scenario (net FOB import price) total expenditure is PRs5.7
billion (table 4). Clearly the government is better off having hedged.
The results also show the effectiveness of hedging for reducing the variability
of import costs. If the government had not hedged, the monthly import bill
would range from PRs460 million in April to PRs589 in October, with the big-
gest month to month change between September and October, when the cost
increases PRs67 million (from PRs522 million to PRs589 million; table 4). If the
government had hedged, the import cost would vary only slightly, with a dif-
ference between the highest and lowest months' payments of less than PRs3O
million.
Assume also that the government sets a release price for the year of PRs4,800
(equivalent to an FOB price of $133 per ton). If the government had not hedged,
Table 4. The Impact of the Futures Hedging Program on Wheat Import Payments
FOB price CiF price, Import cost Release Illustrative government subsidy or taXb
(dollars per ton) (rupees per ton) (millions of rupees) price Rupees per ton Millions of rupees
Without With Without With Without With (rupees Without With Without With
Month hedging hedging hedging hedging hedging hedging per ton) hedging hedging hedging hedging
January 133.4 133.7 4,800 4,810 480.0 481.0 4,800 0 -10 0.0 -1.0
February 130.4 134.6 4,710 4,839 471.0 483.9 4,800 90 -39 9.0 -3.9
March 127.5 134.7 4,621 4,842 462.1 484.2 4,800 179 -42 17.9 -4.2
April 126.8 132.7 4,598 4,778 459.8 477.8 4,800 202 22 20.2 2.2
May 137.8 135.1 4,935 4,852 493.5 485.2 4,800 -135 -52 -13.5 -5.2
June 134.1 134.9 4,822 4,847 482.2 484.7 4,800 -22 -47 -2.2 -4.7
00 July 132.3 132.4 4,767 4,771 476.7 477.1 4,800 33 29 3.3 2.9
August 127.9 126.3 4,632 4,585 463.2 458.5 4,800 168 215 16.8 21.5
September 147.0 129.5 5,215 4,683 521.5 468.3 4,800 -415 117 -41.5 11.7
October 169.0 128.6 5,887 4,655 588.7 465.5 4,800 -1,087 145 -108.7 14.5
November 166.4 130.7 5,806 4,718 580.6 471.8 4,800 -1,006 82 -100.6 8.2
December 164.2 131.3 5,741 4,735 574.1 473.5 4,800 -941 65 -94.1 6.5
Total 6,053.4 5,711.4 -293.4 48.6
Average 141.4 132.0 5,044 4,756 504.5 475.9 4,800 -244.5 40.5 -24.5 4.1
Note: The government purchases 100,000 tons of wheat each month. FOB denotes free on board (used to value exports), and CIF denotes cost, insurance, and
freight (used to value imports).
a. FOB price is converted into a CIF price by adding a freight cost of $24 per ton, assuming no export subsidies and using an exchange rate of PRs30.5 per dollar.
b. A negative value is a subsidy; a positive value is a tax.
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues), and authors' calculations.
Faruqee, Coleman, and Scott 281
Figure 1. Hedging Effectiveness Using Futures
Rupees per ton
6,000
CIF price without hedging
5,500
5,000
/ / Re~~~~~~~lease price
4,500 CIF price with hedging
4,000 I
January March May July September November
February April June August October December
Source. Table 4.
the unexpected rise in prices toward the end of the year would result in huge
subsidy payments of PRs293 million for the entire year. These payments would
be required because of the increase in international prices of more than $40 per
ton between August and October. However, having hedged and locked in a
price, the government does not incur subsidy payments.
This hedging program results in a lower total import cost because market
prices increase in the later half of the year. However, lowering the import bill is
not the goal of the hedging program, and the program should not be considered
successful because it is profitable. If international prices fall, the government
would end up paying more in subsidies than if it had not hedged. Payment of
additional subsidies would not indicate failure of the strategy, however. The
hedging strategy is a success because it reduces the volatility of international
prices and the cost of imports, enabling the government to manage its finances
better, and not because it reduces the overall cost of imports and thereby saves
the government money. In the long run the government can expect neither to
gain nor to lose money through hedging, and the cost of using hedging instru-
ments is equal to the brokers' fees on the contracts.
Strategy II: Hedging with Options Contracts
A second strategy involves the purchase of call options. This example in-
volves a slightly different purchasing arrangement than in the futures illustra-
282 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
tion. However, this difference does not invalidate the overall result and message
of the article-that hedging with any of the three instruments would help the
government to manage its price risk. The choice between using futures vis-a-vis
options depends on the preferred risk of the government. Options are different
from futures in that the former hedges against price movements in one direction
only (buying an option is much like buying insurance), while futures insulate
hedgers from price movements in both directions. Therefore, perhaps it is inap-
propriate to make direct comparisons between options and futures.
Call options give the holder the right to buy a specific commodity at a speci-
fied strike price. In a sense, call options provide insurance against prices rising at
a later date. The use of call options is appropriate for the government of Paki-
stan in managing future wheat imports. A call option differs from a futures
contracts strategy in that a futures contract locks in the import price. A down-
side of the futures hedging strategy is that if prices decline, the government can-
not take advantage of lower prices and incurs higher subsidy payments than if it
had not hedged. Purchasing a call option enables the buyer to establish a maxi-
mum price for a commodity by providing protection from upward price move-
ments while at the same time allowing the buyer to participate in the benefits of
downward price movements of the underlying commodity. The premium paid
for the option is the cost of receiving the upward price protection and can be
viewed as the "insurance policy" premium.
A specific hedging strategy using options contracts would work as follows. In
January 1994 the government decides to import 1.2 million tons of wheat dur-
ing the calendar year. It also decides to make half the purchases during the sec-
ond quarter and half during the fourth quarter (with purchases of 600,000 tons
in each case). In early January, U.S. exporters offer a price of $135 per ton. If the
government believes that prices could move lower during the year, it would
prefer to delay its purchase of the wheat until the time of actual delivery. By
waiting, however, the government risks the possibility that prices will rise, in-
creasing the cost of imports. In effect, the government would like to participate
in any downward move in prices while at the same time protecting itself against
upward changes in price. In early February 1994 the government decides to
import the first 600,000 tons of wheat during the month of April. It would like
to lock in the $135 per ton price being offered for April delivery but be able to
benefit if wheat prices fall. To accomplish this objective, on February 7, 1994,
the government purchases 4,410 May white wheat call options (equivalent to
600,000 tons) with a strike price of $132.28 per ton. The cost of these options is
$3.86 per ton. As noted above, physical cash white wheat FOB Portland for April
at this time is trading at $135 per ton, and the May futures contract is trading
for $133.19 per ton. In purchasing these call options, the government buys the
right (or option) to purchase white wheat futures at $132.28 per ton.
By April, May futures are trading at $130.07 per ton, cash white wheat FOB
Portland is trading at $126.77 per ton, and May white wheat options with a
$132.28 per ton strike price are worth $0.37 per ton. Given that the May fu-
Faruqee, Coleman, and Scott 283
tures price ($130.07 per ton) is below the $132.28 per ton strike price, the op-
tions held by the government have little value and will likely expire worthless.
Physical cash prices, however, have followed the general price decline, and the
government can purchase its wheat at prices that are substantially lower than
those that prevailed in January. Although in this case the price protection is not
exercised, the government has the flexibility to wait for lower prices, because its
upside risk is covered by the options.
The government pays the cash price of $126.77 per ton, plus the net cost of
options of $3.49 per ton ($3.86 purchase price less $0.37 sale price) plus a $0.10
per ton brokerage fee, yielding a net of $130.36 per ton-$3.59 per ton more
than if it does not hedge. Although hedging results in a higher price paid, the
strategy is nevertheless appropriate because it protects the government against
an increase in prices. The difference between the $130.36 per ton paid and the
cash price of $126.77 per ton ($3.59 per ton) represents the insurance premium
for guaranteeing a price of no more than $132.28 per ton. Relative to the Febru-
ary forward price of $135 per ton, the effective price of $130.36 per ton repre-
sents a saving of $4.64 per ton, a reduction of $2.8 million in import costs.
In April 1994 the government wishes to purchase the remaining 600,000 tons
of white wheat for delivery in November of 1994. The situation is such that no
FOB offers for white wheat in Portland are currently available for November or
December delivery. However, the December white wheat futures on the Minne-
apolis Grain Exchange are trading at $129.97 per ton. In addition, December
white wheat call options with a $128.60 per ton strike price are trading at $4.78
per ton. The government decides to purchase 4,410 December call options with
the $128.60 strike price as protection against a price increase. As in the April
hedge, the government has protected itself from upside price risk but is still able
to reap the benefit of price declines in the physical cash market.
When November 1994 arrives, December futures are trading at $169.39 per
ton, cash white wheat FOB Portland is worth $166.36 per ton, and December
white wheat options with a $128.60 strike price are worth $40.79 per ton. Ob-
viously, white wheat prices have increased substantially, as reflected in the fu-
tures price, the options price, and the physical cost of FOB white wheat. Because
the futures price exceeds the strike price, the government exercises its right to
purchase futures at $128.60 per ton, because these contracts are now worth
$169.39 per ton. The government can then sell the futures contracts for a profit
of $40.79 and purchase the physical cash wheat for $166.36 per ton.
To evaluate the actual cost of the wheat purchase taking the options hedging
operation into consideration, we subtract the price of the option ($40.79 per
ton) from the wheat purchase price ($166.36 per ton) and add back the original
cost of the option ($4.78) and brokerage fee ($0.10), yielding a net purchase
price of $130.45 per ton. The use of options contracts enables the purchaser to
protect itself against upside price risk. If prices fall, the buyer of the option also
benefits. In this case, the government has reduced its import bill by $21.5 mil-
lion by hedging.
Table 5. The Impact of the Options Hedging Program on Wheat Import Payments
FOB price CIF pricea Import cost Release Illustrative government subsidy or tax"
(dollars per ton) (rupees per ton) (millions of rupees) price Rupees per ton Millions of rupees
Without With Without With Without With (rupees Without With Without With
Month hedging hedging hedging hedging hedging hedging per ton) hedging hedging hedging hedging
April 126.8 130.4 4,598 4,709 2,759 2,826 4,800 202 91 121 54
November 166.4 130.5 5,806 4,712 3,484 2,827 4,800 -1,006 88 -604 53
00 Total 6,243 5,653 -483 107
Average 146.6 130.5 5,202 4,711 3,121 2,826 4,800 -402 89 -241 54
Note: The government purchases 600,000 tons of wheat in April and 600,000 tons in November. FOB denotes free on board (used to value exports), and CIF
denotes cost, insurance, and freight (used to value imports).
a. FOB price is converted into a CIF price by adding a freight cost of $24 per ton, assuming no export subsidies and using an exchange rate of PRs3O.5 per dollar.
b. A negative value is a subsidy; a positive value is a tax.
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues), and authors' calculations.
Faruqee, Coleman, and Scott 285
Table 5 shows the impact of the options hedging program on government
subsidy payments. Over the year, the average FOB price with hedging is $130.50
per ton compared with $146.60 per ton without hedging, resulting in PRs590
million savings on imports. The policy of establishing a PRs4,800 per ton re-
lease price would cost the government PRs483 million without the options hedg-
ing, compared with PRs107 million in revenues with hedging.
Strategy III: Hedging with Swaps
A third hedging alternative is to use a commodity swap. Swaps were devel-
oped to manage relatively long-term risk and are generally available on the over-
the-counter market (that is, they are negotiated between parties rather than traded
on an exchange). Swaps are purely financial instruments in that no exchange of
physical goods takes place. This feature distinguishes swaps from futures and
options contracts, in which the parties can make or take delivery of the physical
(agricultural) commodity. (In practice, of course, only the net amounts change
hands.)
The hedger utilizes swaps to shift price risk to the investment community and
to manage the price risk of the commodity portfolio of the business. A swap
transaction accomplishes this by establishing three variables: the amount or vol-
ume of the swap, a fixed price level, and a variable price level. Fluctuations of
the variable price around the fixed price are used to establish a stream of pay-
ments to each party to the swap. A swap with two parties typically involves a
consumer of the commodity and a producer; a bank or other type of financial
institution acts as intermediary. The consumer pays the fixed price amount and
receives the variable price amount. The producer receives the fixed price amount
and pays the variable price amount.
The great advantage of swaps is that they afford great flexibility by decoupling
the hedging activity from the physical trading activities of an organization. Swaps
also enable an organization to manage price risk for relatively long periods of
time. Their major drawback is that they require cash flow and are very credit
intensive. Because swap transactions involve a high counterparty risk, banks
may require up-front cash collateral (in an escrow offshore account that could
be earning interest) to cover a predetermined level of risk exposure. The under-
developed market for swaps in the agricultural area presents another drawback;
to date most swaps of physical commodities have been in metals and petroleum.
To see how a swap would work, assume that the government wants to secure
a long-term price of wheat equal to $135 per ton. It enters into a swap agree-
ment with a bank such that the fixed price of the swap is $135 per ton and the
variable price used is the monthly average price of the nearby white wheat fu-
tures contract traded on the Minneapolis Grain Exchange. The amount is 100,000
tons per month. At the end of each month the price of white wheat on the ex-
change is averaged, and the fixed price of the swap ($135 per ton) is subtracted
from the variable price to determine the payment to be made to or received from
the government. Assume that prices average $127 per ton in the first month and
286 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
$145 per ton in the second month. The government pays the bank $8 per ton, or
$800,000, the first month and receives $10 per ton, or $1,000,000, the second
month. The cash flows from the swap transaction apply against the actual physical
market transactions the government undertakes in the white wheat market. Pre-
sumably in the first month the government purchases 100,000 tons of white
wheat at $8 per ton less than the fixed price; the next month the price in the
physical market is $10 per ton higher.
Applying the swap concept to wheat purchases during 1994 yields the results
shown in tables 6 and 7 and in figure 2. The example uses spot cash white wheat
values from 1994 and assumes a desired import price of $135 per ton (this price
constitutes the fixed price level).
The financial intermediary charges a 1 percent commission for the service of
arranging the swap. The variable price used as a reference is the average Minne-
apolis nearby futures price (column one in table 6). The government imports
100,000 tons of wheat per month, or 1.2 million tons for the entire 1994 year.
In table 6, the price paid in the actual physical cash market (fourth column) is
adjusted by the net payment to achieve a net (or effective) price close to the
target price of $135 per ton (fifth column). In fact, the average cash price paid
for all of 1994 in the actual physical cash market is $141 per ton.
Table 7 shows the impact of using the swap agreement on import costs. With-
out the swap mechanism the government would pay a total of PRs6.05 billion
Figure 2. Hedging Effectiveness Using Swaps
Rupees per ton
6,000
CIF price without hedging
5,500
5,000
/ _ / ~~~~~~Release price
4,500
CIF price with hedging
4,000
January March May July September November
Febrary April June August October December
Source, Table 7.
Faruqee, Coleman, and Scott 287
Table 6. Calculation of the Net Wheat Import Price for the Strategy
of Hedging with Swaps
(U.S. dollars per ton)
Average
Minneapolis Fixed target
Month (1994) futures price price Net paymentr Cash priceb Net price'
January 136.4 135.0 1.4 133.4 133.3
February 132.8 135.0 -2.2 130.4 134.0
March 128.6 135.0 -6.4 127.5 135.2
April 134.3 135.0 -0.7 126.8 128.8
May 137.4 135.0 2.4 137.8 136.7
June 133.6 135.0 -1.4 134.1 136.9
July 133.0 135.0 -2.0 132.3 135.6
August 142.4 135.0 7.4 127.9 121.8
September 161.7 135.0 26.7 147.0 121.6
October 173.9 135.0 38.9 169.0 131.5
November 168.3 135.0 33.3 166.4 134.4
December 167.3 135.0 32.3 164.2 133.3
Note: Values are the net free on board (FOB) wheat import price.
a. Average Minneapolis futures price minus fixed price of $135 per ton.
b. Price actually paid in the market.
c. Price actually paid in the market minus net payment plus 1 percent brokerage fee ($1.35 per ton).
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues),
and authors' calculations.
for wheat purchases in 1994; utilizing a swap mechanism the payment would be
only PRs5.71 billion. Assuming a release price of PRs4,800 per ton as in the
previous two examples, the wheat swap significantly reduces the variability of
subsidy payments and tax revenues. With the swap, the monthly payments and
revenues range from PRs10.8 million in subsidies to PRs35.8 million in taxes;
without the swap, payments and revenues range from PRs108.7 million in sub-
sidies to PRs20.2 million in taxes. More important, the swap arrangement al-
lows the government to avoid the net cost of PRs293.4 million in subsidy pay-
ments that would result if the government had not hedged.
In this example, one advantage of utilizing a swap rather than futures and
options is that the consuming entity (in this case the government) does not have
to worry about liquidity problems on the exchange or the mechanics and strat-
egy of executing futures and options contracts. Care must be taken, however, to
ensure that the variable price used has a strong relationship with the actual
physical cash market and cannot be manipulated.
V. CONCLUSIONS
This article has some important implications for future wheat policy in Paki-
stan. Domestic wheat prices have been largely isolated from world markets, and
the government has succeeded in reducing price fluctuations. However, the policy
does not remove risk from the economy as a whole but merely transfers the risk
from wheat market participants in the form of unstable prices to the govern-
Table 7. 'The Impact of the Swaps Futures Hedging Program on Wheat Import Payments
FOB price CIF pricea Import cost Release Illustrative government subsidy or taXb
(dollars per ton) (rupees per ton) (millions of rupees) price Rupees per ton Millions of rupees
Without With Without With Without With (rupees Without With Without With
Month hedging hedging hedging hedging hedging hedging per ton) hedging hedging hedging hedging
January 133.4 133.3 4,800 4,798 480.5 479.8 4,800 0 2 0.0 0.2
February 130.4 134.0 4,710 4,820 471.0 482.0 4,800 90 -20 9.0 -2.0
March 127.5 135.2 4,621 4,856 462.1 485.6 4,800 179 -56 17.9 -5.6
April 126.8 128.8 4,598 4,660 459.8 466.0 4,800 202 140 20.2 14.0
May 137.8 136.7 4,935 4,902 493.5 490.2 4,800 -135 -102 -13.5 -10.2
June 134.1 136.9 4,822 4,908 482.2 490.8 4,800 -22 -108 -2.2 -10.8
N.) July 132.3 135.6 4,767 4,868 476.7 486.8 4,800 33 -68 3.3 -6.8
Go August 127.9 121.8 4,632 4,446 463.2 444.6 4,800 168 354 16.8 35.4
September 147.0 121.6 5,215 4,442 521.5 444.2 4,800 -415 358 -41.5 35.8
October 169.0 131.5 5,887 4,743 588.7 474.3 4,800 -1,087 57 -108.7 5.7
November 166.4 134.4 5,806 4,831 580.6 483.1 4,800 -1,006 -31 -100.6 -3.1
December 164.2 133.3 5,741 4,797 574.1 479.7 4,800 -941 3 -94.1 0.3
Total 6,053.4 5,707.0 -293.4 53.0
Average 141.4 131.9 5,044 4,755.9 504.5 475.6 4,800 -244.5 44.1 -24.5 4.4
Note: The government purchases 100,000 tons of wheat each month. FOB denotes free on board (used to value exports), and CIF denotes cost, insurance, and
freight (used to value imports).
a. FOB price is converted into a CIF price by adding a freight cost of $24 per ton, assuming no export subsidies, and using an exchange rate of PRs3O.5 per dollar.
b. A negative value is a subsidy; a positive value is a tax.
Source: International Wheat Council (various issues), U.S. Department of Agriculture (various issues), and authors' calculations.
Faruqee, Coleman, and Scott 289
ment (and ultimately taxpayers throughout the economy) in the form of un-
stable subsidy payments. The current system provides farmers and millers with
little incentive to undertake risk management on their own behalf. The private
sector has little need to worry about fluctuating prices when making production
and investment decisions. In effect, the government has crowded out private
sector risk management activities.
Overall, the government is the entity most exposed to price variability. Given
the large number of relatively small wheat farmers and traders, market partici-
pants cannot pursue risk management strategies on their own. The current struc-
ture of risk distribution, whereby the government pools the risk of small pro-
ducers and traders, may therefore be appropriate. However, having assumed the
price risk, the government needs to manage it by taking advantage of mecha-
nisms to externalize the price risk or transfer it to other entities.
Commodity hedging could be a useful method of managing commodity price
risks as long as the market participants understand the mechanisms and the
government keeps regulatory, legal, and institutional barriers to a minimum.
The government and potential market participants must well understand the
nature of hedging, the various instruments available, the potential obstacles,
and practical considerations. In particular, commodity hedging using futures,
options, and swaps could significantly reduce the variability of the cost of im-
ports. The simulations of actual hedging strategies indicate that hedging would
reduce the variability of import costs, thereby facilitating the management of
public expenditures and planning. Other mechanisms for price stabilization gen-
erally cost more than hedging. If the government needs to borrow to finance
additional subsidies resulting from unforeseen increases in international wheat
prices, the cost of borrowing represents the cost of not hedging. Although hedg-
ing involves risks and costs, not hedging may be riskier and costlier. However,
commodity hedging operations, which involve simultaneous transactions in cash
and futures markets, can be complex and hence require specialized expertise. Of
the instruments evaluated, swaps could be more attractive than futures and op-
tions because they are easier to implement and financial intermediaries are avail-
able to facilitate the transactions.
In searching for alternatives to the current system, the government of Paki-
stan should consider hedging using futures, options, and swaps, as well as other
methods of price stabilization. The government has already considered develop-
ing an agricultural buffer fund (Afzal and others 1993). Another possible alter-
native would be to borrow and lend in international credit markets to cover
unexpected subsidy payments associated with fluctuating commodity prices. Such
self-insurance schemes differ from hedging in that they are ex post and require
action once unfavorable movements in commodity prices have occurred, while
hedging provides ex ante insurance against such price movements. Deaton (1992),
however, argues that implementation of self-insurance schemes may be prob-
lematic due to the time series properties of commodity prices. Commodity prices
tend to have persistent and large swings and a significant element of uncertainty.
290 THE WORLD BANK ECONOMIC REVIEW, VOL. Ii, NO. 2
Therefore, the government would have to borrow an uncertain amount for an
uncertain period of time. During periods of persistently low commodity prices,
Deaton argues, the amount required could be substantial. Also, when commod-
ity prices are low, borrowing countries are less creditworthy and therefore are
more risky to the lender. However, stabilization funds and buffer stock schemes
can be used in conjunction with hedging. For example, Claessens and Varangis
(1994) show how a stabilization fund can be significantly cheaper to operate if
hedging instruments are used to cover extreme movements in commodity prices,
thereby allowing a buffer fund to cover price movements within a narrow range
of prices.
The future configuration of the global wheat environment holds a special
significance for wheat import practices in Pakistan. Changes in this environment
could affect how Pakistan imports wheat, from whom it imports, and at what
price it imports. Transformation of the global environment could result from
changes in both the policy environment and the fundamental supply-demand
situation. From a policy standpoint, the general trend in global economies is to
reduce government spending and adopt more market-oriented policies. The con-
text for these changes was the GATT negotiations that led to the formation of the
World Trade Organization and a phased reduction of agricultural subsidies.
This reduction has already led to reforms in the Common Agriculture Policy of
the European Union, which were first instituted in 1992 and continued to be
implemented through at least 1996. These reforms have lowered guaranteed
wheat prices and have led to lower planted acreage, lower production, and lower
intervention stocks of wheat.
The volatility of prices and the absolute price level of wheat are likely to
increase relative to the level of the 1980s. As a result, the cost of imported wheat
will likely be higher for Pakistan than it has been in the past 10 years. Higher
world wheat prices in themselves could lead to lower subsidies for export wheat
in that as prices rise governments need to provide less in subsidies to make their
own wheat competitive in world markets. Most likely, the market changes we
describe here will produce a more amenable environment for hedging world
prices of wheat on U.S.-based futures exchanges. With U.S. prices less isolated
from global factors (in part because of lower subsidies), U.S. wheat futures prices
should be more highly correlated with world wheat prices than has been the case
in the past. This could mean that countries such as Pakistan should find hedging
price risk on U.S. exchanges a more viable option.
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Afzal, Mohammad, Abdus Salam, Mohammad Iftichas, Ahmed Khan, and Mohammad
Ashiq. 1993. "Support Pricing Structure in Pakistan." Pakistan Journal of Agricul-
tural Economics 2:68-97.
Faruqee, Coleman, and Scott 291
Akiyama, Takamasa, and Panayotis Varangis. 1991. "Price Stabilization of Raw Jute in
Bangladesh." Policy Research Working Paper 813. International Economics Depart-
ment, World Bank, Washington, D.C. Processed.
Claessens, Stijn, and Jonathan R. Coleman. 1993. "Hedging Commodity Price Risks in
Papua New Guinea." In Stijn Claessens and Ronald C. Duncan, eds., Managing Com-
modity Price Risk in Developing Countries. Baltimore, Md.: Johns Hopkins Univer-
sity Press.
Claessens, Stijn, and Ronald C. Duncan. 1993. "Overview." In Stijn Claessens and Ronald
C. Duncan, eds., Managing Commodity Price Risk in Developing Countries. Balti-
more, Md.: Johns Hopkins University Press.
Claessens, Stijn, and Panayotis Varangis. 1993. "Implementing Risk Management Strat-
egies in Costa Rica's Coffee Sector." In Stijn Claessens and Ronald C. Duncan, eds.,
Managing Commodity Price Risk in Developing Countries. Baltimore, Md.: Johns
Hopkins University Press.
. 1994. "Oil Price Instability, Hedging, and Oil Stabilization Fund. The Case of
Venezuela." Working Paper 1290. Policy Research Department, World Bank, Wash-
ington, D.C. Processed.
Coleman, Jonathan R., and Donald F. Larson. 1993. "Tariff-Based Stabilization of Com-
modity Prices in Venezuela." In Stijn Claessens and Ronald C. Duncan, eds., Manag-
ing Commodity Price Risk in Developing Countries. Baltimore, Md.: Johns Hopkins
University Press.
Deaton, A. S. 1992. "Commodity Prices, Stabilization, and Growth in Africa." Discus-
sion Paper 166. Research Program in Development Studies, Woodrow Wilson School,
Princeton University, Princeton, N.J. Processed.
Ederington, Louis H. 1979. "The Hedging Performance of the New Futures Markets."
Journal of Finance 34(1):157-70.
Faruqee, Rashid, and Jonathan R. Coleman. 1996. Managing Price Risk in the Pakistan
Wheat Markets. World Bank Discussion Paper 334. Washington, D.C.: World
Bank.
Gilbert, C. L. 1993. "Domestic Price Stabilization Schemes for Developing Countries."
In Stijn Claessens and Ronald C. Duncan, eds., Managing Commodity Price Risk in
Developing Countries. Baltimore, Md.: Johns Hopkins University Press.
International Wheat Council. Various issues. World Wheat Statistics. London.
Knudsen, Odin, and John Nash. 1990. "Domestic Price Stabilization Schemes in Devel-
oping Countries." Economic Development and Cultural Change 38(April):539-58.
Larson, Donald F. 1993. "Policies for Coping with Price Uncertainty for Mexican Wheat."
Policy Research Working Paper 1120. International Economics Department, World
Bank, Washington, D.C. Processed.
Massell, B. F. 1969. "Price Stabilization and Welfare." Quarterly Journal of Economics
83(May):285-98.
Myers, Robert. 1993. "Strategies for Managing Coffee Price Risks in Costa Rica." In
Stijn Claessens and Ronald C. Duncan, eds., Managing Commodity Price Risk in
Developing Countries. Baltimore, Md.: Johns Hopkins University Press.
Newbery, D. M. G., and Joseph E. Stiglitz. 1981. The Theory of Commodity Price Sta-
bilization. Oxford: Clarendon University Press.
Pakistan, Government of. 1995. Economic Survey 1994-95. Islamabad: Finance Divi-
sion, Economic Adviser's Wing.
292 THE WORLD BANK ECONOMIC REVIEW, VOL. 1X, NO. 2
Palaskas, Theo, and Panayotis Varangis. 1991. "Is There Excess Co-Movement of Pri-
mary Commodity Prices? A Co-Integration Test." Policy Research Working Paper
758. International Economics Department, World Bank, Washington, D.C. Processed.
Satyanarayan, Sudaka, Maurice E. Thigpen, and Panayotis Varangis. 1993. "Hedging
Cotton Price Risk in Francophone African Countries." Policy Research Working
Paper 1233. International Economics Department, World Bank, Washington, D.C.
Processed.
Sheales, T. C., and W. G. Tomek. 1987. "Hedging Australian Wheat Exports Using
Futures Markets." Journal of Futures Markets 7(5):519-33.
Thomas, Vinod. 1985. Linking Macroeconomic and Agricultural Policies for Adjust-
ment and Growth: The Colombian Experience. Baltimore, Md.: Johns Hopkins Uni-
versity Press.
U.S. Department of Agriculture. Various issues. Wheat: Situation and Outlook. Wash-
ington, D.C.: Economic Research Service.
Varangis, Panayotis. 1994. "Recent Developments in Agricultural Markets in Develop-
ing Countries." Paper presented at the Summit on Risk Management in American
Agriculture, Farm Foundation and Commodity Futures Trading Commission, No-
vember 29, Washington, D.C. Processed.
Varangis, Panayotis, and Donald F. Larson. 1996. "Dealing with Commodity Price
Uncertainty." Policy Research Working Paper 1667. International Economics De-
partment, World Bank, Washington, D.C. Processed.
Varangis, Panayotis, Maurice E. Thigpen, and Takamasa Akiyama. 1993. "Risk Man-
agement Prospects for Egyptian Cotton." Policy Research Working Paper 1077. In-
ternational Economics Department, World Bank, Washington, D.C. Processed.
World Bank. 1994. "Pakistan: Strategy for Sustainable Agricultural Growth." Gray Cover
Report 13092-Pak. South Asia Region 1, Washington, D.C. Processed.
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 293-325
Explaining Industrial Growth in Coastal China:
Economic Reforms . . . and What Else?
Ashoka Mody and Fang-Yi Wang
In the 1980s China experienced "an explosion of pent-up entrepreneurship" facili-
tated by wide-ranging, although often unorthodox, economic reforms. This article uses
data on the output of 23 industrial sectors in seven coastal regions (provinces and
counties) over the period 1985 to 1989 to study the correlates of growth. Although
industry-specific features-the degree of specialization and competition-bad some
influence on growth, much of the action came from region-specific influences and re-
gional spillovers. Regional influences included the open-door policies and special eco-
nomic zones that successfully attracted investments from overseas Chinese to particu-
lar locations. Existing regional strengths, especially high-quality human capital and
infrastructure, also contributed to growth. The results illuminate the interplay between
conditions conducive for growth-for example, the contribution of foreign expertise is
greatly enhanced by available human capital. China made judicious use of the advan-
tages of backwardness by targeting areas that were less developed and less encumbered
by the legacy of existing institutions, although it was fortunate in this regard that the
backward regions were in close proximity to Hong Kong and Taiwan (China). Impor-
tant also was the transmission of growth impulses across the provinces and counties,
possibly through prereform cadre and administrative networks.
In the 1980s China experienced an explosion of pent-up entrepreneurship facili-
tated by wide-ranging, although often unorthodox, economic reforms. Walker's
(1993) apt metaphor rightly focuses the spotlight on China's entrepreneurs who
include not just factory managers but also local government officials, especially
mayors of cities and counties. Growth in gross domestic product (GDP) jumped
from 6.4 percent a year between 1965 and 1980 to 10.1 percent between 1980
and 1989. From 1985 to 1989, the years on which we focus, the pace of eco-
nomic reforms was stepped up and performance was especially outstanding: GDP
grew at 11.5 percent a year, and industrial output, the principal engine of growth,
grew at a yearly rate of 14.4 percent. Moreover, factor productivity-which
made virtually no contribution to growth in the three decades before 1980-
Ashoka Mody is with the Cofinancing and Project Finance Department at the World Bank; Fang-Yi
Wang is with De Anza College, Cupertino, California, and was a consultant at the World Bank when this
research was conducted. The authors thank Elinor Berg, Michael Klein, Jenny Laniouw, Bart Verspagen,
and especially Edward Glaeser and Paul Romer for many helpful comments. The authors are grateful to
Shahid Yusuf for both his comments and his help with initial financing and organization of this project.
The project has been partly financed by the World Bank Research Committee (RPo 677-50).
� 1997 The International Bank for Reconstruction and Development / THE WORLD BANK
293
294 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
grew at an annual rate of 2.4 percent for state-owned enterprises and 4.6 per-
cent for collectively owned enterprises and accounted for 27 percent of growth
between 1980 and 1988 (Chow 1993; Jefferson, Rawski, and Zheng 1990). At
the same time China's share of world markets jumped dramatically between
1985 and 1989, particularly (but not exclusively) in light manufactured goods,
such as shoes, clothing, toys, and small electrical appliances.
Gains in industrial output were especially marked in the coastal region, where
growth during 1985-89 was significantly higher than that in other regions and
was also substantially above its own growth rate in the previous five years (see
table 1). Five coastal provinces (Fujian, Guangdong, Jiangsu, Shandong, and
Zhejiang) were at the center of the "miracle," registering growth rates of about
20 percent a year between 1985 and 1989. The performance of the three coastal
counties (Beijing, Shanghai, and Tianjin) was less impressive. Throughout China,
but especially in the coastal provinces, enterprises in the nonstate sector were
the star performers. In the Chinese context, the nonstate sector includes collec-
tive enterprises, which are typically owned by local governments-that is, by
governments below the provincial or county level-whose officials have been a
key source of domestic entrepreneurship (see Bateman and Mody 1991 and Oi
1992). Table 2 provides the share of industry by ownership for the eight coastal
provinces and counties.
To examine China's exceptional growth experience, this article attempts to
explain the variation in the growth of 23 industrial sectors in each of seven
provinces and counties along the east coast of China during the period 1985 to
1989. The unit of analysis is the growth rate of an industrial sector in a specific
Table 1. Growth in Industrial Output by Ownership in Coastal China,
1980-89
(average annual percent)
Total State-owned Collectively owned Othersa
Region 1980-85 1985-89 1980-85 1985-89 1980-85 1985-89 1980-85 1985-89
Coastal counties
Beijing 8.7 12.9 6.2 8.3 12.0 12.1 37.9 36.2
Tianjin 9.1 11.4 7.2 4.9 11.6 12.0 24.5 35.8
Shanghai 7.3 6.6 5.1 2.2 15.7 8.8 23.3 30.8
Coastal provinces
Jiangsu 15.1 17.3 8.6 9.9 19.3 18.4 27.5 26.6
Zhejiang 18.7 17.8 10.4 8.2 23.4 18.0 33.0 28.6
Fujian 13.5 20.5 9.2 11.4 13.8 16.6 33.7 37.5
Shandong 11.2 21.5 6.8 10.2 15.6 21.9 26.8 55.3
Guangdong 14.4 23.5 11.2 15.1 16.6 22.2 23.4 40.7
Total 12.0 16.5 7.4 8.0 17.6 18.0 27.7 34.6
China 11.3 14.4 7.9 8.7 17.9 19.0 - -
- Not available.
a. Includes mainly collectively owned enterprises below the township level, private enterprises,
partnerships, individuals, and joint ventures with foreigners.
Source: China, State Statistical Bureau (1990).
Mody and Wang 295
Table 2. The Share of Industry by Ownership in Coastal China, 1980, 1985,
and 1989
(percent)
State-owned Collectively owned Othersa
Region 1980 198S 1989 1980 1985 1989 1980 1985 1989
Coastal counties
Beijing 80.6 71.1 59.6 17.4 20.5 19.6 2.1 8.4 20.8
Tianjin 80.1 72.1 55.8 15.6 17.7 18.4 4.4 10.2 25.8
Shanghai 87.5 77.5 65.8 9.6 15.4 16.3 2.9 7.1 18.0
Coastal provinces
Jiangsu 57.2 40.5 30.6 33.4 40.6 42.4 9.4 18.9 27.1
Zhejiang 56.1 35.5 24.6 34.5 44.4 44.6 9.4 20.1 30.9
Fujian 71.2 56.8 40.1 21.1 21.8 18.7 7.7 21.4 41.2
Shandong 67.6 54.6 38.5 26.6 32.4 32.6 5.9 13.0 28.9
Guangdong 63.0 52.5 37.6 27.1 30.6 28.6 9.9 17.0 33.9
Total 71.4 55.7 40.8 22.5 29.9 31.4 6.1 14.3 27.8
a. Includes mainly collectively owned enterprises below the township level, private enterprises,
partnerships, individuals, and joint ventures with foreigners.
Source: China, State Statistical Bureau (1990).
region in a specific year. Three sets of influences on the growth rate are exam-
ined:
* Industry-specific features: the degree of specialization and competition
* Regional growth factors: the availability of infrastructure, educational levels,
and direct foreign investment; also, the initial per capita income of the
province or county measures the extent of backwardness and hence the
catch-up potential
* Regional spillover effects: the relationship between growth in a region and
growth in other regions.
Certain distinguishing features of this analysis, as well as its limitations, are
worth noting. First, by comparing growth rates within a relatively homogeneous
region (the Chinese east coast), the study overcomes some concerns in interpret-
ing cross-country growth regressions, where it is difficult to control for widely
different economic, social, and political regimes.1 Second, studies of developing-
country growth focus principally on a country's GDP (Mankiw 1995 surveys that
literature); our focus on individual industrial sectors is likely to yield more reli-
able estimates. In this respect, we follow Glaeser and others (1992) and
Henderson, Kuncoro, and Turner (1995) who study industrial growth within
the United States. Third, we build on the analysis of Glaeser and others (1992)
by including the possibility of regional spillovers along with regional influences.
Fourth, although covering only a short time span of four years, we are able to
exploit the panel features of the data to examine factors influencing growth
1. Islam (1995) uses country dummies to control for country-specific features but finds them correlated
with the traditional explanatory variables.
296 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
within and across the provinces and counties. Finally, we also study whether
heavy and light industries have been subject to different growth impulses.
The main limitation of the study arises from the concern that the industrial
output data used may have built-in biases. We are reassured, however, by the
significant variation in growth rates across sectors, regions, and time, suggesting
that measured growth rates are not merely a reflection of some bureaucratic
data-recording process. Moreover, we conduct a number of sensitivity analyses
running regressions for different samples, checking for the presence of influen-
tial observations, and testing the robustness of important explanatory variables.
However, we have attempted to interpret the results conservatively, highlighting
the most quantitatively and statistically significant findings.
Section I decomposes output growth into time-dependent, regionwide, and
industry-specific components, as well as their interactions, to identify the proxi-
mate sources of growth. Section II describes the approach to studying the corre-
lates of growth used and our explanatory variables. Section III presents and
interprets our findings. Section IV summarizes our major findings and also draws
some lessons for other countries.
I. A DECOMPOSITION OF GROWTH OF OUTPUT
Did growth occur across the board or only in certain regions or industries?
Within regions or industries, did growth vary substantially from year to year?
Variance analysis allows us to quantitatively decompose output growth into
time, region, and industry-specific effects and their interactions. Identifying the
main sources of variance in the data through decomposition analysis helps in a
preliminary quantitative assessment of the different sources of growth. The find-
ing of significant time and regional differences in growth rates after controlling
for sectoral growth patterns also provides some reassurance that the industrial
output data are not being generated in a bureaucratically mechanical manner.
Growth in time period t, region r, and industry i, Gtri, is assumed to be the
additive result of main and interaction effects.
(1) Gtri = m + at + Pr +Ti + atr + bti + Cri + etri
where m is a constant, at, Pr, and xi are the main time, region, and industry
effects, respectively, atr, b,i, and cri are the second-order interaction terms be-
tween two main effects, and Etri is the interaction term for the three main effects.
Following Schankerman (1991), the variance of output growth can therefore
be expressed as:
(2) Var (Gtri) = Var (at) + Var (A) + Var (ti) + Var (a1r) + Var (b,i) + Var (cr) + Var (Etri)-
4.98 5.72 2.88 1.33 48.14 2.18 34.78
The numbers below the variables in equation 2 are the results derived by equat-
ing the expected values of the variance components with their observed values
(see the appendix for the derivation). Because this is a decomposition, the values
Mody and Wang 297
add up to 100 percent. The small variance of �ct-4.98 percent-implies that
during 1985 to 1989, time-varying factors had only a minor effect on growth.
Thus, although the overall pace of reforms accelerated, the effect was not felt
uniformly in all regions and industries.
Purely regional effects, Pr, were also small-5.72 percent-implying that across
years and industrial sectors, there was no consistent ranking of regional growth.
Together with their interaction, time, and region effects, Cxi, fr, and a,r explain
12 percent of the variation in growth. Thus reforms did not manifest themselves
primarily through general coastal expansion or through growth in specific coastal
provinces or counties.
Industry-specific factors, cr, were small as well, accounting for 2.88 percent of
the variation in growth. Hence, no industrial group grew uniformly rapidly or
slowly throughout the period. For example, the electronics and telecommunica-
tions sector grew only 5.6 percent in 1985-86, whereas in 1987-88, it rose a
remarkable 19.3 percent.
The dominant source of variation in the data comes from the interaction of
time and industry (bt1), which explains 48.14 percent of the total variance in
growth. Thus output growth rates for specific sectors varied from year to year,
but within a year they were strongly correlated across regions. This effect cap-
tures an industry-specific wave phenomenon evident from a visual examination
of the time pattern of sectoral growth rates for miscellaneous light industries in
table 3. The term wave is used here not to suggest any predictable sequence of
industries experiencing successive surges in growth, but only to indicate that
specific industries achieved high rates of growth across regions at the same time.
Different industries led in different years; some of the most labor-intensive sec-
tors, such as garments, achieved their biggest spurt only very late. The wave
phenomenon was not restricted to light industries. In 1985-86 rapid growth
was evident in leather products, pharmaceutical products, chemical fibers, and
metallic products in most of the coastal region. In 1986-87 electronics and chemi-
cals replaced leather and metallic products. In 1987-88 paper products, trans-
portation equipment, electronics, and pharmaceutical products expanded rap-
idly, only to lose their position to the apparel industry in 1988-89.
Such synchronization could be accounted for by shifts in buyers' preferences
for goods, industry-specific technological improvements rapidly transmitted along
the coast, or coordinated strategies among decisionmakers to promote growth
of specific industries at specific times. Also formal and informal interactions
among firms and labor turnover, particularly of highly skilled managers and
engineers, may have extended technology and skills learned in the open areas to
the rest of the coastal region (Ho and Huenemann 1984, p. 55). Another in-
triguing possibility is that decisionmakers (whether in the communist party or
industrial administration) maintained close ties that led to the rapid diffusion of
development strategies along the coast (see Yusuf 1993 and the literature he
cites). This network of decisionmakers could provide a grid for information
flows leading to replication of sectoral targeting strategies, among other things.
298 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 3. Growth in Output of Light Industries in Coastal China, 1985-89
(average annual percent)
Years and
industries Guangdong Fujian Jiangsu Zhejiang Shandong Beijing Tianjin Shanghai
1985-86
Apparel 27.0 20.6 11.8 1.1 12.7 -5.6 6.7 -14.7
Leather products 31.2 28.8 22.7 15.0 26.3 5.7 9.8 8.3
Wood products 3.1 9.3 13.1 17.5 5.0 -0.7 -7.4 0.5
Furniture 4.4 -1.9 11.2 11.0 18.6 3.3 -4.3 2.1
Paper products 15.8 11.8 18.8 15.8 18.0 8.8 5.4 2.4
Art products 14.9 39.5 -7.6 11.8 25.6 -36.7 -3.0 -16.6
Plastic 17.3 17.5 8.4 11.3 19.7 3.6 6.7 5.1
1986-87
Apparel 36.1 19.1 14.5 17.6 13.8 4.3 -0.1 11.3
Leather products 54.5 13.5 24.1 13.0 18.7 -2.3 -2.8 3.9
Wood products 15.3 16.2 24.1 1.1 16.3 -3.0 -15.7 -0.7
Furniture 26.1 12.6 19.8 14.6 21.0 13.1 0.4 5.8
Paper products 28.1 16.8 27.8 20.2 20.8 10.2 6.2 15.5
Art products 20.8 12.1 21.1 20.4 41.1 14.4 8.6 2.1
Plastic 27.3 13.4 18.7 17.6 21.6 -8.7 6.5 4.0
1987-88
Apparel 15.4 21.0 12.7 17.4 20.1 13.2 -8.9 7.1
Leather products 17.2 17.0 9.1 9.1 9.7 -22.2 -3.4 -2.2
Wood products 25.1 20.0 -7.1 8.8 16.3 -7.6 -27.2 -16.0
Furniture 10.5 6.0 6.6 14.2 17.5 1.9 -2.7 0.5
Paper products 95.7 56.0 79.5 64.3 60.7 155.6 66.4 103.2
Art products 0.8 21.2 13.2 18.5 28.7 18.0 11.9 6.2
Plastic 14.8 18.4 6.8 16.7 26.5 -1.4 -3.2 -4.6
1988-89
Apparel 28.2 49.8 13.1 16.9 15.2 12.4 14.5 7.3
Leather products 26.6 5.0 -1.8 10.7 10.4 -3.0 -4.0 -3.5
Wood products 5.1 13.4 -5.5 5.0 23.6 -10.0 -12.0 -3.7
Furniture 6.9 1.7 -11.1 -7.0 10.3 3.8 -6.6 -1.3
Paper products 5.7 11.1 -0.4 4.7 10.4 2.5 -5.2 -2.3
Art products 10.7 8.7 11.5 20.8 27.0 -3.6 8.9 18.4
Plastic 15.5 14.2 2.6 4.0 17.5 0.8 -8.9 0.9
Note: The underlined values indicate an industry-specific wave phenomenon evident from a visual
examination of the time pattern of sectoral growth rates for miscellaneous light industries. Although
output growth rates for specific sectors varied from year to year, certain sectors achieved high rates of
growth across regions.
Source: China, State Statistical Bureau, China Statistical Yearbook (various years).
In a field study of major decisionmakers, Oi (1995) found considerable support
for this hypothesis.
But the synchronization could also reflect data limitations. If price deflators
for particular industrial groups are biased in different directions in different
years, then high synchronization would be built into the data, making it appear
that certain sectors grew more rapidly than others in a given year when, in fact,
they did not. Such biases in industrial price deflators would exaggerate the ex-
tent of synchronization. Here, in this variance decomposition, the limited objec-
Mody and Wang 299
tive is to describe the variance in the data, whether it arises from data artifacts
or from interesting economic forces. In the regression results reported below,
however, interpretation is more critical. A conditioning term-growth of the
same industry outside the region-could be viewed as a control variable for this
deficiency in the quality of data. But in this case, we would have to downplay its
interpretation as a measure of regional spillovers. The continued plausibility of
the regional spillover hypothesis arises from the differences in degree of the cross-
regional synchronization for heavy and light industries and the findings of a
field study (Oi 1995). Such synchronization is also evident in the study by Glaeser
and others (1992).
The remaining 34.78 percent of the variance in output growth is attributable
to the third-order interaction between time, region, and industry (t,i). We inter-
pret this as a regional effect conditional on time-varying industry-specific fac-
tors. Although the changing identity of high-growth sectors is a major source of
variation in growth, this third-order interaction indicates that growth in an in-
dustrial sector during a particular year is not uniform in every region. Regional
differences in initial conditions, human capital endowment, and infrastructure
availability, among others, may cause industries in some regions to grow faster
than those in others. Thus even though the own effect of regional differences
(13r), and its interactions with time (atr) or industry (cri), do not explain much of
the variation in growth, after adjusting for time-varying and industry-specific
factors, significant regional effects remain.
Unconditional time, industry, and regional factors do not carry significant
explanatory power for variation in growth. Instead, about half the variation in
growth during 1985-89 is associated with time-varying sectoral growth differ-
ences (bti). Regional effects, by contrast, emerge only after controlling for the
time and industry effects.
II. INVESTIGATING THE CORRELATES OF GROWTH
The annual growth rate of output in an industrial sector in a given region is
our dependent variable. Various industry-specific, regionwide, and cross-regional
factors are the independent variables whose correlation with growth we seek to
examine. The goal here is not to test any specific model of growth but to de-
scribe its most robust partial correlates.
Following Glaeser and others (1992), we focus on growth itself rather than
on increases in productivity. (Their dependent variable was employment growth;
we use output growth.) Although it may be more appropriate to use increases in
labor or total factor productivity as the dependent variable, this is not possible
in our case because consistent labor force data by industrial sector are not con-
sistently available. We find, however, that the data on growth of industrial out-
put are so rich that considerable insights can be obtained even in the absence of
information on labor and capital inputs. Indeed, if we believe that enterprise-
level decisions to acquire or invest in labor or capital inputs are influenced by
300 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Figure 1. A FrameworkforIndustrial Growth in Coastal China
Region A Region B
Infrastructure, congestion, Infrastructure, congestion,
education, foreign investment, education, foreign investment,
initial income initial income
< ~~~~~~~~spillovers
Regional
spillovers
available knowledge, infrastructure, human capital, and industrial organization,
then not only productivity but also a considerable amount of output growth can
be attributed to these factors.
The basic framework of analysis is described in figure 1 for two regions (A
and B) and three industrial sectors (1, 2, and 3). The most proximate influences
on an industry's growth rate are industry-specific variables that condition the
extent of knowledge flows within an industry and the incentives to invest in the
development and appropriation of knowledge. The variables we use-the de-
gree of industry specialization and entrepreneurship-are the same as those used
by Glaeser and others (1992). We add a set of regional variables to the regres-
sions. The assumption is that having controlled for industry-specific characteris-
tics, the effect of region-specific variables (such as infrastructure) will be simi-
larly felt by all industries (for a similar assumption, see Waldmann and De Long
1990, who analyze growth across industries in different countries, and Stock-
man 1988). Finally, if there are regional spillovers, an industry in a particular
region will be influenced by growth in other regions.
Table 4 provides descriptive statistics for the seven provinces and counties
used in the regression analysis below for 1986-89. The regression analysis, un-
Mody and Wang 301
Table 4. Descriptive Statistics for the Data on Industry in Coastal China,
1986-89
Standard Minimum Maximum
Variable Mean deviation value value
Industry specific
Specialization indexa, S 1.028 0.466 0.000 4.347
Entrepreneurship indexb, E 1.203 0.839 0.018 5.292
Region specific
Secondary school enrollment rate 0.444 0.125 0.300 0.731
Accumulated foreign direct investment
per person (thousands of dollars) 0.015 0.019 0.000 0.067
Roads' (kilometers) 0.434 0.174 0.262 0.882
Interaction between roads and congestion
(population per square kilometer) 389.543 532.696 75.648 1,797.400
Telephones per 1,000 persons 13.462 9.473 4.712 36.194
GDP per capita (current yuans) 1,845.360 1,188.910 708.000 5,161.000
Regional spillover
Growth in industry in region, G (percent) 0.112 0.151 -0.423 1.032
Growth in industry outside region (percent) 0.104 0.119 -0.195 0.886
Note: Statistics are for beginning-of-period values, based on annual data for 1985-86 to 1988-89
for seven provinces and counties: Fujian, Guangdong,-Jiangsu, Shandong, Shanghai, Tianjin, and Zhejiang.
a. The specialization index for industry i in region r, at time t is Si,, = (output in industry i / total
output) for region r / (output in industry i / total output) for all regions.
b. The entrepreneurship index for industry i, in region r, at time t is Esr, = [(number of firms/total
output) for industry i in region r] / [(number of firms / total output) for industry i in all regions].
c. Length of road routes (kilometers) is normalized by area (square kilometers).
Source: Authors' calculations based on data from China, State Statistical Bureau (1990); China
Statistical Yearbook (various years); Hayase and Kawamata (1990); Statistical Yearbook of Fujian (various
years); Statistical Yearbook of Guangdong (various years).
like the variance decomposition, is based on data only for seven regions (the five
coastal provinces and two counties-Shanghai and Tianjin); foreign investment
data for Beijing are not available.
Industry-Specific Variables
An important structural feature of an industry is its degree of regional spe-
cialization. Presumably greater specialization is good if the relevant knowledge
is best acquired within the industry, but deleterious when diverse skills and in-
formation from other industries are important. Another important industry char-
acteristic is the degree of entrepreneurship and competition that can spur invest-
ment, although too much competition can lead to diminished investible surpluses.
With the data at hand, the existence of entrepreneurship or competition is in-
ferred only indirectly from the size of firms in industry i in region r relative to its
average in all seven regions.
As in Glaeser and others (1992), we calculate the following measures of spe-
cialization, Sirt, and entrepreneurship, Eirt.
Sirt = [(output in industry i / total output) for region r] / [(output in industry i
/ total output) for all regions]
302 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Eirt = [(number of firms / total output) for industry i in region r] / [(number of
firms / total output) for industry i in all regions].
The time subscript indicates that these measures are different for each year.
Si,, is the ratio of the share of industry i in region r to its average share across
the seven regions. S greater than 1 implies that the industry commands a larger
share of the region's output than the average share that industry enjoys in the
seven regions. We interpret a rising Sirt for a region-industry as an indication of
increasing specialization of that industry in that region. As S increases, knowl-
edge flows will be increasingly restricted to sources within that industry. Learn-
ing from other industrial sectors is likely to be greater when S is low. Jacobs
(1969), who contends that exchanges of information between different sectors
are more productive than exchanges within a sector, predicts that high-S indus-
tries will grow more slowly than low-S industries. Porter (1990) makes the op-
posite prediction. Interpretations other than knowledge flows can also be used
to explain the link between S and growth. For example, suitability of regional
factor endowments to the sector may contribute to a positive relationship be-
tween S and growth.
We interpret Eirt as a possible measure of entrepreneurial strength, but it could
also measure the degree of competition. If small firms are synonymous with
more competition, and more firms imply the existence of entrepreneurship, then
the interpretations of the variable will be indistinguishable. A high E for a region-
industry implies more firms for a given output in that region relative to the
average number of firms divided by output in the industry across all seven re-
gions. A high E could, therefore, be interpreted as more entrepreneurship or
greater competition. In any case a high E indicates smaller average firm size. In
terms of effects on growth, an unresolved debate centers around whether com-
petition or monopoly is more effective in encouraging inrnovation. Similarly, the
effect of size on growth remains controversial.
Region-Specific Variables
The region-specific variables are the beginning-of-year values for GDP per capita,
secondary school enrollment rate, foreign direct investment per person, road
network (the length of roads in the region divided by the region's area), telecom-
munications availability (telephone lines per capita), and congestion (popula-
tion density).
These regional growth-related factors not only are important in their own
right but also can have important spillover effects. Lucas (1988) notes that
human capital is twice blessed: first because it is inherently productive and
second because interactions among well-educated people further increase effi-
ciency. Shleifer (1990) suggests that good infrastructure provides the focal point
for the development of agglomerations, which in turn create the environment
for knowledge spillovers. Foreign investors bring knowledge on international
best practices in production technologies but also provide links to interna-
tional markets.
Mody and Wang 303
Regional Spillovers
A control variable, growth of the industry outside the province or county
(that is, growth of the industry in the other six counties and provinces), is also
included in the regressions. By construction, it is a time-varying region- and
industry-specific variable. However, as a practical matter, because it captures
across-the-board industrial growth, it is close to being a time-varying industry-
specific factor with little regional variation in a given time period. In view of the
discussion above that cross-regional correlation may be built into the data on
account of biases in price indexes, the interpretation of the coefficients on this
variable requires some care.
Before reporting our growth regression results, it is important to note the
difficulty of identifying causality in these regressions (Mankiw 1995 has an ex-
tended discussion). Our goal, as a first step, is to identify the bundle of influ-
ences that coexist through a growth process. However, certain techniques are
used that bear on the issue of endogeneity. The potential endogeneity of the
industry-specific variables, competition and specialization, is addressed by using
their lagged (beginning-of-period) values. The endogeneity of regional variables
poses a less serious problem. First, beginning-of-period values are used in the
regression, and second, our dependent variable is not growth in a region, but
rather growth of a specific industry within the region. Infrastructure, education,
and flows of foreign investment are likely to be influenced by overall regional
growth rather than by the expansion of a particular industry.
III. CORRELATES OF GROWTH
The regressions are run for 23 industrial sectors: food processing, textiles,
apparel, leather products, wood products, furniture, paper products, art prod-
ucts, plastic products, electronics, petroleum, chemicals, pharmaceutical prod-
ucts, chemical fibers, electricity, rubber products, nonmetal products, ferrous
products, nonferrous products, metallic products, transportation equipment,
electrical machinery, and other machinery. The eight regions for which indus-
trial output data are available include the three coastal counties-Beijing, Shang-
hai, and Tianjin-and five coastal provinces-Fujian, Guangdong, Jiangsu,
Shandong, and Zhejiang. However, Beijing, although considered in our output
decomposition analysis, cannot be included in the regression analysis because of
incomplete availability of explanatory variables.
Limited degrees of freedom prevent running separate regressions for each in-
dustrial sector. This poses a problem because growth has not been uniform across
sectors, raising the possibility that independent variables have very different in-
fluences on the different sectors. Because a significant feature of China's recent
growth has been the rapidly growing share of light industrial sectors, our inter-
mediate solution to this problem is to group sectors into light and heavy indus-
tries and to reestimate the basic regression. Although the coefficients show inter-
304 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
esting differences in magnitudes, we find the basic results unchanged and hence
focus on the pooled results, noting the differences that do arise when light and
heavy industries are considered separately.
Pooling Time Series and Cross-Sectional Data
Because we are pooling time series and cross-sectional data, we first test for
serial correlation in growth rates. If observations in growth rates in four adjoin-
ing years in a specific industry in a particular region are not independent of one
another, the standard errors will be biased and the inferences drawn will be
stronger than warranted. Recall that we have three dimensions in our data: in-
dustry, region, and year. Our interest is in the correlation over time. We can,
therefore, sort the data by region, then by industry, and finally by year; alterna-
tively, we can sort by industry, followed by region and year. Both procedures
ensure that the adjoining observations are for four successive years. In either
case, the Durbin-Watson statistic for the base regression is 1.87, implying that
the autocorrelation problem is not serious.
The base regression is estimated with time and industry dummies, but with-
out regional dummies. When dummy variables for regions are added to the re-
gression, we, in effect, remove from the data the variation due to differences in
the levels of variables across regions. The coefficients thus obtained are weighted
averages of within-region relationships, which are sometimes described as short-
run effects. When region dummies are not included, we are able to compare
across regions. Because interregional differences occur over a longer period of
time than do variations within a region, dropping regional dummies, as we do in
our principal regressions, captures the long-run effects. We also report the more
interesting short-run estimates.
The time, region, and industry dummies are not reported in the tables that
present our regression results. But the main patterns of the results for the time
and regional dummy variables are worth noting. Table 5 presents the results for
the time dummies. The size of the coefficients for the time dummies in the first
column in table 5 shows an upward trend through the period under consider-
ation, and the coefficients are significantly different from the constant term for
the base year. However, excluding the time dummies only reduces the R2 mar-
ginally, indicating their limited explanatory power. The statistical significance
of the time dummies is sensitive to whether the observations are weighted by the
population of the region; when observations are not weighted by population,
the influence of the two counties, Shanghai and Tianjin, increases and the time
dummies are not significantly different from 0, suggesting that the time effects
were felt primarily in the five coastal provinces.
The inference we draw from the increasing coefficients on the time dummies,
consistent with the variance decomposition analysis, is that there were indepen-
dent, though limited, time effects during this period. In other words, the gradual
move toward a more market-oriented economy appears to have had some secu-
lar effects independent of the region and industrial sector. The second column in
Mody and Wang 305
Table 5. Time Dummy Coefficients, 1986-88
Base Regression without regional
Year regressiona explanatory variablesb
1986 -0.08 0.02
(-3.2) (1.80)
1987 -0.05 0.05
(-1.89) (4.15)
1988 -0.04 0.03
(-1.97) (2.41)
Note: Results are relative to the base year 1989. The t-statistics are in parentheses.
a. Overall regression results for the base case are given in column 4 in table 7.
b. Overall regression results are not reported for the regression without regional explanatory variables.
It is the model reported in column 4 in table 7 excluding the following variables: secondary school
enrollment, foreign direct investment, roads, population density, telephones, and GDP per capita.
Source: Authors' calculations.
table 5 gives the results for the time dummies after dropping the region-specific
variables (secondary school enrollment, foreign direct investment, roads, popu-
lation density, telephones, and GDP per capita). Again, the estimated coefficients
on the time dummies are statistically different from the base year.
Table 6 presents the coefficients for the regional dummy variables, using Shang-
hai as the base for comparison. The coefficients in the first column are positive
but not significantly different from 0 (even at the 10 percent level of confidence).
The second column gives the results when we drop the region-specific variables
from the regression. The pattern of regional dummies in the second column
mirrors more closely the statistics of regional growth presented in table 1, with
the regional coefficient higher for Guangdong than for Fujian, followed by
Jiangsu, Zhejiang, and Tianjin. The exception in the regional order of growth is
Table 6. Regional Dummy Coefficients, 1986-89
Base Regression without regional
Region regressiona explanatory variablesb
Guangdong 1.8 0.11
(1.1) (6.4)
Fujian 1.8 0.08
(1.0) (3.8)
Jiangsu 1.6 0.07
(1.1) (3.7)
Zhejiang 2.0 0.05
(1.2) (2.2)
Shandong 1.7 0.12
(1.1) (7.4)
Tianjin 2.4 -0.02
(1.8) (-0.6)
Note: Results are relative to the base region Shanghai. The t-statistics are in parentheses.
a. Overall regression results for the base case are given in column 1 in table 7.
b. Overall regression results are not reported for the regression without regional explanatory variables.
It is the model reported in column 1 in table 7 excluding the following variables: secondary school
enrollment, foreign direct investment, roads, population density, telephones, and GDP per capita.
Source: Authors' calculations.
306 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Shandong, whose coefficient indicates a higher growth rate than Guangdong's,
although the F-test shows that the two coefficients are not significantly different
from each other. These results give some confidence that the regional variables,
such as foreign investment (and accompanying know-how), domestic investment
(especially in infrastructure), human capital, and initial per capita income levels
are good explanatory variables for differences in regional growth. Industry dum-
mies show no interesting pattern, and including or excluding them makes little
difference to the results.
The Principal Results
Because results in this type of analysis are sensitive to the variables included
(Levine and Renelt 1992), table 7 reports those results that appear robust to vari-
ous specifications based on sensitivity tests (described at the end of this section).
INDUSTRY-SPECIFIC VARIABLES. After controlling for other variables, industrial
specialization has a largely negative effect on growth. This result suggests that
the flow of knowledge across industries is more conducive to growth than is the
flow within an industrial sector (Jacobs 1969). Less-specialized industrial sectors
gain from knowledge spillovers from other sectors. The short- and long-run
effects are not very different (table 7). Recall that at S equal to 1, the output
share of industry i in region r equals the average output share of industry i in the
seven regions. A decline in S to 0.9 increases the growth rate 0.5 percentage
points (for example, from 6 to 6.5 percent). However, the relationship between
industrial specialization and growth does not appear to be linear. Beyond S equal
to about 2, specialization enhances growth.2 The evidence, therefore, is also
consistent with Porter's (1990) hypothesis on the benefits of knowledge flows
within the same industry, although the degree of specialization must be large
enough. We report below that specialization promotes growth in the heavy
industrial sectors.
The statistical significance of the coefficients for E and E2 is weak. The gen-
eral thrust of the results is similar across various specifications and hence worth
noting: increasing relative firm size has a deleterious effect on growth. The val-
ues of the coefficients, however, also suggest that when the average size of firms
in an industrial sector in a particular region is smaller than a third of the average
firm size in all regions, growth in that region-industry suffers (possibly through
excessive competition and/or diminished investible surpluses).
FOREIGN DIRECT INVESTMENT. We begin the discussion of the regional influences
with the role of foreign investment. A key element of economic reform in China
has been the open door to foreign investment. Although triggered by government
policy, growth in foreign investment has taken on a life of its own, reaching
close to $20 billion in 1993. Many overseas Chinese have invested large amounts
2. The nonlinear specification includes the squared term. The relation is summarized with the elasticity
at the mean values of the variables (see table 4).
Mody and Wang 307
Table 7. Determinants of Industrial Growth in Coastal China, 1986-89
Short run Long run
Variable 1 2 3 4 5 6
Industry specific
Specialization index, S -0.061 -0.060 -0.060 -0.061 -0.061 -0.057
(-2.477) (-2.462) (-2.416) (-2.470) (-2.501) (-2.326)
Specialization index 0.014 0.014 0.014 0.014 0.015 0.013
squared, S2 (1.899) (1.864) (1.855) (1.902) (2.015) (1.788)
Entrepreneurship index, E 0.035 0.038 0.034 0.033 0.031 0.042
(1.683) (1.863) (1.638) (1.714) (1.646) (2.199)
Entrepreneurship index -0.006 -0.007 -0.005 -0.006 -0.005 -0.007
squared, E2 (-1.267) (-1.432) (-1.188) (-1.244) (-1.117) (-1.507)
Region specific
Secondary school -3.617 -1.449 0.779 1.373 1.121
enrollment rate (-3.496) (-1.366) (2.962) (2.371) (2.219)
Secondary school -0.663 -0.281
enrollment rate squared (-1.151) (-0.552)
Foreign direct investment 7.815 4.130 6.068 1.628 1.655 1.946
(4.360) (2.822) (3.502) (4.747) (4.816) (6.199)
Roads 6.647 5.089 -0.489 -0.587
(3.087) (2.393) (-2.657) (-2.896)
Roads squared -13.569 -10.532 0.688 0.832
(-3.335) (-2.625) (3.530) (3.593)
Interaction between roads 0.006 0.005
and congestion (3.302) (2.803)
Telephones -0.015 -0.017
(-1.513) (-4.248)
Telephones squared 0.0005 0.0003
(2.169) (3.078)
GDP per capita -0.0006 -0.0003 -0.0006 -0.0002 -0.0002 -0.0001
(-5.003) (-3.548) (-3.897) (-5.005) (-4.969) (-2.136)
Regional spillover
Growth in industry 0.785 0.782 0.783 0.777 0.778 0.780
outside region (20.033) (19.776) (19.923) (19.492) (19.519) (19.675)
Regional dummy variables
included? Yes Yes Yes No No No
Adjusted R2 0.617 0.610 0.615 0.604 0.604 0.609
Number of observations 640 640 640 640 640 640
Note: The dependent variable is growth of industry i in region r at time t (G.,r). All regressions
include time and industry dummy variables and observations are weighted by regional population.
Short-run regressions report within-region relationships; long-run regressions drop regional dummies
and report interregional relationships. The t-statistics are in parentheses. See table 4 for more complete
definitions of variables and descriptive statistics.
Source: Authors' calculations.
of capital and know-how, despite what, by Western standards, would be con-
sidered a great deal of uncertainty regarding property rights and enforcement of
contractual obligations (see Yusuf 1993).
Our results show that foreign direct investment has a strong impact on growth,
particularly in the short run (column 1 in table 7). The short-run elasticity of
growth with respect to foreign direct investment, calculated at the mean value of
308 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
the foreign direct investment variable, is 0.10, indicating that a 10 percent in-
crease in foreign investment can raise the growth rate 1 percent. However, the
apparent effect of foreign investment is influenced by trends in secondary school
enrollment rates, which, as noted below, fell during this period of rapid growth.
Hence human capital is seen to have a perverse effect on growth in the short run
(see column 1, table 7). Because a change in school enrollment rates is not a
good measure of change in the stock of human capital, the perverse effect is
overstated, and to that extent, the positive effect of foreign investment is prob-
ably exaggerated in the short-run estimates.
When the secondary school enrollment rate is dropped, the coefficient for
foreign investment falls by about half (see column 2, table 7). If we assume that
there was little change in human capital within any region during the period
under consideration, then the new estimate for foreign investment is closer to
being right, and hence the elasticity, of the growth rate with respect to foreign
investment is closer to 0.06. When the secondary enrollment rates are dropped
from the equation, the coefficient on foreign investment declines, but other coef-
ficients remain essentially unchanged (see column 2, table 7). The effect of for-
eign investment declines in the long run (and hence is a less potent source of
differences in growth between regions), but still remains statistically significant
and quantitatively important. The foreign investment coefficient decreases from
about 4 to about 2, and the growth elasticity falls from 0.06 to 0.03.
One interpretation of these results is that in the short run, foreign investment
is the most mobile factor and hence is a dominant driver of growth. In the longer
run, such variables as education and infrastructure respond to increased de-
mand for complementary assets, and the contribution of foreign investment
declines. There is also a complementary relationship between domestic human
capital formation and foreign investment flows, as discussed below.
HUMAN CAPITAL: EDUCATION AND FOREIGN KNOWLEDGE. Measurement of the
stock of knowledge available for productive use is a complex task even under
normal conditions and is especially difficult in a dynamic situation when
knowledge from many different sources is being utilized. Traditionally, second-
ary school enrollment rates have been used as proxies for the domestic stock
of knowledge, or domestic human capital, and serve well as long-run
approximations. Using data for the only year available-1987-we compare
enrollment rates with the more appropriate proxy, average years of schooling in
the labor force, and find a very high correlation coefficient (0.965, significant at
99 percent) between the two indicators.3 If this finding applies to the whole
3. The labor force includes population in the age group 15 to 54. The average length of education is
calculated as (16U + 12H + 9M + 6E + 01) / T, where U, H, M, and E are the number of persons with
university, high school, middle high, and elementary school education, respectively. I stands for illiterate.
T is the total population in the working age group. The relevant data were obtained from the 1987
population census. Data are available for seven provinces and counties in the coastal region-Beijing,
Fujian, Guangdong, Jiangsu, Shandong, Tianjin, and Zhejiang.
Mody and Wang 309
period between 1985 and 1989, then secondary school enrollment is a good
surrogate for at least the part of human capital endowment due to years of
formal education, and our long-run estimates can be considered reasonably
reliable.
However, short-run changes in human capital are more difficult to measure.
The extensive reforms that began in 1984 were accompanied by an actual fall in
school enrollment rates in most of the coastal provinces and counties. This is not
altogether surprising during a period of rapid growth accompanied by increases
in the demand for labor. Many of the new entrants to the labor force were young
women who probably dropped out of school to take up newly available jobs.
Over the short period under consideration, the stock of domestic human capital
is unlikely to have changed as a consequence of such labor force responses, al-
though unless the trend is reversed, human capital will deplete over time.
The short-run, or within, estimate shows a negative coefficient for secondary
education (first column in table 7), reflecting the cyclical shift out of education
described above. The finding tells us little about the relationship between do-
mestic human capital and growth in the short run, because, as noted, changes in
secondary enrollment rates greatly overstate the depletion of human capital. In
the long run, that is, when the comparison is across regions, education has the
expected positive effect on growth. However, returns to secondary education
diminish beyond a point.4 Similar results have been obtained for cross-country
regressions; see Pritchett (1996) for a recent review. The coefficients for the
education variables in column 5 show that when enrollment increases from 30
to 35 percent (that is, approximately from the Fujian enrollment rate to the
Guangdong enrollment rate), growth rises 5 percentage points. However, when
enrollment increases from 55 to 60 percent, the increase in growth is only 3
percentage points. Thus Tianjin gets a smaller effect from raising its enrollment
rate than does Fujian; Shanghai, with an enrollment rate in the mid-60 percent
range, gains even less.
Education becomes even more effective when it is associated with foreign
knowledge. Column 1 in table 8 shows that the interaction between school en-
rollment rates and foreign investment is significantly positive, suggesting mutual
reinforcement between domestic human capital and foreign knowledge that ac-
companies the investment. Also, the coefficient on foreign investment becomes
negative when the interaction term is introduced, implying that much of the
power of foreign knowledge may come through the local base of human capital.
Perhaps exposure to foreign knowledge breaks the isolation of the local economy
and brings experience-based practices that are rarely available in textbooks and
are best communicated in a hands-on manner in a production setting (Romer
1993).
4. The coefficient of the square of secondary enrollment rate in column S of table 7 is negative but
not statistically significant; however, we find that this result is sensitive to the specification and, in
certain cases, the squared term is statistically significant. Thus we believe that the nonlinearity needs to
be taken seriously.
310 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 8. The Interaction of Foreign Investment with Infrastructure
and Education in Explaining Growth in Coastal China, 1986-89
Variable 1 2 3
Industry specific
Specialization index, S -0.061 -0.061 -0.056
(-2.480) (-2.517) (-2.309)
Specialization index squared, S2 0.015 0.016 0.013
(1.976) (1.999) (1.764)
Entrepreneurship index, E 0.033 0.033 0.042
(1.757) (1.737) (2.162)
Entrepreneurship index squared, E2 -0.005 -0.005 -0.007
(-1.209) (-1.199) (-1.508)
Region specific
Secondary school enrollment rate 1.021 1.076 0.879
(3.546) (3.619) (4.229)
Interaction between secondary school enrollment rate 8.631
and foreign direct investment (2.041)
Foreign direct investment -1.373 0.241 1.660
(-0.910) (0.328) (2.688)
Roads -0.152 -0.118
(-0.617) (-0.465)
Roads squared 0.325 0.249
(1.233) (0.878)
Interaction between roads and foreign direct investment 4.608
(2.121)
Telephones -0.016
(-3.961)
Telephones squared 0.0003
(1.798)
Interaction between telephones and foreign direct 0.023
investment (0.494)
GDP per capita -0.0002 -0.0002 -0.0001
(-5.374) (-5.448) (-2.413)
Regional spillover
Growth in industry outside region 0.779 0.779 0.779
(19.591) (19.594) (19.671)
Adjusted R2 0.606 0.606 0.609
Number of observations 640 640 640
Note: The dependent variable is growth of industry i in region r at time t (Gi,,). All regressions
include-time and industry dummy variables but no regional dummy variables and observations are
weighted by regional population. t-statistics are in parentheses. See table 4 for more complete definitions
of variables and descriptive statistics.
Source: Authors' calculations.
INFRASTRUCTURE. Good infrastructure not only facilitates the flow of
information but also provides the focal point for the development of
agglomerations (Shleifer 1990). We consider two types of infrastructure: roads
and telecommunications. Roads represent the traditional infrastructure, and their
stock has grown only slowly to date. Phone lines, in contrast, have grown rapidly
to meet the needs of the international trading community-much, possibly all,
of the new telecommunications investment uses modern digital technology.
Mody and Wang 311
The results show that a network of roads has a positive effect on growth but
is subject to diminishing returns in the short run (column 1, table 7), possibly
reflecting indivisibilities in infrastructure investment (Weitzman 1970). Roads
are more productive in high-density areas (as reflected in the positive coefficient
on the interaction term between roads and population density). The long-run
increasing returns are possibly related to network effects: gains from an increase
in the length of a route rise as the route interconnects new areas and multiplies
the connections possible. The effectiveness of foreign investment flows also ap-
pears to depend on the availability of infrastructure, as is shown in the strong
positive interaction between foreign investment and the roads network (column
2, table 8).
Telecommunications growth has an even stronger effect; telephones per 1,000
residents show increasing returns both in the short run and in the long run (col-
umns 3 and 6, table 7). The short- and long-run elasticities are both approxi-
mately 0.10.
INITIAL CONDITIONS. The initial per capita income of a region turns out to be
an important variable in explaining subsequent growth. When initial per capita
income is not included in the regressions, the partial correlations between growth
and the other variables change significantly; as noted below in our discussion of
sensitivity tests, variables other than per capita income do not have a similar
influence when added or dropped from the analysis.
The strongly negative relationship between industrial growth rates in a region
and the initial per capita income of the region suggests that growth is being
influenced not just by steady-state factors but also by transitory influences. If
steady-state growth had been achieved in the different industrial sectors and
regions, both neoclassical and endogenous growth models predict that the initial
levels of backwardness will have no influence on subsequent growth (Mankiw,
Romer, and Weil 1992). Only when an economy is moving to a new steady state
will initial levels of backwardness provide an additional impetus to growth. This
seems particularly appropriate for coastal China, which has indeed been shaken
up and put on a new growth trajectory.
Figure 2 shows a strong inverse relationship between the rate of growth of
industrial output during 1985-89 and the log of per capita GDP in 1985. In the
terminology suggested by Mankiw, Romer, and Weil (1992) and by Barro and
Sala-i-Martin (1992), there is evidence of absolute convergence. In other words,
even without controlling for other variables that may affect steady-state growth,
the relatively backward provinces grew faster than the more advanced regions.
For example, initial backwardness partly explains why Fujian grew so fast de-
spite low educational attainment and limited infrastructure.
Absolute convergence applies not only to industrial growth (as described in
figure 2) but also to per capita GDP. Over the 1980s the per capita GDP of the five,
relatively poor, coastal provinces increased relative to that of the three richer
counties (the ratio of GDP per capita in the five provinces to that in the three
312 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Figure 2. Absolute Convergence versus Conditional Convergence
Growth rate of gross value of industrial output, 1985-90
40
Conditional
30 convergence
30 -"
20
Fujian Guangdong "
0 Jiangsu "
10 Shandong
Zhejiang S<
Shanghai
0
Absolute
convergence
-10
6.5 7.0 7.5 8.0 8.5
Log per capita GDP, 1985
Note: Absolute convergence is plotted by fitting the observed values of the five provinces and three
counties. Conditional convergence is plotted according to table 6, equation 4. Its slope is based on the
coefficient of per capita GDP, YIN. Its intercept is the sum of the products of the mean values of the
independent variables (except Y/AN and their respective coefficients.
Source: Authors' calculations.
counties rose from 0.23 in 1980 to 0.38 in 1988, see China, State Statistical
Bureau, 1990). But while there was convergence within the coastal region, there
was divergence between the coast and the rest of China. The per capita income
was higher in the coastal region than in the rest of China when the reforms were
launched, and the gap has increased over time. The region's GDP per capita was
50 percent higher than the average in the rest of the nation in 1980; it was 74
percent higher in 1988.
These observations point to an interesting international parallel. In cross-
country comparisons, absolute convergence is observed among advanced in-
dustrial countries but not among poor economies. Poor economies converge
conditionally, that is, after controlling for education and investment rates.
Within the group of industrial nations, the rate of conditional convergence is
higher than the rate of absolute convergence, because the richer ones typically
also have higher education and investment rates (see Mankiw, Romer, and
Weil 1992).
Mody and Wang 313
We have not investigated the possibility of conditional convergence outside
the coastal region. However, not surprisingly, conditional convergence within
the coastal region, as within the industrial economies, is more rapid than abso-
lute convergence. The richer coastal regions also tend to have better education
and infrastructure, and thus it may be supposed that they have higher steady-
state growth rates. The fact that the poorer regions are growing faster despite
their lack of endowment indicates that they are benefiting from their back-
wardness.
The common interpretation of this catching-up phenomenon is that regions
with low per capita income also have low capital per worker and so have a
higher marginal product of capital than regions that are well endowed with
capital. Thus the poorer regions potentially attract new capital (along with new
ideas). Our evidence certainly supports this view: the poorer regions have at-
tracted huge amounts of foreign capital and knowledge. But in addition, as dis-
cussed above, the more advanced regions have been burdened by an institu-
tional setup that has been a drag on growth. But the Chinese are also fortunate
in this regard that the backward regions are in proximity to Hong Kong and
Taiwan (China), both major centers of knowledge and capital.
GROWTH OF THE INDUSTRY OUTSIDE THE REGION. Results show that the growth
of an industrial sector in any region is powerfully influenced by the growth of
the same industry in other regions during the same year. On average, a 1 percent
increase in the growth rate of an industrial sector outside the region is associated
with a 0.78 percent increase in the growth rate of that industry within the region.
Unlike other variables, this variable not only passes the test of significance but
also accounts for 49 percent of the total sum of squares. This is another way of
capturing the wave phenomenon noted in the variance decomposition exercise.
The high t-statistics for the coefficient for growth outside the region are also
obtained by Glaeser and others (1992), who interpret the result as a demand
effect-exogenous growth in demand, in this view, conditions the growth of
specific sectors irrespective of the region. As noted above in the discussion on
variance decomposition, synchronization across regions can also occur as a result
of technology diffusion or networking among decisionmakers. A concern arises,
however, because there may be biases in the data-gathering process, which build
in cross-regional correlations. To that extent, this variable conditions for these
correlations.5
We try two extensions of the basic regression to gain further insight into the
cross-regional influences at work. First, we interact growth outside the region
with the specialization variable. The results indicate that growth outside the
region has a stronger effect in conditions lacking industrial specialization (table
9, column 1). In other words, the more a sector is specialized within a region, the
5. Coe and Helpman (1995) find large international research and development (R&D) spillovers.
However, Jaffe and Trajtenberg (1996), examining patent data, find limited cross-national citations,
which they interpret as evidence of limited geographical spillovers.
314 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 9. Cross-Regional Influences on Industrial Growth in Coastal China,
1986-89
Variable 1 2 3
Industry specific
Specialization index, S -0.028 -0.069 -0.042
(-1.092) (-2.834) (-1.622)
Specialization index squared, S2 0.010 0.016 0.010
(1.328) (2.152) (1.303)
Entrepreneurship index, E 0.030 0.032 0.032
(1.634) (1.694) (1.571)
Entrepreneurship index squared, E2 -0.005 -0.005 -0.003
(-1.214) (-1.172) (-0.724)
Region specific
Secondary school enrollment rate 0.728 0.638 1.311
(2.785) (2.206) (3.464)
Foreign direct investment 1.616 0.994 4.053
(4.750) (2.478) (3.020)
Roads -0.475 -0.919 -0.132
(-2.605) (-4.121) (-0.546)
Roads squared 0.678 0.982 0.355
(3.510) (4.106) (1.493)
GDP per capita -0.0001 -0.0001 -0.0002
(-4.914) (-3.619) (-4.674)
Growth in Guangdong 0.458
(13.302)
Regional spillover
Growth in industry outside region 1.009
(12.136)
Interaction between growth in industry outside -0.211
region and specialization index (-3.179)
Interaction between growth in industry outside
region and regional dummy variable
Guangdong 0.969
(13.714)
Fujian 0.797
(8.751)
Zhejiang 0.700
(9.297)
Jiangsu 0.902
(13.271)
Shandong 0.607
(9.908)
Tianjin 0.750
(4.471)
Shanghai 0.714
(4.840)
Adjusted R2 0.610 0.613 0.550
Number of observations 640 640 548
Note: The dependent variable is growth of industry i in region r at time t (GC,,). All regressions
include time and industry dummy variables but no regional dummy variables, and observations are
weighted by regional population. The t-statistics are in parentheses. See table 4 for more complete
definitions of variables and descriptive statistics.
Source: Authors' calculations.
Mody and Wang 315
less it is affected by growth of that same industry outside the region. This is not
surprising. With scale economies, certain industries will be concentrated in par-
ticular regions. At the same time, they will also develop certain technical special-
izations or market niches that limit the usefulness of the experience of firms in
the same industrial sector but located in other regions. Also, intellectual prop-
erty is likely to be more protected in such specialized sectors. In general, light
industrial sectors, with lower capital intensity and less specialization than the
heavier sectors, are likely to be more able to absorb external influences rapidly,
as discussed below.
Second, in view of the policy attention accorded to Guangdong (and more
recently to Fujian), and also given their physical proximity to Hong Kong
and Taiwan (China), a question of interest is whether these regions are con-
duits of growth impulses. For the time span studied, no evidence to this ef-
fect is found. When growth outside the region is interacted with region dum-
mies, the coefficients show that Guangdong benefited most from growth
outside the province and Fujian was third on the list, with Jiangsu in be-
tween (table 9, column 2). This indicates that Guangdong, Jiangsu, and Fujian
are most responsive to macro influences, such as changes in buyer percep-
tions and changes in government policies. We then replace growth outside
the region (which is growth in all outside regions) with growth in Guangdong
as an independent variable to isolate the effects that Guangdong may have
had on growth in other regions (Guangdong itself is not included in this
regression). Guangdong's growth does have a statistically significant impact
on other regions, but the magnitude of the effect is much smaller than when
growth in all other regions is considered (table 9, column 3). Similar conclu-
sions apply to Fujian.
It is, however, likely that spillovers from Guangdong and Fujian to the other
provinces will be significant in the long run. Within the coastal region, these
provinces have the greatest flexibility to respond to external stimuli. As other
regions become more receptive to change, Guangdong and Fujian can be ex-
pected to have greater spillover effects. Field surveys in Guangdong and Fujian
show unambiguously that modern production techniques, including sophisti-
cated methods of quality control, are being rapidly adopted in these provinces.
As such experience accumulates, increasing labor mobility will complement ex-
isting administrative communication networks to diffuse the knowledge gained
to other parts of China.
Light and Heavy Industries
Thus far we have assumed that all industrial sectors respond to the explana-
tory variables in the same manner. Here we note some differences between light
and heavy industrial sectors (table 10). Although the differences are of interest,
the exercise also gives us confidence in the results reported so far-the signs of
the coefficients are similar, and the key variables (barring education) continue to
be statistically significant for both light and heavy industries.
Table 10. Growth in Light and Heavy Industries in Coastal China, 1986-89
All industries Light industries- Heavy industriesb
Variable 1 2 3 4 5 6
Industry specific
Specialization index, S -0.061 -0.061 -0.109 -0.109 -0.112 -0.117
(-2.470) (-2.501) (-2.936) (-2.939) (-1.687) (-1.765)
Specialization index squared, S2 0.014 0.015 0.022 0.022 0.043 0.046
(1.902) (2.015) (2.267) (2.240) (1.497) (1.583)
Entrepreneurship index, E 0.033 0.031 0.142 0.143 0.017 0.015
(1.714) (1.646) (1.924) (1.933) (0.859) (0.761)
Entrepreneurship index squared, E2 -0.006 -0.005 -0.034 -0.035 -0.002 -0.002
(-1.244) (-1.117) (-1.483) (-1.496) (-0.514) (-0.343)
Region specific
Secondary school enrollment rate 0.779 1.373 0.813 0.611 0.778 1.654
(2.962) (2.371) (1.800) (0.622) (2.454) (2.368)
Secondary school enrollment rate squared -0.663 0.223 -0.981
(-1.151) (0.231) (-1.406)
Foreign direct investment 1.628 1.655 2.234 2.225 1.254 1.296
(4.747) (4.816) (3.843) (3.816) (3.004) (3.101)
Roads -0.489 -0.587 -1.084 -1.053 -0.097 -0.241
(-2.657) (-2.896) (-3.528) (-3.132) (-0.422) (-0.961)
Roads squared 0.688 0.832 1.269 1.223 0.295 0.507
(3.530) (3.593) (3.925) (3.195) (1.218) (1.779)
GDP per capita -0.0002 -0.0002 -0.0002 -0.0002 -0.0002 -0.0001
(-5.005) (-4.969) (-3.184) (-3.181) (-3.775) (-3.730)
Regional spillover
Growth in industry outside region 0.777 0.778 0.827 0.827 0.583 0.590
(19.492) (19.519) (16.898) (16.864) (6.260) (6.339)
Adjusted R2 0.604 0.604 0.671 0.670 0.493 0.491
Number of observations 640 640 280 280 360 360
Note: The dependent.variable is growth of industry i in region r at time t (Gi,). All regressions include time and industry dummy variables but no regional
dummy variables, and observations are weighted by regional population. The t-statistics are in parentheses. See table 4 for more complete definitions of variables
and descriptive statistics.
a. The light industry group includes food processing, textiles, apparel, leather products, wood products, furniture, paper products, art products, plastic
products, and electronics.
b. The heavy industry group includes petroleum, chemicals, pharmaceutical products, chemical fibers, electricity, rubber products, nonmetal products, ferrous
products, nonferrous products, metallic products, transportation equipment, electrical machinery, and other machinery.
Source: Authors' calculations.
Mody and Wang 317
The estimated equation does a better job of explaining growth in light indus-
tries (R2 = 0.67) than in heavy industries (R2 = 0.49). Of special interest is the
finding that growth outside the region, which measures the degree of synchroni-
zation or diffusion across regions, has a higher coefficient for light industries.
This is to be expected given the lower capital intensity and hence higher mobility
of light industrial sectors. Guangdong, Fujian, and Jiangsu benefit especially
from growth outside the region in both heavy and light industries; this differ-
ence is measured by interacting growth outside the region with regional dum-
mies.6 In light industries, other regions also benefit strongly from the diffusion
process, whereas the effect for heavy industries falls off in other regions and is
not statistically different from 0 for Shanghai and Tianjin. In both light and
heavy industries, when growth in Guangdong is used as an explanatory vari-
able, the partial correlation is positive and significant, but smaller in magnitude
than the coefficient obtained for growth outside the region, implying again that
Guangdong is more an imitator than a leader.
Foreign investment provides a bigger effect in light industries, although it has
a significant coefficient for heavy industries. Similarly, infrastructure does more
for light than for heavy industries. In contrast, education has a positive effect on
growth in light industries, but the effect is not statistically different from 0.
Thus, although formal education is important, its relationship with growth is
imprecise, and tacit knowledge based on experience (and channeled through
foreign sources) appears to be a somewhat firmer predictor of growth. For heavy
industries, we observe diminishing returns to education, as was seen above for
all industries; within the range of observed secondary school enrollment rates,
this implies a positive, though declining, effect of education on growth of heavy
industries.
The lack of specialization has a stronger association with growth among
light industries, which is not surprising; skills are likely to be more mobile in
such sectors. When all observations are pooled, we note that specialization is
an aid to growth only beyond S = 2. For light industries, the positive effects of
specialization are felt at even higher levels of specialization (beyond S = 2.5);
in comparison, for heavy industries, specialization is conducive to growth af-
ter S = 1.3. The implication is that specialized sectors, which have also grown
rapidly, are principally in the heavy industry group. Although an exact corre-
spondence cannot be easily made, our results bear some similarity to those of
Henderson, Kuncoro, and Turner (1995). They find that for the mature sec-
tors, specialization promotes growth (in the Chinese case, heavy industry has
been the more traditional focus of state investment); in contrast, they find that
a diverse industrial environment fosters new industrial sectors (while the light
industries studied here are not new in the sense of being high-technology, they
have required many new skills to meet the exacting demands of the interna-
tional market).
6. These results are not presented to conserve space but can be provided on request.
Table 11. Sensitivity Analysis of Growth Determinants for Coastal China, 1986-89
Variable 1 2 3 4 s 6 7
Industry specific
Specialization index, S -0.061 -0.052 -0.066 -0.062 -0.049 -0.079
(-2.501) (-2.093) (-2.688) (-2.530) (-1.860) (-3.540)
Specialization index squared, S2 0.015 0.012 0.016 0.013 0.014 0.018
(2.015) (1.559) (2.096) (1.750) (1.723) (2.981)
Entrepreneurship index, E 0.031 0.017 0.030 0.020 0.059 0.022
(1.646) (0.885) (1.564) (1.086) (3.351) (1.205)
Entrepreneurship index squared, E2 -0.005 -0.004 -0.006 -0.004 -0.012 -0.003
(-1.117) (-0.820) (-1.211) (-0.979) (-2.756) (-0.662)
Region specific
Secondary school enrollment rate 1.373 0.301 0.821 1.301
(2.371) (0.633) (1.190) (2.556)
Secondary school enrollment rate squared -0.663 0.577 -0.407 -0.888
(-1.151) (1.213) (-0.570) (-2.081)
Foreign direct investment 1.655 1.486 0.837 1.115 1.432
(4.816) (5.222) (3.866) (2.729) (4.004)
Roads -0.5 87 -0.684 -0.741 -0.551 -0.620
(-2.896) (-4.145) (-4.521) (-2.130) (-3.435)
Roads squared 0.832 0.698 0.834 0.759 0.818
(3.593) (3.702) (4.395) (2.573) (4.995)
GDP per capita -0.0002 -0.0001 -0.0001 -0.0001 -0.0001 -0.0001
(-4.969) (-4.319) (-6.275) (-5.452) (-3.176) (-4.458)
Regional spillover
Growth in industry outside region 0.778 0.771 0.775 0.774 0.7s4 0.818
(19.519) (19.022) (19.262) (19.297) (17.496) (20.161)
Adjusted R2 0.604 0.590 0.596 0.599 0.351 0.535 0.616
Number of observations 640 640 640 640 640 640 640
Note: The dependent variable is growth of industry i in region r at time t (G,). All regressions include time and industry dummy variables but no regional
dummy variables and observations are weighted by regional population. Observations are weighted by the population in the region for the regressions reported in
columns 1-6. No weights are used for the regression reported in column 7. The t-statistics are in parentheses. See table 4 for more complete definitions of variables
and descriptive statistics.
Source: Authors' calculations.
Mody and Wang 319
Sensitivity and Misspecification
Our sensitivity analysis uses the methods of Belsley, Kuh, and Welsch (1980).
We first drop one observation at a time and find that no single observation
influences the coefficients significantly. This result could have been expected,
given the large sample of 640 observations. We then drop specific sets of obser-
vations, excluding from regressions a province, a year, an industry, a region-
industry, a year-industry, and a year-region. The distributions of the coefficients
show a very strong concentration around the mean value. We can therefore rule
out the possibility of outliers driving our regression results.
In the regressions reported, we weight the observations by the population of
the region, which gives more weight to the provinces and less to the counties,
reducing the influence of the counties in the regression results. To see how much
the results are influenced by this weighting procedure, we also run our basic
regression by treating every observation equally (column 7, table 11). The re-
sults do not change qualitatively, except that diminishing returns to education
are now more evident: this is as expected because the more educated counties
that recorded relatively modest economic performance now have greater weight
in the regression.
Another type of sensitivity analysis is done by adding or dropping inde-
pendent variables (table 11). Omitting secondary school enrollment rates has
little effect on the sign and magnitude of the remaining coefficients (column
4). Similarly, the regression results are not sensitive to specifications that
exclude an entire set of industry- or region-specific variables, as columns 5
and 6 demonstrate.
If there is no serious misspecification problem, regional factors other than
initial per capita income predict that the counties (Shanghai and Tianjin) should
have done especially well because they had better than average access to foreign
investment, education, and infrastructure. But instead, growth in these counties
was slow, possibly because of the significant presence of state-owned enterprises,
which is not captured in the regressions. When we include the share of state-
owned enterprises as an independent variable, it does not generate significant
results because the share of these enterprises is correlated with per capita income
(and also with the variable E, which is the inverse of average firm size). Thus the
relatively slow growth in recent years of the two richer regions reflects diminish-
ing returns, which arise not merely from a technological source but also from
the constraining effects of the institutional structure within which past industri-
alization occurred.
IV. CONCLUSIONS
We have examined three sets of influences on industrial growth along the
eastern coast of China: factors specific to an industrial sector, regional influ-
ences, and regional spillovers.
320 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Overall, industry-specific influences explain only a small portion of variance
in growth. A low level of specialization, perhaps allowing for absorption of
influences from other industrial sectors, seems to promote more rapid growth
for light industries, whereas specialization seems to be conducive to growth in
heavy industries. Our findings on the role of competition are statistically weak,
possibly because of the very crude statistical proxy used for competition.
A number of regional influences are important. Higher levels of education
differentiate good performers from poor performers over the long haul: gains of
even a few percentage points in secondary school enrollment rates have an im-
portant effect on growth. When only light industries are considered, however,
the relationship between growth and secondary school enrollment is potentially
influential but imprecise. For heavy industries, education has diminishing re-
turns, although the positive effects continue well into the range observed in the
sample (as well as the range spanned by most middle-income countries).
The role of secondary school education, however, cannot be considered sepa-
rately from knowledge acquired through international links. Foreign investment
showed consistently as a spur to growth, especially in the short run and in light
industries. Moreover, we found that foreign investment and education interact
positively. It is worth noting that secondary school enrollment rates in Fujian
province at 31 percent are close to the average for low-income countries (World
Bank 1991). Our results suggest that China's coastal provinces were able to
exploit their educational attainment better than other low-income regions be-
cause the complementary effects of foreign knowledge enhanced the educational
level of the work force.
Infrastructure investment, particularly in telecommunications but also in roads,
yields increasing returns. There is some question whether infrastructure is a true
enabling factor; although it accelerates output growth, it also responds to growth.
Large infrastructure investments are occurring along the coast in the wake of the
huge growth of the past several years. Thus although good infrastructure is valu-
able, conditions that enable externality-generating infrastructure investments to
be put in place as demand emerges are equally important.
Almost half of the variation in industrial growth along the coast is attribut-
able to the synchronization in growth of particular industries across provincial
and county boundaries. The identity of the most rapidly growing sectors changed
from year to year across the entire region. Such synchronization was more pro-
nounced in light than in heavy industries. Although many substantive possibili-
ties exist to explain the synchronization-and a recent field study documents
that the perception of synchronization exists among decisionmakers on the Chi-
nese coast (Oi 1995)-we have noted that, on account of data construction and
reporting, the extent of regional spillovers is likely to be less than the statistical
analysis may suggest.
China has pursued a decentralized economic reform program. Particular re-
forms have been tried in specific regions-sometimes with and sometimes with-
out the blessing of the central government. To complement reforms for increas-
Mody and Wang 321
ing allocative efficiency, China has pursued a long-term strategy for encourag-
ing investments by specific new entrepreneurs. The open-door policies and spe-
cial economic zones have successfully attracted investments from overseas Chi-
nese to the southeastern coast. At the same time, local governments have been
given greater autonomy to invest in new business ventures (for example, the so-
called collectively owned enterprises) and in infrastructure. Although many of
the experiments are considered innovative, the lack of coordination and waste-
ful regional competition have resulted in damaging macroeconomic effects.
In any case, synchronization across regions has been quite strong. The source
of this synchronization cannot be discerned from the data at hand, but it is clear
that a network of communication channels exists across the country. Such a
network could reflect the links between the cadres of the communist party or
could even predate the party, reflecting much older economic and social ties
(Yusuf 1993 and Oi 1995). Success has, thus, required a combination of cen-
trally approved local experiments, local government entrepreneurship, and an
effective network for diffusing success across different regions.
An interesting aspect of the decentralization has been that regions with rela-
tively low per capita income and hence a large catch-up potential were targeted
early on. These regions were relatively unencumbered by state-owned enterprises,
planning bureaucracies, and other mechanisms that guided output in the prereform
era. Indeed, some of the counties in Guangdong province that experienced the
most spectacular growth rates, such as Shenzen and the neighboring areas, were
essentially agricultural communities (or even wastelands) 15 or 20 years ago.
Although the successes of the strategy have been evident, questions have been
raised about policy reversals and setbacks and the consequent lack of govern-
ment credibility (see Sung 1991 and Chen, Jefferson, and Singh 1992). Such
credibility lapses are generally viewed as expensive, inasmuch as they create
investor uncertainty and reduce investment. Yet investors, especially foreign in-
vestors, have rarely been deterred. Foreign investment has been almost an exog-
enous force, dampened only occasionally by policy conditions. At the same time,
locally financed infrastructure and human capital investments plus job training
within enterprises have proceeded with vigor, fueling growth.
We suggest two related possibilities. First, the credibility of government poli-
cies as a determinant of investment is overrated; it is likely that credibility and
certainty derive from overall economic performance rather than from govern-
ment actions per se. Second, investors may accept contradictions and reversals
as a reflection of the government's response to evolving conditions.
If this analysis of China's recent experience is in any respect correct, what
lessons does it hold for other countries? Decentralized experiments are valuable,
but they may well require local governments that are entrepreneurial. Human
capital and infrastructure aid the process of transformation but in more com-
plex interactive modes than usually assumed. A steady flow of foreign invest-
ment and skills provides a strong advantage. For wider impact, the lessons from
decentralized experiments must flow to other regions. Mechanisms to ensure
322 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
information transfers are essential, but difficult to establish. In a complex re-
form process, credible commitments may be desirable, but governments also
need to stay flexible.
APPENDIX. DETAILS OF THE DECOMPOSITION OF GROWTH OF OUTPUT
The standard variance components method assumes as the first approxima-
tion a growth equation of additive main and interaction effects with zero means
and covariances. Let Gtri denote growth in industry i in region r at time t. Vari-
ance of Gtri can be decomposed to the main effects of time (ax), region (,r),
industry (ti), and their respective interaction effects (atr, bti, cri, and etri).
Formally,
Gtri = m + at + 1r +i + atr+abti+Cri+ etrn
where m is a constant, t = 1, -. . ., nt, r = 1, . . . , nr, i = ,*** ni.
With the assumption of zero covariance, VarG can be expressed as follows:
Var(G) = Var(a) + Var(p) + Var('T) + Var(a) + Var(b) + Var(c) + Var(e)
or
s2(G) = s2a + s20 + S2t + S2a + S2b + S2c + S2e.
Variance components are estimated by equating observed values of variances
to their expected values (Searle 1971). Let N = ninrnt.
Define
To = StSrSi G ti
E(TO) = ninnrt (m2 + S2a + S+ S2, + S2a + S2b + S2, + S2e)
=N(m2 + s2a + s2 +S2t+Sa+S b+ S2c+ Se)
T, = G2/ ninnrf = (StSrSjG)2 l ninnrt
= (ninrntm + Stninra + Srninfp + Sinftnrt
+ S,Sra + SSib + SrSic + StSrSie)2 / nin,nt
E(T1) = Nm2 + ninflS2a, + nintS2 p + ntnrs2' + nis2a + nfs2b + nts2c + S2e
T2 = S,Gt2/ nin, = St(SSiGG)2 / ninr
= St(nin,m + nin,a + Snij + Sinrj + niSra + n,Sib +SrSic + SrSie)2/ ninf
E(T2) = Nm2 + Ns2, + ninflS23 + ntnrS2T + ntnis2a + ntnfs2b + nts2c + ns2e.
Similarly, let
T3 = SrGr2 I nfnl = S,(StSjG)2 / n,ni
E(T3) = Nm2 + nrn s2" + Ns25 + ntnrs 2t + nrnis2 a + nrs2b + ntnrs2c + ns2e
T4 = SiGi2 / ntn, = Si(StSrG)2 / ntnr
E(T4) = Nm2 + nrniS2a + n,nis2P + Ns2, + n s2a + nrnis2b + ntnis2c + nS2e
Ts = SSiGti 2/ nr = StSi(SrG)2 / n,
E(T5) = Nm2 + Ns2a + ntnjs2P + NS2, + nftns2a + NS2b + ntn S2, + ntn S2,
Mody and Wang 323
T6 = StSrGtr2/l i = StSr(SiG)2/ ni
E(T6) = Nm2 + NS2a + Ns2 + ntnrs21 + NS2a + n,tnrS2b + ntnS2c + ntnrS2e
T7 = SrSiGr, / 1n = S,Si(S,G)2 / nt
E(T7) = Nm2 + nrnfS2,, + Ns2, + NS2T + nrn S2 f + nnfiS2b + Ns2c + nrniS2,.
The system therefore contains eight equations with eight unknowns. The eight
equations refer to the expressions of E(TO), . . . , E(T7) and the eight unknowns
are mi2, 52a) s2 S2T S2a, S2b, s2c, and S2e. We can solve the system by equating
sample values of TO, . . . , T7 to their expected values, E(TO), . . . , E(T7). The
solutions are the following:
S2a ={ni[(T1- T2) - (T3- T6)] - [(TO- TS) + (T4- T7)]} 1 [ni(ni- 1) (nt- 1) (nr- 1)]
s2b= {fr[(Ti T4) - (T2- T5)1 - [(TO- T7) + (T3- T6)] 1 1[nr(nr- 1) (ne-l) (ni-1)]
s2= {nt[(T1- T3)- (T4-T7) - [(TO- T6) +(T2- T5)]} I [n,(n,- 1) (nr l)(ni- 1)]
S2e=(To-T1+T2+T3+T4-T5-T6-T7) [(nf- 1) (nr- 1) (ni- 1)]
S2a = [(T3- T6) - (nrni- N)S2a - (nr, ntr)s2 b - (nr- nntr)S2e] 1 (nrni- N)
S2 = [(T4 - T7) - (ni- nrni)S2a - (n,ni - N)s2 - (ni - nrni )S2e1 /(nni - N)
s2 = [(T2- T5) - (ntn,r- N)S2 - (ne- nni)s2c - (n,- ntni)S2e] / (ntnr- N).
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Barro, Robert, and Xavier Sala-i-Martin. 1992. "Convergence." Journal of Political
Economy 100(2, April):223-51.
Bateman, Deborah, and Ashoka Mody. 1991. "Growth in an Inefficient Economy: A
Chinese Case Study." Private Sector Development Department, World Bank, Wash-
ington, D.C. Processed.
Belsley, David A., Edwin Kuh, and Roy E. Welsch. 1980. Regression Diagnostics. New
York: Wiley and Sons.
Chen, Kaug, Gary H. Jefferson, and Inderjit Singh. 1992. "Lessons from China's Eco-
nomic Reform." Journal of Comparative Economics 16(2):201-25.
China, State Statistical Bureau. Various years. China Statistical Yearbook. Beijing: China
Statistical Information and Consultancy Service Center.
. Various years. Statistical Yearbook of Fujian. Beijing: China Statistical Informa-
tion and Consultancy Service Center.
. Various years. Statistical Yearbook of Guangdong. Beijing: China Statistical
Information and Consultancy Service Center.
-. 1990. Compendium of Historical Data, 1949-89. Beijing: China Statistical In-
formation and Consultancy Service Center.
Chow, Gregory C. 1993. "Capital Formation and Economic Growth in China." Quar-
terly Journal of Economics 108(3):809-42.
324 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Coe, David T., and Elhanan Helpman. 1995. "International R&D Spillovers." European
Economic Review 39(5):859-87.
Glaeser, E. L., H. D. Kallal, J. A. Scheinkman, and Andrei Shleifer. 1992. "Growth in
Cities." Journal of Political Economy 100(6):1126-52.
Hayase, Yasuko, and Seiko Kawamata. 1990. Population Statistics of China. Institute
of Developing Economies, Tokyo. Processed.
Henderson, Vernon, Ari Kuncoro, and Matt Turner. 1995. "Industrial Development in
Cities." Journal of Political Economy 103(5):1067-90.
Ho, S. P. S., and Ralph Huenemann. 1984. China's Open Door Policy. Vancouver:
University of British Columbia Press.
Islam, Nazrul. 1995. "Growth Empirics: A Panel Data Approach." Quarterly Journal of
Economics 110(4):1127-70.
Jacobs, Jane. 1969. The Economy of Cities. New York: Vintage Books.
Jaffe, Adam, and Manuel Trajtenberg. 1996. "Flows of Knowledge from Universities
and Federal Labs: Modelling the Flow of Patent Citations over Time and across Insti-
tutional and Geographic Boundaries." NBER Working Paper 5712. National Bureau
of Economic Research, Cambridge, Mass. Processed.
Jefferson, Gary H., Thomas G. Rawski, and Yuxin Zheng. 1990. "Growth, Efficiency,
and Convergence in China's State and Collective Industry." Economic Development
and Cultural Change 40(2):239-66.
Levine, Ross, and David Renelt. 1992. "A Sensitivity Analysis of Cross-Country Growth
Regressions." American Economic Review 82(4):942-63.
Lucas, Robert. 1988. "On the Mechanics of Economic Development." Journal of Mon-
etary Economics 22(1, July):3-42.
Mankiw, Gregory N. 1995. "The Growth of Nations." Brookings Papers on Economic
Activity 1:275-376.
Mankiw, Gregory N., David Romer, and David Weil. 1992. "A Contribution to the Empirics
of Economic Growth." Quarterly Journal of Economics 107(2, May):407-38.
Oi, Jean C. 1992. "Fiscal Reform and the Economic Foundations of Local State
Corporatism in China." World Politics 45(1):99-126.
. 1995. "Cadre Networks, Information Diffusion, and Market Production in
Coastal China." PSD Occasional Paper 20. Private Sector Development Department,
World Bank, Washington, D.C. Processed.
Porter, M. E. 1990. The Competitive Advantage of Nations. New York: Free Press.
Pritchett, Lant. 1996. "Where Has All the Education Gone?" Policy Research Working
Paper 1581. Policy Research Department, World Bank, Washington, D.C. Processed.
Romer, Paul M. 1993. "Two Strategies for Economic Development: Using Ideas and
Producing Ideas." In Proceedings of the World Bank Annual Conference on Devel-
opment Economics 1992. Washington, D.C.: World Bank.
Schankerman, Mark. 1991. "Revisions of Investment Plans and the Stock Market Rate
of Return." NBER Working Paper 3937. National Bureau of Economic Research, Cam-
bridge, Mass. Processed.
Searle, S. R. 1971. Linear Models. New York: Wiley and Sons.
Shleifer, Andrei. 1990. "Externalities and Economic Growth: Lessons from Recent Work."
Background paper 3 to the World Development Report 1991: The Challenge of De-
velopment. World Bank, Washington, D. C. Processed.
Mody and Wang 325
Stockman, A. C. 1988. "Sectoral and National Aggregate Disturbances to Industrial
Output in Seven European Countries." Journal of Monetary Economics 21(2/3):387-
409.
Sung, Yun-win. 1991. The China-Hong Kong Connection: The Key to China's Open-
Door Policy. Cambridge, U.K.: Cambridge University Press.
Waldmann, R. J., and J. Bradford De Long. 1990. "Interpreting Procyclical Productiv-
ity: Evidence from a Cross-Nation Cross-Industry Panel." Discussion Paper 1495.
Harvard Institute of Economic Research, Harvard University, Cambridge, Mass. Pro-
cessed.
Walker, Tony. 1993. "Hats off to the Revolution." Financial Times, March 30, p. 21.
Weitzman, M. L. 1970. "Optimal Growth with Scale Economies in the Creation of
Overhead Capital." Review of Economic Studies 37(4):555-70.
World Bank. 1991. World Development Report 1991: The Challenge of Development.
New York: Oxford University Press.
Yusuf, Shahid. 1993. "The Rise of China's Nonstate Sector." China Department, World
Bank, Washington, D.C. Processed.
I
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 327-55
Organized Labor and the Political Economy
of Product Market Distortions
Martin Rama
Why are economic reforms reversed through strikes and demonstrations in some coun-
tries, but backed by the labor movement in other countries? Why do product and labor
market distortions differ so much across countries? This article addresses these ques-
tions by means of a simple, heuristic model of the economy that replicates in an inte-
grated manner several independent results from the recent political economy litera-
ture. Unlike most of this literature, however, the model focuses on the role played by
organized labor, rather than by rent-seeking firms and guilds. A two-stage game be-
tween the government and organized labor determines the level of product market
distortions (for example, import tariffs). In the first stage, the players may undertake
costly actions, such as redistributing income or striking, in order to increase their bar-
gaining power. In the second stage, they negotiate over product market distortions
and wages. Under very general assumptions, several policy regimes exist: Changes in
the key parameters of the economy may trigger a switch in the strategy of trade unions
from confrontation to cooperation and hence change the policy regime. Cross-country
data highlight that, in spite of its simplicity, the model reproduces some observed em-
pirical regularities.
Since the early 198 Os, removing product market distortions has been at the core
of economic policy, both in industrial and in developing countries. Industrial
countries have made efforts at fostering competition based on the deregulation
of the markets for goods and services. Developing countries have implemented
structural and sectoral adjustment programs, usually with the support of the
World Bank. These reform programs include liberalizing foreign trade, prob-
ably the most frequent goal, as well as curtailing subsidies, suppressing legal
monopolies, and eliminating the direct allocation of credit and foreign exchange.
Overall, these reforms have been positive, although a few disturbing facts
remain. First, recidivism appears to be a widespread problem. Some countries
that had begun the process of liberalizing eventually built up new distortions
that represented a major departure from the initial program. The reinforcement
Martin Rama is with the Policy Research Department at the World Bank. This article is part of a
broader research effort entitled "The Impact of Labor Market Policies and Institutions on Economic
Performance," supported by the Research Committee of the World Bank through grants RPO 678-49
and RPO 680-96. The author gratefully acknowledges comments by Dani Rodrik, Guido Tabellini,
participants at the Latin American Econometric Society Meetings, and four anonymous referees and the
excellent research assistance provided by Praveen Kumar.
� 1997 The International Bank for Reconstruction and Development / THE WORLD BANK
327
328 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
of Australia's antidumping legislation provides a clear example of this shift. In
other countries, like Nigeria and Venezuela, street riots and political turmoil led
to a reversal of the reform program. In the Ukraine, strikes by coal miners forced
the government to put an approved plan of pit closures on hold. In Zambia, the
reform program derailed because of opposition by trade unions.
Equally disturbing, some countries launched significant economic reforms amid
the worst conceivable scenario, in spite of the likely adjustment costs. In Mexico,
for example, during the 1980s, the price of oil exports collapsed, access to for-
eign lending vanished, interest rates on foreign debt soared, and the country
suffered one of the worst earthquakes in the century. Yet the country undertook
an ambitious liberalization program. On top of this, it backed up the program
with a series of explicit agreements (known as the Pacto) between the govern-
ment and the main social partners, including organized labor.
Last but not least, the welfare costs of economic reform were often higher and
more long-lasting than expected. In Chile, for instance, unemployment jumped
to two-digit rates for an entire decade after trade liberalization. In the European
Union, the fear of high unemployment has deterred the removal of product mar-
ket distortions in sectors such as steel, automobiles, and textiles. In some devel-
oping countries, those who criticize structural adjustment on the grounds of its
high social cost may actually have a point. By and large, the welfare costs of
reform have made the expression "adjustment fatigue" a meaningful description
of reality.
An obvious explanation for these anomalies is that the reforms under con-
sideration focused mainly on product market distortions but did little to elimi-
nate labor market distortions. For instance, far fewer structural adjustment
loans have attached conditionality to labor market deregulation than to trade
liberalization. Distortions such as mandatory minimum wages and indexation
clauses could therefore be responsible for real wage stickiness in formerly dis-
torted sectors. Similarly, hiring and firing restrictions could be to blame for
insufficient labor reallocation to the competitive sectors of the economy. As a
result, the welfare costs from the reform would be higher than expected. The
appearance that the costs exceed the discounted benefits provides a rationale
for recidivism.
Based on this explanation, success would require the removal of policy-
induced distortions in both product and labor markets. However, labor market
rigidities cannot easily explain why governments undertook reforms under se-
vere adverse shocks, as was the case in Mexico. Furthermore, this explanation
cannot deal with a more fundamental problem: the clearly suboptimal regula-
tion of product markets prior to reform. What trapped so many countries, both
industrial and developing, in such wild distortions to competition?
Within the Pigouvian framework, welfare-maximizing governments unilater-
ally decide economic policies. Thus, perhaps policymakers around the world
had a wrong (say, populist) model of the economy. This answer is appealing in
the case of formerly planned economies. But for other countries, the political
Rama 329
economy approach provides a more promising answer. Within this approach,
economic policies are the endogenous outcome of a distributive conflict, shaped
by the institutional setting in which decisionmaking occurs.
In this article, I adopt the political economy approach to account for product
market distortions as well as for the reported anomalies in the reform process. I
develop a simple, heuristic model of the economy to articulate in a consistent
fashion some of the main messages from the recent political economy literature.
However, unlike this literature, the model gives a prominent role to labor mar-
ket policies and institutions, which have so far been conspicuously missing from
the debate (see, for instance, the surveys by Rodrik 1995 and Helpman 1995).
The contribution of this article is therefore to extract the labor market implica-
tions of the political economy approach in an intuitive manner, still backed by
the more rigorous analysis in the original models in this literature.
The heuristic model belongs to the bargaining variety of model in the political
economy literature. Product market distortions result from a policy game be-
tween a benevolent government, representing all society, and organized labor,
representing some of the workers. Note that organized labor is not the only
economic agent to benefit from product market distortions. The literature has
given a much more prominent role to individual firms and sectoral guilds. How-
ever, by focusing on organized labor the model can explain the incidence of
labor market features in product market distortions and, therefore, in the suc-
cess or failure of economic reforms.
This bargaining model involves two stages. In the first one, players have to
decide whether to undertake actions that are to some extent indivisible, such as
redistributing income or striking. These actions entail a cost, but they also in-
crease the players' bargaining power. In the second stage, the players bargain
over the level of product market distortions and wages in the distorted sector.
Their net payoffs thus depend on parameters such as the costs of striking and
redistributing income but also on the determinants of gains and losses from
distortions, which include the elasticity of labor demand, the deadweight loss
created by the distortions, and the size of the sector they affect.
Even though the structure of the model may seem unrealistically simple, it
captures some of the regularities observed in practice. An empirical section of
the article shows that countries differ significantly in the extent to which gov-
ernments redistribute income and organized labor strikes. More important, these
differences are correlated with distortions in product and labor markets in the
way predicted by the model. Of course, I do not interpret the empirical evidence
presented here as a real test of the model. But the data show that in spite of the
criticism that some of the model's assumptions may raise, particularly regarding
the choice of players and nature of the game, the model provides useful insights
into the design of economic reforms.
Three main messages emerge from the exercise. First, different levels of prod-
uct and labor market distortions characterize several policy regimes. The policy
regime in which organized labor strikes and the government does not redistrib-
330 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
ute income (called Latin American, for short) has higher product market distor-
tions and sectoral wage differentials than the regime with income redistribution
and no strikes (Scandinavian). The regimes with strikes and income redistribu-
tion (European) or no strikes and no income redistribution (East Asian) occupy
an intermediate position. A successful economic reform therefore requires a
change in the policy regime rather than a mere change in the level of some policy
instrument within a given regime.
Second, removing product market distortions reduces welfare as long as the
players have no incentive to switch their strategies. The very fact that the initial
(distorted) situation was an equilibrium means that it was not in the interest of
any of the players to deviate from it, given the actions undertaken by the other
player. Particularly, government would settle for high tariffs because other means
of appeasing unions, such as redistributing income, were even more costly. There-
fore, the welfare cost associated with product market reforms is not necessarily
the result of labor market rigidities (for example, minimum wages or hiring and
firing costs), but rather the outcome of strategic interaction between the players.
Third, changes in some key parameters of the economy may trigger a switch
in the strategy of organized labor from confrontation to cooperation and, there-
fore, a reduction in the equilibrium level of distortions. In the discussion of these
changes, the model reproduces some interesting results of the recent political
economy literature, applied now to labor issues in developing countries. In par-
ticular, I analyze how income redistribution, the flexibility of labor demand, the
severity of macroeconomic shocks, and the scope of the reforms undertaken
may all affect the equilibrium level of product market distortions.
Section I presents the main argument in an intuitive way, under the form of a
heuristic model of the economy, and discusses how this model fits in the politi-
cal economy literature. It shows, under very general assumptions, that several
policy regimes exist. Section II makes more specific assumptions concerning the
structure of the economy and identifies how changes in key parameters trigger a
switch in the strategies of the players. The section gives only the results of this
more elaborate version of the model; the appendix provides their derivation.
Section III presents cross-country evidence from the 1980s to suggest that the
model is empirically relevant. The data indicate that the model correctly predicts
the relationship between income redistribution and labor conflict, on the one
hand, and the level of product and labor market distortions, on the other hand.
Section IV draws the main policy implications of the analysis and shows how
they mimic some of the recent results of the political economy literature, while
introducing the labor dimension that is too often missing. Section V concludes.
I. THE ARGUMENT
Several approaches in the literature account for distortive economic policies.
A relevant difference among them concerns the role played by the government.
At one end of the spectrum, the rent-seeking approach, pioneered by Tullock
Rama 331
(1967) and Krueger (1974), assumes that the government unilaterally sets
distortive policies. Given these policies, interest groups compete for appropria-
tion of the ensuing rents. At the other end, the war-of-attrition model, first ana-
lyzed by Alesina and Drazen (1991), assumes no government at all. Economic
policies, distortive or otherwise, merely reflect the nature of the equilibrium
between rival interest groups. In between these two extremes, the bargaining
model introduced by Barro and Gordon (1983) and the common agency ap-
proach used by Grossman and Helpman (1994) assume some form of interac-
tion between the government and the private sector.
To explain why I use a bargaining model, I first summarize some key features
of the other approaches. The main contribution of the rent-seeking approach
has been to show the potential magnitude of the waste of resources triggered by
distortive economic policies. The rent-seeking approach usually assumes a com-
petitive labor market. Rama (forthcoming) evaluates the social costs of distor-
tions when workers are unionized. However, this approach often takes distortive
policies as given. In this respect, there is a similarity with the Pigouvian ap-
proach to economic policy, except that in the rent-seeking approach govern-
ment policies reduce, rather than increase, welfare. Attempts at introducing pres-
sures from interest groups into policymaking (as in Magee, Brock, and Young
1989 and Hillman 1989) have been quite insightful, but the mechanisms at work
vary substantially depending on the policy issue.
Assuming that there is no government at all is, from the policy perspective, as
extreme as assuming that the government unilaterally implements distortive
policies. In the war-of-attrition approach, each of the interest groups in conflict
has to choose between different economic policies from a totally exogenous policy
menu. Faced with an unexpected negative shock, interest groups can either fa-
vor a permanent adjustment based on nondistortionary taxation or a temporary
adjustment resorting to distortionary instruments. Such groups would prefer a
temporary adjustment if the first group to sign on for permanent adjustment
most likely would bear a disproportionate share of the new, nondistortionary
taxes. No doubt, the analysis of this dilemma has produced very interesting
results, some of which I replicate in an intuitive manner within the bargaining
approach (see section IV).
The bargaining model and the common agency approach both include a gov-
ernment that controls the policy instruments and has a well-defined objective
function. If this government does not adopt an optimal policy, it is because the
private sector confronts it with costs (in the case of the bargaining approach) or
contributions (in the cast of the common agency approach) that make
distortionary policy a more attractive alternative. These two approaches thus
can be used to analyze how changes in the key parameters of the economy affect
the value of the costs or contributions facing the government, hence the level at
which the government sets policy instruments.
The common agency approach has provided solid microeconomic foundations
to suboptimal policymaking. Rama and Tabellini (1997) analyze the implica-
332 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
tions of this approach when organized labor tries to influence government deci-
sions. In the model in that paper, factors of production in the formal sector of the
economy are organized in interest groups, while workers in the informal sector
are not. The paper shows that in spite of the conflict of interests between orga-
nized labor and capital, product and labor market distortions move in the same
direction in response to changes in the key economic and political parameters of
the model. It also shows that policy conditionality by multilateral institutions
should target product market distortions, not labor market distortions.
In this article I explore the potential of the bargaining model to yield new
insights. The approach I use follows up on the simple but insightful model by
Rodrik (1992). In that model, economic policies result from a game between the
government and an interest group, with one of the players behaving as the leader
and the other as a subordinate. In Rodrik's interpretation, better economic per-
formance occurs in East Asia, where interest groups are subordinate, compared
with Latin America, where they are not. Rather than assuming that the condi-
tion of leader is exogenously given, however, the model I use here goes one step
further and makes leadership endogenous.
The political economy literature also includes a voting variety, in which some
heterogeneity of the citizens leads, through the majority rule, to suboptimal policy
choices. One of the most obvious heterogeneities is in endowments: the wealth
of the median voter is usually lower than the average wealth indeed. This citizen
may hence favor redistribution, even if it entails some inefficiency (see Persson
and Tabellini 1990). Heterogeneity may also concern information: some indi-
viduals do not know whether they will be winners or losers from the reforms. In
this case, a status-quo bias may emerge in which a majority of the population
does not support the reforms, even when everybody knows that they will in-
crease aggregate income (see Fernandez and Rodrik 1991).
Although the literature on voting has provided very interesting results, it is
not well suited to the analysis of interest groups and their influence on economic
policies. Influence implies that some individuals carry more weight than others
in the decisionmaking process. This asymmetry between individuals, in turn,
conflicts with the very idea of majority rule. Moreover, in many developing
countries the assumption that decisionmaking is based on majority rule is clearly
unrealistic. For these reasons, the model in this article will not draw from the
voting variety of the new political economy literature.
The Rules of the Game
Product and labor market distortions result from a full-information two-stage
game between the government and organized labor. The government is modeled
as in the Pigouvian approach to economic policy, that is, as a benevolent social
planner whose objective is to maximize national income (the size of the pie).
Therefore, if only the government got involved in the policymaking process, no
product market distortions would occur. The model implicitly assumes that there
are no externalities in production or consumption nor any departures from per-
Rama 333
fect competition that would make the distortion of product markets a second-
best policy. (For an analysis of the case with market imperfections, see Rama
1997.) A difference with the Pigouvian approach arises, however, because the
game involves interaction between the government and an agent that benefits
from the distortions and that aims to maximize its own income (its slice of the
pie).
Organized labor is not the only group to benefit from product market distor-
tions. Firms also likely gain from these distortions. Thus, firms play a crucial
role in the rent-seeking and the common agency approaches. In the model in this
article firms are supposed to be passive. This choice, aimed at analytical tracta-
bility, is also consistent with a widespread practice in the wage bargaining litera-
ture. The most popular model in this literature, namely the monopoly union
model, assumes indeed that firms have no strategic power beyond the "right to
manage," which implies that they always operate on their labor demand curve
(see Oswald 1985). By contrast, most of the political economy literature, which
often assumes perfect competition in factor markets, conspicuously excludes
organized labor. Papers dealing with macroeconomic policies, such as those by
Horn and Persson (1988), Holden (1991), Driffill and Schultz (1992), and Forteza
(forthcoming), are exceptions to this rule.
The control variables of the players differ in the two stages of the game. In the
second stage, the government and organized labor negotiate over sectoral wages
and product market distortions. In the first stage, they aim at building up bar-
gaining power for the negotiation. As in the war-of-attrition model, the players
have to decide whether to undertake a costly indivisible action. More specifi-
cally, I assume that in the first stage organized labor has to decide whether to
strike, while the government has to decide whether to redistribute income. Both
players have perfect information on what the other player does. Note that imperfect
information is not a necessary condition for strikes (or other costly actions) to be
decided by the players, as shown by Fernandez and Glazer (1991).
Because both the government and organized labor can-choose between two
different strategies, there are four potential outcomes or policy regimes. Each of
them leads, in the second stage of the game, to an equilibrium level of the prod-
uct market distortion and the sectoral wage. Table 1 defines the labels used here
to identify the four policy regimes. The labels capture the idea that labor mar-
kets have distinct features in different regions of the world (see Nelson 1991)
and that these differences may be large enough to think in terms of clusters,
rather than in terms of a continuous distribution.
Table 1. A Typology of Policy Regimes
Organized labor
Government Strikes Does not strike
Redistributes income European Scandinavian
Does not redistribute income Latin American East Asian
334 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
The labels should not be taken literally. For instance, a complete lack of in-
come redistribution in the Latin American and East Asian policy regimes does
not describe reality. Moreover, the chosen geographical regions are not actually
homogeneous in terms of their labor market features. For example, Austria should
belong to the Scandinavian rather than the European type of regime. In addi-
tion, other countries and regions fit the cases in table 1. Thus the label United
States could replace the East Asian label. Section III presents a more careful
assessment of the validity of this kind of clustering.
The actions undertaken in the first stage have an incidence on the bargaining
power of the players in the second stage, when they decide about wages and
product market distortions. If the government redistributes income and workers
do not strike (the Scandinavian case), I assume that the government is the
Stackelberg leader in the second stage. This corresponds to an autonomous state,
in terms of Rodrik's (1992) analysis of development policy. If, on the contrary,
the government does not redistribute income and workers do strike (the Latin
American case), then organized labor is the Stackelberg leader. This reflects a
subordinate state, according to Rodrik's terminology. As regards the two other
cases (European and East Asian), both players achieve a similar bargaining power.
Thus a Cournot equilibrium characterizes the second stage of the game.
The Players' Strategies
To ensure that decisions by the players are consistent over time, the game has
to be solved backwards. In the second stage, product market distortions and the
wage differential between the distorted sector and the rest of the economy lead
to a loss Yi for the government, where i indicates the prevailing policy regime
(i = EU for European, SN for Scandinavian, LA for Latin American, or EA for
East Asian). Departures from the Walrasian equilibrium also imply a gain of Z,
for organized workers. The assumptions concerning the bargaining power of the
players in the second stage of the game can be summarized as follows:
(1) YEA > YSN, YLA > YEU, ZEU > ZSN, ZLA > ZEA.
However, the loss Yi and the gain Zi are not yet the appropriate measures for the
sizes of the pie and the slice, respectively. The costs of strikes and income redis-
tribution during the first stage of the game also need to be taken into account.
The cost of income redistribution comes from the inefficiencies associated
with transfers. Even with redistribution based on nondistortionary taxation, losses
associated with revenue collection, resource administration, and the like would
occur. The weaker the administrative capabilities of the government, the more
significant these losses are. Assume that loss Yi is multiplied by a factor Q > 1
when the government redistributes income. Parameter 4 is an indicator of the
inefficiency of the state. When 4 is close to 1, income redistribution resembles
textbook compensatory payments, but as it increases, the waste component of
income redistribution becomes more important. At the same time, redistribution
increases the gain Zi of union members by a factor O' > 1.
Rama 335
The unions' decisions also affect the gain by union members. Assume that Z,
is multiplied by a factor E < 1 when organized labor strikes. The cost 1 - E) may
be associated with workdays lost, with casualties in union ranks during demon-
strations, or with the harassment, imprisonment, or death of union members.
Parameter ( is thus an indicator of union rights. When E0 is close to 1, strikes do
not entail any significant losses for union members, but as it decreases, confron-
tation becomes more costly. Strikes increase the government's loss Y, by a factor
0' > 1. Parameter 0' measures the disruption of economic activity created by
labor conflicts.
Consider the case where organized labor strikes. The government faces a loss
equal to 4E'YEU if it redistributes income and equal to O'YLA if it does not.
Therefore, it is indifferent between the two strategies for a level Os of state inef-
ficiency such that
(2) =LA
YEU
For a higher level of state inefficiency (4 > Os), the government prefers not to
redistribute income, even if the second-stage outcome is worse in terms of sectoral
wage differentials and product market distortions.
Similarly, when organized labor does not strike, the government faces a loss
equal to OYSN if it redistributes income and equal to YEA if it does not. The
inefficiency of the state �0 for which the government is indifferent between the
two strategies verifies that
(3) o = YEA
-SN
As before, for higher levels of state inefficiency (o > 00), the government prefers
not to redistribute income.
From the point of view of organized labor, when there is income redistribu-
tion the gain from striking is 4'0ZEu, while the gain from not striking is O'ZSN.
At some level OR of union rights, the labor movement is indifferent between the
two strategies. This level satisfies
(4) OR = ZSN
ZEU
If union rights are below OR, then unions prefer not to strike.
Finally, in the case with no income redistribution, the gains for union mem-
bers are E)ZLA and ZEA, depending on whether they strike. At the critical value
0o, the labor movement is indifferent between the two strategies:
(5) 0 = ZEA
ZLA
If union rights are below 00, the labor movement prefers not to strike.
336 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
The Policy Regimes
The prevailing policy regime depends on the relationship between the actual
values of parameters o and e and the critical thresholds �s, 00, OR, and 00. Table
2 shows that under the very general assumptions adopted so far, all four regimes
can be sustained as subgame perfect equilibriums in pure strategies. There is,
however, no one-to-one mapping from the level of parameters 0 and e to the
equilibrium policy regime. Multiple equilibriums characterize cases 5 and 6 in
table 2. Case 6, particularly, implies that the same levels of inefficiency of the
state and of union rights may be associated with either an autonomous or a
subordinate government. The existence of an equilibrium in mixed strategies
(case 7) reinforces the idea that the same level of parameters 0 and E may lead to
different policy regimes. However, in this article I deal with equilibriums in pure
strategies only, as a way to account for stable levels of distortions in product
and labor markets.
Starting from a given policy regime, can the economy switch to a regime char-
acterized by a lower loss for the government, that is, by a larger size of the pie?
The discussion above made clear that larger welfare losses characterize regimes
where the labor movement chooses to confront than those where it prefers to
cooperate. What could induce a shift in the strategy of the labor movement from
confrontation to cooperation? To analyze this issue requires a more detailed
Table 2. Feasible Policy Regimes
Parameter thresholds for inefficiency Policy regime
Case of the state, 4, and union rights, E) with Nash equilibrium
1 0 < min ( 0, , s) and (9 > E)R European
or
E > min 0,)R I and <Os
2 4 < min f 0,, 0,) and 9 < 9, Scandinavian
or
O < max {\ 0R I E and 4 <
3 ) > max { 0,, }and 9 > 90 Latin American
or
E > min {00 ,E) and 4> p
4 o > max { 0, 4)}and () < 0, East Asian
or
O < max { 0,R I and 0 >0
5 0, < O < 0, and (R < E < 0, East Asian and European
6 0, < b < %0 and E0 < e < 0R Latin American and Scandinavian
7 Os < b < %, and ER < E < 00 No equilibrium in pure strategies
or
%, <4 < 4), and 0, < 0 <OR
Note: The level of inefficiency of the state at which the government is indifferent about redistributing
income is 9o when organized labor does not strike and o, when organized labor strikes. The level of
union rights at which the labor movement is indifferent about striking is 00 when government does not
redistribute income and 0R when government redistributes income.
Rama 337
model. Thus section II presents an example in which a few additional assump-
tions on the second stage of the game make the functions Yi and Zi depend on
meaningful parameters, such as the flexibility of labor demand, the prevailing
macroeconomic conditions, or the size of the sector affected by product market
distortions.
II. AN EXAMPLE
Import tariffs provide a useful example of a product market distortion. By
making foreign goods more expensive, they shift domestic demand toward home-
produced substitutes. Labor demand in the import-substitution sector increases,
which allows firms in this sector to hire a larger number of workers, to pay
incumbent workers more than their alternative wage, or both. Organized labor
can therefore benefit from the tariffs. But two well-known welfare costs arise in
the process. First, employment in the import-substitution sector typically in-
creases, causing an inefficient allocation of labor. And second, consumption
decisions, based on distorted relative prices, cause a loss of consumer surplus.
Although these two welfare costs are neat in the case of tariffs, they also
obtain for other product market distortions. Output or export subsidies, cheap
inputs from publicly owned firms, or any other policy that shifts up labor de-
mand in some specific sector lead to the same kind of inefficiency in labor allo-
cation as an import tariff. Moreover, all of these transfers have to be financed in
one way or another, thus giving rise to a tax burden. The loss of consumer
surplus mentioned above is just a particular case of the tax burden. But some
deadweight loss exists whatever the tax base. For instance, if the government
finances the transfer by printing money, then the welfare cost comes from the
drop in the demand for real balances.
The appendix works out the analytical expression of the functions Yi and Z,
in the chosen example, provided that three simplifying assumptions hold. First,
I assume that a linear function of the product and labor market distortions ap-
proximates demand for labor, L, in the distorted sector:
(6) L = 1- ci(W - D -1)
with ox > 0. In this expression, W is the level of the sectoral wage, and D is the
level of the product market distortion (say, the tariff rate). Wages in the rest of
the economy are normalized to 1, so that if there were no distortions in product
and labor markets, L would equal 1.
With the chosen specification, parameter oc measures labor market flexibility.
The level of ax is an upward function of the wage elasticity of labor demand
indeed (they both coincide for L = 1). The more stringent the hiring and firing
restrictions facing the firm, the lower is oc. In terms of this simple example, a
deregulation of the labor market would lead to an increase in the value of pa-
rameter oc, which thus captures the labor market rigidity stemming from the
labor market policies and institutions in force.
338 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Second, I assume that the tax burden associated with the product market
distortions is a quadratic function of the level of the distortion. This assumption
corresponds to the case in which the tax base is a downward linear function of
the tax rate. Consider the import tariff example again. If the domestic demand,
C, for goods had the form C = C0 - O3D, where D denotes the tariff rate, then the
loss of consumer surplus would amount to OD2 / 2, the area of the correspond-
ing Harberger triangle. The calculation would be similar if C were money de-
mand and D were the additional inflation tax required to finance the transfer, in
which case a would measure the decline in demand as the inflation rate increases.
The inflation tax example suggests that parameter a depends on macroeconomic
conditions. Indeed, the elasticity of a tax base with respect to the corresponding
tax rate is likely to increase with fiscal pressure. For instance, announcing a 1
percent increase in public expenditures could have little impact on the demand for
real balances if the initial budget was in equilibrium but could lead to its collapse if
the country was on the verge of hyperinflation. Hereafter, I assume that adverse
shocks such as a deterioration in the terms of trade, an increase in the debt service
burden, or a natural disaster lead to a higher value of parameter ,, which in effect
summarizes the prevailing macroeconomic conditions.
Third, I assume that workers in the protected sector bear only a fraction a of
the social cost resulting from product and labor market distortions. If the social
cost were uniformly distributed across all the population, the fraction 6 would
boil down to the percentage of the labor force employed in that sector. Assum-
ing that these workers are unionized, a would also measure the size of the trade
union with which the government has to deal. But from a policy perspective, it is
convenient to interpret parameter 6 as the scope of the product market reforms
under consideration. An attempt to liberalize trade across the board is thus asso-
ciated with a larger value of a than a reduction of import tariffs in a narrowly
defined sector.
A Diagrammatical Representation
Figure 1 depicts the objective functions Y and Z in the plane (W - 1 , D) of
sectoral wage differentials and product market distortions. The appendix pro-
vides the derivation of this figure. Curves like Y = YSN in figure 1 represent the
combinations of distortions W - 1 and D that achieve a constant welfare loss in
the second stage of the game. The closer these curves are to the origin, the lower
are the distortions and, therefore, the lower is the welfare loss. Curves like Y =
YSN become vertical at the intersection with the upward-sloping line labeled
"government's reaction function." This line indicates the lowest welfare loss the
government can achieve through the product market distortion, D, for a given
wage differential, W - 1. It consequently represents the best policy response of a
subordinate government to the wage decisions made by a trade union that be-
haves as a leader.
Similarly, curves like Z = ZLA in figure 1 represent the combinations of distor-
tions W - 1 and D that secure a constant gain Z to union members in the second
Rama 339
Figure 1. Product and Labor Market Distortions
Product market distortion (D) Union's
reaction
Z = LEA /function
Government's
reaction
function
01
Sectoral wage differential (W- 1)
stage of the game. The farther is this curve from the origin, the larger are the
market distortions and, therefore, the higher is the gain to organized labor. Curves
like Z = ZLA become horizontal at the intersection with the upward-sloping line
labeled "union's reaction function." This line indicates the maximum gain the
labor movement can achieve through the labor market distortion, W - 1, for a
given product market distortion, D. It consequently represents the best reply of
a subordinate labor movement to the policy decisions made by a government
that behaves as a leader.
Equilibrium Distortions
Based on figure 1, it is relatively straightforward to characterize the different
policy regimes allowed by the model. When both players have similar bargain-
ing power in the second stage of the game (that is, in the East Asian and Euro-
pean regimes), the equilibrium distortions lie at the intersection of the corre-
sponding reaction curves. The level of the labor market distortion therefore
maximizes the gain for trade union members given the level of the tariff set by
the government, while the level of the tariff minimizes the welfare loss given the
sectoral wage differential created by union activities. Note, however, that aggre-
gate welfare in the East Asian regime exceeds that in the European regime, be-
cause the former regime does not entail deadweight losses from income redistri-
bution and labor conflicts.
When one of the players is strong enough to behave as the leader, it can take
advantage of the fact that its own decisions affect the decisions made by the
340 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
other player. For instance, a government behaving as a leader (the Scandinavian
regime) does not treat the sectoral wage differential W - 1 as given. It rather
incorporates in its decisionmaking process the way organized labor adjusts this
wage differential to changes in the product market distortion D. The govern-
ment therefore sets the level of D so as to pick up the point of the union's reac-
tion function that minimizes the welfare loss Y. In figure 1, that point is SN.
The analysis is symmetrical when organized labor behaves as the leader. In
this case, the trade union can take advantage of the fact that the government
tries to minimize the welfare loss from a sectoral wage differential, W - 1, by
adjusting the level of the product market distortion, D. The optimal wage differ-
ential, from the point of view of organized labor, corresponds to the point of the
government's reaction function that maximizes the gain, Z. In figure 1, that
point is LA.
More specifically, the appendix shows that
(7) � < DSN< DEA= DEU< DLA, 1 < WSN < WEA = WEU < WLA-
The biggest product market distortion is therefore associated with the Latin
American case and the smallest one with the Scandinavian case, whereas the
European and East Asian cases occupy an intermediate position. Similarly, wage
differentials are narrow in Scandinavia and wide in Latin America. Again, Eu-
rope and East Asia lie between these two extremes.
Although the ranking of product and labor market distortions was derived
under simplifying assumptions, its validity is more general. Equation 7 holds
true provided that aggregate welfare decreases in line with distortions, while the
well-being of union members increases. Indeed, none of the players can be worse
off by taking advantage of the way the other player makes its decisions. This
additional bargaining power could be irrelevant, but never harmful. In general,
it allows the player who gets it to attain a higher value of its objective function.
If the government gets this extra power, it can reduce the aggregate welfare loss,
which in turn amounts to reducing the size of product and labor market distor-
tions. By contrast, if unionized workers get it, they can secure a larger slice of
the pie, and this in turn requires that product and labor market distortions in-
crease. In more formal terms, the ranking of distortions summarized in equation
7 is verified provided that the reaction functions of the two players are upward
sloping, which is likely.
III. SOME EMPIRICAL EVIDENCE
In spite of its simplicity, the model in this article reproduces quite well some
regularities observed in practice. Table 3 reports interesting data in this respect.
These data, arranged under the form of six variables, refer to income redistribu-
tion, labor conflicts, and market distortions in different regions of the world. All
the countries in the table belong to one of the four regions considered in the
model (EU, SN, LA, and EA). Regions are not always defined in the conven-
Rama 341
tional way though, as shown by the inclusion of Austria in Scandinavia and the
exclusion of Denmark, Finland, Greece, Sweden, and Portugal from Europe.
Also, table 3 excludes countries with missing data for three or more of the six
variables. Table 3 reports regional averages for all variables. These averages
suggest that regions differ by more than a matter of degrees.
Column A in table 3 reports expenditure on social sectors, measured as a
fraction of total expenditure by the central government. Social sectors include
education, health, social security, and housing, whereas total expenditure in-
cludes defense. I use population weights to calculate regional averages. Column
A shows that social expenditures represent around two-thirds of the budget in
Europe and Scandinavia, but only one-third in most countries in East Asia and
Latin America. The exceptions are Indonesia, where the fraction of social ex-
penditures is even lower (about one-fifth of total expenditures), and Costa Rica
and Uruguay, where the fraction is much closer to the European and Scandina-
vian figures. Except for Costa Rica and Uruguay, it is fair to conclude that in-
come redistribution by the government is not a salient feature of the two regions
in the South.
Labor conflicts, by contrast, occur both in the North and in the South. Many
countries in Europe and Latin America were characterized by a significant num-
ber of strikes in the second half of the 1980s. Column B in table 3 shows that
roughly 10 percent of nonagricultural workers in these two regions participated
in strikes. The figures in column C suggest that the resulting economic cost was
far from trivial: more than half a day per worker was lost in Europe every year
and one full day was lost in Latin America. At the other end of the spectrum,
Scandinavia and East Asia display very peaceful labor markets. Only in Finland
and the Republic of Korea is the number of days lost significant, although it is
still much lower than the European and Latin American averages.
The regional ranking in terms of labor market distortions matches the one
predicted by the model. Column D of table 3 reports the ratio between the legal
minimum wage and gross domestic product (GDP) per capita, a useful indicator
of labor market distortions. Based on this ratio, Latin America has the most
distorted labor markets, and Scandinavia has the least distorted ones, with East
Asia and Europe lying between these two extremes. Although the regional aver-
ages suggest little difference between Europe and Scandinavia, the similarity is
mostly due to one country (Sweden) driving up the Scandinavian average.
The very existence of a nationwide legal minimum wage provides yet another
way to confirm that the model correctly predicts the regional ranking of labor
market distortions. All Latin American countries for which there are data have a
legal minimum wage, and the variance in the ratios between the minimum and
GDP per capita across countries is quite narrow. At the other end, no country in
Scandinavia, except Sweden, has a legal minimum wage. The picture is mixed in
East Asia and Europe.
Table 3 reports two different measures of product market distortions. Col-
umn E reports the trade intensity ratio constructed by Pritchett (1993). This is
Table 3. Indicators of Policy Regimes across Countries, 1980s
Government Workers Minimum waged Trade
expenditure on involved in Worker-days lost (percentage of intensity Average tariff
social sectorsa strikesb to strikes' GDP ratio' ratef
(percent) (per 100 workers) (per 100 workers) per capita) (residual) (unweighted)
Country A B C D E F
East Asia
Indonesia 0.12 0.1 0:4 34.5 19.7 14
Japan - 0.2 0.5 28.1 -40.0 -
Korea, Rep. of 0.28 1.5 19.1 40.5 0.9 23
Malaysia 0.33 0.3 1.1 0.0 24.1 14
Singapore 0.34 - - 0.0 145.9 0
Thailand 0.30 0.1 1.0 79.2 -10.4 31
East Asian average 0.19 0.3 2.5 37.4 -30.4 20
Europe
Belgium 0.58 0.6 3.6 63.8 62.5 5
France 0.69 0.8 5.3 52.7 -27.3 5
Germany 0.69 - - 0.0 -17.1 5
Ireland 0.54 3.7 26.1 - 70.9 5
Italy 0.53 24.1 191.0 0.0 -19.2 5
Netherlands 0.63 0.4 1.4 72.4 30.9 5
Spain 0.64 15.5 36.2 31.1 - 5
United Kingdom 0.49 3.4 23.5 0.0 -19.4 5
European average 0.61 9.5 59.4 18.8 -14.0 5
Latin America
Argentina 0.43 - - - -3S.9 27
Bolivia 0.45 - - 37.6 - 20
Brazil 0.39 19.2 142.1 34.0 -2.0 55
Colombia 0.49 0.4 0.1 80.0 -34.5 38
Costa Rica 0.62 2.0 15.1 87.7 0.2 24
Ecuador - 1.1 22.7 59.9 3.0 38
Guatemala - - - 58.6 -29.1 24
Mexico 0.26 0.5 9.7 40.9 -23.0 30
Peru 0.22 10.8 440.7 - -24.7 36
Uruguay 0.62 - - 31.2 -38.0 31
Venezuela 0.38 0.3 4.7 53.7 -12.6 30
Latin American average 0.36 9.9 99.6 43.9 -16.7 39
Scandinavia
Austria 0.70 0.2 0.2 0.0 - 12
Denmark 0.52 3.9 13.0 0.0 -12.2 5
Finland 0.58 12.5 33.7 0.0 -17.0 9
Norway 0.56 1.8 10.1 0.0 2.0 5
Sweden 0.60 - - 60.7 -8.7 4
Scandinavian average 0.60 4.1 12.6 16.9 -8.8 7.4
-Not available.
a. Percentage of total central government expenlditure, including defense, for all available years in the 1980s. The numerator includes central government
expenditures on education, health, social security-and welfare, and housing and community amenities. Regional averages were obtained using population weights.
b. Percentage of the nonagricultural labor force in the 1980s. Regional averages were obtained using population weights.
c. Values are for the nonagricultural labor force in the 1980s. Regional averages were obtained using population weights.
d. Values are for 1985-89. Regional averages were obtained using population weights.
e. The trade intensity ratio is the residual obtained upon regressing the ratio of total merchandise trade to GDP on population, area, GDP per capita, GDP per
capita squared, the transport costs, and a dummy for oil-exporting countries. Regional averages were obtained using the 1985 GDP in U.S. dollars as weights.
f. For Europe and Scandinavia, these are pre-Uruguay round average most favored nation rates. Regional averages were obtained using imports from all
countries except free trade areas (for Europe and Scandinavia) and 1985 GDP in U.S. dollars (for East Asia and Latin America) as weights.
Source: For column A, International Monetary Fund (various years); for columns B and C, International Labour Organisation (various years); for column D,
World Bank data; for column E, Pritchett (1993); and for column F, Erzan and others (1989).
344 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
the residual of a cross-country regression explaining the openness coefficient of
the economy as a function of an array of country characteristics, including total
population, the level of development, geographic area, transport costs, and
whether the country is an oil exporter. The more negative the residual, the more
closed the economy. It seems thus natural to assume that the country deviation
from its predicted openness reflects, at least to some extent, the trade orienta-
tion of government policies. The more negative the deviation, the stronger the
antitrade bias. The regional ranking of product market distortions derived from
this indicator is mostly in accordance with the predictions of the model. The
exception is East Asia, which appears to be the least open to trade. But the result
is due exclusively to the case of Japan, a country characterized by a strong
antitrade bias and a significant weight at the regional level. The antitrade bias of
Japan is not surpassed by that of any other country in table 3.
Column F in table 3 reports the other measure of product market distortions,
the average tariff rate. Based on this measure, Latin America emerges as the
region with the most distorted product markets (at least in the 1980s), whereas
Scandinavia and Europe have the least distorted product markets. As predicted,
East Asia stands in an intermediate position. The only aspect in which the fig-
ures in this column do not fit the predictions of the model is the observed simi-
larity between Europe and Scandinavia. Nontariff barriers, which are common
in Europe but much less so in Scandinavia, suggest that this similarity may be
fictitious.
All things considered, the data reported in table 3 show that the simple, heu-
ristic model presented above reproduces quite well some important empirical
regularities. Of course, these data do not provide a real test of the model. They
only illustrate the plausibility of the story told by the model, which makes its
policy implications worth analyzing. The next section shows that these implica-
tions are consistent with previous results from the new political economy litera-
ture. The evidence reported in this section implies that they could also be rel-
evant in practice.
IV. THE POLICY IMPLICATIONS
The example developed in section II shows that all feasible regimes have strictly
positive product market distortions and sectoral wage differentials. Product and
labor market distortions do not occur because policymakers have a wrong model
of the economy; the government aims at maximizing national income and cor-
rectly identifies the way market distortions affect the size of the pie. The reason
for bad policies is the strategic interaction between the government and an eco-
nomic agent trying to get a larger slice of the pie. This interaction accounts for
one of the anomalies reported in the introduction, namely the pervasiveness of
market distortions across countries.
The largest distortions and wage differentials obtain in the Latin American
case, where there is no income redistribution. This result is consistent with the
Rama 345
findings by Sachs (1989), who argues that high income inequality in Latin America
contributes to intense political pressure to raise the incomes of lower-income
groups, which in turn contributes to populist policies and weak economic per-
formance. To the extent that large market distortions reduce the growth rate of
the economy, the result is also consistent with the findings by Alesina and Rodrik
(1994) and Persson and Tabellini (1994) on the negative relationship between
inequality and long-run growth.
However, these results do not imply that income redistribution will alleviate
market distortions. In the model, government decides to redistribute income in
an effort to maximize national income. The government might prefer not to
redistribute (as in the Latin American case) because of the high inefficiency of
the state, which implies that the deadweight losses from redistribution surpass
the deadweight losses from large product market distortions and sectoral wage
differentials. Therefore, as long as the efficiency of the state remains unchanged,
trying to reduce distortions by redistributing income would only decrease
welfare.
Removing Distortions versus Changing the Policy Regime
The equilibrium level of product and labor market distortions maximizes each
player's payoff, given the actions undertaken by the other player. The exact level
of these distortions depends on the values of parameters such as a, f, and (. It
also depends, for any given value of these parameters, on the prevailing policy
regime, that is, on the players' bargaining power in the second stage of the game.
But as long as neither the policy regime nor the level of these parameters is
modified, removing the product market distortion necessarily reduces the play-
ers' payoffs.
The government being benevolent, a drop in its payoff equals a reduction in
the welfare level of the representative individual. The model can therefore ac-
count for two of the anomalies reported in the introduction, namely the unex-
pectedly high costs of economic reforms and widespread recidivism. A govern-
ment with a naive view of policymaking would assume that product market
distortions reduce welfare and go ahead with economic reform. But this naive
government would also realize ex post that welfare is reduced and would prefer
to withdraw the reform.
In the model, the only sustainable changes in the level of product market
distortions and sectoral wage differentials come from changes in parameters o,
E, a, 3, and 6. Modifying the first two seems out of the reach of policymakers,
at least within the time frame usually considered in discussing economic reforms.
The inefficiency of the state is better seen as a constraint in the short run. And
union rights are usually shaped by historical and cultural factors. Therefore, the
rest of the discussion deals with the consequences of changes in parameters a, P,B
and (.
These changes have two effects. As long as the feasible policy regime is not
affected, they produce smooth variations in the equilibrium level of the endog-
346 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 4. Comparative Statics by Policy Regime
(change in parameter)
Type of policy regime Labor market Macroeconomic Scope of product
and variable flexibility, a conditions, a market reforms, a
East Asian and European policy regimes
Product market distortion, D -- -
Sectoral wage differential, W - 1 -
Welfare loss, Y - +
Scandinavian policy regime
Product market distortion, D -
Sectoral wage differential, W - 1 -
Welfare loss, Y - +
Latin American policy regime
Product market distortion, D 0
Sectoral wage differential, W - 1 -
Welfare loss, Y
Note: - indicates a negative, + indicates a positive, and 0 indicates a zero value for the partial
derivative of the market distortion or welfare loss with respect to the parameter.
Source: Author's calculations.
enous variables of the model. Table 4 summarizes these variations. It shows the
sign of the partial derivatives of the market distortions D and W - 1 and of the
welfare loss Y, with respect to a, I, and 6 in each of the policy regimes. After
some threshold, however, changes in parameters a, g3, and 6 may also lead to a
regime switch and, therefore, to a jump in the equilibrium level of D, W- 1, and
Y. These parameters have an incidence on the critical values Os, 00, OR, and e0
that determine the strategy of the players in the first stage of the game.'
Discontinuities in the equilibrium level of the endogenous variables can be
used to account for policy reforms. The rest of this section deals with economic
reform in countries in the South, focusing on the switch from the Latin American
to the East Asian policy regime. As a necessary condition for this switch, workers
must prefer not to strike in the first stage of the game. This requires an increase in
the critical threshold e0 for any given level E) of union rights. Because the thresh-
old depends on parameters a, pj, and a, it is possible to assess how labor demand
flexibility, macroeconomic conditions, and the size of the protected sector affect
the final outcome in terms of product and labor market distortions.
The Scope of the Reforms
Parameter a is higher the larger the share of the labor force employed in the
protected sector. Consequently, its level is an indication of both the scope of the
1. The distinction between changes within a regime and changes in the prevailing regime is reminiscent
of the fixed-price equilibrium theory (see, particularly, Malinvaud 1977 and Benassy 1982). However,
the existence of several regimes results here from the indivisibility of the choices the players face in the
first stage of the game, not from any price rigidity. Particularly, wages are endogenous in the model in
this article, while they are exogenously determined in the fixed-price equilibrium theory.
Rama 347
reforms under consideration and the size of the labor movement with which the
government has to deal. The appendix shows that
(8) deo >0
da
which means that the labor movement is less willing to confront as c increases.
This is because union members are consumers too. As workers, they stand to
gain from higher import tariffs in their own sector of activity, regardless of what
the tariff level is for other sectors. But as consumers, they are better off if tariffs
for all other sectors are low. Unions will thus favor a higher protection rate
when the issue is the tariff rate for a specific sector, but not necessarily when the
general import tariff is at stake.
The policy implication is that ambitious economic reforms, aimed at removing
product market distortions across the board, may get more support from work-
ers than reforms characterized by a narrow sectoral focus. This conclusion reso-
nates with the messages from the literature on the virtues of corporatism (see
Bruno and Sachs 1985, Freeman 1988, and Rama 1994, among others). In this
literature, an encompassing labor movement internalizes to a larger extent the
aggregate effects of its decisions than a set of smaller and uncoordinated trade
unions. As a result, this larger labor movement is also more likely to favor coop-
eration and wage moderation. Cooperation may take the form of a social pact.
But the argument holds even in the absence of an explicit agreement. A reform of
many sectors at a time may indeed replicate, for each of the sectoral unions, the
benefits that would be internalized by an encompassing labor movement.
At the same time, this argument provides a warning against ambitious reform
programs that are not backed by an explicit negotiation with the involved social
partners. Although sectoral unions are expected to internalize a larger share of
the benefits from the reform when the reform affects many sectors at once, the
share might not be large enough to induce the unions to cooperate. In analytical
terms, 0 would remain above 00 for each of the individual unions, so that all of
them would choose simultaneously to incur the costs of striking. This possibility
corresponds to a scenario of widespread unrest and resistance to economic
reform.
The Benefit of Adverse Shocks for Economic Reform
Intuitively, adverse shocks, such as a deterioration in the terms of trade, an
increase in the debt service burden, or a natural disaster, reduce welfare. How-
ever, such shocks might increase welfare because of the strategic interaction
between the players. The economy is not in the first-best equilibrium in any of
the policy regimes. Accordingly, changes in the value of the parameters that
would reduce welfare in the absence of any pecuniary externalities can actually
improve welfare in their presence. In the new political economy literature, this
paradox has been illustrated by Drazen and Grilli (1993).
348 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Adverse shocks reduce welfare by increasing the value of parameter ,B and,
consequently, the marginal tax burden associated with product market dis-
tortions. But adverse shocks also modify the incentives facing the players.
Because product market distortions become more costly, the equilibrium lev-
els of both D and W - 1 decrease in all of the regimes (see table 4). The net
effect of a higher : on aggregate welfare thus reflects two opposite forces.
On the one hand, the inefficiency associated with any given level of the prod-
uct market distortion is higher. On the other hand, the equilibrium levels of
both the product market distortion and the sectoral wages are lower. Under
the assumptions of the model, the first effect dominates in all regions, except
the Latin American case.
Adverse shocks may, however, trigger economic reform. In terms of the model,
the critical threshold 00 verifies
(9) dE0 >0
dfP
(see the appendix). Therefore, organized labor is less likely to strike the more
severe are the shocks suffered by the economy. Confrontation becomes a costly
way to secure a higher level of product and labor market distortions, and the net
payoff from these distortions falls as j increases.
Significant reforms may be launched in bad times and, furthermore, be sup-
ported by trade unions in the distorted sectors. Thus the model accounts for yet
another of the anomalies reported in the introduction. This relationship between
adverse shocks and successful reforms, however, raises the issue of whether re-
forms may survive once the shocks recede. In this sense, labor market reforms
may be crucial.
The Role of Labor Flexibility
Lower hiring and firing restrictions increase the value of parameter a. As a
result, changes in the level of either the product market distortion or the sectoral
wage differential have larger effects on sectoral employment. From the point of
view of trade unions, the higher flexibility of labor demand leads to a lower
monopoly power, so that the optimal sectoral wage associated with any given
level of the product market distortion decreases. From the government's point
of view, the higher flexibility of sectoral labor demand increases allocation costs,
so that the optimal product market distortion associated with any given level of
the sectoral wage differential increases.
The fall in the equilibrium wage differential is large enough to lead to a de-
crease in the equilibrium tariff, despite the greater willingness of the government
to distort product markets. Under the assumptions of the model, tariffs remain
unchanged only in the Latin American policy regime. The equilibrium level of D
is a decreasing function of parameter a in all of the other policy regimes. Be-
cause W - 1 is a decreasing function of cx in all of the regimes, the welfare loss Y
Rama 349
is unambiguously reduced as labor demand becomes more flexible. Put differ-
ently, within any given regime aggregate welfare is higher the higher the flexibil-
ity of the labor market.
However, increased labor market flexibility may also trigger a switch in the
economic regime. The appendix shows that
(10) d-o < 0
dax
which means that, other things being equal, organized labor is more likely to
cooperate the less flexible is labor demand. This is because incumbent workers
lose from a higher labor demand flexibility both in the Latin American and the
East Asian cases, but the loss is larger in the latter than in the former.
The possibility that reform will fail because of increased labor market flex-
ibility questions the intuitive explanation discussed in the introduction, ac-
cording to which labor market rigidities were to blame for recidivism and
bad economic performance. In terms of the model, this explanation makes
sense only as long as the policy regime remains unchanged. It cannot be ap-
plied to regime switches, that is, to discontinuous jumps in the level of prod-
uct market distortions. Equation 10 provides some support to Freeman's
(1993) idea that labor market interventions should be considered as a basic
ingredient in the political economy of reforms. Interventions that are costly
and distortive at first glance can influence the strategy of those who stand to
lose from reforms.
V. CONCLUSIONS
Most economists acknowledge that first-best policies are seldom implemented
because of politics. Yet politics are absent from the standard tools of the trade.
Economists tend to evaluate the merits and demerits of different policies assum-
ing that factor markets are perfectly competitive and governments are in full
control of the policy instruments. Although many agree that these are not realis-
tic assumptions, they fear that introducing political considerations could under-
mine analytical rigor. Basing policy advice on first-best solutions and then hav-
ing policymakers struggle with the real-world pressures to drop them therefore
appears to be a safer alternative. The problem is that this alternative may not
even lead to a second-best outcome.
The main contribution of the recent political economy literature has been to
bring these real-world pressures into the picture, without compromising analyti-
cal rigor. The key in this respect is to evaluate what different groups stand to
gain or to lose from the policies under consideration and to link their actions to
these gains and losses. Utility maximization by rational individuals remains the
cornerstone of the analysis, but the control variables are not restricted to prices
and quantities anymore. Moreover, because the number of active groups is much
350 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
smaller than the number of consumers or producers, optimal decisions by each
of them may depend on what the other groups do.
Although I adopt the political economy approach in this article, my ambition
is deliberately modest. I present no new technique or analytical breakthrough.
Instead I highlight the role of organized labor as an active interest group and
derive some implications for economic reform. Organized labor has played a
very minor role indeed in the new political economy literature. Although the
literature has explained some macroeconomic policies based on the interaction
between the government and trade unions, it usually has analyzed microeconomic
distortions under the assumption of perfectly competitive factor markets. Yet
policymakers across the world would agree that organized labor is one of the
most relevant interest groups they have to face.
The simple, heuristic model of the economy presented here can reproduce
within a unified framework a series of results derived elsewhere in a more rigor-
ous way, but with no reference to organized labor. Some of these results, such as
the existence of several policy regimes and the need to induce a regime switch to
have a successful reform, obtain under very general assumptions. I introduce
more specific assumptions to analyze the links between the key parameters of
the economy and the equilibrium level of product and labor market distortions.
But even these more specific assumptions are fairly plausible, thus suggesting a
general validity of the results.
The model accounts in an intuitive way for some of the anomalies observed in
the process of economic reform, both in industrial and in developing countries.
Reforms that seem unambiguously welfare-improving from a Pigouvian perspec-
tive may actually entail large social costs if the policy regime remains unchanged.
These unexpected costs, in turn, may account for widespread recidivism. Con-
versely, adverse shocks that would be expected to reduce welfare may actually
increase it, by triggering a change in the strategy of organized labor from con-
frontation to cooperation. Hence explicit or implicit social pacts emerge under
unfavorable circumstances.
The labor market implications of this simple model provide qualified support
to the prevalent view in the development community. The World Bank (1995)
argues in favor of limited government intervention in labor markets and more
reliance on collective bargaining between workers and employers. In terms of
the model, less stringent hiring and firing restrictions (that is, a higher value of
ax) and a larger percentage of the labor force enrolled in trade unions (that is, a
higher value of a) are associated with lower levels of distortions in every policy
regime. A higher level of union membership may also make organized labor
more willing to cooperate, hence facilitating a favorable switch in the policy
regime. More flexible labor demand may trigger confrontation, though, thus
reducing the chances of successful economic reform.
Rama 351
APPENDIX. DERIVATION OF THE RESULTS OF THE MODEL
The social cost Y of product and labor market distortions can be calculated
based on the assumptions that the labor demand schedule is linear and the tax
burden is quadratic. The linearity of labor demand, L, implies that the Harberger
triangle measuring the loss of producer surplus can be approximated as [(L - 1)
(1 - X)] / 2, with X being the marginal productivity of labor in the distorted
sector. (In analytical terms, X = W - D is the wage level for which L would
remain unchanged, if the distortion were removed.) The quadratic nature of the
tax burden means the loss of consumer surplus equals j3D2 /2. The loss function
Y can therefore be written as
(A-1) Y=a(l-W-D)2/2+ PD2/2
with Y> 0, and Y = 0 when there are no sectoral wage differentials and the
product market is not distorted.2 In equation A-1, a is a parameter that mea-
sures labor market flexibility, W is the level of the sectoral wage, D is the level of
the product market distortion, and ,6 is a parameter that summarizes the prevail-
ing macroeconomic conditions.
The rent that the sectoral trade union extracts by pushing wages up equals
L(W - 1). But product and labor market distortions entail a loss aY for union
members (a is a parameter that measures the scope of the product market re-
forms under consideration). Using equations 6 and A-1, the gain function Z can
be written as follows:
(A-2) Z = [1 + a(l - W+ D)](W- 1) - a[x(l - W+ D)2 /2 + f D2 / 2].
Economically meaningful equilibriums are characterized by levels of W and D
such that Z> 0, with Z = 0 when the product market is not distorted and wages
are equal across sectors.
The combinations of W - 1 and D for which the loss function Y remains
constant verify
(A-3) dD _ a[D- (W -1)]
dY (a+,B)D-a(W-1)
Equation A-3 corresponds to curves like Y = YSN in figure 1. These curves are
vertical for
(A-4) D= (W -1).
Equation A-4 represents the government's reaction function in the second stage
of the game. It indicates the level of the product market distortion D that mini-
2. Note that when the distortion is a trade barrier, it may be feasible to put some numbers in equation
A-1. If wages were equal across sectors, then the loss of producer surplus would be equal to aD2/2.
Because the loss of consumer surplus is f3D2/2, the ratio between the two welfare losses (usually calculated
when assessing the costs of protection) provides an estimate of the ratio cc / P.
352 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
mizes the loss Y for a given level of the sectoral wage differential W - 1. By
contrast, the curves are horizontal when
(A-5) D = W- 1.
The closer these curves are to the origin, the lower is the loss Y.
The combinations of W - 1 and D for which the gain function Z of the labor
movement is constant, in turn, verify
(A-6) dD ax(2 + a)(W -1) -1- a(1 + a)D
dY (1 + c+)(W -1) - (a + ,3)aD
Equation A-6 corresponds to curves like Z = ZLA in figure 1. These curves are
horizontal when
(A-7) W I = 1 + o(l + )D
(A-7) a(2 + a)
Equation A-7 represents the union's reaction function in the second stage of the
game. It indicates the level of the sectoral wage differential W - 1, which maxi-
mizes Z for a given level of the product market distortion D. The constant-gain
curves become vertical for
(A-8) D= (+ (W-1).
The farther these curves are from the origin, the larger is the union's gain.
In the Scandinavian policy regime (SN), the government behaves as a
Stackelberg leader. It therefore minimizes the welfare loss Y, given by equation
A-1, under the constraint represented by the union's reaction function, given by
equation A-7. It follows that
(A-9) DSN = ) WSN 1 + a[a + (2 + -a)_
In the Latin American policy regime (LA), organized labor is the Stackelberg
leader. It takes the government's reaction function, given by equation A-4, as a
constraint in the maximization of its objective function Z, given by equation
A-2. The solutions to this problem verify
(A-10) DLA = 2 WLA = 1 + cc3(2+
P3(2 + a7)' x(
Finally, in the European and East Asian policy regimes (EU and EA, respec-
tively), a Cournot equilibrium emerges. The latter corresponds to the intersec-
Rama 353
tion of the reaction curves represented by equations A-4 and A-7. The market
distortions are
(A-li) DEA =DEU W 1 ' EA =WEU = 1 + a+
ax�+ (2 + c)' WxEoA -.W2
Having solved for the equilibrium levels of sectoral wage differentials and
product market distortions, it is possible to determine the critical thresholds of
parameters 0 and 0 (the efficiency of the state and the cost of unions striking,
respectively) in the first stage of the game. When organized labor strikes, the
level Os of the inefficiency of the state for which the government is indifferent
regarding income redistribution is
(A-12) O 2 [a+ (2+o)2
132(2 + )2
(see equations 2, A-1, A-10, and A-11). When the union does not strike, the
critical threshold becomes p0, with:
(A-13) =o (a+= )[a+1(2+a)2]
[ac+ P(2+0)]2
(see equations 3, A-1, A-9, and A-11).
Similarly, when the government redistributes income, the critical threshold
0ER of labor movement rights for which the trade union is indifferent regarding
strikes is given by
(A-14) 0ER [a + J3(2 + a)]2[1(2 + 6)3 + 2a1x(2 + 0)2 + a(2a - Doa)]
(a+ 3)[2a+ P(2+a)][cx + P3(2+G)2]2
(see equations 4, A-2, A-9, and A-11). When there is no income redistribution,
the critical threshold becomes 00, with
(A-IS5) 00 = 1(2 + a)[2a + ,(2 + a)]
[a + P(2 +)]2
(see equations 5, A-2, A-10, and A-11).
Equations A-12 and A-13 imply Os > Oo, which means that case 6 in table 2
cannot take place. In other words, there is no combination of the inefficiency of
the state and union rights such that the government could be indistinctly subor-
dinate or autonomous. However, case 5 still holds. Therefore, the example con-
firms that there is no one-to-one relationship between the efficiency of the state
and union rights, on the one hand, and the policy regime, on the other hand.
354 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Alesina, Alberto, and Allan Drazen. 1991. "Why Are Stabilizations Delayed?" Ameri-
can Economic Review 81 (December) :1170-8 8.
Alesina, Alberto, and Dani Rodrik. 1994. "Distributive Politics and Economic Growth."
Quarterly Journal of Economics 108 (May):465-90.
Barro, Robert, and Robert Gordon. 1983. "Rules, Discretion, and Reputation in a Model
of Monetary Policy." Journal of Monetary Economics 12(July):101-21.
Benassy, Jean-Pascal. 1982. The Economics of Market Disequilibrium. New York: Aca-
demic Press.
Bruno, Michael, and Jeffrey Sachs. 1985. The Economics of Worldwide Stagflation.
Cambridge, Mass.: Harvard University Press.
Drazen, Allan, and Vittorio Grilli. 1993. "The Benefit of Crises for Economic Reform."
American Economic Review 83(3, June):598-607.
Driffill, John, and Christian Schultz. 1992. "Wage Setting and Stabilization Policy in a
Game with Renegotiation." Oxford Economic Papers 44(July):440-59.
Erzan, Refik, H. Kuwahara, S. Marchese, and R. Vossenaar. 1989. "Profile of Protec-
tion in Developing Countries." Unctad Review 1(1):29-49.
Fernandez, Raquel, and Jacob Glazer. 1991. "Striking for a Bargain between Two Per-
fectly Informed Agents." American Economic Review 81(March):240-52.
Fernandez, Raquel, and Dani Rodrik. 1991. "Resistance to Reform: Status Quo Bias in
the Presence of Individual-Specific Uncertainty." American Economic Review 81(De-
cember):1146-55.
Forteza, Alvaro. Forthcoming. "The Wage Bargaining Structure and the Inflationary
Bias." Journal of Macroeconomics.
Freeman, Richard. 1988. "Labor Market Institutions and Economic Performance." Eco-
nomic Policy 6(April):63-80.
.1993. 'Labor Market Institutions and Policies: Help or Hindrance to Economic
Development?" In Proceedings of the World Bank Annual Conference on Develop-
ment Economics 1992, pp. 117-44. Washington, D.C.: World Bank.
Grossman, Gene, and Elhanan Helpman. 1994. "Protection for Sale." American Eco-
nomic Review 84(4, September):833-50.
Helpman, Elhanan. 1995. "Politics and Trade Policy." NBER Working Paper 5309. Na-
tional Bureau of Economic Research, Cambridge, Mass. Processed.
Hillman, Arye. 1989. The Political Economy of Protection. Fundamentals of Pure and
Applied Economics Series 32. London: Harwood Academic Publishers.
Holden, Steinar. 1991 "Exchange Rate Policy and Wage Formation in an Economy with
Many Trade Unions." European Economic Review 35(December):1543-57.
Horn, Henrik, and Torsten Persson. 1988. "Exchange Rate Policy, Wage Formation,
and Credibility." European Economic Review 32(October):1621-36.
International Labour Organisation. Various years. Yearbook of Labour Statistics. Geneva.
International Monetary Fund. Various years. Government Finance Statistics Yearbook.
Washington, D.C.
Rama 355
Krueger, Anne. 1974. "The Political Economy of the Rent-Seeking Society." American
Economic Review 64(3, June):291-303.
Magee, Stephen, William Brock, and Leslie Young. 1989. Black Hole Tariffs and En-
dogenous Policy Theory. Cambridge, U.K.: Cambridge University Press.
Malinvaud, Edmond. 1977. The Theory of Unemployment Reconsidered. Cambridge,
U.K.: Basil Blackwell.
Nelson, Joan. 1991. "Organized Labor, Politics, and Labor Market Flexibility in Devel-
oping Countries." The World Bank Research Observer 6(1, January):37-56.
Oswald, Andrew. 1985. "The Economic Theory of Trade Unions: An Introductory Sur-
vey." Scandinavian Journal of Economics 87(2):160-93.
Persson, Torsten, and Guido Tabellini. 1990. Macroeconomic Policy, Credibility, and
Politics. Fundamentals of Pure and Applied Economics Series 38. London: Harwood
Academic Publishers.
1994. "Is Inequality Harmful for Growth? Theory and Evidence." American
Economic Review 84(3, June):600-21.
Pritchett, Lant. 1993. "Measuring Outward Orientation in Developing Countries." Policy
Research Working Paper 566. Policy Research Department, World Bank, Washing-
ton, D.C. Processed.
Rama, Martin. 1994. "Bargaining Structure and Economic Performance in the Open
Economy." European Economic Review 38(2, February):403-15.
. 1997. "Labor Market Institutions and the Second-Best Tariff." Scandinavian
Journal of Economics.
. Forthcoming. "Imperfect Rent Dissipation with Unionized Workers." Public
Choice.
Rama, Martin, and Guido Tabellini. 1997. "Endogenous Distortions in Product and
Labor Markets." Forthcoming in European Economic Review.
Rodrik, Dani. 1992. "Political Economy and Development Policy." European Economic
Review 36(April):329-36.
1. 995. "Political Economy of Trade Policy." In Gene M. Grossman and Kenneth
Rogoff, eds., Handbook of International Economics. Volume III. Amsterdam: Elsevier.
Sachs, Jeffrey. 1989. "Social Conflict and Populist Policies in Latin America." NBER
Working Paper 2897. National Bureau of Economic Research, Cambridge, Mass.
Processed.
Tullock, Gordon. 1967. "The Welfare Costs of Tariffs, Monopolies, and Theft." West-
ern Economic Journal 5(June):224-32.
World Bank. 1995. World Development Report 1995: Workers in an Integrating World.
New York: Oxford University Press.
THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2: 357-82
What Can New Survey Data Tell Us about Recent
Changes in Distribution and Poverty?
Martin Ravallion and Shaohua Chen
It has been claimed that in recent times the poor have lost ground, both relatively and
absolutely, even when average levels of living have risen. This article tests that claim
using household surveys for 67 developing and transitional economies over 1981-94.
It finds that changes in inequality and polarization were uncorrelated with changes in
average living standards. Distribution improved as often as it worsened in growing
economies, and negative growth was often more detrimental to distribution than posi-
tive growth. Overall, there was a small decrease in absolute poverty, although with
diverse experiences across regions and countries. Almost always, poverty fell with growth
in average living standards and rose with contraction.
Are the incidence and depth of poverty rising? Does inequality increase with
rising average standards of living? Do richer societies become more polarized?
Do the poor share in the benefits of higher average levels of living? How much
do they lose from falling average living standards? These questions are often
asked, but they are hard to answer convincingly.
In principle, household surveys can address such questions, but the coverage
and quality of surveys are uneven. As a rule, the poorer a country, the more
difficult it is to know just how poor its people are and whether their living
standards are improving over time. Other factors, such as the openness and size
of the country, influence the availability and quality of data. For example, the
average cost of a representative household survey falls with the size of the popu-
lation represented. Data on poor people have historically been wanting relative
to most other data. For example, the World Bank's World Development Re-
ports for 1979 and for many years after only give distributional data from house-
hold surveys for 20 or so developing countries. Yet macroeconomic aggregates
are available for almost all countries.
Analysts have used estimates of distributional statistics (such as the well-known
Gini index of inequality) for the 1960s and 1970s as both dependent and inde-
Martin Ravallion and Shaohua Chen are with the Policy Research Department at the World Bank.
This project received financial assistance from the World Bank's Poverty and Social Policy Department
and the joint British-Dutch-Swedish trust fund for studying the social and environmental consequences
of growth-oriented policies. For discussions on this topic and comments on the paper, the authors are
grateful to Gaurav Datt, Jyotsna Jalan, Emmanuel Jimenez, Michael Lipton, Oey Meesook, Binayak
Sen, Lynne Sherburne-Benz, Dominique van de Walle, Quentin Wodon, participants at various
presentations, and three anonymous referees.
C) 1997 The International Bank for Reconstruction and Development/ THE WORLD BANK
357
358 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
pendent variables in cross-country regressions. Yet some of these statistics are
not based on nationally representative household surveys; rather they are syn-
thetic estimates built up from other sources, including nonsurvey data (Fields
1994). Even among the survey-based estimates, the surveys have varied greatly
in, for example, the measure of living standards used, with implications for sum-
mary statistics on distribution such as the Gini index.
The availability of distributional data for developing countries has improved
over the past 10 years. For example, the World Development Report 1996 pre-
sents distributional data for 67 low- and middle-income countries (World Bank
1996b). The timeliness of data has also improved. In the World Development
Report 1985, the average lag is 11 years, so the average survey date is 1974
(World Bank 1985). The lag is now five years. Nationally representative house-
hold surveys underlie all the distributional data given in the World Develop-
ment Reports for recent years. Many countries and international agencies, in-
cluding the World Bank, have sought to improve the quality of data and the
coverage of survey data. Despite these efforts, we have a long way to go before
all poor countries have a good-quality survey for monitoring poverty and even
further before data can be compared with confidence across countries. But we
have made progress.
This article aims to provide a broad picture of the evolution of measures of
distribution and poverty since the mid-1980s and to analyze the correlation of
these changes with growth and contraction in average levels of living. Our ap-
proach is largely descriptive. Although distributional data have improved, we
remain skeptical of attempts to use these data to test seemingly sophisticated
multivariate models. We draw out some of the simple bivariate relationships
and test their robustness to the underlying measurement problems. By carefully
assembling the data set and choosing appropriate econometric methods for esti-
mating the relationships of interest, we hope to extract the signal from the noise
in these data.
Section I discusses the data and econometric methods for estimating the main
relationships of interest. Section II discusses the study's results concerning how
distribution has changed over time. Section III examines progress in reducing
poverty. Section IV presents our conclusions.
I. DATA AND METHODS
Although data have improved, international comparisons of distributional
statistics are still plagued by both conceptual and practical problems. We survey
some of the issues and discuss their implications for estimating the main rela-
tionships of interest. We then describe the data set developed for this study.
International Comparisons of Statistics on Poverty and Distribution
Official exchange rates are clearly deceptive in making international com-
parisons of absolute levels of living. But the problems of making purchasing-
Ravallion and Chen 359
power-parity currency conversions should not be understated. Estimates of the
purchasing-power-parity exchange rate have varied widely, with implications
for (among other things) international comparisons of poverty rates.
Given that we want to include the countries of Eastern Europe in this study,
absolute level comparisons of poverty across countries pose an extra problem.
Applying a developing-country poverty line to Eastern Europe would imply very
low poverty rates in that region, while applying an Eastern European poverty
line would give very high poverty rates in many low-income countries. Measure-
ments at extremes of the distribution are problematic in conventional sample
surveys.
A further issue is that of comparing different survey-based measures of living
standards. For example, some surveys only obtain income and others only ob-
tain consumption. An income-based measure is bound to show higher inequality
than one based on consumption. (At one survey date, income will be unusually
low for some households and unusually high for others; with some opportuni-
ties for saving or borrowing, consumption will be less unequal.) Also, in devel-
oping countries particularly, measurement errors are thought to be greater for
income, which tends to inflate measured inequality. Differences between coun-
tries in measured inequality may thus reflect in part differences in the welfare
indicators used.
Survey questionnaires can also differ widely in, for example, the number of
distinct categories of consumer goods that they identify and the order in which
they ask questions. Some income surveys still rely on questions such as "What is
your income from self-employment? " that are clearly very difficult to answer. A
convincing questionnaire requires a careful and complete accounting of revenues
and costs in the household enterprises (recognizing that these may be tangled up
with other activities). Survey quality varies, and even seemingly similar surveys
might not be comparable. This could be a serious problem for cross-country
comparisons of the levels of incomes and of summary measures based on their
distribution. Most of the empirical literature compares the levels of summary
measures (such as inequality measures or poverty rates) across countries; the
existence of country-level fixed effects in distribution-arising from, among other
things, survey design-can make such comparisons deceptive.
Comparisons across countries at different overall levels of development also
pose a potential problem given variations in the relative importance of con-
sumption of nonmarket goods. The local market value of all consumption in
kind (including consumption from own production, which is particularly im-
portant in relatively underdeveloped rural economies) should ideally be included
in the measure of total consumption expenditure; similarly, the imputed profit
from production of nonmarket goods should be included as part of income. This
is not always done. However, this is a far bigger problem in the surveys con-
ducted prior to 1980 or so than in those conducted since then. It has become
routine for survey data for developing countries to include valuations for con-
sumption or income from own production, following guidelines of the U.N.
360 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Household Survey Capability Programme or advice from the World Bank or
elsewhere. Nonetheless, the methods of valuation do vary; for example, some
current surveys use the price at the nearest market, while others use the average
farm-gate selling price.
Econometric Methods for Cross-Country Regressions Using Survey Data
The data problems summarized above clearly throw doubt on simple cross-
country comparisons of the measured levels of inequality and poverty. How-
ever, it can still be possible to detect the true relationship between (say) poverty
and aggregate affluence. Indeed, some quite simple econometric methods can
retrieve the true relationship of interest, provided that the structure of measure-
ment errors satisfies certain assumptions.
We want to know whether a measure of inequality or poverty responds sys-
tematically to growth in average levels of living. (For concreteness we focus on
poverty in the following discussion.) However, the data are riddled with mea-
surement errors and noncomparabilities. To some extent these behave like
country-level fixed effects, although they also induce artificial variation over
time. So there is latent heterogeneity in distribution, reflecting in part differences
in the type of data. There may also be a common time trend. Combining these
features, let measured poverty, P, in country i at date t be given by:
(1) log Pi, = ai + i log Fi + 7t + �it (i =t = 1,..,Tj)
where ci is a fixed effect reflecting the time-persistent differences between coun-
tries in distribution, , is the "growth elasticity" of poverty with respect to mean
consumption given by g0;, y is trend rate of change over time t, and cit is a white-
noise-error process that includes errors in the poverty measure.1
Notice that 3 is not the same as the growth elasticity that can be derived
analytically under the assumption that the Lorenz curve does not change (Kakwani
1993). The latter elasticity must be negative, and indeed it has a unique
(nonstochastic) value for any poverty measure, mean, and distribution. By con-
trast, ,B is an empirical elasticity in which the Lorenz curve shifts consistently
with the data. In principle it could take any sign or magnitude, depending on
how distribution changes with growth, and it has its own distribution. In esti-
mating 3 our interest is whether actual growth processes typically reduce pov-
erty, not whether some hypothetical growth process does so.
We do not, however, observe the true mean jg, but we do observe the follow-
ing estimate:
(2) log pit = log Wit, + Vat.
Equation 2 contains a country-specific, time-varying error term (vi,) that is as-
sumed to be white noise, as in the standard errors-in-variables model (see, for
example, Greene 1991, chap. 9). However, unlike the standard errors-in-
1. A white-noise error is one that has zero mean, is independent over time and between countries,
and has constant variance.
Ravallion and Chen 361
variables model, vi, is allowed to be contemporaneously correlated with Eit in
equation 1, recognizing that both the poverty measure and mean consumption
are derived from a common household survey. Using equation 2, equation 1
takes the form:
(3) log Pi, = oai + i log Rti, + yt + Eit - l5Vit.
Taking first differences, we can eliminate (xi and obtain:
(4) Alog Pi, = y + PAlog Sit + AEit - I3AVit
(where AX,t - Xt - Xit_,).2 So, roughly speaking, the rate of poverty reduction is
regressed on the rate of growth in mean consumption.3
However, the standard ordinary least squares (OLS) regression method does
not in general give unbiased estimates of either , or y even in very large samples;
in other words OLS is inconsistent under the above assumptions. It can be shown
that, as the number of countries (N) approaches infinity, the OLS estimate of L
converges to:4
(5)plim = + 2[Cov(i,vi) - PVar(vit)]
(5) plimi3=13+ Var(Aloggti)
The second term on the right-hand side of equation 5 is the asymptotic bias in
the OLS estimate. This is made up of the usual attenuation bias when an explana-
tory variable is measured with error, plus an extra common-survey bias caused
by the correlated measurement errors. Surveys that overestimate (underestimate)
mean consumption presumably tend to underestimate (overestimate) poverty
measures; so it is plausible that Cov(eFi,vit) < 0. Thus, as long as growth does in
fact reduce poverty (3 < 0), both Cov(eit,vi,) and PVar(vi,) are negative and hence
offsetting. Whether on balance there is over- or underestimation of the true value
of the growth elasticity cannot be determined without imposing further struc-
ture on the measurement errors.
One way to add structure is by noting that the error term in equation 1 in-
cludes effects of measurement errors in both the mean and the Lorenz curve, for
both can induce errors in measured poverty. A natural assumption to make is
that overestimating the mean by (say) 10 percent has the same effect on mea-
sured poverty as a 10 percent increase in the true mean. Also allowing for other
2. Alternatively, we could take deviations from the means over time (giving the "within" or "fixed
effects" estimator). However, this requires stronger assumptions for consistency under the present
structure of measurement errors. Under certain conditions, we can assure consistency by combining the
estimates obtained from the two methods of transforming the data (Hsiao 1986). However, those
conditions include that the time-varying measurement error in the right-hand-side variable is uncorrelated
with that in the left-hand-side variable, which is implausible in this setting.
3. Note, however, that using growth rates rather than changes in logs gives biased estimates of
equation 4 for all except small changes.
4. This is proved by taking the probability limit (plim) of the formula for the OLS regression coefficient
as N approaches infinity.
362 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
(distributional) errors in measured poverty, we can postulate the following de-
composition of the error term in equation 1:
(6) eit = ivit + tit
where 4it is another white-noise process, interpretable as the error in the poverty
measure caused by mismeasurement of distribution. Then the asymptotic bias in
the OLS estimate simplifies to:
(7) plim _= 2Cov&t, v -o)
Var(A log gt )
as long as the distributional error git) is uncorrelated with the growth error (vit).
Then the common-survey bias exactly offsets the attenuation bias. There is no
obvious reason why the growth and distributional errors are correlated. Over-
estimation of the mean might be due to overestimation of the incomes of the
nonpoor in a survey (such as by oversampling a rich area), but it does not seem
plausible that this is typically the case. Sometimes the problem is caused by
overestimation of the incomes of the poor.
So under these assumptions about the structure of measurement errors in this
setting, and allowing for latent heterogeneity caused by lack of strict data com-
parability across countries, we can obtain consistent estimates (unbiased as N
approaches infinity) of the growth elasticity by simply applying OLS to equation
4. That is the approach followed here.
But that does not give us the correct standard errors. Notice that the differ-
ence transformation used to obtain equation 4 also changes the properties of the
error term. In addition to eliminating the unobserved fixed effects, the transfor-
mation introduces a first difference in the original error term (ei,). If the latter is
white noise, then the new error process in equation 4 is correlated within coun-
tries and over time, although not between countries. Successive spells for a given
country are not statistically independent, because they have one survey in com-
mon. Conventional methods of calculating standard errors then have to be modi-
fied. Specifically, the variance-covariance matrix of the error process AEi, has a
block diagonal structure (with a separate block for each country) in which non-
zero off-diagonal elements only appear within the blocks, because of the com-
mon surveys for adjacent spells. In this article we correct all standard errors and
t-ratios to take account of the structure of the error covariance matrix of this
specification. We also correct them for any general type of heteroscedasticity
that might be present, after first correcting for the block diagonal structure of
the covariance matrix.
Would it be better to replace g, by the private consumption component of the
national accounts? This component, too, is measured with error; in addition to
the existing error in the national accounts' estimate of consumption for a given
year, there are new errors in matching the survey period used to measure pov-
erty. Those errors are presumably uncorrelated with the error in measured pov-
Ravallion and Chen 363
erty. However, as we have shown above, that correlation actually works in our
favor, by counterbalancing the usual attenuation bias arising from the measure-
ment error in the explanatory variable. Replacing the survey mean with mean
consumption from national accounts data thus creates an inconsistent estimate
of the growth elasticity; the attenuation bias remains, but we can no longer rely
on the offsetting common-survey bias.
The Data
We developed a data set for this study that greatly expands the data set docu-
mented in Chen, Datt, and Ravallion (1994), which uses national household
surveys for 44 countries, 19 for more than one point in time. The present article
uses data for 67 countries, of which 42 have at least two surveys during the
period since 1980.5 Table 1 gives the countries and dates covered by region in
the new data set. We include as many surveys as available that satisfy our com-
parability standards (discussed below). Relative to Chen, Datt, and Ravallion,
the new data set gains in coverage for all regions. Overall, 85 percent of the
population (in the countries included) is represented by at least one survey. The
coverage varies, though; the thinnest coverage is for the Middle East and North
Africa (47 percent of the population represented), followed by Sub-Saharan Africa
(66 percent).
All measures of household living standards are normalized by household size.
The distributions are also weighted by household size. So, for example, we esti-
mate the percentage of people living in households with consumption per person
below the poverty line, not the percentage of households. Similarly the empirical
Lorenz curves are weighted by household size, so they correspond to fractiles of
persons, not households.
In all cases we estimate our measures of living standards from the primary
data source (tabulations or household-level data) rather than relying on ex-
isting estimates. The estimation from tabulations requires an interpolation
method. We mainly use parameterized Lorenz curves with flexible functional
forms, which have proved reliable in past work (Ravallion, Datt, and van de
Walle 1991 and Datt and Ravallion 1992). Also, we only use nationally rep-
resentative surveys.
Two surveys for one country define what we term a "spell." Both measures of
living standards used in a given spell are estimated the same way from the source
data. In particular, in constructing the spells we use the same living standards
indicator-either expenditure or income per person-over time. So we do not
compare an income measure at one date with an expenditure measure for the
same country at another date. In some cases, different subperiods use different
measures for a given country; for example, surveys may switch from income to
5. The data set has been used for various recent compilations of regional and country-level
distributional and poverty data, including World Bank (1996a, 1996b, 1997). The data set overlaps
that used by Deininger and Squire (1996), which focuses solely on inequality but goes back further in
time.
364 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 1. Coverage of the Data Set
Percentage of
1993 population
Region represented Country Survey dates Welfare indicator
East Asia 88.0 China 1985, 1990,
1992, 1993 Income
Indonesia 1984, 1987,
1990, 1993 Expenditure
Malaysia 1984, 1989 Income
Philippines 1985, 1988 Expenditure
Thailand 1981, 1988 Income
1988, 1992 Expenditure
Eastern Europe 85.9 Belarus 1988, 1993 Income
and Central Asia Bulgaria 1988, 1992 Income
Czech Republic 1988, 1993 Income
Estonia 1988, 1993 Income
Hungary 1989, 1993 Income
Kazakstan 1988, 1993 Income
Kyrgyz Republic 1988, 1993 Income
Latvia 1988, 1993 Income
Lithuania 1988, 1993 Income
Moldova 1988, 1992 Income
Poland 1985, 1987,
1989, 1993 Income
1990, 1992 Expenditure
Romania 1989, 1992 Income
Russia 1988, 1993 Income
Slovak Republic 1988, 1992 Income
Slovenia 1987, 1993 Income
Turkmenistan 1988, 1993 Income
Ukraine 1988, 1992 Income
Yugoslavia 1985, 1989 Income
Latin America and 83.9 Bolivia 1990 Income
the Caribbean Brazil 1985, 1989 Income
Chile 1990, 1992 Income
Colombia 1988, 1991 Income
Costa Rica 1981, 1989 Income
Dominican Republic 1989 Income
Ecuador 1994 Expenditure
Guatemala 1986/87, 1989 Income
Honduras 1989, 1992 Income
Jamaica 1988, 1989, 1990,
1991, 1992, 1993 Expenditure
Mexico 1984, 1992 Expenditure
Nicaragua 1993 Expenditure
Panama 1989 Income
Peru 1985/86, 1994 Expenditure
Venezuela 1981,1987,
1989, 1991 Income
Middle East and 46.7 Algeria 1988 Expenditure
North Africa Egypt 1991 Expenditure
Ravallion and Chen 365
Table 1. (continued)
Percentage of
1993 population
Region represented Country Survey dates Welfare indicator
Jordan 1986/87, 1992 Expenditure
Morocco 1984/85, 1990 Expenditure
Tunisia 1985, 1990 Expenditure
South Asia 98.4 Bangladesh 1983/84, 1985/86,
1988/89, 1991/92 Expenditure
India 1983, 1986/87,
1987/88,1988/89,
1989/90, 1990/91,
1992 Expenditure
Nepal 1984/85 Income
Pakistan 1991 Expenditure
Sri Lanka 1985, 1990 Expenditure
Sub-Saharan Africa 65.9 Botswana 1985/86 Expenditure
C6te d'Ivoire 1985, 1986,
1987, 1988 Expenditure
Ethiopia 1981/82 Expenditure
Ghana 1987, 1988, 1992 Expenditure
Guinea 1991 Expenditure
Guinea-Bissau 1991 Expenditure
Kenya 1992 Expenditure
Lesotho 1986/87 Expenditure
Madagascar 1993 Expenditure
Mauritania 1988 Expenditure
Niger 1992 Expenditure
Nigeria 1985, 1992 Expenditure
Rwanda 1983/85 Expenditure
Senegal 1991/92 Expenditure
South Africa 1993 Expenditure
Tanzania 1991, 1993 Expenditure
Uganda 1989/90, 1992 Expenditure
Zambia 1991, 1993 Expenditure
Zimbabwe 1990 Expenditure
Total 85.0
Note: Income denotes household income per person, and expenditure denotes household consumption
expenditure per person. The 1991-92 and 1991-93 spells for Jamaica and Tanzania, respectively, were
not used because of serious comparability problems. The 1993 China survey, the 1992 Honduras survey,
and the 1992 Uganda survey arrived too late to be used in constructing the spells but were used for other
calculations (tables 2, 4, and 5).
Source: Household surveys done for individual countries, mostly by government statistical agencies.
consumption. We then swap the measure at one survey date. (If this is impos-
sible, then the spell is dropped.) When there is a choice we use consumption in
preference to income.
The data set allows us to construct 64 spells for 67 countries between 1981
and 1994 (using 109 surveys). Table 1 gives the distribution of the spells across
regions and presents details on the specific countries and periods for each spell.
The coverage deteriorates markedly for Sub-Saharan Africa when we construct
366 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
the spells; although we have 28 surveys spanning 19 countries in Sub-Saharan
Africa, only 7 spells are possible for 4 countries. So we are less confident about
results for that region.
One-third of all spells are for Eastern Europe and Central Asia, reflecting in
part the breakup of the Soviet Union. The data for Eastern Europe and Central
Asia should probably be treated differently than the data for the other regions.
For one thing, the countries in Eastern Europe and Central Asia are undergoing
major structural changes that also have implications for the comparability of
data on household living standards over time and across countries. For example,
standard welfare measures do not allow for the rationing of consumer goods;
relaxing rationing in the transition to a market economy has entailed welfare
gains that are not easily captured by conventional surveys. Similarly, subsidies
on publicly provided goods are often ignored and may have changed during the
transition. Some nonmarket goods have become market goods during the transi-
tion. And the methods used for valuing consumption in kind may not have
changed so as to reflect properly the changes in the economy; old planning prices
may now bear little relationship to opportunity cost. The survey data and the
consumer price index may not properly reflect these facts. There are also sam-
pling biases in a number of these surveys; for example, some are likely to have
undersampled (growing) informal segments of the economy (Atkinson and
Micklewright 1992). It is beyond our scope here to fix these problems. We do,
however, take some care to note differences between the data for Eastern Europe
and Central Asia and those for other regions.
II. CHANGES IN DISTRIBUTION
We use these data to address the set of questions posed at the beginning of
this article. But we must first be more precise about the distributional measures
used.
What Do We Mean by "Distribution" and How Should It Be Measured?
Conventional measures of inequality satisfy the "transfer principle" whereby
inequality is said to have fallen if the new distribution can be obtained from the
old one by a set of transfers in which the gainers are poorer than the losers.
Several measures satisfy the transfer principle (for a survey of standard measures
of inequality and their properties see Sen 1973). Here we use the most common
measure of inequality found in practice, namely the Gini index.
However, a conventional inequality measure may not pick up distribu-
tional changes of concern to policymakers. Impacts on the middle strata can
be important to the political feasibility of policy reform, yet an inequality
measure such as the Gini index may not capture changes in the share of
income held by the middle stratum. This calls for a measure of polarization,
that is, the extent to which the society is divided into the "haves" and "have-
nots." Roughly speaking, distribution A is said to be more polarized than B
Ravallion and Chen 367
if the incomes in A tend to be more bimodal, in that there are more poor and
rich, but fewer people in the middle (Wolfson 1994). For example, if we
transfer income within the poorest half such that the gainers are poorer than
the losers, and we do the same within the richest half, then polarization will
have increased-there will be fewer people at middle incomes-yet inequal-
ity will have decreased. (Suppose that there are four people with incomes $1,
$2, $3, and $4. We take $0.50 from the person with $2 and give it to the
person with $1, and we take $0.50 from the person with $4 and give it to the
one with $3. The new distribution is $1.50, $1.50, $3.50, and $3.50. In-
equality has fallen, because the gainers were poorer than the losers, but po-
larization has risen, because the new distribution is more sharply divided
into "rich" and "poor.")
To illustrate how inequality and polarization can diverge in a developing-
country context, consider the effects of a shift in the domestic terms of trade in
favor of the rural sector. Suppose (to simplify the exposition) that there are four
income groups: ranked from lowest to highest income, they are the rural poor,
the urban poor, the rural rich, and the urban rich. The rural poor and the rural
rich gain from the shift in the terms of trade (at least in the long run), while both
the urban groups lose. To simplify the exposition, we assume that the gain to the
rural poor is roughly equal to the loss to the urban poor; similarly, the gain to
the rural rich is about equal to the loss to the urban rich. We also assume that
the rankings of the four groups are preserved. The prorural shift in the terms of
trade reduces inequality by any measure satisfying the transfer principle-the
new distribution can be obtained from the old one by a set of transfers in which
the recipient is poorer than the donor. But the change increases polarization, by
the above definition; the overall distribution becomes more bimodal, due to the
lower inequality both among the poor (due to the convergence in incomes be-
tween the rural and urban poor) and among the rich (with the rural rich gaining
relative to the better-off urban rich).
Thus an analysis that is concerned solely with inequality as conventionally
defined may miss relevant aspects of how distribution has changed. Claims about
how inequality changes during a growth process could well have more to do
with polarization. It is possible, for example, that in our attempts to understand
the political economy of distributional impacts of policy reform we have been
looking at the wrong measures. Inequality may well decrease with reform-and
the change would be judged a social welfare improvement by conventional ethi-
cal criteria used in economics-and yet the society may become more polarized,
with heightened social tensions arising from the polarizing effects of diverse
impacts among middle-income groups, whereby some become poorer, while oth-
ers prosper.
To measure polarization we use the index proposed by Wolfson (1994). Like
the Gini index, it is between 0 (no polarization) and 1 (complete polarization).
When there is complete equality there is also zero polarization. However, while
maximum inequality entails that the richest person has all of the income, maxi-
Table 2. Regional Summary of Changes in the Distribution of Income or Consumption
Real survey mean per capita household
income or consumption Inequalitya Polarizationb
Number of spells Mean rate Number of spells Mean rate Number of spells Mean rate
Number of for which it of change for which it of change for which it of change
Region spells Fell Rose (percent per year) Fell Rose (percent per year) Fell Rose (percent per year)
East Asia 9 0 9 3.6 3 6 1.1 3 6 1.5
Eastern Europe and
Central Asia 21 18 3 -6.9 3 18 5.0 3 18 4.6
Latin America and
the Caribbean 14 5 9 1.5 10 4 -0.3 8 6 -0.5
0> Middle East and
�� North Africa 3 1 2 1.3 1 2 0.7 1 2 1.3
South Asia 10 6 4 0.2 6 4 0.0 4 6 -0.2
Sub-Saharan Africa 7 5 2 -6.0 4 3 -1.5 5 2 -2.1
Total 64 35 29 -2.0 27 37 1.6 24 40 1.4
Total excluding Eastern
Europe and
Central Asia 43 17 26 0.4 24 19 -0.1 21 22 -0.2
Note: See table 1 for countries and survey dates.
a. Measured by the Gini index.
b. Measured by the Wolfson (1994) polarization index.
Source: Authors' calculations.
Ravallion and Chen 369
mum polarization occurs when half the population has zero income and the other
half has twice the mean. The Wolfson polarization index (W) can be written as:
(8) W = 2(g * - tL)Im
where g is the distribution-corrected mean income (given by the actual mean
times 1 minus the Gini index), pL iS the mean income of the poorest half of the
population, and m is the median income. Like inequality, this is not the only
available measure of polarization, but it appears adequate for the present purpose.
Changes in Inequality and Polarization
Table 2 gives a regional summary of the changes in distribution. Inequality
rose in 37 of the 64 spells, while polarization rose in 40. Both measures indicate
a worsening in 6 out of 9 spells for East Asia and in 18 out of 21 spells for
Eastern Europe and Central Asia. In Latin America and the Caribbean and in
Sub-Saharan Africa distribution improved more often than it worsened; inequality
fell in 10 of 14 spells for Latin America and the Caribbean, and polarization fell
in 8 of 14 spells, while for Sub-Saharan Africa the Gini index fell in 4 of 7 spells,
and polarization fell in 5 spells. In South Asia inequality fell in 6 of 10 spells,
while polarization fell in 4 spells. Of the 3 spells for the Middle East and North
Africa, distribution worsened in 2.
Combining all the spells, the average rate of increase in both the Gini index
and the polarization index was significantly positive; for the Gini index, the mean
rate of increase was 1.6 percent a year with a standard deviation of 0.48 percent;
for the polarization index, the mean rate of increase was 1.4 percent with a stan-
dard deviation of 0.52 percent. However, this worsening of distribution on aver-
age is largely due to the experience of Eastern Europe and Central Asia. If we
exclude that region from the calculations, neither in the Gini index nor in the
polarization index was the mean rate of change significantly different from 0.
Although there is a clear conceptual distinction between our measures of in-
equality and polarization, there is a surprisingly close correspondence between
them for these data. The relationship is quite strong and significant (the overall
correlation coefficient between the rates of change is 0.83). In all but 7 of the 64
spells the two measures of distribution moved in the same direction. In 32 cases
both inequality and polarization increased, while both fell in 23 cases. In the
largest deviation from the least squares regression line (estimated on the full
sample of spells), the Gini index fell 2.6 percent, while the polarization index
rose 8.3 percent. The bulk of the points for Eastern Europe and Central Asia
were in the region of both increasing polarization and increasing inequality.
And the measured rates of increase in inequality and polarization in Eastern
Europe and Central Asia were high by any standards.
Growth and Distributional Change
Is there any systematic tendency for distribution to change in the process of
rising average household living standards? The distribution of the benefits of
370 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
economic growth is a long-standing issue in development economics (for a re-
cent overview of the arguments see Bruno, Ravallion, and Squire 1996). In re-
cent times economists have been much concerned about the distributional impli-
cations of the types of growth processes in poor countries. It is difficult to predict
the effect of growth on distribution on a priori grounds. We turn instead to
empirical evidence.
Figure 1 plots the changes in the (log) Gini index against the changes in (log)
real household consumption (or income) per person. (The picture looks very
similar for the polarization index, which is to be expected given their high corre-
lation.) Over the 64 spells, the correlation is negative. On regressing the change
in the log Gini index on that in mean consumption and allowing a trend (by
adding the number of years between surveys as an additional explanatory vari-
able), we obtain the results reported in table 3. Higher mean consumption has a
significant negative effect on inequality. We also find a significant underlying
trend increase in inequality. However, when we remove the spells for Eastern
Europe and Central Asia, both effects vanish (figure 1). When we try adding a
complete set of regional dummy variables (both slope and intercept), we find no
other significant regional differences.
Figure 1. Inequality and Growth
Change in log Gini index (xlOO)
60
* Contracting mean Growth in mean
50 - with rising inequality with rising inequality
40 -
30 - -O
20 - 0 El~~=
20 * � <_
--_ O
10 4 O
0~~~~~~~~~~~
0~ El
-10 Contracting mean Growth in mean
with falling inequality with falling inequality
-20 I I I
-120 -100 -80 -60 -40 -20 0 20 40 60
Change in log mean consumption or income between surveys (x100)
* Eastern Europe and Central Asia r0 Other countries
Note: See table 1 for countries and survey dates.
Source: Authors' calculations.
Ravallion and Chen 371
Table 3. Trends and Growth Elasticities of Inequality and Polarization
Growth
Measure of distribution and sample Trend (y) (xloo) elasticity (:) R2
Gini index of inequality
Full sample 1.10 -0.24 0.54
(3.21) (6.07)
Excluding Eastern Europe and Central Asia 0.13 -0.01 0.01
(0.58) (0.23)
Eastern Europe and Central Asia 3.71 -0.11 0.75
(3.18) (1.21)
Wolfson polarization index
Full sample 1.00 -0.21 0.40
(2.55) (4.51)
Excluding Eastern Europe and Central Asia 0.00 -0.01 0.00
(0.22) (0.12)
Eastern Europe and Central Asia 3.82 -0.05 0.68
(3.08) (0.56)
Note: Estimates were obtained using OLS, regressing the difference between household surveys in the
log of the measure of distribution on the time elapsed between the surveys and the difference in the log
of the real value of the survey mean. Absolute t-ratios are in parentheses, based on robust standard
errors corrected for heteroscedasticity and serial correlation due to common surveys across sequential
spells. Sample sizes are 64 spells for the full sample, 43 spells for the sample excluding Eastern Europe
and Central Asia, and 21 spells for Eastern Europe and Central Asia. See table 1 for countries and
survey dates.
Source: Authors' calculations.
So these data do not indicate that higher average consumption tends to be
associated with higher inequality or that inequality tends to increase indepen-
dently of growth. For Eastern Europe and Central Asia, growth still negatively
affects inequality, but the effect is not significant. There is a trend increase in
inequality in the Eastern Europe and Central Asia countries. The same conclu-
sions hold for polarization (table 3). We find no evidence here that some middle-
income households have become worse off during spells of growth while others
have gained.
III. PROGRESS IN REDUCING POVERTY
This section looks at how poverty measures have been changing and what
relationship those changes have had with changes in average living standards.
Assessing and Comparing Progress in Reducing Poverty
All our poverty comparisons over time use poverty lines that have constant
real value, according to country-specific consumer price indexes. When we also
want to compare the level of poverty between countries we use purchasing-power-
parity exchange rates. However, these are not available for several countries in
our data set (particularly, but not only, in the Former Soviet Union). Therefore,
we expand the number of data points considerably by using poverty lines that
are relative across countries, but absolute over time; because we only compare
Table 4. Regional Summary of Changes in Poverty
Number of spells
Poverty, fell for Poverty rose Mean rate of change in povertya
all three Trend is for all three (percent per year)
Region Total poverty lines ambiguous poverty lines SO percent 75 percent 100 percent
East Asia 9 7 1 1 -6.1 -4.6 -2.7
Eastern Europe and Central Asia 21 2 2 17 109.2 25.4 9.4
Latin America and the Caribbean 14 7 1 6 -1.2 -0.8 -0.4
Middle East and North Africa 3 2 0 1 1.3 -0.5 -0.9
South Asia 10 4 2 4 2.6 0.7 0.2
Sub-Saharan Africa 7 2 0 5 6.8 6.0 4.4
Total 64 24 6 34 35.9 8.3 3.1
Total excluding Eastern Europe
and Central Asia
Before 1990 30 16 4 10 -0.6 -0.7 -0.4
After 1990 13 6 0 7 1.7 1.2 0.9
Note: See table 1 for countries and survey dates.
a. The three poverty lines are set at 50, 75, and 100 percent of the mean household income or consumption expenditure per person for the first survey date in
each country.
Source: Authors' calculations.
Ravallion and Chen 373
rates of change, the lack of absolute comparability of the levels is not too worry-
ing. However, we do test the robustness of this practice by also comparing rates
of change in level-comparable poverty measures and mean consumptions.
We first examine poverty lines that are absolute over time, but relative be-
tween countries. The initial value for the poverty line (at the beginning of the
first spell) is set at a common proportion of the mean living standards indicator
from the first survey. The poverty line is then updated over time using the local
consumer price index. We present summary results in table 4 for three such
poverty lines, set at 50, 75, and 100 percent of the initial survey mean in each
country. Poverty lines for European countries are typically around 50 percent of
the mean, and this is also a common figure in middle-income developing coun-
tries, while a figure closer to 75-100 percent of the mean is more common in
low-income countries (Ravallion, Datt, and van de Walle 1991). The range of
50-100 percent appears to embrace the range of poverty lines found in practice.
As shown in table 4, poverty fell in 24 of the 64 spells for all three poverty
lines, while it increased in 34 spells for all three lines; in only 6 spells was the
trend ambiguous (poverty increased for some poverty lines and decreased for
others). The table also gives the results by region. The regions in which poverty
fell unambiguously in half or more of the spells were East Asia (7 of the 9 spells)
and Latin America (7 of the 14 spells). The regions in which poverty rose in half
or more of the spells were Eastern Europe and Central Asia (17 of 21 spells
showed an unambiguous increase) and Sub-Saharan Africa (5 out of 7). In South
Asia an unambiguous increase in poverty was as common as a decrease (4 of the
10 spells in each case, with two ambiguous spells). Although there seem to be
some regional patterns, the variation within regions is notable; indeed, in no
case did all spells for a region indicate the same direction of change.
The sharp increase in the poverty measures for most of Eastern Europe and
Central Asia is striking. (See Milanovic 1995 for further discussion.) We find
that the impact was particularly pronounced at the lower end of the distribu-
tion. However, there is one glaring outlier for Eastern Europe and Central Asia.
Poverty measures for Poland fell sharply in 1987-89; indeed, this is the spell
with the largest drop in poverty among all 64 spells. However, the Poland spells
were erratic; for example, the (income-based) 1989-93 spell showed a sharp
increase in poverty, while the (expenditure-based) 1990-92 spell showed little
change. There may be comparability problems here.
Next we attempt to fix the absolute value of the poverty line across countries.
Table S gives our estimates of the percentage of the population living on less than
$1 a day at 1985 international prices. This is a typical poverty line among low-
income countries (World Bank 1990 and Ravallion, Datt, and van de Walle 1991).
The table also gives the poverty gap index, the mean shortfall below the poverty
line (counting the nonpoor as having zero shortfall) expressed as a percentage of
the poverty line. The table updates past estimates available for 1990, including
those in Chen, Datt, and Ravallion (1994). There are a number of differences be-
tween these numbers and previous estimates published in World Bank (1990, 1992,
Table 5. Poverty Measures Using an International Poverty Line of $1 a Day per Person at 1985 Purchasing Power Parity,
1987-93
Percentage of population Poverty gap index Mean poverty gap
consuming less than $1 a day (percent) of the poor (cents)
Region 1987 1990 1993 1987 1990 1993 1987 1990 1993
East Asia 29.7 28.5 26.0 8.3 8.0 7.8 27.9 28.1 29.9
Eastern Europe and Central Asia 0.6 - 3.5 0.2 - 1.1 27.1 - 30.8
Latin America and the Caribbean 22.0 23.0 23.5 8.2 9.0 9.1 37.2 39.3 38.8
Middle East and North Africa 4.7 4.3 4.1 0.9 0.7 0.6 18.3 15.9 15.7
South Asia 45.4 43.0 43.1 14.1 12.3 12.6 31.1 28.6 29.1
Sub-Saharan Africa 38.5 39.3 39.1 14.4 14.5 15.3 37.3 37.0 39.1
Total 30.7 - 29.4 9.5 - 9.2 30.9 - 31.3
Total excluding Eastern Europe
and Central Asia 33.9 32.9 31.9 10.8 10.3 10.5 31.7 31.2 32.8
- Not available.
Note: The poverty measures are population-weighted means over all countries in the data set within each region. See table 1 for countries and survey dates.
Source: Authors' calculations.
Ravallion and Chen 375
1993). Aside from new data, the main difference is that, unlike past estimates, no
model-based extrapolations are used for countries without survey data. The num-
bers used here are only for countries with appropriate household surveys.
Half of the 122 surveys used are household consumption surveys. We use
consumption expenditure (including the imputed value of consumption in kind)
per person as the indicator of household welfare. When only an income survey is
available, we rescale mean income per person according to the estimated con-
sumption share from the national accounts. As in Chen, Datt, and Ravallion
(1994), we make adjustments to line up the surveys in time. Of the 67 countries
represented for 1981-94, 22 have only one survey; 35 have two surveys, and 10
have three or more surveys. If there is a survey within one year of the target date,
then we use that survey. If there is not, then we use the closest survey, adjusting
the survey mean consumption or income according to the rate of growth in real
private consumption per person from the national accounts. When the target
date is between two surveys, we make the adjustment for both and use a time-
weighted average. We cannot make the adjustment for Eastern Europe and Cen-
tral Asia in 1990 because the World Bank's data base is missing a substantial
amount of data.
As for past estimates, we do not convert dollars into local currencies at offi-
cial exchange rates, but rather at rates that attempt to assure purchasing power
parity (PPP)-so that $1 is worth roughly the same in different countries. For
currency conversions, we use the Ppp rate for consumption in 1985 in the Penn
World Tables 5.6. (Summers and Heston 1991 describe the Penn World Tables,
Mark 5.) This is the latest available comprehensive set of consumption PPP rates
and is widely considered to be the most reliable source for consumption Ppps.
However, Penn World Tables 5.6 entails some important revisions to past PPps.
The main change is a substantial increase in the estimated proportion of people
living on less than $1 a day in East Asia, mainly arising from an upward revision
in the number for China. This increase is due entirely to the revision in the PPP
rate for China. If we use instead Penn World Tables 5.0, the East Asia percent-
ages fall to 14.0 (1987), 14.0 (1990), and 11.6 (1993). Other changes caused by
the revised Ppp rates include a lower estimate for India, bringing down the South
Asia aggregate, and lower rates for the Middle East and North Africa. Holding
the survey data set constant, the numbers for Latin America and Sub-Saharan
Africa are affected very little by the revisions to the Ppp values.
From table 5, the results indicate a small drop in aggregate poverty between
1987 and 1993. This holds for both the headcount index (percentage of the
population consuming less than $1 a day) and the poverty gap index (average
distance in cents below $1 a day, when averaged over the whole population,
with 0 for the nonpoor). The regional breakdown indicates a fall in poverty for
East Asia, the Middle East and North Africa, and South Asia (with signs of a
slight reversal from 1990 to 1993) and increases in poverty for Eastern Europe
and Central Asia, Latin America, and Sub-Saharan Africa. For 1993 the regional
ranking from highest to lowest percentages of the population living on less than
376 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
$1 a day is South Asia, Sub-Saharan Africa, East Asia, Latin America, the Middle
East and North Africa, and Eastern Europe and Central Asia; for the poverty
gap index the ordering is Sub-Saharan Africa, South Asia, Latin America, East
Asia, Eastern Europe and Central Asia, and the Middle East and North Africa.
So, for example, while South Asia has the highest overall poverty incidence,
Sub-Saharan Africa has the highest depth of poverty (so that at some lower
poverty line, the incidence of poverty is highest in Sub-Saharan Africa).
Table 5 also gives the mean poverty gap of the poor as a percentage of the
poverty line (which is simply the poverty gap index divided by the headcount
index). Although the aggregate proportion living on less than $1 a day is falling,
the average distance below $1 a day among the poor has remained close to
$0.31 over the period.
Poverty and Growth
The extent to which poor people share in a rising average standard of living
has been much debated. A still common view is that the poor are generally left
behind. Several recent studies have challenged this view, suggesting that a rising
(falling) overall mean standard of living is typically associated with falling (ris-
ing) absolute poverty (Fields 1989; World Bank 1990, 1995; Squire 1993;
Ravallion 1995). Here we apply our updated and expanded data set to this issue.
Figure 2 plots the change in the log poverty rate between surveys against that
in average consumption. We set the poverty line at 75 percent of the initial mean
standard of living; the pattern is similar for other poverty lines. Higher rates of
growth in average living standards are associated with higher rates of poverty
reduction. Unlike the distributional measures, the slope is similar for Eastern
Europe and Central Asia and the other regions.
To estimate the overall growth elasticities and distributional trends for vari-
ous poverty measures, we use the data on spells to estimate equation 4. To allow
for the uneven spacing of the surveys, the constant term in equation 4 is replaced
by the lapsed time in years between surveys (and the usual constant term is
suppressed). OLS gives consistent estimates under our assumptions about the struc-
ture of measurement errors, although the standard errors have to be corrected
(see section I). The results are given in table 6.
Regressing the first difference of the log of the proportion of the population
living on less than 50 percent of the initial mean standard of living against the
difference in the log of the real value of the mean for the 64 spells, we obtain a
growth elasticity of -2.6. Thus, a 10 percent increase in the mean standard of
living can be expected to result in a 26 percent drop in the proportion of people
living on less than half the initial mean. For higher poverty lines, the growth
elasticity falls (in absolute value). Regressing the rates of change in the propor-
tion of the population living on less than 75 percent of the initial mean standard
of living against the percentage change in the real value of the survey mean, the
regression coefficient is -1.3. At 100 percent of the initial mean, the elasticity
falls to -0.7 (table 6).
Ravallion and Chen 377
Figure 2. Poverty and Growth
Change in log poverty rate (xlOO)
150
100 Pot 'Q Poverty and mean
both rising
Poverty rising, 0
mean falling
50
-50 _-� � z
Poverty and mean Poverty falling,
both falling mean rising
-100 I I I lI I I I
-120 -100 -80 -60 -40 -20 0 20 40 60
Change in log mean consumption or income between surveys (xlOO)
+ Eastern Europe and Central Asia a Other countries
Note: See table 1 for countries and survey dates.
Source: Authors' calculations.
If we use instead the international $1 a day poverty line, then we find a larger
variance across countries in both the levels and rates of poverty reduction. The
estimated growth in the elasticity of the proportion of the population living on
less than $1 a day is -3.1 (table 6). We obtain a slightly higher elasticity for the
poverty gap index.
Thus the relationship between rates of poverty reduction and rates of growth
in average consumption becomes flatter and more precisely estimated for higher
poverty lines. The incidence of extreme poverty does not tend to be less respon-
sive to growth in average living standards than does the incidence of only mod-
erate poverty. If anything, these data point to the opposite conclusion. Similarly,
the depth of poverty, as reflected in the poverty gap index, is more responsive to
growth than is the incidence of poverty.
There is no sign of a significant distributional trend overall, except for the
poverty line set at 50 percent of the initial mean standard of living (table 6).
The trend for the 50 percent poverty line is due to Eastern Europe and Cen-
tral Asia, where there is a strong trend increase in poverty independent of
growth, as seen in section II; distribution is clearly worsening in these transi-
378 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
Table 6. Distributional Trends and Growth Elasticities of Various Poverty
Measures
Distributional trend Growth
Poverty measurea (y) (xl 00) elasticity (B) R2
Poverty line at 50 percent
Full sample 3.52 -2.59 0.84
(2.37) (15.01)
Excluding Eastern Europe and Central Asia -0.95 -1.57 0.58
(0.87) (6.37)
Eastern Europe and Central Asia 16.66 -1.91 0.93
(2.88) (4.43)
Poverty line at 75 percent
Full sample 0.87 -1.29 0.83
(1.40) (13.24)
Excluding Eastern Europe and Central Asia -0.87 -0.95 0.72
(1.54) (10.23)
Eastern Europe and Central Asia 6.75 -0.97 0.92
(2.46) (4.05)
Poverty line at 100 percent
Full sample 0.15 -0.69 0.84
(0.51) (11.81)
Excluding Eastern Europe and Central Asia -0.38 -0.64 0.85
(1.38) (10.50)
Eastern Europe and Central Asia 2.68 -0.53 0.88
(1.64) (3.59)
Proportion consuming less than $1 a day,
1985 ppp -3.86 -3.12 0.37
(1.40) (2.62)
Poverty gap index in cents per day -6.04 -3.69 0.36
(1.63) (2.61)
Note: Estimates were obtained using OLS, regressing the difference in the log of the poverty measure
between household surveys on the time elapsed between the surveys and the difference in the log of the
real value of the survey mean. Absolute t-ratios are in parentheses, based on robust standard errors
corrected for heteroscedasticity and serial correlation due to common surveys across sequential spells.
Sample sizes for the poverty lines are 64 spells for the full sample, 43 spells for the sample excluding
Eastern Europe and Central Asia, and 21 spells for Eastern Europe and Central Asia. Sample sizes are
both 42 for the proportion living below $1 a day and for the poverty gap index. See table 1 for countries
and survey dates.
a. The proportion of the population with income or consumption below 50, 75, or 100 percent of
the mean household income or consumption expenditure per person for the first survey date in each
country.
Source: Authors' calculations.
tional economies. For the developing countries there is no sign of a trend
independent of growth; zero is our best estimate of the rate of change in
poverty at zero growth.
Are there other significant regional differences in the impact on poverty of a
given rate of growth in average living standards? We add a set of intercept dummy
variables for the regions. (We also tried an intercept dummy variable for whether
the survey data for a given spell were for incomes or expenditures, but this was
insignificant.) At a given rate of growth, the only region that has a rate of pov-
erty reduction significantly different from that of East Asia (taken as the arbi-
Ravallion and Chen 379
trary reference) is Eastern Europe and Central Asia, where the rate of increase in
poverty is significantly higher than we would expect given the rate of change in
average living standards.
We also test whether the impact of growth is any different among regions, by
adding to our regressions the interaction effects between the rate of change in
the mean standard of living and the regional dummy variables. None of these
dummy variables is significant. Thus, for the set of countries in our data set, we
can find no significant differences between regions in the responsiveness of the
poverty measures to growth.
In summary, we find strong evidence that higher rates of growth in average
living standards are associated with higher rates of poverty reduction. The ad-
verse distributional effect of recent growth in a number of the developing coun-
tries has not been strong enough to change the conclusion that growth has ben-
efited the poor. For the developing countries as a whole, there is no significant
trend distributional effect for or against the poor. So at zero growth, the ex-
pected rate of poverty reduction is also zero. For Eastern Europe and Central
Asia there is an adverse distributional effect.
IV. CONCLUSIONS
The main body of our analysis used distributional data from 119 household
surveys since 1980. We constructed spells of distributional change for 42 devel-
oping and transitional economies using two surveys for each spell that satisfy
minimal criteria for comparability, including being nationally representative and
using ostensibly the same indicator of welfare. We estimated various summary
statistics on how distribution and poverty have changed. We mainly looked at
rates of change. However, we also offered an overall assessment of the absolute
levels of poverty (at constant international prices) and how this changed over
1987-93. For that assessment, we used 122 surveys (including countries with
only one survey) and extrapolated over time when necessary.
There are numerous sources of measurement errors and comparability prob-
lems in these data, even after the quality controls were applied. This is particu-
larly worrying for the comparisons of absolute levels of poverty. Although com-
paring only changes avoids some of the difficulties of making level comparisons,
the measures of change over time undoubtedly include noise caused by errors or
inconsistencies of measurement. We argue, however, that the main sources of
bias in our estimation methods for testing the effect of growth on distribution
and poverty are likely to be offsetting and (under certain assumptions about the
structure of measurement errors) to cancel each other out, leaving an unbiased
estimate of the relationship of interest. So we can reasonably hope to have ex-
tracted the signal from the noise in these data.
Our results suggest that both inequality and polarization increased more of-
ten than they decreased among the 64 spells. However, the experience of East-
ern Europe and Central Asia is not typical; if we excluded this region from the
380 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
analysis, then both inequality and polarization fell more often than they rose.
Distribution deteriorated more often than not in East Asia, and it improved
more often than not in Sub-Saharan Africa and Latin America.
For the sample as a whole, we found no support for the view that higher
growth rates in average living standards tended to accompany worsening distri-
bution. Indeed, over the whole sample, rising average consumption was associ-
ated with lower inequality and polarization. However, this conclusion is not
robust to excluding the countries of Eastern Europe and Central Asia, where
there has been a tendency for both inequality and polarization to increase dur-
ing a time of overall economic contraction. Excluding this set of countries from
the analysis, we found that neither inequality nor polarization was correlated
with growth in average consumption; nor did either have an underlying trend, in
either direction.
Turning to performance at reducing absolute poverty, we calculated rates of
change in the proportions of the population living on less than 50, 75, and 100
percent of the initial survey mean for each country. For all three of these cutoff
points, poverty fell in 24 of the 64 spells, and it rose for all three cutoff points in
34 spells (the remaining six being ambiguous according to which cutoff is used).
In East Asia, poverty fell in all except one spell, while it rose in almost all cases
in Eastern Europe and Central Asia. Poverty rose during five of the seven spells
in Sub-Saharan Africa. In South Asia and Latin America, poverty rose about as
often as it fell.
When we forced level comparability, we found that the overall percentage of
people living on less than $1 a day (at 1985 international prices) fell between
1987 and 1993, from 31 to 29 percent. The depth of poverty, as measured by
average distance below the poverty line, remained static in the aggregate over
this period. Progress was uneven across regions, with falling poverty incidence
in East Asia, South Asia, and the Middle East and North Africa, but with rising
poverty incidence in Eastern and Central Europe, Latin America, and Sub-
Saharan Africa.
There is a strong association between the rate of growth in average living
standards and the rate at which absolute poverty fell. In terms of elasticities, the
response of the poverty measures to changes in average consumption is even
stronger for lower poverty lines. The benefits of higher total consumption ap-
pear to be spread quite widely, on average. Structural changes going on in the
transitional economies entail rising poverty even at zero growth. But for the
developing economies as a whole, stagnation in average living standards entails
stagnation for the poor, too. We found no significant regional differences in the
responsiveness of the poverty measures to growth.
REFERENCES
The word "processed" describes informally reproduced works that may not be com-
monly available through library systems.
Ravallion and Chen 381
Atkinson, Anthony B., and John Micklewright. 1992. Economic Transformation in East-
ern Europe and the Distribution of Income. Cambridge, U.K.: Cambridge University
Press.
Bruno, Michael, Martin Ravallion, and Lyn Squire. 1996. "Equity and Growth in De-
veloping Countries: Old and New Perspectives on the Policy Issues." Policy Research
Working Paper 1563. Policy Research Department, World Bank, Washington, D.C.
Processed.
Chen, Shaohua, Gaurav Datt, and Martin Ravallion. 1994. "Is Poverty Increasing or
Decreasing in the Developing World?" Review of Income and Wealth 40(Decem-
ber):359-76.
Datt, Gaurav, and Martin Ravallion. 1992. "Growth and Redistribution Components
of Changes in Poverty Measures: A Decomposition with Applications to Brazil and
India in the 1980s." Journal of Development Economics 38(April):275-95.
Deininger, Klaus, and Lyn Squire. 1996. "A New Data Set Measuring Income Inequal-
ity." The World Bank Economic Review 10(September):565-92.
Fields, Gary. 1989. "Changes in Poverty and Inequality in Developing Countries." The
World Bank Research Observer 4(July): 167-86.
. 1994. "Data for Measuring Poverty and Inequality Changes in the Developing
Countries." Journal of Development Economics 44(June):87-102.
Greene, William H. 1991. Econometric Methods. New York: Macmillan Publishing
Company.
Hsiao, Cheng. 1986. Analysis of Panel Data. Cambridge, U.K.: Cambridge University
Press.
Kakwani, Nanak. 1993. "Poverty and Economic Growth with Application to C6te
d'Ivoire." Review of Income and Wealth 39(June):121-39.
Milanovic, Branko. 1995. "Poverty, Inequality, and Social Policy in Transition Econo-
mies." Research Paper 9. Policy Research Department, World Bank, Washington,
D.C. Processed.
Ravallion, Martin. 1995. "Growth and Poverty: Evidence for the Developing World."
Economics Letters 48(July):411-17.
Ravallion, Martin, Gaurav Datt, and Dominique van de Walle. 1991. "Quantifying
Absolute Poverty in the Developing World." Review of Income and Wealth 37(De-
cember):345-61.
Sen, Amartya. 1973. On Economic Inequality. Oxford: Oxford University Press.
Squire, Lyn. 1993. "Fighting Poverty." American Economic Review 83(May):377-82.
Summers, Robert, and Alan Heston. 1991. "The Penn World Table (Mark 5): An Ex-
tended Set of International Comparisons, 1950-1988." Quarterly Journal of Eco-
nomics 106(May):327-68.
Wolfson, Michael. 1994. "Diverging Inequalities." American Economic Review 84(2,
May):353-58.
World Bank. 1979. World Development Report 1979. New York: Oxford University
Press.
.1985. World Development Report 1985: International Capital and Foreign Trade.
New York: Oxford University Press.
. 1990. World Development Report 1990: Poverty. New York: Oxford Univer-
sity Press.
382 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 2
1992. World Development Report 1992: Development and the Environment.
New York: Oxford University Press.
. 1993. Implementing the World Bank's Strategy to Reduce Poverty: Progress
and Challenges. Washington, D.C.
. 1995. World Development Report 1995: Workers in an Integrating World. New
York: Oxford University Press.
. 1996a. Poverty Reduction and the World Bank: Progress and Challenges in the
Early 1990s. Poverty and Social Policy Department, Washington, D.C.
. 1996b. World Development Report 1996: From Plan to Market. New York:
Oxford University Press.
1997. World Development Indicators 1997. Washington, D.C.
Announcing an important new journal...
Environment and
Development Economics
Published in association with the BeijerInstitute, Royal SwedishAcademy ofSciences
Editor: Charles Perrings, University of York
Aims and Scope Coverage includes:
The aim of Environment and Development * optimal environmental regulation and
Economics is to provide a forum for the results incentives in a development context
ofthe growing body ofresearch into the linkages . the valuation of environmental resources
between economic development and in developing countries and incentives in
environmental change. It publishes papers environmental policy
addressed equally to the research and policy environmental effects ofinstitutional change
communities and is designed to be accessible to
a broad readership. The journal will publish * modelling interdependent economic and
not only scientific papers in a section on theory environmental processes
and applications but also environmental policy Subscription Information
discussion papers published in a regular Policy Pblihd quarterly in Fbruary, May, July and
Options section. Each issue will include non- October. Volume 2: 1997: �81 institutions; �42
technical summaries of all papers in these two individuals. Airmail included. Special
sections, and regular book reviews and review arrangements exist for subscribers in most low-
essays. income countries. ISSN 1355-770X
Recent contents rFurther Information 5267
Environmental and trade policies: sonie I Please send me further information on
methodological lessons I Environment and Development Economics I
V. KERRY SMITH AND J. ANDRESS ESPINOSA Name l
Macroeconomic policies, second best theory Address _
and the environment
M. MUNASINGHE AND K. G. MALER l l
l I
Natural resource degradation effects of poverty Send to: Journals Marketing Department,
and the population growth are largely policy- Cambridge University Press, The Edinburgh
induced: the case of Colombia Building, Cambridge CB2 2RU, UK.
J. HEATH & H. BINSWANGER | Tel: +44 (0)1223 325806 Fax: +44 (0)1223 315052 |
Email: journals_marketing@cup.cam.ac.uk l
Measuring sustainable development: progress l l
on indicators I UNIVCAMBRIDGES I
D. PEARCE, K. HAMILTON AND G. ATKINSON L UNIVES PRESS
L_-_-_-_-_____
Distributors of COLOMBIA GERMANY ISRAEL NEPAL 00-677 Warzawa SWITZERLAND
fnfoenlace Ltda. UNO-Vedag Yozmot Literature Ltd. Everest Media Intemational Services (P) Tel: (48 2) 628-6089 Librairie Payot Service tnstiutionnel
W orld Bank Carrera6 No. 51-21 PoppelsdorterAlleeS5 PO. Byx 56055 Ltd. C6tes-de-Montbenon 30
Apartado Aereo 34270 53115 Bonn 3 Yohanan Hasandlar Street GPO Box 5443 PORTUGAL 1002 Lausanne
Publications Santale de Bogota, D.C. Tel: (49 228) 212940 Tel Aviv 61560 Kathmandu Livrana Portugal Tel: (41 21) 341-3229
Pri ce and credit terms vary Tel: (571) 285-2798 Tel: (972 3) 5285-397 Tel: (977 1) 472 152 Apartado 2601, Rua Do Carmo 70-74
fror cvtvnry to country. GREECE 1200 Lisbon ADECO Van Diermen
Consvult your local diotribultor COTE D'IVOtRE Papasotiriou S.A. ROY Intemational NETHERLANDS Tel: (1) 347-4982 EditionsTechniriues
CenterdEddion etde Ditusion 35, StoumaraSir. PO Box 13056 De Lindeboon/tInOr-Publikaties Ch.deLacuez41
before placing an order. Alncaines (CEDA) 10682 Athens Tel Aviv 61130 PO. Box 202, 7480 AE Haaksbergen ROMANIA CH1807 Bbonay
04 B.P 541 Tel: (30 1)364-1826 Tet: (9723) 5461423 Tel: (31 53( 574-0004 CompaniDe LibrariiBucuresti S.A. Tel: (41 21)043 2673
ARGENTINA Abidjan 04 Str Lipscani no. 26, sector 3
Oficina del Libro Inlemacional Tel: (225) 24 6510;24 6511 HAITI Palestinian Authority/Middle Easl NEWZEALAND Bucharest TANZANIA
Av. Cordoba 1877 CuOure Dfflusion Index Inlormation Services EBSCO NZ Ltd. Tel: (40 1) 613 9645 Oxford University Press
1120 Buenos Aires CYPRUS 5, Rue Capois PO.B.19502 Jerusalem Private Mail Bag 99914 Maktaba Street, PO Box 5299
Tel: (541) 815-8354 Cenlerlor Applied Research C.P 257 Tel: (972 2)6271219 New Markel RUSSIAN FEDERATION Dares Salaam
Cyprus College Port-au-Prince Auckland Isdalelstvo Ves Mir> Tel: (255 51) 29209
AUSTRALIA, FIJI, PAPUA NEW 6, Diogenes Street, Engomi Tel: (509 1) 39260 ITALY Ili: (04 9(524-0119 9a, Lolpachniy Pereslok
GUINEA, SOLOMON ISLANDS, PO Boo 20060 Licosa Commissionaria Sansoni SPA Moscow 101031 THAILAND
D.NA.TU,lAmatiWESTERvic SAO Nicosia HDNG KONG, MACAD Via DuGa Di Calobda, 1/1 NIGERIA Tel: (7 095) 917 87 49 Central Books Distribution
D.A. Information Services Tel: (357 2) 44-1730 Asia 2000 Ltd. Casella Poslole 552 University Press Limited 306 Silom Road
648 WhRehorse Road Sales & Circulation Department 50125 Firenze Three Crowns Building Jericho SINGAPORE, TAIWAN, Bangkok 10500
Micham 3132 CZECH REPUBLIC Seabird House, unil 1101-02 Tel: (55) 645-415 Private Mail Bag 0095 MYANMAR, BRUNEI Tel: (662) 235-5400
Victoria Nalional Inloormabion Center 22-28 Wyndham Street, Cenlral Ibadan Asahgate Pubftintg Asia tatffic Pte Ud.
Tel: (61)3 92107777 prodejno, Konviktska S Hong Kong JAMAICA Tel: (234 22) 41-1356 41 Kallang Pudding Road #04-03 TRINIDAD & TOBAGO,
AUSTRIA CS-113 57 Prague 1 Tel: (852) 2530-1409 Ian Randle Publishers Ltd. Golden Wheel Building AND TE CARRIBBEAN
AUeTRoI and Co.Tel: (422) 2422-9433 206 Old Hope Road, Kingston 6 NORWAY Singapore 349316 Systemratics Studies Unn
Gerold and Co. HUNGARY Tel: 809-927-2085 NIC Info AVS Tel: (65) 741-5166 9 Wafts Street
Weihburggasse 26 DENMARK Euro Into Service Book Depaement, Postboks 6512 Curepe
A-1011 Wien SamfundsLitteratur Margiszgeti Europa Haz JAPAN Eoerstad SLOVENIA Trinbiad, Wet Indies
Tel: (43 1)512-47-31-0 Rosenoems Alie 11 H-1138 Budapest Easlem Book Service N-0606 Oslo GospodarskiVestnikPublishingGroup Tel:(809)662-5654
DK-1970 Frederiksberg C Tel: (361)111 6061 3-13 Hongo3-chome, Bunkyo-ku Tel: (47 22) 974500 Dunajskacesta S
MicroIndustriesDevelopment Tel: (45 31)351942 Tokyo 113 1000 Ljubljana UGANDA
Micro Industries Development ~~INDIA Tel: (81 3) 3818-0861 PAKISTAN Tel: (3836 61) 133 83 47: 132 12 30 Gustro Ltd.
Assistance Society (MIDAS) ECUADOR Allied Publishers Ltd. Mirza Book Agency PO Box 9997, Madhvani Building
House 5nd Ae Libn Mundi 751 Mount Road KENYA 65, Shahrah-e-Ouaid-e-Azam SOUTH AFRICA, BOTSWANA Plot 16/4 Jinja Rd.
Dhanmondk 10 Area Libreria Intemacional Madras - 630 002 Aflca Book Service (E.A.) Ltd. Lahore 54000 Intemational Subscription Service Kampala
Dhakae1209 PO. Box 17-01-3029 Tel: (91 44) 852-3930 Quaran House, Mfangano Street Tel: (92 42) 735 3601 PO. Box 41095 Tel: (25641)254763
Tel: (880 2) 326427 Juan Leon Mera 851 PO. Box 45245 Craighall
uito Nairobi Oxlord Univershy Press Johannesburg 2024 UNITED KINGDOM
BELGIUM Tel: (5932)521-606; (5932)544-185 INDONESIA Tel: (254 2) 223 641 SBangalore Town Tel: (27 11) 880-1448 MicroinfoUd.
Joan De Lannoy t Pt. Indira Limied Sharae Faisal PO. Bax 3, Afton, Hampshire GU34 2PG
An. du Rol 202 EGYPT, ARAB REPUBLIC OF Jalan Borobudur 20 KOREA, REPUBLIC OF PO Box 13033 SPAIN England
1060 Brussels Al Ahram Distnbulion Agency PO. Box 181 Daejon Trading Co. LId. Karachi-75350 Mundi-Prensa Libros, S.A. Tel: (44 1420) 86848
Tel: (32 2)530-5109 Al Galaa Street Jakara 10320 PO. Box 34, Youida, 706 Seoun Bldg Tel: (92 21) 446307 Castello 37
Cairo Tel: (62 21) 390-4290 44-6 Youido-Dong, Yeongchengpo-Ku 28001 Madrid VENEZUELA
BRAZIL Tel: (202) 578-6083 Seoul Pak Book Corporation Tel: (34 1)431-3399 Tecni-Ciencia Libos, S.A.
Publicacies Tecnicas Inlemacionais IRAN Tel: (82 2) 785-1631/4 Aziz Chambers 21, Queen's Road Centw Cuidad Comercia Tamanco
Ltda. The Middle East Observer Ketab Sara Co. Publishers Lahore Mundi-Prensa Barcelona Nivel C2, Caracas
RuaPeixolo Gomide, 209 41, Sherd Street Khaled Eslamboli Ave., 6th Street MALAYSIA Tel: (92 42) 636 3222; 636 0885 Consellde Cent, 391 Tel: (58 2) 959 5547; 5035: 0016
01409 San Paulo, S Cairo Kusohe Delairooz No. 8 University of Malaya Cooperative 08009 Barcelona
Tel: (551 i) 259-6644 Tel: (202)393-9732 PO. Box 15745-733 Bookshop, Limited PERU Tel: (34 3)488-3492 ZAMBIA
Tehran PO. Box 1127 Ediorial Desarrollo SA University Bookshop, University of
CANADA FINLAND Tel: (98 21) 8717819; 8716104 Jalan Pantai Baru Apartado 3824, Uma 1 SRI LANKA, THE MALDIVES Zambia
Renouf Publishing Co. Ltd. Akateeminen Kiriakauppa 59700 Kuala Lumpur Tel: (51 14) 285380 Lake House Bookshop Great East Road Campus
5369 Canotek Rood PO. Box 128 Kowkab Publishers Tel: (603) 756-5000 100, Sir Chiatmpalam Gardiner PO. Box 32379
Ontawa, Ontano K1J 9,13 FIN-00101 Helsinki PO. Box 19575511 PHILIPPINES Mawaths Lusaka
Tel: (613) 745-2665 Tel: (358 0)121 4418 Tehran MEXICO lotemalional Booksource Center Inc. Colombo 2 Tel: (260 1) 252 576
Tel: (98 21) 258-3723 INFOTEC 1127-A Antipolo St, Barangay, Tel: (94 1)32105
CHINA FRANCE Avn San Femando No.37 Venezuela ZIMBABWE
Cina inancia Economic Wodd Bank Publications IRELAND Col. Tonello Guerra Makati City SWEDEN Longman Zimbabwe (Pvt.)Ltd.
Pub7173hingHouse 66, avenue dlena Govemment Supplies Agency 14050 Mexico, D.F Tel: (63 2) 896 6501; 6505; 6507 Wennergren-Willioms AB Toure Road, Ardbennie
8, Do Fe Si Dong Jie 75116 Pars Oig an tSoldthair Tel: (525)624-2800 P0. Box 1305 PO. Box ST125
Beijing Tel: (33 1) 40-69-30-56157 4-5 Horceort Road POLAND S-171 25 Solna Southenton
Tel: (86 10) 6333-8257 Dublin 2 Intemational Publishing Service Tel: (466)705-97-50 Harare
Tel: (353 1) 661-3111 Ul. Piekna 31/37 Tel: (263 4) 6216617
6wooer
Coming in the next issue of
THE WORLD BANK
ECONOMIC REVIEW
September 1997
Volume 11, Number 3
* Rationing Can Backfire: The "Day without a Car"
in Mexico City
by Gunnar S. Eskeland and Tarhan Feyzioglu
* A New Database on State-Owned Enterprises
by Luke Haggarty and Mary M. Shirley
* Prices and Protocols in Public Health Care
by Jeffrey S. Hammer
* Imports of Inputs, Foreign Investment, and Change
in the Structure of East European Exports
by Bernard Hoekman and Simeon Djankov
* Formal and Informal Regulation of Industrial Pollution:
Comparative Evidence from Indonesia
and the United States
by Sheoli Pargal, Hemamala Hettige, Manjula Singh,
and David Wheeler
* Capital Flows to Developing Countries:
Long-Term and Short-Term Determinants
by Mark Taylor and Lucio Sarno
m 11~I ii11 1.
1 Ifll
I 5ur