INTERNATIONALBANK
FOR
WORLD BANK
R
E T
C N
O E
N
STRUCTION PM
AND DEVELO
September 2003 No. 30
A regular series of notes highlighting recent lessons emerging from the operational and
analytical program of the World Bank`s Latin America and Caribbean Region
ECONOMIC ANALYSIS OF URBAN
UPGRADING PROJECTS
Maria Ang�lica Sotomayor
Latin America's urban population jumped from 50 percent in economic prices. Embedded taxes in material and labor
1960 to 75 percent in 2000 (108 and 389 million respec- should be deducted, using conversion factors if available. If
tively), making it the most urbanized region in the world. not, the analyst should investigate major input market distor-
Municipalities throughout Latin tions, such as unskilled labor and
America are struggling to keep up with imported goods in economies with
this growth, in difficult economic envi- protectionist policies, and adjust
ronments, making sustainable provision market prices to correct for major
of basic services a huge challenge. The market distortions. Cash flows (in
World Bank is working closely with economic prices) should be dis-
national, regional and municipal govern- counted using the country's esti-
ments in developing and financing urban mated opportunity cost of capital
upgrading projects. This article shares when available (e.g. Peru's Ministry
some experiences with different tech- of Finance uses 14 percent); other-
niques for estimating economic benefits wise a 12% discount rate is com-
and costs used by the Bank in appraising monly used in Bank projects.
urban upgrading projects1 . The projects
usually combine infrastructure invest-
ments and complementary activities to Methodologies for
strengthen, build capacity and organize
assessing economic
communities and government at differ-
ent levels to ensure appropriate use of benefits
services provided and sustainability.
Like costs, economic benefits have
to be valued at economic prices. Fi-
Assessing economic costs nancial income/revenue generated
by the consumption of a good (price
times quantity) needs to be adjusted
Project economic costs include capital
to reflect the "true value" of consumption for the consumer,
costs, operation and maintenance costs, rehabilitation, re-
or "willingness to pay", represented by the area under the
settlement, and costs associated with complementary actions
demand curve for any given quantity. This includes the
and contingencies. Costs of providing additional infrastruc-
actual payment and consumer surplus. In the example, if
ture (such as sewerage connections) necessary for users to
price is $3, consumers will demand 40 units and pay $120,
realize intended benefits, should be included, even if not
but are willing to pay $160 (area 0, 5, A, 40).
financed by the project.
Revealed preferences (for goods and services sold in mar-
Net incremental costs and benefits need to be assessed under
kets)
two scenarios: with, and without the project. They should be
adjusted for taxes, subsidies, and externalities, to arrive at
When a project provides goods or services sold in markets,
such as water and electricity, a demand curve can be derived
1 Mass transport and mobility components should be analyzed separately.
1
and (ii) consumers in the market,
Figura 1: Willingness to Pay conscious of the attribute being
(Financial Revenue + Consumer Surplus) valued. Multicollinearity is
almost always present and
5 important, and affects estimated
values. To assess estimate
reliability, different functional
4 Consumer forms and different variables
Surplus should be analyzed.
A market price
e 3 Contingent valuation
pric Financial
Revenue The contingent valuation method
2 (CVM) involves directly asking
people, in surveys, how much
demand
they would be willing to pay for
1
cur specific services, although
ve commonly used . Iit is the most
controversial non-market method.
0 20 40 60 80 100 It is called "contingent" valuation,
because people are asked their
Quantity
willingness to pay, contingent on
a specific hypothetical scenario
In the example, if price is $3, consumers will demand 40 units and description of the service.
and pay $120, but are willing to pay $160 (area 0, 5, A, 40).
Applying CVM is generally complicated, lengthy, and
from consumers' revealed preferences observed in the mar- expensive. To collect useful data and provide meaningful
ket, and used to calculate willingness to pay. Demand can be results, the CV survey must be properly designed, pre-tested
estimated using data from companies or household surveys and implemented. CV survey results are often highly
(such as ENCOVI in Guatemala) on income, quantities sensitive to what people believe they are being asked to value,
consumed, and associated monetary and non-monetary costs as well as the context described in the survey. This makes it
(e.g. time spend fetching water). essential for CV researchers to clearly define the specific
service(s) and context, and to demonstrate that respondents
understand them and are actually stating their values for these
Non-market methods services when answering the valuation questions.
For slum upgrading services such as drainage and parks for
which market data are non-existent or insufficient, non-direct
methods must be used to assess economic benefits. Method Advantages and Disadvantages
Contingent valuation and hedonic pricing are commonly used
in urban projects, others random utility and travel cost
models. Non-market methods involve more difficulties and Hedonic Pricing Model
sources of error. � Estimates values from data on actual choices. Data on
property sales and characteristics are readily available,
Hedonic Pricing and can be related to other data sources to obtain
Hedonic pricing is a good method for estimating the descriptive variables.
economic benefits of access to improved urban services.
Econometric analysis of data from observed transactions in � Property markets are relatively efficient in responding to
the housing market yields estimated marginal values for information, so can be good indications of value.
particular non-market attributes of interest (such as access to Property records are often reliable.
services, distance to center, etc.) embodied in houses and
reflected in housing prices. � Only benefits related to housing prices can be measured.
Most slum-upgrading interventions do affect housing
This method is relatively straightforward and prices, so the method is generally useful. It can only
uncontroversial, because it is based on actual market prices capture willingness to pay for perceived differences in
and easily measured data. It is relatively inexpensive if data attributes, and their direct consequences. If
are readily available. If data must be gathered and compiled, beneficiaries are unaware of linkages between the
costs increase substantially. For this method to work attribute and benefits to them or their property, the value
effectively: (i) the housing market should function well (it is will not be reflected in home prices.
important to discuss results and values with local realtors);
2
� Hedonic pricing assumes that beneficiaries have oppor- pressing value for the good, respondents might be
tunities to select their preferred combination of features, expressing feelings about the scenario or the valuation
given their incomes. However, the hous- exercise itself. For example, respondents may
ing market may be affected by outside state a positive willingness to pay because
influences, like taxes, interest rates, etc. they feel good about the act of paying for a
social good (the "warm glow" effect), al-
� The method is relatively complex to though they may believe that the good itself is
implement and interpret, requiring unimportant. Respondents may state a posi-
statistical expertise. Results depend tive willingness to pay to signal that improved
heavily on model specification. Much environmental quality is important to them.
data must be gathered and manipulated. Alternatively, some respondents may value
The time and expense of applying the good, but state low willingness to pay,
hedonic pricing depend on data because they disapprove of some aspect of the
availability and accessibility. scenario, such as increased taxes.
Contingent Valuation � Respondents may make associations
� Contingent valuation is enormously that the researcher had not intended. For
flexible and can be used to estimate the example, if asked for willingness to pay for
economic value of virtually anything. It improved visibility (through reduced
is best for estimating values for goods pollution), respondents may answer based on
and services that are easily identified and health risks that they associate with air
understood by users and consumed in pollution.
discrete units (e.g., user days of
recreation). � Some researchers argue that people
make hypothetical and actual decisions in
� It requires competent survey analysts to fundamentally different ways. For example,
achieve defensible estimates, but results respondents may fail to take questions
are easy to analyze and describe. seriously because they will not actually have
Monetary values can be presented as to pay the stated amount. Responses may be
means or median values per capita or per unrealistically high if respondents believe
household, or as aggregate values for they will not have to pay for the good or
affected populations. service but their answer may influence future
supply. Responses may be unrealistically low
� CV has been widely used for two if respondents believe they will have to pay.
decades, and considerable research is
being conducted to improve the � If respondents are first asked for their
methodology, make results more valid willingness to pay for one part of an
and reliable, and better understand its environmental asset (e.g. one lake in an entire
strengths and limitations. lake system) and then asked to value the
whole asset (e.g. the whole lake system), the
� Controversy remains over whether it amounts stated may be similar �the
adequately measures willingness to pay "embedding effect." Respondents' expressed
for environmental quality. CV assumes willingness to pay sometimes depends on
that people understand the good in where something is placed on a list of things
question and will state contingent being valued �the "ordering problem."
preferences just as they would reveal Respondents may give different willingness to
them in a real market. However, most pay amounts, depending on the payment
people are unfamiliar with placing vehicle chosen. For example, taxes may elicit
monetary values on environmental goods protest responses from those who oppose tax
and services and may lack an adequate increases. Contributions or donations may
basis for stating their true value. CVM is elicit answers that reflect people's sense of
difficult to apply in valuing continuous their "fair share", rather than expressing their
variables such as water consumed actual value for the good.
through a piped system. � Strategic bias arises when the respondent
provides an biased answer tailored, to influ-
� Answers to willingness to pay questions ence a particular outcome. Information bias
may be biased because respondents answer a different may arise when respondents are asked to
question than the surveyor intended. Rather than ex- value attributes with which they have little or no experi-
3
Approaches for Analyzing Urban Upgrading Projects
Method for estimation of NPV EIRR
Project and components
economic benefits (US$ 000) (%)
The Venezuela-Caracas Slum Upgrading Project is financing basic urban services and infrastructure in two informal
settlement areas, including water, sanitation, access, drainage, community centers, public lighting, electricity and
resettlement housing. Communities determine their own neighborhood improvement plans.
19,645 a/
Potable Water Contingent Valuation -
Sewerage Contingent Valuation 1,863 a/ -
Access
Travel time and vehicle
Principal 9,986 a/ -
operating costs
Secondary 34,843 a/
Colombia's Bogot� Urban Services Project is providing improved mobility and access through major urban transport
investments and water, sanitation and related basic services, particularly for 1.2 million residents in low-income areas.
Travel time and vehicle
Mobility 122,300 25%
operating costs
Urban Upgrading:
I. Perdomo Hedonic-property values b/ 1,456 19%
G. Yomasa 13,624 60%
Brazil's Recife Urban Upgrading Project is improving water, sanitation, lighting, transport, land tenure security, macro
and micro drainage, offering parks and recreation areas in low income areas in the Beberibe basin.
Beberibe basin (indirect
Hedonic-property values 12,176 34%
impact area)
Urban upgrading in low-
Hedonic-property values 51,079 38%
income areas
Resettlement (risk prone
Hedonic-property values 2,380 25%
areas)
Water Revealed preferences 8,238 37%
a/ NPV calculated using a 10% discount rate.
b/ Economic analysis based on a sample of investments in two areas (I. Perdomo and G. Yomasa).
ence; the amount and type of information presented to
respondents may affect their answers. ********
� Non-response bias is a concern when sampling
respondents, since individuals who do not respond may
have, on average, different values than individuals who
do respond. Learn More
http://www.worldbank.org/urban/
World Bank's recent experience in LAC and about our projects...
http://www4.worldbank.org/sprojects/
Given the complexity of urban upgrading projects, and their
importance for achieving poverty reduction goals in the About the Author
region, efforts have been made by the Bank to integrate
Maria Ang�lica Sotomayor is an Economist in the Water &
different approaches in analyzing these projects to account
Sanitation Cluster of the Finance, Private Sector and Infra-
for attributes of goods and services and user attitudes not
structure Department in the World Bank's Latin America
captured by traditional market methods of valuation. The
Caribbean Region.
following table summarizes the results of the economic
analysis of some of the latest urban upgrading projects
Special thanks to Maria Emilia Freire, Urban Regional Advisor,
financed by the Bank in the region. The complexity of the Dean Cira, task manager of the Caracas and Recife projects,
analysis and the use of appropriate valuation techniques has Thakoor Persaud task manger of the Bogot� project and Vitor
evolved and will continue to do so as projects have become Serra, for their valuable inputs and comments.
more complex to better respond to new development
challenges in the region. In making the decision of what About "en breve"
method to use the only rule is to use revealed preferences
whenever possible. When revealed preferences are not used, "en breve" appears every two weeks in English and Span-
ish (occasionally Portuguese). To subscribe send and email
hedonic pricing works well if the project is expected to to "en_breve@worldbank.org" Visit the archive at:
increase overall property values, whereas for environmental
benefits contingent valuation is better. http://www.worldbank.org/en_breve
4