Middle East Gainers and Losers
and from Trade Reform
North Africa in Morocco
Working Paper
Series
No. 37
August 2004
The World Bank by
Martin Ravallion
Produced by the
and
Office of the
Chief Economist Michael Lokshin
Summaries in
Arabic and French
30057
Gainers and Losers from Trade Reform in Morocco
by
Martin Ravallion
and
Michael Lokshin
Development Research Group, World Bank
August 2004
Discussion papers are not formal publications of the World Bank. They represent preliminary and often unpolished results
of country analysis and research. Circulation is intended to encourage discussion and comments; citation and the use of the
paper should take account of its provisional character. The findings and conclusions of the paper are entirely those of the
authors and should not be attributed to the World Bank, its affiliated organizations, or to members of its Board of Executive
Directors or the countries they represent.
Acknowledgements
For helpful discussions on this topic we are grateful to Touhami Abdelkhalek, Jennie
Litvack, Rachid Doukkali and Chris Ward. The assistance of Nithin Umapathi and Dimitri
Kaltsas is gratefully acknowledged.
Table of Contents
Summary
1. Introduction.............................................................................................................................................1
2. Measuring and explaining the welfare impacts of reform using micro data...........................................2
3. Measured welfare impacts of trade reform in Morocco..........................................................................6
3.1 The predicted price changes and the survey data..........................................................................6
3.2 Implied welfare impacts at household level..................................................................................7
3.3 Who are the net producers of cereals in Morocco?.......................................................................9
3.4 Vertical versus horizontal impacts on inequality .........................................................................9
3.5 Two caveats.................................................................................................................................12
4. Conclusions...........................................................................................................................................12
References ...................................................................................................................................................29
Tables
Table 1. Predicted price changes due to agricultural trade reform in Morocco ........................................14
Table 2a. Consumption shares and welfare impacts through consumption................................................15
Table 2b. Percentage gains from each policy: Production component .......................................................16
Table 3. Household impacts of four trade reforms....................................................................................17
Table 4. Mean gains from Policy 4 by region...........................................................................................18
Table 5. Decomposition of the impact on inequality ................................................................................18
Table 6. Summary statistics on explanatory variables in the regression analysis.....................................19
Table 7. Regression of per capita gain/loss on selected household characteristics...................................20
Table 8. Urban-rural split of regressions for per capita gains...................................................................22
Figures
Figure 1. Impacts on poverty......................................................................................................................24
Figure 2. Frequency distributions of gains/losses for Policies 1 and 4.......................................................25
Figure 3. Absolute and proportionate gains for Policies 1 and 4 plotted against percentile of
consumption................................................................................................................................26
Figure 4. Production/consumption decomposition of the welfare impacts for Policy 4, plotted against
percentile of consumption per person .........................................................................................27
Figure 5. Net producers of cereals in the distribution of total consumption per person in rural areas......28
Summary
We use Morocco's national survey of living standards to measure the short-term welfare
impacts of prior estimates of the price changes attributed to various agricultural trade reform
scenarios for de-protecting cereals -- the country's main food staple. We find small impacts on mean
consumption and inequality in the aggregate. There are both gainers and losers and (contrary to past
claims), the rural poor are worse off on average after de-protection. We decompose the aggregate
impact on inequality into a "vertical" component (between people at different pre-reform welfare
levels) and "horizontal" component (between people at the same pre-reform welfare). There is a
large horizontal component, which dominates the vertical impact of full de-protection. The diverse
impacts reflect a degree of observable heterogeneity in consumption behavior and income sources,
with implications for social protection policies.
:
�
. .
( )
) "" .
.( ) "" (
.
.
R�sum�
Nous nous servons de l'enqu�te nationale marocaine sur les conditions de vie pour mesurer
les impacts de bien-�tre � court terme d'estimations ant�rieures des fluctuations de prix attribu�es aux
divers sc�narios de r�forme du commerce agricole visant � soustraire les c�r�ales -- principale denr�e
de premi�re n�cessit� du pays -- au r�gime de protection. Nous constatons des impacts mineurs sur
la consommation moyenne et l'in�galit� dans l'agr�gat. Il y a � la fois des gagnants et des perdants et
(contrairement aux revendications pass�es), les pauvres du milieu rural se trouvent plus mal nantis
apr�s suppression de la protection. Nous d�composons l'impact global sur l'in�galit� en une
composante �verticale� (entre populations se situant � des niveaux diff�rents de bien-�tre avant la
r�forme) et une composante �horizontale� (entre populations se situant au m�me niveau de bien-�tre
avant la r�forme). La composante horizontale est importante et domine l'impact vertical d'une totale
suppression de la protection. Les impacts diff�rents refl�tent un degr� observable d'h�t�rog�n�it�
dans le comportement de consommation et les sources de revenu, avec des implications pour les
politiques de protection sociale.
1. Introduction
As a water-scarce country, Morocco does not have much natural advantage in the production
of water-intensive crops such as most cereals, including wheat, which is used to produce the country's
main food staples. The desire for aggregate self-sufficiency in the production of food staples has led
in the past to governmental efforts to foster domestic cereal production, even though cereals can be
imported more cheaply. Since the 1980s, cereal producers have been protected by tariffs on imports
as high as 100%.
There have been concerns that the consequent reallocation of resources has hurt consumers
and constrained the growth of production and trade. Reform to the current incentive system for
cereals has emerged as an important issue on the policy agenda for Morocco (World Bank, 2003).
The major obstacles to reform stem from concerns about the impacts on household welfare,
particularly for the poor. There has been very little careful research into who will gain and who will
lose from such reforms.
Nonetheless, there has been much debate about the equity implications. It is generally agreed
that urban consumers are likely to gain from lower cereal prices. More contentious are the welfare
distributional impacts in rural areas. Defenders of the existing protection system have argued that
there will be large welfare losses to the rural economy from trade reform. Critics have argued against
this view, claiming that the bulk of the rural poor tend to be net consumers, and so lose out from the
higher prices due to trade protection. They argue that the rural poor are likely to gain from the
reform, while it will be the well off in rural areas who tend to be net producers who will lose; see for
example, Abdelkhalek (2002) and World Bank (2003).
This paper studies the household welfare impacts of the relative price changes induced by
specific trade policy reform scenarios for cereals in Morocco. Past analyses of the welfare impacts
have been highly aggregated, focusing on just one or a few categories of households. Here we
estimate impacts across 5,000 sampled households in the Morocco Living Standards Survey for
1998/99. This allows us to provide a detailed picture of the welfare impacts, so as to better inform
discussions of the social protection policy response to trade liberalization.
Past approaches to studying the welfare impacts of specific trade reforms have tended to be
either partial equilibrium analyses, in which the welfare impacts of the direct price changes due to
tariff changes are measured at household level, or general equilibrium analyses, in which second-
round responses are captured in a theoretically consist way but with considerable aggregation across
household types. In general terms, the economics involved in both approaches is well known. And
both approaches have found numerous applications.
We combine these two approaches. In particular, the price changes induced by the trade-
policy change are simulated from a general equilibrium analysis done for a Joint Government of
Morocco and World Bank Working Group. We take the methods and results of that analysis as given
and carry them to the Moroccan Living Standards survey. Our approach respects the richness of
detail available from a modern integrated household survey, allowing us to go well beyond the highly
aggregative types of analysis one often finds. We not only measure expected impacts across the
distribution of initial levels of living, but we also look at how they vary by other characteristics, such
as location. We are thus able to provide a reasonably detailed "map" of the predicted welfare impacts
by location and socio-economic characteristics.
In studying the distributional impacts of trade reform we make a distinction between the
"vertical impact" and "horizontal impact." The former concerns the way the mean impacts vary with
1
level of pre-reform income; how does the reform affect people at different pre-reform incomes? The
horizontal impact relates to the disparities in impact between people at the same pre-reform income.
As argued in Ravallion (2004), many past discussions of the distributional impacts of trade and other
economy-wide reforms have tended to focus more on the vertical impacts, analogously to standard
practices in studying the "benefit incidence" of tax and spending policies. However, as we will
demonstrate here, this focus may well miss an important component of a policy's distributional
impact, arising from the horizontal dispersion in impacts at given pre-reform incomes. We show how
the impact of a policy on a standard inequality measure can be straightforwardly decomposed into its
vertical and horizontal components. The former tells us how much of the change in total inequality
can be accounted for by the way in which mean impacts conditional on pre-reform income vary with
the latter. If there is no difference in the proportionate impact by level of income then the vertical
component is zero. The horizontal component tells us the contribution of the deviations in impacts
from their conditional means. Only when the impact of the reform is predicted perfectly by pre-
reform income will the horizontal component be zero. We study the relative importance of these two
components of our predicted distributional impact of trade reform in Morocco.
The following section discusses our approach in general terms. Section 3 presents our results
in detail, while section 4 reviews the main findings.
2. Measuring and explaining the welfare impacts of reform using micro data
We use pre-existing estimates of the household-level welfare impacts of the price changes
generated by a Computable General Equilibrium (CGE) model. The CGE analysis generates a set of
price changes; these embody both the direct price effects of the trade-policy change and indirect
effects on the prices of both traded and non-traded goods once all markets respond to the reform.
Standard methods of first-order welfare analysis are used to measure the gains and losses at
household level.
Our focus here is very much on the short-term welfare impacts. In keeping with the
limitations of the preceding general equilibrium analysis, our approach does not capture the dynamic
effects of trade reform through labor market adjustment and technological innovation. Nor does it
capture potential gains to the environment.1
The specifics of our approach to estimating welfare impacts at the household level can be
outlined as follows.2 Each household has preferences over consumption and work effort represented
by the utility function ui(qi , Li) where qi is a vector of the quantities of commodities demanded
d d
by household i and Li is a vector of labor supplies by activity, including supply to the household's
1 Though it is not a subject of the present analysis, arguments are also made about adverse environmental
impacts arising from the expansion of protected cereal production into marginal areas, It is claimed that
scarce water resources have also been diverted into soft wheat production. For further discussion see
World Bank (2003).
2 There are many antecedents of our approach in the literatures on both tax reform and trade reform, though
there are surprisingly few applications to point to in the ex ante assessment of actual reform proposals. For
another example see Chen and Ravallion (2004). Hertel and Reimer (2004) provide a useful overview of
the strengths and weaknesses of alternative approaches to assessing the welfare impacts of trade-policies,
including references to empirical examples for developing countries.
2
own production activities.3 The household is assumed to be free to choose its preferred combinations
of qi and Li subject to its budget constraint.
d
The household owns a production activity that generates a profit
i ( pi ) = max[ pi qi - ci (qi )] where pi is the vector of supply prices, and ci (qi ) is the
s s s s s s
household-specific cost function.4 The indirect utility function of household i is given by:
vi[pi , pi ,wi] = max[ui(qi , Li) pi qi = wiLi + i(pi )]
s d d d d s (1)
(qi ,Li )
d
where pi is the price vector for consumption, wi is the vector of wage rates.
d
We take the predicted price impacts from the CGE model as given for the analysis of
household-level impacts. In measuring the impacts we are constrained of course by the data, which
do not include prices and wages. However, this limitation does not matter to calculating a first-order
approximation to the welfare impact in a neighborhood of the household's optimum. Taking the
differential of (1) and using the envelope property (whereby the welfare impacts in a neighborhood of
an optimum can be evaluated by treating the quantity choices as given), the gain to household i
(denoted gi ) is given by the money metric of the change in utility:
gi dui m
= [pijqij
s s dpij
s
d d dpij
d n
]+(wkLsik dwk ) (2)
vi s - pij qij
j=1 pij pij
d
k=1 wk
where vi is the marginal utility of income for household i (the multiplier on the budget constraint in
equation 1) and Lsik is the household's "external" labor supply to activity k. (Notice that gains in
earnings from labor used in own production are exactly matched by the higher cost of this input to
own-production). The proportionate changes in prices are weighted by their corresponding
expenditure shares; the weight for the proportionate change in the j'th selling price is pijqij , the s s
revenue (selling value) from household production activities in sector j; similarly - pij qij is the
d d
(negative) weight for demand price changes and wkLsik is the weight for changes in the wage rate for
activity k. The term pijqij - pij qij gives (to a first-order approximation) the welfare impact of an
s s d d
equi-proportionate increase in the price of commodity j.
Equation (2) is the key formula we will use for calculating the welfare impacts at household
level, given the predicted price changes. In the specific model we will use (as discussed later), real
wage rates are fixed. So the last term on the right hand side of (2) drops out. (We discuss likely
implications of relaxing this assumption in section 3.5).
3 We make the standard assumptions that goods have positive marginal utilities while labor supplies have
negative marginal utilities.
4 On can readily include input prices in this cost function; see Chen and Ravallion (2004) for a more general
formulation. In the present context this makes no difference to the subsequent analysis so we subsume
factor prices in the cost function to simplify notation.
3
Notice that by applying the calculus in deriving (2) we are implicitly assuming small changes
in prices. Relaxing this requires more information on the structure of the demand and supply system;
see for example Ravallion and van de Walle (1991). This would entail considerable further effort,
and the reliability of the results will be questionable given the aforementioned problem of incomplete
price and wage data. For the same reason, we will have little choice but to largely ignore geographic
differences in the prices faced, or in the extent to which border price changes are passed on locally.
Having estimated the impacts at household level, we can study how they vary with pre-
reform welfare, and what impact the reform has on poverty and inequality. Let yi denote the pre-
reform welfare per person in household i while yi = yi + gi is its post-reform value, where gi is
*
the gain to household i. (Ideally, yi will be an exact money-metric of utility, though in practice it can
be expected that it is an approximation given omitted prices or characteristics). The distribution of
post-reform welfare levels is y1 , y2,...yn . By comparing standard summary measures of poverty or
* * *
inequality for this distribution with those for the pre-reform distribution, y1 , y2 ,...yn , we can assess
overall impacts.
Of obvious interest is to see how the gains vary with pre-reform welfare. Is it the poor who
tend to gain, or is it middle-income groups or the rich? However, it is important to recognize that the
assignment of impacts to the pre-reform distribution is very unlikely to be a degenerate distribution,
with no distribution of its own. There will almost certainly be a dispersion in impact at given pre-
reform welfare. This will arise from (observable and unobservable) heterogeneity in characteristics
and prices. It could also arise from errors in the welfare measure. Averaging across the distribution
of impacts at given pre-reform welfare, one can calculate the conditional mean impact given by:
gi = Ei(gi y = yi )
c (3)
where the expectation is formed over the conditional distributions of impacts. By including a
subscript i in the expectations operator in (3), we allow the possibility that the horizontal dispersion in
impacts is not identically distributed. In our empirical implementation, equation (3) will be estimated
using a non-parametric regression.
Taking these observations a step further, we can think of the overall impact on inequality as
having both vertical and horizontal components.5 This is straightforward for the Mean Log Deviation
(MLD) -- an inequality measure known to have a number of desirable features.6 The mean log
deviation defined on the distribution of post-reform welfares y1 , y2,...yn is given by:
* * *
I* = 1 n
ln( * (4)
n y* / yi )
i=1
5 Antecedents to this type of decomposition can be found in the literature on horizontal equity in taxation. In
the context of assessing a tax system, Auerbach and Hassett (2002) show how changes in an index of social
welfare can be decomposed into terms reflecting changes in the level and distribution of income, the burden
and progressivity of the tax system and a measure of the change in horizontal equity.
6 For further discussion of the MLD see Bourguignon (1979) and Cowell (2000). MLD is a member of the
General Entropy class of inequality measures.
4
n
where y* = yi * / n is mean post-reform welfare. Similarly,
i=1
n
I = 1 ln( (5)
n y / yi )
i=1
is the pre-reform MLD. (In both (4) and (5) it is assumed that yi > 0 and yi > 0 for all i. Thus I*-
*
I is the change in inequality attributable to the reform. The proposed decomposition of the overall
change in inequality can then be written as:
I* - I = 1 n 1+ g / y = 1 n 1+ g / y + 1 n 1+ gi / yi
c
(6)
n ln1+gi c
i=1 / yi n ln1+gi
i=1 / yi n ln1+gi
i=1 / yi
vertical component + horizontal component
The vertical component is the contribution to the change in total inequality (I*- I ) of the way in which
mean impacts vary with pre-reform welfare levels. If there is no difference in the proportionate
impact by level of welfare ( gi / yi = g / y for all i) then the vertical component is zero. The
c
horizontal component is the contribution of the deviations in impacts from their conditional means. If
the impact of the reform is predicted perfectly by pre-reform welfare ( gi = gi for all i) then the
c
horizontal component is zero.
We also want to try and explain the differences in impacts in terms of observable
characteristics of potential relevance to social protection policies. The way we have formulated the
problem of measuring welfare impacts above allows utility and profit functions to vary between
households at given prices. To try to explain the heterogeneity in measured welfare impacts we can
suppose instead that these functions vary with observed household characteristics. The indirect utility
function becomes:
vi(pi , pi , wi ) = v(pi , pi , wi, x1 , x2 ) = max[u(qi , Li, x1 ) pi qi - wiLi = i]
s d s d d d d
i i i (7)
where i = ( pi , x2 ) = max[ pi qi - c(qi , x2 )] .
s s s s
i i Note that we allow the characteristics that
influence preferences over consumption ( x1 ) to differ from those that influence the profits from own-
i
production activities ( x2 ).
i
The gain from the price changes induced by trade reform, as given by equation (7), depends
on the consumption, labor supply and production choices of the household, which depend in turn on
prices and characteristics, x1 and x2 . For example, households with a higher proportion of children
i i
will naturally spend more on food, so if the relative price of food changes then the welfare impacts
will be correlated with this aspect of household demographics. Similarly, there may be differences in
tastes associated with stage of the life cycle and education. There are also likely to be systematic
covariates of the composition of welfare.
Generically, we can now write the gain as:
gi = g( pi , pi ,wi, x1 , x2 )
s d
i i (8)
5
However, we do not observe the household-specific wages and prices. So we must make further
assumptions. In explaining the variation across households in the predicted gains from trade reform
we assume that: (i) the wage rates are a function of prices and characteristics as
wi = w(pi , pi , x1 , x2 ) and (ii) differences in prices faced can be adequately captured by a
d s
i i
complete set of regional dummy variables.
Under these assumptions, and linearizing (8) with an additive innovation error term, we can
write down the following regression model for the gains:
gi = 1x1i + 2x2i + kDki + i (9)
k
where Dki =1 if household i lives in county k and Dki = 0 otherwise and i is the error term.
3. Measured welfare impacts of trade reform in Morocco
3.1 The predicted price changes and the survey data
The price changes (implied by trade reform) we use here were generated by a CGE model
that was commissioned by a joint working group of the Ministry of Agriculture, Government of
Morocco, and the World Bank, as documented in Doukkali (2003). The model was constructed with
the aim of realistically representing the functioning of the Moroccan economy around 1997-98. The
model was explicitly designed to assess the aggregate impacts of de-protecting cereals in Morocco.
In addition to allowing for interactions between agriculture and the rest of the economy (represented
by six sectors), the model is quite detailed in its representation of the agricultural sector. It allows for
16 different crops or groups of crops, three different livestock activities, 13 major agro-industrial
activities, six agro-ecological regions, and within each region the model distinguishes between
rainfed agriculture and four types of irrigated agriculture. The model has two types of labor, both
with fixed real wage rates.
Four policy simulations are undertaken. The simulations then differ in the extent of the tariff
reductions for cereals, namely 10% (Policy 1), 30% (Policy 2), 50% (Policy 3) and 100% (Policy 4).
In all cases, the government's existing open-market operations, which attempt to keep down
consumer prices by selling subsidized cereals, are also removed.7 The loss of revenue from a 50%
tariff cut approximately equals the saving on subsidies.
Table 1 gives the predicted prices changes for various trade liberalization scenarios, based on
Doukkali (2003).8 As one would expect, the largest price impact is for cereals, though there are some
non-negligible spillovers into other markets, reflecting substitutions in consumption and production
and welfare effects on demand. Some of these spillover effects are compensatory. For example,
some producer prices rise with the de-protection of cereals.
7 In addition to administering the tariffs on imported soft wheat, the Government of Morocco buys, mills and
sells around one million tons of soft wheat in the form of low grade flour, which is sold on the open market
to help consumers.
8 Rachid Doukkali kindly provided price predictions from the CGE model mapped into the categories of
consumption and production identified in the survey. The production revenues were calculated from the
survey data by matching these consumption categories to the variables containing information about
household production of the corresponding goods.
6
The survey data set used here is the Enqu�te National sur le Niveau de Vie M�nages
(ENNVM) for 1998 done by the government's Department of Statistics, which kindly provided the
data set for the purpose of this study. This is a comprehensive multi-purpose survey following the
practices of the World Bank's Living Standards Measurement Study.9 The ENNVM has a sample of
5,117 households (of which 2,154 are rural) spanning 14 of Morocco's 16 regions (the low density
southernmost region -- the former Spanish Sahara -- was excluded). The sample is clustered and
stratified by region and urban/rural areas. The survey did not include households without a fixed
residence ("sans abris"). The survey allows calculation of a comprehensive consumption aggregate
(including imputed values for consumption from own production). We used the consumption
numbers calculated by the Department of Statistics. This is our money metric of welfare. Ideally this
would be deflated by a geographic cost-of-living index, but no such index was available, given the
aforementioned lack of geographic price data.
3.2 Implied welfare impacts at household level
Tables 2a, 2b which give the budget and income shares at mean points and the mean welfare
impacts broken down by commodity based on the ENNVM; Table 2a is for consumption while 2b is
for production. Notice how different consumption patterns are between urban and rural areas; for
example, rural households have twice the budget share for cereals as urban households. Strikingly,
while there is a 1.7% gain to urban consumers a s whole, this is largely offset by the general
equilibrium effects through other price changes (Table 2a). Also notice that income obtained directly
from production accounts for about one quarter of consumption; the rest is labor earnings, transfers
and savings. Of course in rural areas, the share is considerably higher, at 87%. And about one third
of this is from cereals.10
Table 3 summarizes the results on the implied welfare impacts. Our results indicate that the
partial trade reforms have only a small positive impact on the national poverty rate, as given by the
percentage of the population living below the official poverty lines for urban and rural areas used by
the Government's statistics office.11 However, a larger impact emerges when we simulate complete
de-protection (Policy 4). Then the national poverty rate rises from 20% to 22%. All four reforms
entail a decrease in urban poverty (though less than 0.4% points) and an increase in rural poverty.
(We will examine impacts over the whole distribution below).
Turning to the impacts on inequality in Table 3, we find that the trade reforms yield a small
increase in inequality, with the Gini index rising from 0.385 in the base case to 0.395 with a complete
de-protection of cereals (Policy 4). Impacts are smaller for the partial reforms (Policies 1-3). The
overall per capita gain is positive for the smaller tariff reduction (Policy 1) but becomes negative for
Policies 2, 3 and 4. As one would expect, there is a net gain to consumers and net loss to producers,
though the amounts involved are small overall. There are small net gains in the urban sector for
Policies 1-3. Larger impacts are found in rural areas, as we would expect. The mean percentage loss
from complete de-protection is a (non-negligible) 5.7% in rural areas.
9 The survey's design and content are similar in most respects to the 1991 Living Standards Survey for
Morocco documented in the LSMS web site: http://www.worldbank.org/lsms/.
10 Notice that there is no income from meat recorded in the data. The most plausible explanation is that
Moroccan farmers sell livestock to butchers or abattoirs rather than selling meat as such. Following
conventional survey processing practices, livestock is treated as an asset, so that proceeds from the selling
of livestock is not treated as income. This is questionable. As a test, we redid our main calculations using
the survey data on the transaction in livestock, and adding net sales into income. This made negligible
difference to the results. Details are available from the authors.
11 These have been updated using the CPI. The poverty lines were 3922 Dirham per year in urban areas and
3037 in rural areas. See World Bank (2001) for details.
7
Table 3 gave our results for the impact on poverty as estimated using the government's
official poverty lines. It is important to test robustness to alternative poverty lines. For this purpose,
we use the "poverty incidence curve," which is simply the cumulative distribution function up to a
reasonable maximum poverty line. The results are given in Figure 1; to make the figure easier to read
we focus on Policies 1 and 4. (The curves for Policies 2 and 3 are between these two).
We see that there is an increase in poverty overall from complete de-protection; this is robust
to the poverty line and poverty measure used (within a broad class of measures; see Atkinson, 1987).
The impact on poverty is almost entirely in rural areas; indeed, there is virtually no impact on urban
poverty. However, in rural areas the results in Figure 1 suggest a sizeable impact on poverty from
complete de-protection. The mean loss as a proportion of consumption for the poorest 15% in rural
areas is about 10%. There is an increase in the proportion of the rural population living below 2000
Dirham per person per year from 6.2% to 9.9%; the proportion living below 3000 Dirham rises from
22.2% to 26.3%. (For the country as a whole, the poverty rate for the former poverty line rises from
2.8% to 4.4% under Policy 4, while it rises from 11.4% to 13.1% for the 3000 Dirham line).
Our finding of adverse impacts on the rural poor contradicts claims made by some observers
who have argued that the rural poor tend to be net consumers of cereals, the commodity that incurs
the largest price decrease with this trade reform (Table 1). We will return to this point when we study
the welfare impacts further.
Table 4 gives the mean impacts of Policy 4 by region, split urban and rural. Impacts in urban
areas are small in all regions, with the highest net gain as a percentage of consumption being 1.3% in
Tanger-Tetouan, closely followed by Tensift Al Haouz and Fes-Boulemane. The rural areas with
largest mean losses from de-protection of cereals are Tasla Azilal, Meknes Tafil, Fes-Boulemane and
Tanger-Tetouan. Table 4 also gives mean impacts for the poorest 15% in rural areas (in terms of
consumption per person). When we focus on the rural poor defined this way, the region incurring the
largest mean loss for rural households is Tanger-Tetouan, followed by Fes-Boulemane and Chaouia-
Ouardigha. The contrast between the small net gains to the urban sector and net losses to the rural
poor is most marked in Tanger-Tetouan.
To begin exploring the heterogeneity in welfare impacts, Figure 2 gives the cumulative
frequency distributions of the gains and losses. To simplify the figure we again focus on Policies 1
and 4. We find that with complete de-protection (Policy 4) about 8.9% of the households incurred
losses greater than 500 Dirhams per year (about 5% of overall mean consumption) while about 5%
lose more than 1000 Dirhams per year. As one would expect, there is a "thicker tail" of negative
gains for rural areas. About 16% of rural households lose more that 500 Dirhams and 10% lose more
than 1000.
In Figure 3 we plot the mean gains against percentiles of consumption per capita for Policies
1 and 4. We give both absolute gains/losses and gains as a % of the household's consumption. For
policy 1, there is a tendency for the mean absolute gain to rise as one moves from the poorest
percentile through to the richest, though the gradient is small. The mean proportionate gain is quite
flat. For Policy 4, mean absolute impacts also rise up to the richest decile or so, but then fall.
Proportionate gains follow the same pattern though (again) the gradient seems small.
However, what is most striking from Figure 3 is the wide spread, particularly downwards
(indicating losers from the reform). The variance in absolute impacts is particularly large at the upper
end of the consumption distribution, though if anything the dispersion in proportionate impacts tends
to be greater at the other end of the distribution, amongst the poorest.
8
In Figure 4 we provide a split between producers and consumers for Policy 4. As we would
expect, to the extent that there is much impact on producers, they tend to lose, though not more so for
poor producers than rich ones. For consumption we tend to see more gainers, and a higher variance
in impact as one moves up the consumption distribution. However, we see that the downward
dispersion in total welfare impacts in Figure 3 is due more to the conditional variance in impacts
through production than through consumption.
There are two quite striking findings in these Figures. Firstly, notice that there are sizeable
losses on the production side amongst the poor. Granted, some large losses are evident for the high
income groups. But the claims that the poor do not lose as producers are clearly false. Furthermore,
the poor are often not seeing compensatory gains as consumers.
Secondly, it is notable that the results in Figures 3 and 4 indicate that the mean gains vary
little with mean consumption. Focusing on the "poor" versus the "rich" is hardly of much interest in
characterizing gainers and losers from this reform. The diversity in impacts tend to be "horizontal" in
the distribution of income, meaning that there tend to be larger differences in impacts at given
consumption than in mean impacts between different levels of consumption.
Next we examine these two findings in greater detail.
3.3 Who are the net producers of cereals in Morocco?
In the population as a whole, we find that 16% of households are net producers (value of
cereals production exceeds consumption). These households are worse off from the fall in cereal
prices due to de-protection. In rural areas, the proportion is 36%.
However, the survey data do not support the claim that the rural poor in Morocco are on
average net consumers of cereals. Figure 5 shows how producers and net producers are spread across
the distribution of total household consumption per person in rural Morocco. We give both the scatter
of points and the conditional means estimated using the local regression method.12 In the first (top
left) panel we give the proportion of producers. Then we give the proportion of net producers (for
whom production exceeds consumption of cereals in value terms). Finally we give net production in
value terms. In each case the horizontal axis gives the percentile of the distribution of consumption
from poorest through to richest.
We find that a majority of the rural poor produce cereals. Naturally much of this is for home
consumption. However, even if we focus solely on net producers, we find that over one third of the
poorest quintile tend to produce more than they consume. Furthermore, the mean net production in
value terms tends to be positive for the poor; in rural areas, the losses to poor producers from falling
cereal prices outweigh the gains to poor consumers. More than any single feature of the survey data,
it is this fact that lies at the heart of our finding that the rural poor lose from the reform.
3.4 Vertical versus horizontal impacts on inequality
To measure the relative importance of the vertical versus horizontal differences in impact, we
can use the decomposition method outlined in section 2. This decomposition requires an estimate of
12 See Cleveland (1979). This is often referred to as LOWESS (Locally Weighted Scatter Plot Smoothing).
We used the LOWESS program in STATA.
9
the conditional mean E(g y) , i.e., the regression function of g on y. We estimated this using the
nonparametric local regression method of Cleveland (1979).
Table 5 gives the results of this decomposition for each policy reform. For the small partial
reform under Policy 1, the vertical component dominates, accounting for 73% of the impact on
inequality. However, as one moves to the bigger reforms, the horizontal component becomes
relatively large. Indeed, we find that 119.8% of the impact of Policy 4 on inequality is attributable to
the horizontal component, while -19.8% is due to the vertical component. So we find that the vertical
component was inequality reducing for Policy 4, even though overall inequality rose (Table 5).
There is clearly a high degree of horizontal inequality in measured impacts at given mean
consumption. Some of this is undoubtedly measurement error, which may well become more
important for larger reforms. But some is attributable to observable covariates of consumption and
production behavior, as discussed in section 2. In trying to explain this variance in welfare impacts,
the characteristics we consider include region of residence, whether the household lives in an urban
area, household size and demographic composition of the household, age and age-squared of the
household head, education and dummy variables describing some key aspects of the occupation and
principle sector of employment; Table 6 gives summary statistics on the variables to be used in the
regressions. We recognize that there are endogeneity concerns about these variables, though we think
those concerns are minor in this context, especially when weighed against the concerns about omitted
variable bias in estimates that exclude these characteristics. Under the usual assumption that the error
term is orthogonal to these regressors we estimate equation (9) by Ordinary Least Squares.
The results are given in Table 7. Recall that these are averages across the impacts of these
characteristics on the consumption and production choices that determine the welfare impact of given
price and wage changes. This makes interpretation difficult. We view these regressions as being
mainly of descriptive interest, to help isolate covariates of potential relevance in thinking about
compensatory policy responses.
Focusing first on the results for Policy 4, we find that larger losses from full de-protection of
cereals are associated with families living in rural areas, that are relatively smaller (the turning point
in the U-shaped relationship is at a household size of about one), have more wage earners, higher
education, work in commerce, transport etc., and live in Chaouia-Ouardigha, Rabat, Tadla Azilal and
Meknes Tafil. Recall that these effects stem from the way household characteristics influence net
trading positions in terms of the commodities for which prices change. So, for example, it appears
that larger families tend to consume more cereals, and so gain more from the lower cereals prices.
Results are similar for partial de-protection, though education becomes insignificant for Policy 1.
In Table 8 we give an urban-rural breakdown of the regressions for Policies 1 and 4. There
are a couple of notable differences. (Again we focus on Policy 4 in the interests of brevity). We find
significant positive effects of having more children and teenagers on the gains from trade reform in
rural areas, presumably because such families are more likely to be cereal consumers. The education
effect at higher levels of schooling is much more pronounced in urban areas. The effect of working in
the transport and commerce sector is more statistically significant in urban areas, though this effect is
still sizeable in rural areas. The regional effects are more statistically significant in urban areas than
in rural areas. Of course there are still sizable regional differences in mean impacts in Table 8,
though they are statistically less significant than we found in Table 7. In fact the quantitative
magnitudes of the regional differences are just as large for rural areas in Table 8 as for urban plus
rural areas in Table 7.
10
It should not be forgotten that the results in Tables 7 and 8 are conditional geographic effects
(conditional on the values taken by other covariates in the regressions). As we saw in Table 4, there
are pronounced (unconditional) geographic differences in mean impacts in rural areas across different
regions. Whether one draws policy lessons more from the conditional or unconditional effects
depends on the type of policy one is using. If it is simply regional targeting then of course the
unconditional geographic effects in Table 4 will be more relevant. However, finer targeting by
household characteristics, in combination with regional targeting, will call for the sorts of results
presented in Tables 7 and 8.
The share of the variance in gains that is accountable to these covariates is generally less than
10%. Values of R2 of this size are common in regressions run on large cross-sectional data sets,
though it remains true that a large share of the variance in impacts is not accountable to these
covariates. (The exception to our low R2 is for Policy 1, for which almost half of the variance in gains
across urban households is explained). It must be expected that there is a sizable degree of
measurement error in the gains, stemming from measurement error in the underlying consumption
and production data. No doubt there are also important idiosyncratic factors in household-specific
tastes or production choices.
These regressions try to explain the variance in the gains from the reform. It is of interest to
see if we can do any better in explaining the incidence of losses from reform amongst the poor. This
is arguably of greater relevance to compensatory policies, which would presumably want to focus on
poor losers. To test how well the same set of regressors could explain who was a poor loser from the
reforms we constructed a dummy variable taking the value unity if a rural household incurred a
negative loss and was "poor"; to assure a sufficient number of observations taking the value unity we
set the poverty line higher than the official line, namely at a consumption per person of 5,000 Dirham
per year (rather than the official line of about 3,000). (We confined this to rural areas since that is
where the losses are concentrated). In the case of full de-protection (Policy 4), we find that about
14% of the variance in this measure can be explained by the set of regressors in Table 8 while for
Policy 1 the share is 20%.13 While there are a number of identifiable covariates for identifying likely
losers amongst the poor, it is also clear that there is a large share of the variance left unexplained.
Another way to assess how effectively this set of covariates can explain the incidence of a net
loss from reform amongst the poor is by comparing the actual value of the dummy variable described
above with its predicted values from the model, using a cut off probability of 0.5. For Policy 4, there
are 472 households out of 2,100 who were both poor and incurred a loss due to the reform. Of these
the model could only correctly predict that this was the case for 18% (86 households). For Policy 1,
the model prediction was correct for 27% of the 463 households who were both poor and were made
worse off by the reform.
Yet most forms of indicator targeting -- whereby transfers are contingent on readily observed
variables, such as location -- would be based on similar variables to those we have used in our
regressions; indeed, if anything targeted policies use fewer dimensions. This suggests that indicator
targeting will be of only limited effectiveness in reaching those in greatest need. Self-targeting
mechanisms that create incentives for people to correctly reveal their status (such as using work
requirements) may be better able to do so.
13 The R2 for OLS regressions are 0.139 and 0.191 for Policy 4 and 1 respectively. Using instead a probit
model to correct for the nonlinearity the pseudo R2's are 0.135 and 0.196.
11
3.5 Two caveats
While the above results are suggestive, two limitations of our analysis should be noted. The
first stems from the fact that the Doukkali (2003) model assumed fixed wage rates. While sensitivity
to alternative labor market assumptions should be checked, we can speculate on the likely impacts of
allowing real wages to adjust to the reforms. Here it can be argued that the export-oriented cash crops
that will replace cereals will tend to be more labor intensive than cereals. Thus we would expect
higher aggregate demand for the relatively unskilled labor used in agriculture, and hence higher real
wages for relatively poorer groups. This will undoubtedly go some way toward compensating the
rural poor, and may even tilt the vertical distributional impacts in favor of the poor.
A second concern is that there may well be dynamic gains from greater trade openness that
are not being captured by the model used to generate the relative price impacts; for example, trade
may well facilitate learning about new agricultural technologies and innovation that brings longer-
term gains in farm productivity. These effects may be revealed better by studying time series
evidence, combined with cross-country comparisons.
4. Conclusions
The welfare impacts of de-protection in developing countries have been much debated. Some
people have argued that external trade liberalizations are beneficial to the poor while others argue that
the benefits will be captured more by the non-poor. Expected impacts on domestic prices have figured
prominently in these debates.
The paper has studied the welfare impacts at household level of the changes in commodity
prices attributed to a proposed trade reform, namely Morocco's de-protection of its cereals sector.
This would entail a sharp reduction in tariffs, with implications for the domestic structure of prices
and hence household welfare. The paper draws out the implications for household welfare of the
previous estimates of the price impacts of reform done for a Joint Government of Morocco and World
Bank Committee. Standard methods of first-order welfare analysis are used to measure the gains and
losses at household level using a large sample survey.
In a number of respects, our detailed household-level analysis throws into question past
claims about the likely welfare impacts of this trade reform. In the aggregate, we find a small
negative impact on mean household consumption and a small increase in inequality. There is a
sizable, and at least partly explicable, variance in impacts across households. Rural families tend to
lose; urban households tend to gain. There are larger impacts in some provinces than others, with
highest negative impacts for rural households in Tasla Azilal, Meknes Tafil, Fes-Boulemane and
Tanger-Tetouan. Mean impacts for rural households in these regions are over 10% or more of
consumption. There are clearly sizeable welfare losses amongst the poor in these specific regions.
The adverse impact on rural poverty stems in large part from the fact that the losses to the net
producers of cereals outweigh the gains to the net consumers amongst the poor. Thus, on balance
rural poverty rises. This contradicts the generalizations that have been made in the past that the rural
poor in Morocco tend to be net consumers of grain, and hence gainers from trade reform. Yes, a
majority are net consumers, but on balance the welfare impacts on the rural poor are negative.
Our results lead us to question the high level of aggregation common in past claims about
welfare impacts of trade reform. We find diverse impacts at given pre-reform consumption levels.
This "horizontal" dispersion becomes more marked as the extent of reform (measured by the size of
12
the tariff cut) increases. Indeed, we estimate that all of the impact of complete de-protection of
cereals on inequality is horizontal rather than vertical; the vertical impact on inequality was actually
inequality reducing. For a modest reform of a 10% cut in tariffs, the vertical component dominates,
though there is still a large horizontal component. It is clear from our results that in understanding the
social impacts of this reform, one should not look solely at income poverty and income inequality as
conventionally measured; rather one needs to look at impacts along "horizontal" dimensions, at given
income.
We have been able to identify some specific types of households whose consumption and
production behavior makes them particularly vulnerable. These results are suggestive of the targeting
priorities for compensatory programs. The fact that we also find a large share of unexplained
variance in impacts also points to the limitations of targeting based on readily observable indicators,
suggesting that self-targeting mechanisms may also be needed.
13
Table 1. Predicted price changes due to agricultural trade reform in Morocco
Sectors Consumers (% change in prices) Producers (% change in prices)
Policy 1 Policy 2 Policy 3 Policy 4 Policy 1 Policy 2 Policy 3 Policy 4
Cereals and cereals products -3.062 -7.786 -12.811 -26.691 -2.858 -7.193 -11.744 -24.107
Fresh vegetables -0.714 -0.884 -1.051 -1.128 -0.580 -0.767 -0.871 -0.756
Fruits -0.637 -0.681 -0.683 -0.139 -0.429 -0.301 -0.104 0.843
Dairy products and eggs -0.472 -0.414 -0.257 0.751 -0.505 -0.487 -0.333 0.637
Meat (red and poultry) -0.320 -0.109 0.332 1.896 -0.306 -0.078 0.357 1.936
Sugar -0.200 0.100 0.400 1.300 -0.368 -0.378 -0.354 -0.094
Edible oils -0.671 -1.064 -1.405 -2.225 -0.632 -0.998 -1.336 -2.061
Fresh and processed fish 0.000 0.696 1.300 2.996 0.000 0.600 1.300 2.881
Other ag. and processed food -0.369 -0.402 -0.421 -0.635 0.268 1.294 2.475 5.388
Services 0.142 0.500 0.758 1.460 0.056 0.500 0.844 1.708
Energy, electricity and water -0.060 0.540 1.140 2.580 -0.051 0.549 1.149 2.597
Other industries 0.000 0.600 1.200 2.800 0.000 0.600 1.200 2.793
Note: The tariff cuts on imported cereals are 10%, 30%, 50% and 100% for Policies 1,2,3 and 4 respectively.
14
Table 2a. Consumption shares and welfare impacts through consumption
Consumption Policy 1 Policy 2 Policy 2 Policy 4
Shares
National
Cereals 0.084 0.2572 0.6540 1.0761 2.2420
Fresh vegetables 0.042 0.0297 0.0368 0.0437 0.0469
Fruits 0.022 0.0139 0.0148 0.0148 0.0030
Dairy products and eggs 0.032 0.0153 0.0134 0.0083 -0.0243
Meat (red and poultry) 0.112 0.0359 0.0122 -0.0373 -0.2129
Sugar 0.015 0.0030 -0.0015 -0.0060 -0.0195
Edible oils 0.032 0.0212 0.0336 0.0444 0.0703
Fresh and processed fish 0.013 0.0000 -0.0089 -0.0166 -0.0383
Ag. and processed food 0.101 0.0371 0.0405 0.0424 0.0640
Services 0.066 -0.0094 -0.0332 -0.0504 -0.0971
Energy, electricity, water 0.148 0.0089 -0.0799 -0.1688 -0.3819
Other industries 0.333 0.0000 -0.2000 -0.4001 -0.9335
Total 1.000 0.4127 0.4817 0.5506 0.7187
Urban
Cereals 0.066 0.2034 0.5172 0.8510 1.7730
Fresh vegetables 0.037 0.0264 0.0327 0.0389 0.0417
Fruits 0.022 0.0139 0.0149 0.0149 0.0030
Dairy products and eggs 0.034 0.0160 0.0141 0.0087 -0.0255
Meat (red and poultry) 0.107 0.0342 0.0116 -0.0355 -0.2027
Sugar 0.011 0.0021 -0.0011 -0.0042 -0.0138
Edible oils 0.024 0.0163 0.0258 0.0341 0.0540
Fresh and processed fish 0.014 0.0000 -0.0096 -0.0180 -0.0414
Ag. and processed food 0.096 0.0354 0.0386 0.0404 0.0610
Services 0.067 -0.0095 -0.0333 -0.0505 -0.0973
Energy, electricity, water 0.155 0.0093 -0.0835 -0.1763 -0.3990
Other industries 0.368 0.0000 -0.2207 -0.4414 -1.0300
Total 1.000 0.3476 0.3067 0.2621 0.1231
Rural
Cereals 0.136 0.4154 1.0565 1.7383 3.6217
Fresh vegetables 0.055 0.0394 0.0487 0.0579 0.0622
Fruits 0.021 0.0137 0.0146 0.0146 0.0030
Dairy products and eggs 0.028 0.0131 0.0114 0.0071 -0.0208
Meat (red and poultry) 0.128 0.0410 0.0139 -0.0425 -0.2427
Sugar 0.028 0.0056 -0.0028 -0.0112 -0.0364
Edible oils 0.053 0.0356 0.0564 0.0746 0.1181
Fresh and processed fish 0.010 0.0000 -0.0068 -0.0126 -0.0291
Ag. and processed food 0.115 0.0422 0.0461 0.0482 0.0728
Services 0.066 -0.0094 -0.0330 -0.0501 -0.0965
Energy, electricity, water 0.129 0.0077 -0.0694 -0.1466 -0.3317
Other industries 0.232 0.0000 -0.1392 -0.2785 -0.6498
Total 1.000 0.6042 0.9964 1.3993 2.4708
15
Table 2b. Percentage gains from each policy: Production component
Production
as a share of Policy 1 Policy 2 Policy 2 Policy 4
total consumption
National
Cereals 0.089 -0.2713 -0.6899 -1.1352 -2.3652
Fresh vegetables 0.053 -0.0381 -0.0471 -0.0560 -0.0601
Fruits 0.041 -0.0261 -0.0279 -0.0280 -0.0057
Dairy products and eggs 0.051 -0.0243 -0.0213 -0.0132 0.0386
Meat (red and poultry) 0.000 0.0000 0.0000 0.0000 0.0000
Sugar 0.000 0.0000 0.0000 0.0000 0.0000
Edible oils 0.025 -0.0169 -0.0268 -0.0354 -0.0560
Fresh and processed fish 0.000 0.0000 0.0000 0.0000 0.0000
Ag. and processed food 0.002 -0.0008 -0.0008 -0.0009 -0.0013
Services 0.000 0.0000 0.0000 0.0000 0.0000
Energy, electricity, water 0.000 0.0000 0.0000 0.0000 0.0000
Other industries 0.000 0.0000 0.0000 0.0000 0.0000
Total 0.262 -0.3774 -0.8139 -1.2687 -2.4498
Urban
Cereals 0.010 -0.0311 -0.0792 -0.1303 -0.2716
Fresh vegetables 0.008 -0.0058 -0.0072 -0.0086 -0.0092
Fruits 0.016 -0.0105 -0.0112 -0.0112 -0.0023
Dairy products and eggs 0.007 -0.0031 -0.0027 -0.0017 0.0049
Meat (red and poultry) 0.000 0.0000 0.0000 0.0000 0.0000
Sugar 0.000 0.0000 0.0000 0.0000 0.0000
Edible oils 0.013 -0.0087 -0.0138 -0.0183 -0.0289
Fresh and processed fish 0.000 0.0000 0.0000 0.0000 0.0000
Ag. and processed food 0.000 0.0000 0.0000 0.0000 0.0000
Services 0.000 0.0000 0.0000 0.0000 0.0000
Energy, electricity, water 0.000 0.0000 0.0000 0.0000 0.0000
Other industries 0.000 0.0000 0.0000 0.0000 0.0000
Total 0.054 -0.0593 -0.1142 -0.1701 -0.3071
Rural
Cereals 0.319 -0.9777 -2.4863 -4.0910 -8.5235
Fresh vegetables 0.186 -0.1329 -0.1645 -0.1955 -0.2099
Fruits 0.113 -0.0722 -0.0771 -0.0773 -0.0158
Dairy products and eggs 0.183 -0.0865 -0.0758 -0.0471 0.1375
Meat (red and poultry) 0.000 0.0000 0.0000 0.0000 0.0000
Sugar 0.000 0.0000 0.0000 0.0000 0.0000
Edible oils 0.061 -0.0409 -0.0649 -0.0857 -0.1357
Fresh and processed fish 0.000 0.0000 0.0000 0.0000 0.0000
Ag. and processed food 0.008 -0.0031 -0.0033 -0.0035 -0.0053
Services 0.000 0.0000 0.0000 0.0000 0.0000
Energy, electricity, water 0.000 0.0000 0.0000 0.0000 0.0000
Other industries 0.000 0.0000 0.0000 0.0000 0.0000
Total 0.870 -1.3131 -2.8719 -4.5000 -8.7527
16
Table 3. Household impacts of four trade reforms
Baseline Policy 1 Policy 2 Policy 3 Policy 4
National
Poverty rate (%) 19.61 20.01 20.33 21.04 22.13
Mean Log Deviation (x100) 28.50 28.92 29.00 29.14 29.17
Gini index 0.385 0.387 0.389 0.391 0.395
Per capita gain 0 6.519 -23.967 -54.816 -133.81
Mean % gain: price changes
weighted by mean shares 0 -0.059 -0.513 -0.971 -2.141
Mean % gain: weighted by
ratios of means (Tables 2a,b) 0 0.035 -0.332 -0.718 -1.731
Production gain 0 -32.078 -69.012 -106.308 -201.017
Consumption gain 0 38.598 45.046 51.492 67.207
Consumption per capita 9350.913 9357.433 9326.947 9296.097 9217.104
Urban
Poverty rate (%) 12.19 12.05 11.96 12.05 11.76
Mean Log Deviation (x100) 25.49 25.41 25.32 25.23 24.93
Gini index 0.366 0.365 0.365 0.364 0.362
Per capita gain 0 35.518 24.8 13.747 -16.491
Mean % gain: price changes
weighted by mean shares 0 0.357 0.374 0.394 0.442
Mean % gain: weighted by
ratios of means (Tables 2a,b) 0 0.288 0.193 0.092 -0.184
Production gain 0 -6.308 -12.103 -17.793 -31.302
Consumption gain 0 41.826 36.903 31.54 14.811
Consumption per capita 12031.2 12066.72 12056 12044.95 12014.71
Rural
Poverty rate (%) 28.28 29.31 30.10 31.54 34.25
Mean Log Deviation (x100) 17.47 17.82 17.82 17.93 17.76
Gini index 0.312 0.313 0.315 0.318 0.328
Per capita gain 0 -33.532 -91.321 -149.512 -295.845
Mean % gain: price changes
weighted by mean shares 0 -0.634 -1.737 -2.855 -5.708
Mean % gain: weighted by
ratios of means (Tables 2a,b) 0 -0.709 -1.875 -3.101 -6.282
Production gain 0 -67.671 -147.612 -228.562 -435.419
Consumption gain 0 34.139 56.291 79.049 139.574
Consumption per capita 5649.034 5615.502 5557.712 5499.522 5353.189
Note: All monetary units are Moroccan Dirham per year. MLD is only calculated over the set of households
for whom consumption is positive. The mean % gains weighted by mean shares are simply the means across
the sample of the % gains at household level. The second mean % gain is weighted by shares at the means
points based on Tables 2a,b.
17
Table 4. Mean gains from Policy 4 by region
Region Total Urban Rural Poorest 15% of
rural
households
Oued Ed-Dahab-Lagouira -0.2 -0.2 . .
Laayoune-Boujdour-Sakia El Hamra -0.34 -0.34 . .
Guelmime Es-Semara -0.96 0.72 -3.47 -0.58
Souss-Massa-Daraa -1.31 0.42 -2.4 -3.09
Gharb-Chrarda-Beni Hssen -2.16 0.02 -3.86 0.1
Chaouia-Ouardigha -4.18 0.32 -8.31 -10.11
Tensift Al Haouz -0.87 1.12 -2.17 0.31
Oriental -0.87 0.38 -2.78 0.25
G.Casablanca 0.48 0.41 2.41 .
Rabat-Sal�-Zemmour-Zaer -0.59 0.33 -4.98 0.23
Doukala Abda -3.13 0.76 -5.92 -3.93
Tadla Azilal -6.93 -0.71 -11.04 -0.95
Meknes Tafil -4.89 -0.19 -11.35 -8.48
Fes-Boulemane -2.4 1.05 -11.52 -13.43
Taza-Al Hoceima-Taounate -4.47 -0.32 -5.78 -8.39
Tanger-Tetouan -2.94 1.31 -9.4 -22.03
Total -2.14 0.45 -5.71 -10.39
Note: Means formed over the household level % gains (equivalent to weighting
proportionate price changes by mean shares).
Table 5. Decomposition of the impact on inequality
Policy 1 Policy 2 Policy 3 Policy 4
Vertical component 72.69 57.57 38.77 -19.77
Horizontal component 27.31 42.43 61.23 119.77
Total 100 100 100 100
Note: The decomposition is only implemented on the sample of households for whom both the
baseline and post-reform consumption is positive
.
18
Table 6. Summary statistics on explanatory variables in the regression analysis
Mean Std. Dev
Urban 0.580 binary
Log household size 1.645 0.550
Log household size 2 3.009 1.621
Female headed household 0.170 binary
If unemployed present 0.248 binary
Number of wage earners 5.912 2.878
Share of children 0-6 0.140 0.162
Share of children 7-17 0.221 0.204
Share of elderly 60+ 0.120 binary
Characteristics of the head
Age of the head 0.505 0.143
Age of the head 2 0.275 0.155
Illiterate head 0.582 binary
Incomplete primary school 0.100 binary
Primary school completed 0.164 binary
Low secondary school 0.058 binary
Upper secondary school 0.059 binary
University 0.036 binary
Industry
Not-employed 0.240 binary
Industrie/B.T.P 0.004 binary
Commerce/Transp./Commun./Admin. 0.273 binary
Service Soci. 0.085 binary
Autres services 0.064 binary
Corps Exter. 0.125 binary
Chomeur 0.012 binary
Femme au foyeur/Eleve/Etudiant 0.037 binary
Jeune enfant 0.009 binary
Vielliard/Retraite/Rentiers 0.074 binary
Infirme/malade 0.068 binary
Autre inactifs 0.010 binary
Regions
Oued Ed-Dahab-Lagouira 0.012 binary
Laayoune-Boujdour-Sakia El Hamra 0.014 binary
Guelmime Es-Semara 0.023 binary
Souss-Massa-Daraa 0.094 binary
Gharb-Chrarda-Beni Hssen 0.058 binary
Chaouia-Ouardigha 0.054 binary
Tensift Al Haouz 0.100 binary
Oriental 0.065 binary
G.Casablanca 0.124 Binary
Rabat-Sal�-Zemmour-Zaer 0.081 Binary
Doukala Abda 0.067 Binary
Tadla Azilal 0.047 Binary
Meknes Tafil 0.072 binary
Fes-Boulemane 0.051 binary
Taza-Al Hoceima-Taounate 0.058 binary
19
Table 7. Regression of per capita gain/loss on selected household characteristics
Policy 1 Policy 2 Policy 3 Policy 4
Coeff. s.e. Coeff. s.e. Coeff. s.e. Coeff. s.e.
Urban 26.139*** 6.275 44.850*** 12.948 64.218** 20.068 113.714** 39.213
Log household size -57.242** 19.583 -78.454* 40.407 -100.548 62.626 -157.373 122.376
Log household size 2 77.337*** 16.806 167.523*** 34.678 260.865*** 53.746 508.026*** 105.023
Female headed household 2.502 7.431 4.072 15.333 5.605 23.765 9.161 46.438
If unemployed present 10.018* 5.909 23.344* 12.192 36.428* 18.896 67.997* 36.924
Number of wage earners -44.722*** 7.019 -101.428*** 14.484 -159.842*** 22.448 -313.541*** 43.865
Share of children 0-6 32.783* 17.72 89.774* 36.564 145.705* 56.67 277.637* 110.736
Share of children 7-17 25.070* 14.155 69.367* 29.206 113.738* 45.266 221.518* 88.453
Share of elderly 60+ -21.3 15.584 -23.551 32.155 -24.389 49.837 -24.334 97.385
Characteristics of the head
Age of the head -38.511 108.759 -151.473 224.41 -272.681 347.809 -624.596 679.642
Age of the head 2 44.097 102.579 142.598 211.658 246.231 328.045 543.07 641.022
Household is literate
only -8.871 7.983 -23.441 16.472 -38.257 25.53 -76.735 49.888
Incomplete primary
education Reference
Primary school completed -14.013* 6.757 -40.623** 13.942 -68.220** 21.608 -141.296*** 42.224
Low secondary school -12.98 10.4 -61.634** 21.458 -112.583*** 33.258 -250.335*** 64.989
Upper secondary school -12.462 10.775 -70.619** 22.233 -130.320*** 34.458 -286.333*** 67.333
University 2.575 13.527 -95.376*** 27.912 -197.887*** 43.26 -476.077*** 84.533
Industry
Not-working/Agriculture Reference
Industrie/B.T.P -3.71 36.465 -0.277 75.242 4.541 116.616 21.281 227.874
Commerce/Transport/
Communications/Admin. -59.926*** 8.198 -122.454*** 16.915 -185.113*** 26.216 -341.751*** 51.228
Service Soci. 4.424 10.036 17.18 20.707 30.536 32.094 66.804 62.714
Autres services -0.2 11.251 9.572 23.214 19.812 35.98 47.874 70.306
Corps Exter. 2.385 8.936 6.785 18.439 10.912 28.579 20.23 55.844
Chomeur 6.627 21.518 27.715 44.399 49.65 68.813 107.951 134.465
Femme au
foyeur/Eleve/Etudiant 2.26 13.49 13.788 27.835 25.401 43.141 55.785 84.301
Jeune enfant 7.629 24.5 -3.891 50.553 -16.336 78.352 -51.207 153.104
Vielliard/Retraite/Rentiers 6.913 11.039 23.527 22.778 40.651 35.303 86.8 68.984
Infirme/malade 3.143 10.96 22.092 22.614 42.489 35.049 100.065 68.488
Autre inactifs -9.955 22.723 1.817 46.885 15.364 72.667 56.497 141.995
20
Table 7. Regression of per capita gain/loss on selected household characteristics (cont.)
Policy 1 Policy 2 Policy 3 Policy 4
Coeff. s.e. Coeff. s.e. Coeff. s.e. Coeff. s.e.
Oued Ed-Dahab-
Lagouira 19.216 22.51 -6.738 46.446 -34.818 71.986 -111.388 140.665
Laayoune-Boujdour-
Sakia El Hamra -1.502 21.067 -20.145 43.47 -40.764 67.374 -98.323 131.652
Guelmime Es-Semara 9.666 16.639 11.901 34.333 12.774 53.212 12.391 103.979
Souss-Massa-Daraa -7.645 10.868 5.611 22.425 22.766 34.756 85.2 67.916
Gharb-Chrarda-Beni
Hssen -10.087 12.229 -7.485 25.232 -3.592 39.107 10.494 76.418
Chaouia-Ouardigha -19.542 12.507 -49.255* 25.807 -81.319* 39.998 -169.114* 78.159
Tensift Al Haouz 2.964 10.696 14.527 22.071 27.258 34.207 65.274 66.842
Oriental -14.038 11.928 -19.198 24.612 -23.918 38.145 -31.056 74.539
G.Casablanca -3.322 10.429 -15.762 21.518 -28.418 33.35 -60.086 65.169
Rabat-Sal�-Zemmour-
Zaer -15.439 11.326 -33.817 23.371 -52.199 36.222 -97.061 70.78
Doukala Abda -13.169 11.76 -23.668 24.265 -34.315 37.607 -59.462 73.487
Tadla Azilal -55.774*** 13.093 -114.700*** 27.016 -174.099*** 41.872 -320.810*** 81.821
Meknes Tafil -37.594** 11.54 -74.192** 23.812 -111.929** 36.906 -209.391** 72.117
Fes-Boulemane -10.249 12.726 -15.356 26.259 -20.651 40.699 -33.326 79.528
Taza-Al Hoceima-
Taounate 5.613 12.367 2.43 25.517 -2.415 39.549 -21.329 77.281
Tanger-Tetouan Reference
Constant 144.096*** 34.638 247.104*** 71.472 354.469** 110.773 642.381** 216.458
R2 0.175 0.080 0.062 0.057
21
Table 8. Urban-rural split of regressions for per capita gains
Urban Rural
Policy 1 Policy 4 Policy 1 Policy 4
Coeff. s.e. Coeff. s.e. Coeff. s.e. Coeff. s.e.
Log household size -32.840* 16.071 45.705 83.159 -89.255* 45.084 -527.017* 294.353
Log household size 2 40.492* 17.841 217.663* 92.32 79.415* 32.524 555.880** 212.348
Female headed household -2.696 6.018 -15.603 31.139 11.984 16.902 27.785 110.356
If unemployed present 2.138 4.668 25.238 24.154 11.086 14.482 35.299 94.551
-
Number of wage earners -23.972** 8.39 -143.745*** 43.414 45.101*** 12.237 -321.182*** 79.894
Share of children 0-6 -15.648 15.206 25.903 78.686 95.815** 36.544 609.370* 238.601
Share of children 7-17 -10.44 11.986 -34.073 62.023 81.378** 29.771 622.563** 194.376
Share of elderly 60+ -17.696 13.328 4.67 68.967 -35.448 32.512 -167.42 212.274
Characteristics of the head
Age of the head -26.02 96.18 -513.051 497.696 -82.081 216.7 -1.00E+03 1414.846
Age of the head 2 33.769 91.377 263.429 472.842 103.772 202.766 1129.226 1323.868
Household is literate only -10.567 6.965 -90.700* 36.042 -8.718 16.11 -75.293 105.182
Incomplete primary education Reference
Primary school completed 0.157 5.566 -44.272 28.804 -31.613* 14.794 -270.881** 96.589
Low secondary school 6.416 7.632 -119.177** 39.494 -73.971* 31.399 -655.218** 205.005
Upper secondary school -5.731 7.551 -249.358*** 39.074 10.925 49.861 -46.655 325.547
University 9.241 9.282 -433.456*** 48.03 20.185 83.244 18.883 543.507
Industry
Not-working/Agriculture Reference
Industrie/B.T.P -4.779 25.641 7.254 132.684 56.769 124.939 366.598 815.737
Commerce/Transp./Commun./
Admin. -96.116*** 10.172 -444.047*** 52.634 -43.789** 15.445 -257.349* 100.843
Service Soci. -1.428 7.574 6.102 39.191 27.61 28.965 247.156 189.116
Autres services -4.7 9.133 6.023 47.259 21.228 25.434 161.257 166.061
Corps Exter. -2.611 6.884 -19.401 35.621 8.742 23.042 57.723 150.44
Chomeur -1.702 15.213 36.377 78.72 60.148 73.543 457.084 480.167
Femme au oyeur/Eleve/Etudiant -4.019 10.145 12.554 52.498 20.295 36.207 110.127 236.4
Jeune enfant -2.268 16.343 -129.322 84.567 107.247 152.23 720.704 993.92
Vielliard/Retraite/Rentiers 1.108 8.138 48.765 42.112 25.588 34.261 154.32 223.691
Infirme/malade 1.847 8.176 63.019 42.308 5.864 30.489 148.543 199.063
Autre inactifs -12.094 16.532 23.685 85.547 22.652 67.323 250.306 439.559
Regions
Oued Ed-Dahab-Lagouira 21.2 15.068 -135.288* 77.973
Laayoune-Boujdour-Sakia El
Hamra -2.496 14.153 -129.348* 73.236
Guelmime Es-Semara 7.558 13.813 -50.41 71.475 23.284 35.563 165.753 232.195
Souss-Massa-Daraa -1.425 10.023 -54.723 51.863 -8.417 21.371 211.302 139.535
Gharb-Chrarda-Beni Hssen -44.733*** 11.143 -204.020*** 57.663 17.31 23.762 208.808 155.141
Chaouia-Ouardigha -15.625 11.08 -89.734 57.333 -19.527 25.012 -201.804 163.304
Tensift Al Haouz -8.763 9.759 -37.2 50.5 8.732 21.097 147.015 137.74
Oriental -18.776* 9.806 -96.129* 50.74 -0.357 25.851 99.206 168.782
22
Table 8. Urban-rural split of regressions for per capita gains (cont.)
Urban Rural
Policy 1 Policy 4 Policy 1 Policy 4
Coeff. s.e. Coeff. s.e. Coeff. s.e. Coeff. s.e.
G.Casablanca -9.23 7.849 -112.350** 40.617 5.551 49.268 79.412 321.673
Rabat-Sal�-Zemmour-Zaer -13.825 8.683 -118.444** 44.931 -36.873 30.677 -142.714 200.295
Doukala Abda -14.916 10.867 -80.126 56.232 -8.244 22.773 -3.679 148.687
Tadla Azilal -50.624*** 12.423 -213.855*** 64.285 -51.570* 24.832 -324.785* 162.13
Meknes Tafil -22.753* 9.622 -126.779* 49.79 -56.111* 24.782 -311.079* 161.8
Fes-Boulemane -11.946 9.954 -38.193 51.509 -2.002 30.661 -5.31 200.186
Taza-Al Hoceima-Taounate -20.264 13.982 -161.597* 72.352 16.747 22.229 80.917 145.137
Tanger-Tetouan Reference
Constant 135.395*** 30.386 463.951** 157.234 162.613* 72.909 959.343* 476.029
R2 0.46 0.08 0.062 0.067
23
Figure 1. Impacts on poverty
Total
.6
Baseline Policy 1 Policy 4
enil .5
ytrevop .4
e
th
w .3
loeb
lepoep .2
% .1
0
0 1000 2000 3000 4000 5000
Annual per capita consumption
Urban Rural
.6 .6
enil .5 enil .5
ytrevop .4 ytrevop .4
e e
th th
w .3 w .3
loeb loeb
lepoep .2 lepoep .2
% .1 % .1
0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Annual per capita consumption Annual per capita consumption
24
Figure 2. Frequency distributions of gains/losses for Policies 1 and 4
Total Total
1 .02
Policy 1
.9
Policy 4
.8
.015
.7
itysned .6 itysned
e
tiv .5 .01
lau .4
mu ityilbabo
C.3 Pr
.005
.2
.1
0 0
-3000 -2000 -1000 -500 0 500 1000 -600 -400 -200 0 200 400
Urban Rural
1 1
.9 .9
.8 .8
.7 .7
itysned .6 itysned .6
e e
tiv .5 tiv .5
lau .4 lau .4
mu mu
C .3 C .3
.2 .2
.1 .1
0 0
-3000 -2000 -1000 -500 0 500 1000 -3000 -2000 -1000 -500 0 500 1000
Absolute gain pre capita Absolute gain per capita
25
Figure 3. Absolute and proportionate gains for Policies 1 and 4 plotted
against percentile of consumption
Policy 1 Policy 4
1500 1500
1000 1000
500 500
atipacr 0 0
pessol -500 -500
in/ag -1000 -1000
teluosba-1500 -1500
-2000 -2000
-2500 -2500
-3000 -3000
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
25 25
10 10
5 5
0 0
tapiacrepss -5 -5
/loinag -20 -20
%
-35 -35
-50 -50
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Per capita consumption percentiles Per capita consumption percentiles
26
Figure 4. Production/consumption decomposition of the welfare impacts for Policy 4,
plotted against percentile of consumption per person
Policy 4, production Policy 4, consumption
1500 1500
1000 1000
500 500
0 0
itapacreps -500 -500
osl/niag -1000 -1000
luteosba-1500 -1500
-2000 -2000
-2500 -2500
-3000 -3000
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Per capita consumption percentiles Per capita consumption percentiles
27
Figure 5. Net producers of cereals in the distribution of total consumption per person
in rural areas
1
.75
sr
ceudorp .5
alerec
%
.25
0
0 10 20 30 40 50 60 70 80 90 100
1
1
slareecfosre .75 slareec noipt .75
ofnoipt um
.5
.5 umsnoc totfo nsocla
.25
producten nusim i on
0
%.25 onit
ucdorP proportasa-.25
0 -.5
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Consumption per capita percentiles Consumption per capita percentiles
28
References
Abdelkhalek, Touhami, 2002, "De I'impact de la Lib�ralization du Marche C�r�alier Marocain:
Enseignements � Partir d'un Mod�le de Comportement des M�nages Rureaux, Critique
Economique 7 : 105-114.
Atkinson, Anthony B., 1987, "On the Measurement of Poverty," Econometrica 55: 749-764.
Auerbach, Alan J., and Kevin A. Hassett, 2002, "A New Measure of Horizontal Equity,"
American Economic Review 92(4): 1116-1125.
Bourguignon, Francois, 1979, "Decomposable Inequality Measures," Econometrica, 47: 901-920.
Chen, Shaohua and Martin Ravallion, 2004, "Welfare Impacts of Morocco's Accession to the
WTO," World Bank Economic Review, in press.
Cleveland, William S., 1979, "Robust Locally Weighted Regression and Smoothing Scatter
Plots," Journal of the American Statistical Association 74: 829-36.
Cowell, Frank, 2000, "Measurement of Inequality," in A.B. Atkinson and F. Bourguignon
(eds) Handbook of Income Distribution, Amsterdam: North-Holland.
Doukkali, Rachid, 2003, "Etude de Effets de la Lib�ralisation des C�r�ales: Resultats des
Simulaions � L'Aide d'un Mod�le Equilibre G�n�ral Calculable," Joint Report of the
Ministry of Agriculture and the World Bank.
Hertel, Thomas W. and Jeffrey Reimer, 2004, "Predicting the Poverty Impacts of Trade
Liberalization: A Survey," Department of Agricultural Economics, Purdue University.
Ravallion, Martin, 2001, "Growth, Inequality and Poverty: Looking Beyond Averages,"
World Development, 29(11), 1803-1815.
______________, 2004, "Competing Concepts of Inequality in the Globalization Debate,"
Policy Research Working Paper, World Bank, Washington DC.
Ravallion, Martin and Dominique van de Walle, 1991, "The Impact of Food Pricing
Reforms on Poverty: A Demand Consistent Welfare Analysis for Indonesia,"
Journal of Policy Modeling, 13: 281-300.
World Bank, 2001, Kingdom of Morocco: Poverty Update. Washington DC, World Bank.
__________, 2003, Kingdom of Morocco: Findings and Recommendations of the
Cereals Working Group: A Critical Review, Washington DC, World Bank.
29
MENA Working Paper Series
No. 1 Has Labor Migration Promoted Economic Integration in the Middle East?
June 1992. Nemat Shafik, The World Bank and Georgetown University.
No. 2 The Welfare Effects of Oil Booms in a Prototypical Small Gulf State.
September 1992. Ahmed Al-Mutuwa, United Arab Emirates University and
John T. Cuddington, Georgetown University.
No. 3 Economic and Social Development in the Middle East and North Africa.
October 1992. Ishac Diwan and Lyn Squire, The World Bank.
No. 4 The Link Between Trade Liberalization and Multi-Factor Productivity:
The Case of Morocco. February 1993. Mona Haddad, The World Bank.
No. 5 Labor Markets in the Middle East and North Africa. February 1993.
Christopher A. Pissarides, The London School of Economics and Political Science.
No. 6 International Competitiveness of the Private Industry and the Constraints
to its Development: The Case of Morocco. June 1993. Hamid Alavi, The World Bank.
No. 7 An Extended RMSM-X Model for Egypt: Quantifications of Market-Oriented
Reforms. September 1993. Karsten Nimb Pedersen, The World Bank.
No. 8 A Report on the Egyptian Tax System. October 1993.
Mark Gersovitz, Roger H. Gordon and Joel Slemrod, The World Bank.
No. 9 Economic Development and Cooperation in the Middle East and North
Africa. November 1993. Ishac Diwan and Lyn Squire, The World Bank.
No. 10 External Finance in the Middle East: Trends and Prospects. December 1993.
Ishac Diwan, John Underwood and Lyn Squire, The World Bank.
No. 11 Tax Incidence on Agriculture in Morocco (1985-1989). April 1994.
Jean-Paul Azam, CERDI, University of Auvergne, Clermont-Ferrand (France)
et CSAE, Oxford (U.K).
No. 12 The Demographic Dimensions of Poverty in Jordan. August 1994.
Chantal Worzala, The World Bank.
No. 13 Fertility and Family Planning in Iran. November 1994. Rodolfo A. Bulatao and Gail
Richardson, The World Bank.
No. 14 Investment Efficiency, Human Capital & Migration A Productivity Analysis
of the Jordanian Economy. May 1995. Gaston Gelos, Yale University,
Department of Economics.
No. 15 Tax Effects on Investment in Morocco. August 1995.
David Sewell, Thomas Tsiopoulos and Jack Mintz, The World Bank.
30
No. 16 Reconstruction in Lebanon: Challenges for Macroeconomic Management.
April 1999. Daniela Gressani and John Page, The World Bank.
No. 17 Towards a Virtuous Circle: A Nutrition Review of the Middle East and North
Africa. August 1999. Regional HNP Knowledge Management, The World Bank.
No. 18 Has Education Had a Growth Payoff in the MENA Region? December 1999.
Lant Pritchett, The World Bank.
No. 19 Rationalizing Public Sector Employment in the MENA Region.
December 2000. Elizabeth Ruppert Bulmer, The World Bank.
No. 20 Achieving Faster Economic Growth in Tunisia. March 2001.
Auguste T. Kouam�, The World Bank.
No. 21 Trade Options for the Palestinian Economy: Some Orders of Magnitude.
March 2001. Claus Astrup and S�bastien Dessus, The World Bank.
No. 22 Human Capital and Growth: The Recovered Role of Educational Systems.
April 2001. S�bastien Dessus, The World Bank.
No. 23 Governance And The Business Environment In West Bank/Gaza.
May 2001. David Sewell, The World Bank.
No. 24 The Impact of Future Labor Policy Options on the Palestinian Labor Market. June
2001. Elizabeth Ruppert Bulmer, The World Bank.
No. 25 Reform and Elusive Growth in the Middle-East � What Has Happened in the
1990s? July 2002. Dipak Dasgupta, Jennifer Keller and T.G. Srinivasan, The World
Bank.
No. 26 Risks and Macro-Economic Impacts of HIV-AIDS in the Middle East and North
Africa: Why waiting to intervene can be costly. July 2002.
David A. Robalino, Carol Jenkins and Karim El Maroufi, The World Bank.
No. 27 Exchange Rate Regime and Competitiveness of Manufactured Exports:
The Case of MENA Countries. August 2002. Mustapha Kamel Nabli and
Marie-Ange V�ganzon�s-Varoudakis, The World Bank.
No. 28 Governance and the Investment Climate in Yemen. September 2002.
Arup Banerji and Caralee McLiesh, The World Bank.
No. 29 Exporting Labor or Goods? Long-term Implications for the Palestinian Economy.
October 2002. Claus Astrup and S�bastien Dessus, The World Bank.
No. 30 Poverty and Transfers in Yemen. December 2002. Dominique van de Walle,
The World Bank.
No. 31 Yemen and the Millennium Development Goals. March 2003. Qaiser Khan and
Susan Chase, The World Bank.
31
No. 32 Making Trade Work for Jobs : International Evidence and Lessons for MENA.
July 2003. Dipak Dasgupta, Mustapha Kamel Nabli, Christopher Pissarides (LSE), and
Aristomene Varoudakis, The World Bank.
No. 33 Opening up Telecommunications to Competition and MENA Integration in the
World Economy. July 2003. Carlo Maria Rossotto, Khalid Sekkat and Aristomene
Varoudakis, The World Bank.
No. 34 Growth, Private Investment and the Cost of Doing Business in Tunisia. February
2004. An�s Casero, Paloma, and Aristomene Varoudakis. The World Bank.
No. 35 Current World Trade Agenda - Issues and Implications for the MENA Region.
May 2004. Dipak Dasgupta,, Mustapha Kamel Nabli, T.G. Srinivasan and Aristomene
Varoudakis, The World Bank.
No. 36 Reforms and Growth in MENA Countries - New Empirical Evidence.
May 2004. Mustapha Kamel Nabli, and Marie-Ange V�ganzon�s-Varoudakis, The
World Bank.
32