Growth, Inequality and Poverty in Madagascar, 2001-2005 (World Bank - 2008)

Growth, Inequality and Poverty in Madagascar, 2001-2005
Africa Region Working Paper Series No. 111

April 2008

Abstract
he paper examines changes in poverty and inequality in Madagascar between the
years 2001 and 2005. During this period Madagascars economic progress has
been notable. Yet the record for poverty and living standards is mixed. Inequality
has declined considerably, the depth of poverty has fallen by almost 25 percent, and
income grew faster for the poor than the average. But poverty remains pervasive in
Madagascar, with more than two thirds of the population below the poverty line. And
though the incidence of poverty has barely changed, the number of the poor has
increased by some two million individuals. Large disparities persist between urban and
rural areas, as well as across provinces. Regression analysis shows that these disparities
persist even after controlling for a wide range of socio-economic and demographic
household characteristics. By matching household-level survey data from the Enqute
Priodique auprs des Mnages to community-level census data we identify three factors
that largely explain the provincial variation in poverty rates: (i) infrastructure, (ii) land
tenure and cropping patterns, and (iii) climate shocks. As for the future, simulations for
benchmark years 2007 and 2010 project incremental reductions in poverty rates on the
order of 0.5-2.0 percent per yearas estimates of earnings functions, provide supporting
evidence of these barriers.
.

JEL Codes: I31, I32, O1, O55.
Key Words: Madagascar; poverty; inequality; growth incidence analysis.

Authors Affiliation and Sponsorship

Nicola Amendola
University of Rome Tor Vergata

Giovanni Vecchi

The Africa Region Working Paper Series expedites dissemination of applied research and policy studies with potential
for improving economic performance and social conditions in Sub-Saharan Africa. The Series publishes papers at
preliminary stages to stimulate timely discussion within the Region and among client countries, donors, and the policy
research community. The editorial board for the Series consists of representatives from professional families appointed
by the Regions Sector Directors. For additional information, please contact Paula White, managing editor of the series,
(81131), Email: pwhite2@worldbank.org or visit the Web site: http://www.worldbank.org/afr/wps/index.htm.

The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s), they
do not necessarily represent the views of the World Bank Group, its Executive Directors, or the countries they
represent and should not be attributed to them.
T
P
u
b
l
i
c

D
i
s
c
l
o
s
u
r
e

A
u
t
h
o
r
i
z
e
d
P
u
b
l
i
c

D
i
s
c
l
o
s
u
r
e

A
u
t
h
o
r
i
z
e
d
P
u
b
l
i
c

D
i
s
c
l
o
s
u
r
e

A
u
t
h
o
r
i
z
e
d
P
u
b
l
i
c

D
i
s
c
l
o
s
u
r
e

A
u
t
h
o
r
i
z
e
d
51700

Growth, Inequality
and Poverty
in Madagascar,
2001-2005
Nicola Amendola

Giovanni Vecchi (*)
April 2008

(*) This paper is the product of a joint AFTH3 and AFTP1 collaboration. Corresponding author:
giovanni.vecchi@uniroma2.it. We would like to thank Benu Bidani, Stefano Paternostro, Ken Simler and David Stifel
whose comments, advice and support have been invaluable. We are grateful to Tiaray Razafimanantena for useful
comments, and to Elena Celada and Laza Razafiarison for help at various stages of the project. The usual disclaimer
applies. Funding from the Japan PHRD preparation grant for Madagascar PRSC V is gratefully acknowledged.
3

Contents

1 Introduction .......................................................................................................................... 5
2 Economic Growth ................................................................................................................. 6
3 Household Surveys in Madagascar ....................................................................................... 8
4 Poverty and Inequality Dynamics ......................................................................................... 9
4.1 The Incidence of Poverty ........................................................................................... 10
4.2 The Depth and Severity of Poverty ............................................................................ 12
4.3 Inequality ................................................................................................................... 14
5 Changes in the Poverty Profile ........................................................................................... 17
6 Growth, Inequality and Poverty .......................................................................................... 20
6.1 Growth Incidence Analysis ........................................................................................ 20
6.2 Growth Elasticities of Poverty ................................................................................... 21
6.3 Growth-Inequality Decomposition ............................................................................ 22
6.4 Sectoral Decomposition of Poverty ........................................................................... 23
7 A Model of Household Consumption ................................................................................. 24
8 Poverty and Growth Projections ......................................................................................... 25
9 Summary and Final Remarks ............................................................................................. 26
List of References........................................................................................................................ 28
Appendix 1 Regression Analysis of Household Consumption ................................................ 30
Appendix 2 Sensitivity Analysis of Poverty Estimates to the Choice of Different Deflators .. 33
Appendix 3 Sectoral Value Added And Population Growth Rates, ......................................... 35

List of Figures

Figure 1 Real GDP (billions of Ariary at 2000 constant prices) and GDP per capita, ............... 6
Figure 2 Sectoral growth rates, 1996-2006 ................................................................................ 7
Figure 3 GDP shares (1984 prices), 1980-2006 ......................................................................... 7
Figure 4 First-order stochastic dominance test, Madagascar 2001-2005 ................................. 11
Figure 5 First-order stochastic dominance by urban/rural area, 2001-2005 ............................. 12
Figure 6 Second-order stochastic dominance test, Madagascar, 2001-05. ............................... 13
Figure 7 Second-order stochastic dominance by urban/rural area ........................................... 13
Figure 8 Lorenz Curves for Madagascar, 2001 and 2005 ........................................................ 15
Figure 9 Changes in the poverty gap index, 2001-2005 ........................................................... 18
Figure 10 The growth incidence curve for Madagascar, 2001-2005 ........................................ 20
Figure 11 Growth incidence curves for urban and rural areas in Madagascar, 2001-05 .......... 21

List of Tables

4
Table 1 EPM 2001 versus EPM 2005: A Comparison ............................................................... 8
Table 2 Poverty and Inequality Trends, Madagascar 2001-2005 (%) ...................................... 10
Table 3 Inequality and mean consumption by province, 2001 and 2005 ................................. 14
Table 4 Inequality Decompositions, Madagascar 2001 and 2005 ............................................ 16
Table 5 Decompositions of the changes in aggregate inequality, 2001 and 2005 .................... 16
Table 6 Poverty estimates by province in 2001 and 2005 ........................................................ 17
Table 7 Comparison of poverty profiles, 2001 and 2005 ......................................................... 19
Table 8 Growth rates (%) among the poor, Madagascar 2001-2005 ....................................... 21
Table 9 Growth elasticities of poverty for Madagascar, 2001 and 2005 .................................. 22
Table 10 Growth-inequality decompositions, Madagascar 2001-2005 .................................... 23
Table 11 Sectoral decomposition of poverty, 2001-2005 ........................................................ 24
Table 12 Impact of growth on poverty in Madagascar............................................................. 26
Table 13 Regression Estimates of Consumption Models for Rural Households, .................... 30
Table 14 Regression Estimates of Consumption Models for Urban Households, ................... 32
Appendix Table 15 Provincial Deflators, Madagascar 2001 and 2005 .................................... 34
Appendix Table 16 Regional vs. provincial deflators: Sensitivity of poverty estimates ...... 34

1 Introduction
Madagascar is one of the worlds poorest countries today, ranked 143 out of 177 according to
UNDP (2006). Yet, the situation has not always been so dire. A number of studies examining
the economic performance of Madagascar in the early 1960s, soon after the country had gained
full independence, show that the Malagasy Republic was once among the richest countries in
Africa see World Bank (2007). Several decades later, living standards have plummeted to
subsistence level for most of Malagasy population, and other social indicators are far below the
Millennium Development Goal targets.
An account of Madagascars more recent economic performance is provided by a study
conducted jointly by the Malagasy National Statistical Institute (INSTAT), Cornell University
and the World Bank see World Bank (2002). The main report contains a comprehensive
assessment of the evolution of poverty and other welfare indicators during the 1990s. Among
the main findings of the report is the fact that national poverty rates remained relatively steady,
while significant swings both in urban and rural areas and across provinces were observed. This
point is further highlighted in the conclusion of the report: further work should be devoted to
understanding what causes geographic variations in poverty; among others, differences in land
quality, infrastructure, and climate should be explored as potential differentiating factors. (p.
35). This paper attempts to perform this task, as part of a broader inquiry into the trends of
living standards, poverty and inequality in Madagascar during the first half of the 2000s.
The paper uses household survey data from the Enqute Priodique auprs des Mnages (EPM)
fielded in 2001 and 2005, with the aims to (i) update the poverty and inequality profiles, (ii) to
identify the changes in the distribution of income and in absolute poverty, and (iii) to project
poverty rates to the present day. Our findings show that while the trend in the headcount ratio is
statistically fragile, and fails to identify the trend in poverty, the use of the poverty gap index
leads to clear-cut results. Let us note incidentally, that the task of identifying the poverty trend is
further complicated by the occurrence of a severe political crisis which started in December
2001 and resulted in a dramatic recession in the year 2002 (GDP per capita dropped by 15
percent). We argue, that the recessions negative impact on the living standards potentially
affects the interpretation of many of the empirical results.
The second aim of this paper is to investigate the determinants of poverty. The availability of
census data from the 2001 Recensement des Communes has allowed us to estimate a household
consumption model in which we control not only for demographic and socio-economic
household characteristics, but also for features of the communities in which households live. In
particular, the paper focuses on the role of infrastructure, structure of the agricultural sector, and
climatic events.
The third and last aim of the paper is to update poverty rates. We use population and sectoral
GDP growth rate projections to forecast poverty in the year 2007. We also venture into longer-
term poverty forecasts for the year 2010.
The paper is organized as follows. Section 2 provides a description of Madagascars recent
performance in terms of economic growth. Section 3 discusses the comparability of EPM 2001
and 2005 data. Section 4 and 5 contain the bulk of the descriptive statistics concerning poverty
and inequality changes; while section 4 focuses on the main trends, section 5 concentrates on
changes in the structure of poverty and inequality. In Section 6 we investigate the mechanics of
the changes in poverty by carrying out standard decomposition techniques, growth incidence
analysis, and estimating poverty elasticities to growth. In section 7 we match survey data with
census data and use regression analysis to explore the determinants of poverty. Section 8
projects poverty rates in the years 2007 and 2010. Section 9 concludes.
6
2 Economic Growth
The trend in real gross per capita domestic product (GDP) is often a useful starting point
element for the analysis of poverty dynamics, inequality and other social indicators. This section
provides an overview of the recent economic development of the Malagasy GDP, based on
national account data.
1

Figure 1 plots the time series of the real gross domestic product (GDP) from 1980 to 2005. In
this period, total GDP increased, on average, by 1.4 percent per year, while the population grew
2.9 percent per year. As a result, per capita GDP decreased, on average, by 1.5 percent per year.
The positive swing in the GDP per capita first occurred in the mid 1990s, but was abruptly
interrupted by the political crisis of 2001-2002. The effects of the crisis on GDP per capita are
clearly visible in Figure 1: during 2002 the economy fell into a deep recession with per capita
GDP shrinking by 15 percent. After the crisis the economy rebounded quickly, with GDP
growth averaging 5 percent per year thereafter. By 2005 the GDP per capita had (almost)
returned to its 2001 level.
Figure 1 Real GDP (billions of Ariary at 2000 constant prices) and GDP per capita,
1980-2006
300000
350000
400000
450000
500000
3500
4000
4500
5000
5500
6000
1980 1985 1990 1995 2000 2006
year
GDP (left y-scale)
GDP per head (right y-scale)

Source: World Bank database.

Figure 2 shows the pattern of sectoral GDP growth rates between 1996 and 2006. During this
decade economic growth was driven by the secondary and the tertiary sectors. The graph shows
that the crisis heavily hit those sectors, while the impact on the agricultural sector was only
modest (minus 1.3 percent). The recovery after the crisis was fast and sustained, largely based
on the performance of the secondary and tertiary sectors. It is worth noting however, that the
growth rates in the secondary sector have declined in the recent years.

1
For more see IMF (2007) and OECD (2007).
7
Figure 2 Sectoral growth rates, 1996-2006
-20
-10
0
10
20
A
n
n
u
a
l

g
r
o
w
t
h

r
a
t
e

(
%
)
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Primary
Secondary
Tertiary

Source: World Bank data.

Figure 3 shows the trends of sectoral GDP shares between the years 1980-2006. While the
evolution of the share of the secondary sector is stable over the whole period, the primary and
tertiary sectors move along different trajectories. Between the mid 1980s and early 1990s, the
share of the primary sector fluctuates mildly around 35 percent. Then, starting in 1996, the
primary share decreases while the share of services begins to increase. This pattern, interrupted
only and temporary reversed by the effects of the 2001 political crisis, resumes after 2002.
Overall, Figure 3 depicts a process of a slowly changing economic structure. Only in recent
years, signs that are typically associated with modern economic growth, such as a significant
decline of the share of agriculture and a rise in the share of secondary and tertiary sectors, have
become visible.
Figure 3 GDP shares (1984 prices), 1980-2006
10
20
30
40
50
60
S
h
a
r
e

o
f

G
D
P

(
%
)
1985 1990 1995 2000 2005 2007
year
share primary sect share secondary sect
share tertiary sect

Source: World Bank database.

8
From the evidence examined in this section, two main conclusions emerge. First, the impact of
economic growth on living standards is likely to be greatly reduced by the high rates of
population growth. While the total GDP growth rate between 2003-2005 has been relatively
high (6.5 percent), it translated only into a 3.5 percent increase in per capita terms. Second, the
sectoral composition of growth may prove to be ineffective in combating poverty. According to
Stifel (2007), 80.1 percent of the population live in households headed by agricultural workers;
yet, productivity growth in agriculture during 2003-2005 was low and slowly improving. The
highest gains accrued to workers employed in services, accounting for 17.4 percent of the labour
force.
3 Household Surveys in Madagascar
The data used for this paper are drawn from the Enqute Priodique auprs des Mnages
(EPM), a nationally representative household-level survey carried out by the Direction des
Statistiques des Menages (DSM) of the national statistical institute (INSTAT). The EPM started
off in 1993 with the aim to assess the living standards of the population. Since its debut, EPM
was repeated in 1997, 1999, 2001, 2002, 2004 and 2005. Given our focus on poverty
comparisons over time, it is important to discuss the comparability of these surveys.
For the early surveys of 1993 to 1999, the World Bank (2002) provides an exhaustive analysis
of the methodological problems arising from different survey designs and different choices
underlying the welfare indicator. Unfortunately, we lack a similar (systematic) assessment for
the surveys subsequent to the World Bank report. Instead, we have to rely on piecemeal
information available in a variety of documents, as well as on a number of personal
communications with INSTAT staff. This section focuses on the EPM for the years 2001 and
2005.
2

Table 1 compares the 2001 and 2005 EPM survey designs, showing that no major differences
exist between the two surveys.
Table 1 EPM 2001 versus EPM 2005: A Comparison
2001 2005
Sample design
Two-stage stratified: 12 strata, (1
st
) 303
ZD (clusters), (2
nd
) 5,080 household (16
per urban cluster, and 18 per rural
cluster).
Two-stage stratified: 44 strata, (1
st
)
561 ZD (clusters), (2
nd
) 11,781
household (21 per cluster).
Rounds One: Oct to Nov 2001. One: Sept 05 to Nov 10, 2005
Actual sample size
23,170 individuals corresponding to
5,080 households (3,040 urban, 2,040
rural)
54,966 individuals corresponding to
11,781 households (5,859 urban,
5,922 rural)
Representativeness
provincial level (6 faritany), and
urban/rural within each province
regional level (22 faritra), and
urban/rural within each region
Time reference, recall
period
Expenditures on food and beverages:
last week, year. Non-food commodities:
last month, year.
Idem
Consumption
Expenditures on both food and non-food
include in-kind consumption items.
Idem

Of special concern to us is the consistency of the welfare indicators definition. This is crucial to
guaranteeing the consistency in the sense of Ravallion and Bidani (1994) of poverty profiles
in different periods. The 2001 and 2005 consumption aggregates were constructed using almost

2
Data from the 2002 survey are reported to be lacking in reliability by most analysts we consulted with,
but the matter is not discussed in any publication we are aware of.
9
identical methodologies. This is the result of a strategy pursued in 2005, when inter-temporal
poverty comparisons were wisely included in the agenda.
3

Three more issues are worth mentioning. First, there is a discrepancy between the EPM-based
estimates of the total population estimates and similar estimates by the IMF. Even though the
discrepancy does not affect the estimates of the class of poverty measures used in the paper, it
matters when comparing the absolute numbers of the poor. Second, while the 2005 EPM
allowed the use of regional deflators, the 2001 EPM was based on provincial deflators. The
(potential) nuisance that arises from the use of different deflators plays a negligible role in the
analysis pursued in the rest of the paper. Appendix 2 illustrates the robustness of the poverty
profile to the choice of different deflators. Third, the procedure used to update the poverty line
between 2001 and 2005 fails to account for substitution effects that may occur in response to
changes in relative prices of basic goods (particularly, in response to the dramatic changes
associated with the dramatic 2002 crisis). Further research is needed to address this issue.
4

The evidence presented in this section leads to the conclusion that there are neither substantial
differences in the designs underlying the two EPM surveys, nor are there inconsistencies in the
construction of the consumption aggregate.
4 Poverty and Inequality Dynamics
In this section we describe the trends in poverty and inequality measures between the years
2001 and 2005. Some caution may be appropriate when interpreting the results in light of the
political crisis of 2002. The 2001 poverty profile describes the living standards immediately
before the crisis; during 2002, the year of the crisis, poverty increased substantially, from 69.7
percent to 75 percent according to INSTAT (2006). It follows that the poverty profile in 2005,
far from being the result of a relatively flat pattern of growth, is the outcome of a buoyant
recovery process towards the pre-crisis level. Poverty comparisons based on 2001 (as opposed
to 2002) are therefore likely to under-estimate the role of economic growth in affecting the
dynamics of poverty. While all this is important in interpreting the estimates in Table 2, we will
revisit this issue in section 6.3.
The main findings summarized in Table 2 are discussed in the remainder of this section. We
first highlight the main trends in a selection of poverty and inequality indicators. Next, we
identify a number of facts that constitute the explicandum for the rest of the paper.

3
In contrast, the comparability of household consumption aggregates for the 1993, 1997 and 1999 EPM
household surveys is not straightforward. See World Bank (2002), Appendix A1.
4
See Ravallion and Lokshin (2006) and Arndt and Simler (2005).
10
Table 2 Poverty and Inequality Trends, Madagascar 2001-2005 (%)
2001 2005
Urban Rural National Urban Rural National
Headcount 44.2 77.3 69.7 52.0 73.5 68.7
95% c. i. [39.9, 48.6] [72.6, 82.0] [65.9, 73.5] [48.0, 55.9] [71.2, 75.7] [66.7, 70.8]
Poverty Gap 18.3 39.8 34.9 19.3 28.9 26.8
95% c. i. [15.7, 21.0] [36.1, 43.5] [32.0, 38.0] [17.4, 21.2] [27.1, 30.6] [25.3, 28.2]
Poverty Gap Squared 9.9 24.2 20.9 9.4 14.5 13.4
95% c. i. [8.1, 11.7] [21.3, 27.1] [18.6, 23.2] [8.3, 10.6] [13.3, 15.7] [12.4, 14.4]

Gini Index 0.439 0.445 0.469 0.405 0.335 0.365
95% c. i. [.419,.459] [.402, .488] [.445, .492] [.382, .429] [.313, .357] [.347, .383]
Theil Index 0.341 0.369 0.402 0.314 0.234 0.273
95% c. i. [.305,.376] [.297,.441] [.362, .441] [.266, .362] [.179, .289] [.232, .314]
Background Statistics
Total population 3,588 12,079 15,667 4,146 14,701 18,847
Population share 22.9 77.1 100 22.0 78.0 100
Mean PCE 475,287 251,770 302,883 401,471 269,663 298,644
Mean PCE among poor 178,737 148,004 152,464 191,731 185,309 186,376
Notes: confidence intervals are based on linearized standard errors. The poverty line used for 2001 is 197,720
Ariary/head/year (988,600 FMG/head/year); for 2005 the poverty line is 305,300 Ariary/head/year. The latter value is
obtained by updating the poverty line for 2001 (duly converted into Ariary) using the 2005 Consumer Price Index.
Source: Authors calculation based on EPM data.
4.1 The Incidence of Poverty
At the national level, the incidence of poverty has not changed. While point estimates in Table 2
show that the headcount ratio decreased from 69.7 percent in 2001 to 68.7 percent in 2005, the
change is not statistically significant. In contrast, the number of poor people has increased by
some 2 million units during the same period. As recently noted by Chakravarti, Kanbur and
Mukherjee (2006), while the economists instinct is probably to conclude that poverty in
Madagascar has decreased this goes against the instinct of those who work directly with the
poor, for whom the absolute numbers notion makes more sense as they cope with more poor on
the streets or in soup kitchens (p. 471). In short, the question of whether poverty in Madagascar
has increased or decreased is a non trivial one.
The result of a first-order stochastic dominance (FOD) analysis adds to the difficulty of
identifying the trend in poverty incidence. We carried out the FOD test, where the null
hypothesis is that the incidence of poverty in 2005 is lower than in 2001, regardless of the
poverty line chosen. Figure 4 shows that this hypothesis is unambiguously rejected by the data.
Not only do the cumulative density functions intersect, but they do so almost exactly at the
poverty line, which suggests that the ranking of poverty would change if the poverty line
changed slightly.
11
Figure 4 First-order stochastic dominance test, Madagascar 2001-2005
0
.
2
.
4
.
6
.
8
1
0 500000 1000000 1500000
per capita expenditure (2005 ariary)
2001
2005

Source: Authors estimates on EPM data.
Poverty changes were not uniformly distributed across the national territory. Table 1 shows an
asymmetric pattern of the incidence of poverty between rural and urban areas. This is a major
issue in a country where poverty in rural areas accounts for 83 percent of total poverty. During
2001-2005 period, the headcount ratio increased from 44.3 percent to 52.0 percent in urban area,
while in rural areas it decreased from 77.3 percent to 73.5.
5
Once again, the analysis of the
numbers of individuals classified as poor leads to opposite conclusions: approximately half a
million people were added to the stock of the poor in urban areas, while the number of poor
people in rural areas increased by 1.5 million. The population growth in rural areas during 2001-
2005 was about 22 percent, compared to 15.5 percent in urban areas: whether this pattern is due
to differences in fertility rates or, instead, to urban-to-rural migratory flows requires further
investigation.
Figure 5 shows the results of a FOD test carried out separately for rural and urban areas. The
cumulative density functions cross in both areas, suggesting that the ordering of poverty as
measured by the headcount ratio is not robust to the choice of the poverty line.

5
Note that while the change in the headcount ratio in urban areas is statistically significant, this does not
hold true for the change in rural areas.
12
Figure 5 First-order stochastic dominance by urban/rural area, 2001-2005
0
.2
.4
.6
.8
1
e
m
p
i
r
i
c
a
l

c
d
f
URBAN
0
.2
.4
.6
.8
1
e
m
p
i
r
i
c
a
l

c
d
f
0 500000 1000000 1500000
2001
2005
RURAL
Source: Authors estimates on EPM data.

When poverty is so pervasive throughout the country, it becomes desirable to consider measures
other than the headcount poverty. The poverty gap index (PG) and poverty gap squared index
(PG2) are obvious choices, and their utilization is investigated in the following section.
4.2 The Depth and Severity of Poverty
According to the estimates in Table 2, in the period between 2001-2005 the poverty gap index
has decreased by one fourth, from 34.9 percent to 26.8 percent, suggesting that the living
standards of the poor have improved significantly during this period.
Figure 6 shows the result of a second-order stochastic dominance (SOD) test. Here, the
hypothesis tested is that poverty in 2005, as measured by the PG index, is lower than in 2001.
We find that the poverty deficit curve for 2005 is always below the poverty deficit curve for
2001, that is the average distance from the poverty line in 2005 is lower than in 2001, regardless
the choice of the poverty line. The reduction in the depth of poverty is substantial and not
affected by the choice of the poverty line.
13
Figure 6 Second-order stochastic dominance test, Madagascar, 2001-05.
0
200000
400000
600000
800000
1000000
A
r
e
a

u
n
d
e
r

t
h
e

p
o
v
e
r
t
y

i
n
c
i
d
e
n
c
e

c
u
r
v
e
0 500000 1000000 1500000
2001
2005

Source: Authors estimation on EPM 2001 and 2005 data.
The pro-rural bias in the reduction of poverty incidence also characterizes the decline in the
depth of poverty. The poverty gap decreased in rural areas (-27.4 percent), while it increased
though only slightly and not significantly in urban areas (+5.5 percent). Figure 7 shows that
this result is robust to the choice of poverty line.
Figure 7 Second-order stochastic dominance by urban/rural area
0
200000
400000
600000
800000
1000000
A
r
e
a

u
n
d
e
r

t
h
e

p
o
v
e
r
t
y

in
c
id
e
n
c
e

c
u
r
v
e
2001
2005
URBAN
0
200000
400000
600000
800000
1000000
A
r
e
a

u
n
d
e
r

t
h
e

p
o
v
e
r
t
y

in
c
id
e
n
c
e

c
u
r
v
e
0 500000 1000000 1500000
RURAL
Source: Authors estimation on EPM 2001 and 2005 data.
The poverty gap squared (PG2) decreased even more dramatically than the PG, from 20.9
percent in 2001 to 13.4 in 2005 (minus 35.9 percent). This indicates that the poorest among the
14
poor benefited even more than the average poor during the period 2001 and 2005, and this
improvement was, once again, more pronounced in rural areas.
4.3 Inequality
The evidence in Table 2 suggests that between 2001 and 2005 a substantial income
redistribution occurred in Madagascar. In this subsection, we investigate the trends in inequality
by focusing on its geographical dynamics.
Table 3 shows the estimates of inequality at the provincial level, separately by urban and rural
areas. This represents the maximum level of disaggregation allowed by the 2001 survey. Three
popular inequality indices are considered, the Gini index, the Theil index and the Mean
Logarithmic Deviation.
According to Table 3 inequality decreased nationally. This finding is robust to the choice of the
inequality index. Similarly, the provincial pattern of inequality indicates that inequality
decreased throughout the country (with the exceptions of Toamasina, Antsiranana and
Mahajanga, when using the Theil index). The provincial trends show, however, a significant
dispersion in inequality reduction rates, with Antananarivo, Fianarantsoa and Toliara standing
out as the provinces in which inequality decreased at the fastest pace.
Table 3 Inequality and mean consumption by province, 2001 and 2005
Province 2001 2005 % Change
GINI
urban rural total urban rural total urban rural total
Antananarivo 0.422 0.470 0.466 0.388 0.309 0.353 -8.1 -34.3 -24.2
Fianarantsoa 0.411 0.369 0.406 0.385 0.278 0.300 -6.3 -24.7 -26.1
Toamasina 0.424 0.356 0.402 0.392 0.390 0.398 -7.5 9.6 -1.0
Mahajanga 0.395 0.365 0.402 0.390 0.366 0.385 -1.3 0.3 -4.2
Toliara 0.450 0.386 0.433 0.431 0.348 0.373 -4.2 -9.8 -13.9
Antsiranana 0.343 0.342 0.389 0.379 0.361 0.384 10.5 5.6 -1.3
Madagascar 0.439 0.445 0.469 0.405 0.335 0.365 -7.7 -24.7 -22.2
THEIL
Antananarivo 0.311 0.382 0.374 0.290 0.199 0.255 -6.8 -47.9 -31.8
Fianarantsoa 0.301 0.244 0.304 0.318 0.140 0.178 5.6 -42.6 -41.4
Toamasina 0.315 0.216 0.290 0.268 0.317 0.314 -14.9 46.8 8.3
Mahajanga 0.277 0.227 0.290 0.296 0.356 0.353 6.9 56.8 21.7
Toliara 0.366 0.267 0.348 0.351 0.217 0.260 -4.1 -18.7 -25.3
Antsiranana 0.193 0.197 0.260 0.290 0.263 0.298 50.3 33.5 14.6
Madagascar 0.341 0.369 0.402 0.314 0.234 0.273 -7.9 -36.6 -32.1
MEAN LOGARITHMIC DEVIATION
Antananarivo 0.315 0.372 0.375 0.249 0.160 0.206 -21.0 -57.0 -45.1
Fianarantsoa 0.283 0.226 0.274 0.248 0.128 0.151 -12.4 -43.4 -44.9
Toamasina 0.311 0.209 0.268 0.265 0.255 0.265 -14.8 22.0 -1.1
Mahajanga 0.255 0.224 0.272 0.250 0.236 0.253 -2.0 5.4 -7.0
Toliara 0.359 0.250 0.316 0.316 0.205 0.235 -12.0 -18.0 -25.6
Antsiranana 0.199 0.194 0.252 0.237 0.215 0.243 19.1 10.8 -3.6
Madagascar 0.338 0.330 0.371 0.276 0.190 0.224 -18.3 -42.4 -39.6
MEAN CONSUMPTION
Antananarivo 626882 412338 475759 463102 304758 352719 -26.1 -26.1 -25.9
Fianarantsoa 345506 172695 201548 296524 238171 247726 -14.2 37.9 22.9
Toamasina 332048 175721 208096 367884 268275 286627 10.8 52.7 37.7
Mahajanga 422524 216732 260147 421211 285374 314709 -0.3 31.7 21.0
Toliara 384209 204392 247397 329001 241362 259095 -14.4 18.1 4.7
Antsiranana 530288 240792 298486 517916 288922 324309 -2.3 20.0 8.7
Madagascar 475287 251770 302883 401471 269663 298644 -15.5 7.1 -1.4
Note: bootstrapped standard errors available from the Authors upon request.

15
Figure 8 illustrates the Lorenz curves for 2001 and 2005 at the national level (top-left graph)
and by urban and rural areas (bottom graphs). In all graphs, the 2005 curve unambiguously
dominates the 2001 curve, which makes the general reduction in inequality measures a well
grounded finding. Inequality reduction is more pronounced in rural areas than in urban areas.
Figure 8 Lorenz Curves for Madagascar, 2001 and 2005
0
.2
.4
.6
.8
1

0
.2
.4
.6
.8
1
P
e
r

c
a
p
i
t
a

e
x
p
e
n
d
i
t
u
r
e

s
h
a
r
e

o
f

p
o
o
r
e
s
t

p
1
0
0
%
0 .2 .4 .6 .8 1
Cumulative population share, p
Legend:

0
.2
.4
.6
.8
1

0
.2
.4
.6
.8
1
P
e
r

c
a
p
i
t
a

e
x
p
e
n
d
i
t
u
r
e

s
h
a
r
e

o
f

p
o
o
r
e
s
t

p
1
0
0
%
0 .2 .4 .6 .8 1
URBAN
0
.2
.4
.6
.8
1

0
.2
.4
.6
.8
1
P
e
r

c
a
p
i
t
a

e
x
p
e
n
d
i
t
u
r
e

s
h
a
r
e

o
f

p
o
o
r
e
s
t

p
1
0
0
%
0 .2 .4 .6 .8 1
RURAL

Source: Authors calculation on EPM 2001 and 2005 data.

In order to investigate the structure of inequality and its dynamics, we have decomposed
inequality levels and changes using the methods described in Shorrocks (1980) and Mookherjee
and Shorrocks (1982).
Table 4 shows the results of the decomposition of the levels of inequality by urban and rural
groups (top panel), and by province (bottom panel). The point of this decomposition is to
separate total inequality (I
TOT
) in the distribution into two components, often referred to as the
within- and the between-components. The within component can be described as the level of
inequality (I
W
) that would be observed if there were no differences in mean levels of
expenditures across population subgroups. Likewise, the between component (I
B
) is the level of
inequality that would be observed in the absence of differences in expenditures within groups.
Shorrocks (1980) derived a class of inequality indices (the so-called Generalized Entropy
Indices) that are additively decomposable, i.e. such that I
W +
I
B
= I
TOT
.
With regard to the decomposition by urban and rural area, Table 4 shows that in 2001 the within
component explains 89 percent of total inequality, which increases to 93 percent for the year
2005. Similarly, the decomposition of inequality by province, shows that the largest
contribution to total inequality is due to the within component (83 percent in 2001, 95 percent in
2005). This pattern suggests that inequality reduction policies in Madagascar should focus on
reducing inequality within population sub-groups (provinces and urban/rural areas), rather than
on narrowing the gap in mean expenditures between the groups.
16
Table 4 Inequality Decompositions, Madagascar 2001 and 2005
2001 2005
Index Within Between Index Within Between
by urban/rural area
MLD
(%)
0.371
(100)
0.332
(89)
0.040
(11)
0.224
(100)
0.209
(93)
0.015
(7)
Theil
(%)
0.402
(100)
0.359
(89)
0.043
(11)
0.273
(100)
0.258
(95)
0.015
(5)
by province
MLD
(%)
0.371
(100)
0.306
(83)
0.065
(17)
0.224
(100)
0.213
(95)
0.010
(5)
Theil
(%)
0.402
(100)
0.335
(83)
0.067
(17)
0.273
(100)
0.263
(96)
0.010
(4)
Source: Authors calculation on EPM data.

With regard to inequality changes, Table 5 shows the results of a popular decomposition
technique, first introduced by Mookherjee and Shorrocks (1982). As with the static
decomposition, we start by partitioning the population into subgroups, say provinces. Next, we
apply the decomposition to the mean logarithmic deviation, thereby following common practice.
The formula used, here omitted, decompose the total change in inequality into three
components: (A) pure inequality effect arising from changes in inequality within groups, (B)
population-share effect (or allocational effect) arising from changes in the number of people
within different groups, and (C) income effect arising from changes in relative expenditures
between groups. Following Jenkins (1995), we decomposed the percentage change of the mean
logarithmic variation. Table 5 shows the results.
For both sub-group partitions, the changes in within-group-inequality (columns A) accounts for
most of the changes in aggregate inequality. The population-share effect (column B) is
negligible. Changes in mean expenditures (column C) across provinces and between urban and
rural areas are significantly equalizing. Overall, Table 6 suggests that what dominated inequality
changes between 2001 and 2005 was the contribution from changes in inequality within
provinces and within urban/rural areas.
Table 5 Decompositions of the changes in aggregate inequality, 2001 and 2005
sub-group
partition
change in
MLD
(%)
% change in MLD accounted for by changes in
within-group
inequalities

(A)
population
shares

(B)
sub-group mean
incomes

(C)
urban/rural -40 -33 0 -7

province -40 -25 0 -15

The main results from the analysis carried out in this section can be summarized as follows.
First, inequality at the national level unambiguously decreased in the period between 2001 and
2005. Second, even if inequality decreased in both rural and urban areas, rural areas experienced
the largest reduction in inequality. Third, a provincial breakdown of inequality reveals large and
persistent differences across the national territory, with the provinces of Antananarivo,
Fianarantsoa and Toliara faring best. Fourth, the decomposition analysis shows that the
observed decline in inequality is largely driven by the decline in inequality within provinces
(urban/rural areas) rather than by the convergence of average consumption incomes between
provinces (urban/rural areas).
17
5 Changes in the Poverty Profile
In this section we provide the poverty profiles for 2001 and 2005 based on the main geographic,
demographic and socio-economic characteristics of the households. The purpose is to examine
the changes in the poverty profiles in order to provide a more in-depth analysis of the structure
of poverty changes during the observed period.
Table 6 provides the geographic profile based on the three Foster-Greer-Thorbecke indices of
poverty. Focussing on the headcount ratio, estimated poverty changes between 2001 and 2005
are found to vary widely across provinces. The incidence of poverty increased in Antananarivo
(+18.7 percent), it decreased in Toamasina, Antsiranana and Fianarantsoa (12.6 percent, 7.4
and 6.7 percent, respectively), while remained relatively stable in the other provinces.
The provincial variation in poverty changes is less pronounced when we consider the depth of
poverty. According to Table 6, the poverty gap index decreased in all provinces during the
observed period, but at different rates. The relative performance in terms of poverty reduction
can be appreciated by looking at the map in Figure 9. The map uses green colours to identify the
best performing provinces (the darker the better) and turns into reds (denoting slowest
provinces) passing through the orange. The map shows that the fastest decline in poverty took
place in the western provinces (Fianarantsoa and Toamasina), and in Mahajanga. Antananarivo
is the laggard province.
Table 6 Poverty estimates by province in 2001 and 2005
2001 2005 % change in index
Headcount ratio
Antananarivo 28.3 57.1 48.6 41.6 64.7 57.7 47.0 13.3 18.7
Fianarantsoa 59.8 87.9 83.2 71.6 78.7 77.6 19.7 -10.5 -6.7
Toamasina 60.8 87.9 82.3 55.8 75.6 71.9 -8.2 -14.0 -12.6
Mahajanga 50.7 78.4 72.5 47.0 76.6 70.2 -7.3 -2.3 -3.2
Toliara 53.3 83.3 76.1 64.3 77.4 74.8 20.6 -7.1 -1.7
Antsiranana 30.1 79.0 69.3 33.8 69.8 64.2 12.3 -11.6 -7.4
Madagascar 44.2 77.3 69.7 52.0 73.5 68.7 17.6 -4.9 -1.4
Poverty Gap
Antananarivo 10.4 25.6 21.1 13.6 21.9 19.4 30.8 -14.5 -8.1
Fianarantsoa 25.5 49.5 45.5 28.8 30.9 30.6 12.9 -37.6 -32.7
Toamasina 28.5 48.3 44.2 21.4 33.1 30.9 -24.9 -31.5 -30.1
Mahajanga 17.4 40.0 35.2 16.1 28.9 26.2 -7.5 -27.8 -25.6
Toliara 25.5 43.6 39.2 28.3 34.0 32.9 11.0 -22.0 -16.1
Antsiranana 8.7 34.1 29.0 9.4 28.1 25.2 8.0 -17.6 -13.1
Madagascar 18.3 39.8 34.9 19.3 28.9 26.8 5.5 -27.4 -23.2
Poverty Gap Squared
Antananarivo 5.3 13.9 11.3 5.8 9.8 8.6 9.4 -29.5 -23.9
Fianarantsoa 13.6 31.7 28.7 14.5 15.1 15.0 6.6 -52.4 -47.7
Toamasina 16.3 30.5 27.6 11.2 18.0 16.7 -31.3 -41.0 -39.5
Mahajanga 8.0 23.9 20.6 7.2 13.9 12.4 -10.0 -41.8 -39.8
Toliara 15.1 26.9 24.1 15.9 19.1 18.4 5.3 -29.0 -23.7
Antsiranana 3.8 18.9 15.8 3.5 14.2 12.5 -7.9 -24.9 -20.9
Madagascar 9.9 24.2 20.9 9.4 14.5 13.4 -5.1 -40.1 -35.9
Note: poverty lines here.

Table 7 describes the poverty profiles for 2001 and 2005 according to a number of demographic
and socio-economic characteristics of the head of household and the household. The trends in
poverty levels mirror the trends identified above in the paper: headcount ratios vary little, while
the poverty gap and the poverty gap squared indices decline substantially. The striking feature
of Table 7, however, is the substantial immobility that emerges from the comparison of the
18
structure of poverty between 2001 and 2005 (column poverty share). For most of the poverty
covariates considered in the table, poverty shares changed very little.
6
There are three notable
exceptions to the general lack of action in Table 7, which we will comment on briefly.

Figure 9 Changes in the poverty gap index, 2001-2005

Source: Authors calculations based on EPM data.

First, the poverty profiles based on the employment status of the head of household change
significantly between 2001 and 2005. The poverty incidence among households headed by self-
employed workers is substantially higher than among wage-earners (and even higher than when
headed by an unemployed person). This is consistent with the higher concentration of self-
employed workers in the agricultural sector (and in rural areas) where productivity is low and
poverty rates are high. Note also, that according to Table 7 the living standards of households
headed by wage-earners decreased markedly between 2001 and 2005, while increasing for
households headed by self-employed workers. Second, the structure of the poverty risks by
economic sector changes in favour of households headed by individuals with employment in
agriculture. As before, this is consistent with the improvements in poverty measures in rural
areas. Third, Table 7 provides support to an argument made by Minten and Stifel (2004):
remoteness and poverty are positively correlated. By comparing 2001 and 2005 we note,
however, a flattening in the structure of poverty risks. This is, perhaps, a sign that, at least to
some extent, infrastructures in rural Madagascar have improved, but other factors may also be
playing a role. It is impossible to distinguish on the basis of simple correlation analysis, but we
will revisit this issue in section 7.
Finally, in contrast with previous findings see World Bank (2002) Table 7 provides no
evidence in support of the argument that gender affects the risk of poverty in Madagascar;
poverty rates for households headed by females are not significantly higher than the average.

6
In this situation, it was deemed needless to carry out formal tests for statistical significance of the
observed differences.
19
Table 7 Comparison of poverty profiles, 2001 and 2005
2001 2005

Pop.
share
Poverty
share
H PG PG2
Pop.
Share
Poverty
share
H PG PG2
Gender Female 15.1 15.0 69.2 35.5 21.9 14.4 14.1 67.6 26.1 13.2
Male 84.9 85.0 69.8 34.8 20.8 85.6 85.9 69.0 26.9 13.4
Age 0-24 5.5 5.1 65.4 30.5 17.0 4.3 3.8 62.0 23.0 11.2
25-44 51.6 52.0 70.3 35.2 21.1 52.0 53.0 70.0 26.9 13.4
45-64 36.0 36.3 70.2 35.7 21.6 36.4 36.3 68.5 27.5 14.0
65+ 6.9 6.6 66.2 32.6 18.8 7.3 6.9 65.1 24.4 11.7
Status Married 83.1 83.5 70.1 34.9 20.8 83.0 83.9 69.5 27.1 13.5
Divorced 7.3 7.7 73.0 36.8 22.5 7.1 6.9 66.7 24.9 12.3
Widower 6.9 6.8 69.0 37.1 23.5 8.3 8.1 66.8 26.4 13.2
Single 2.6 2.0 53.0 22.9 12.5 1.7 1.2 48.5 19.8 10.6
Education None 47.0 57.1 84.6 46.2 29.1 52.3 58.9 77.4 31.9 16.6
Primary 29.5 30.3 71.5 33.7 19.2 30.0 30.3 69.3 25.7 12.4
Lower secondary 10.1 7.1 49.1 19.1 10.0 9.3 7.0 51.5 16.6 7.3
Upper secondary 7.5 4.4 41.1 14.4 6.5 4.9 3.0 41.5 13.4 5.8
Post secondary 5.8 1.1 13.4 4.1 1.7 3.4 0.9 17.6 4.0 1.4
Employment status Not employed 4.2 2.9 48.1 20.3 10.8 3.5 2.8 54.9 22.8 12.0
Wage earner 21.5 11.9 38.4 14.7 7.5 17.8 13.6 52.7 18.9 8.8
Self-employed 74.2 85.2 80.0 41.6 25.4 78.8 83.6 73.0 28.7 14.5
Economic sector Agriculture 70.8 85.1 85.0 44.7 27.4 77.9 84.8 75.4 30.1 15.2
Industry 5.9 3.1 37.8 14.1 7.1 2.8 1.8 44.8 14.7 6.5
Services 23.4 11.8 35.6 13.3 6.8 19.3 13.4 47.9 16.1 7.2
Size 1-2 6.1 3.2 37.2 12.5 5.7 5.1 2.0 27.5 7.5 3.4
3-4 26.5 22.5 59.2 26.6 14.6 25.0 19.5 53.6 17.6 7.9
5-6 31.5 31.1 68.8 32.9 19.3 33.0 32.7 68.3 25.1 12.0
7+ 35.9 43.1 83.8 46.6 29.6 36.9 45.7 85.2 37.2 19.8
Remoteness Most remote 7.2 8.4 81.0 41.0 23.7 14.6 16.2 76.3 30.8 15.5
4 15.0 18.9 88.0 48.6 31.1 15.2 16.9 76.4 32.8 17.9
3 17.4 22.6 90.3 50.0 31.6 14.6 16.9 79.7 33.0 17.2
2 10.9 12.5 79.8 39.8 24.1 14.5 14.8 70.5 24.8 11.4
Least remote 26.5 23.1 60.7 27.8 15.6 19.2 18.6 66.6 24.3 11.4
Second. city 11.9 9.8 57.1 25.3 14.2 12.1 11.2 63.3 24.8 12.6
Large city 10.9 4.7 30.2 10.7 5.2 9.9 5.4 37.9 12.6 5.5
Note: H stands for headcount ratio, PG is the poverty gap index, PG2 is the poverty gap squared index.
20
6 Growth, Inequality and Poverty
After examining the changes in growth (section 2), and the trends in poverty and inequality
(section 4), this section examines their interplay. In section 6.1 we examine growth incidence
curves to assess the extent to which, during 2001-2005, economic growth was pro-poor. In
section 6.2 we estimate the elasticity of poverty measures to economic growth. In sections 6.3
and 6.4 we implement both growth-inequality and sectoral decomposition techniques to identify
the mechanics of the observed poverty decline. Why did poverty decline? Was it caused by the
buoyant economic growth which followed the 2002 crisis? Or was it a consequence of the
income redistribution documented in section 4? Or was it a combination of both growth and
distribution shifts?
6.1 Growth Incidence Analysis
Growth incidence curves (GIC), proposed by Ravallion and Chen (2003), plot per capita
expenditure growth rates against quintiles ranked by per capita expenditure. The GIC provides
an intuitive measure of how much the observed growth has favored the poor relative to the non
poor.
Figure 10 shows the GIC for Madagascar, based on household per capita expenditures from the
2001 and 2005 EPM. Note that during this period the national average growth rate of PCE has
been negative (-1.4 percent). According to the GIC estimated in Figure 10, the poorest 70
percent of the population experienced larger than the average growth. This indicates that process
of economic growth has been unambiguously and strongly pro-poor.
Figure 10 The growth incidence curve for Madagascar, 2001-2005
Average growth rate: -1.4%
-20
0
20
40
60
G
r
o
w
t
h

r
a
t
e

o
f

p
e
r

c
a
p
i
t
a

e
x
p
e
n
d
i
t
u
r
e

2
0
0
1

t
o

2
0
0
5
(
%
)
0 10 20 30 40 50 60 70 80 90 100
Percentiles

Figure 11 shows GICs estimated separately for the urban and rural areas. The shape of the
curves indicates that growth during the 2001-05 period was clearly pro-poor in both rural and
urban areas. However, in urban areas the average growth rate of expenditures was strongly
negative. Table 8 shows the growth rates for selected expenditure percentiles; the comparison
between the rural and urban patterns in growth rates is almost self-explanatory. While in rural
areas the poor experience large and positive growth rates, in urban areas the poor are less
penalized by the decline in average income.
21
Figure 11 Growth incidence curves for urban and rural areas in Madagascar, 2001-05
Average growth rate: -15.5%
-20
0
20
40
60
G
r
o
w
t
h

r
a
t
e

o
f

p
e
r

c
a
p
it
a

e
x
p
e
n
d
it
u
r
e

2
0
0
1

t
o

2
0
0
5
(
%
)
0 10 20 30 40 50 60 70 80 90 100
Percentiles
URBAN
Average growth rate: +7.1%
-20
0
20
40
60
G
r
o
w
t
h

r
a
t
e

o
f

p
e
r

c
a
p
it
a

e
x
p
e
n
d
i
t
u
r
e

2
0
0
1

t
o

2
0
0
5
(
%
)
0 10 20 30 40 50 60 70 80 90 100
Percentiles
RURAL

Table 8 Growth rates (%) among the poor, Madagascar 2001-2005
percentile Madagascar Urban Rural
10 50.9 17.4 56.0
15 48.0 15.5 53.4
20 45.9 12.5 51.4
25 44.2 9.4 49.9
30 42.7 7.0 48.6
50 18.8 -12.8 32.9
mean -1.4 -15.5 7.1
6.2 Growth Elasticities of Poverty
In this section we estimate growth elasticities of poverty based on the (estimated) per capita
expenditure density function. Following Kakwani (1993), we use the following formula:
(1) ( )
( ) ( )
( )
( )
( )
, , 1
0
,
,
0
,
G
P t P t
if
P t
t
zf t z
if
F t z
o
o o o
o
o
n
o

=

where P(t,) is the Foster-Greer-Thorbecke poverty measure with parameter in period t, f(t,z)
and F(t,z) denote, respectively, the probability density function and the cumulative density
function of per capita expenditure in period t, and z is the absolute poverty line.
7

Table 9 shows the non-parametric estimates of the elasticities defined in equation (1), calculated
for t = 2001 and t = 2005. If compared to estimates by other studies for other countries, the
elasticities in Table 9 are low. For instance, Ravallion and Chen (1997) estimated the growth
elasticity of the incidence of poverty to be between -2.0 and -3.0.
While it is hard to comment on the absolute magnitude of the elasticities, it is worth noting that
in Madagascar between 2001 and 2005 elasticities have almost doubled, regardless of the
poverty measure considered. Kakwani and Son (2004) showed that growth elasticity of poverty
decreases with the initial level of economic development and increases with the initial level of
inequality.
8
This implies that economic growth is more effective in reducing poverty in rich

7
See also Duclos and Araar (2006).
8
The result does not hold true for the headcount ratio, according to proposition 1.
22
countries (as opposed to poor countries) and in countries with low levels of inequality.
Madagascar qualifies as a poor country, that has experienced both low economic growth and
decreasing inequality. Hence, the observed increase in the growth elasticities of poverty is likely
to stem from changes in inequality. Thanks to the reduction in inequality which occurred
between 2001 and 2005, poverty in 2005 is more sensitive to economic growth than it was
2001.
Table 9 Growth elasticities of poverty for Madagascar, 2001 and 2005
Elasticity to growth 2001 2005
Headcount ratio 0.553 0.930
Poverty Gap 0.998 1.567
Poverty Gap Squared 1.337 2.002
Source: Authors estimates based on EPM data.
6.3 Growth-Inequality Decomposition
A recurrent theme on poverty reduction debates is the relative contribution of economic growth
and inequality to poverty reduction. In this section we decompose the observed changes in
poverty indices between 2001 and 2005 into two components: (i) the growth component (GC),
which identifies the poverty change due to the growth of mean per capita expenditure, and (ii)
the inequality component (IC), which identifies the poverty change due to a more equal
distribution of income.
Let P(t) be a poverty measure of the Foster, Green and Thorbeke (1984) class in period t.
Following Muller (2006), the ideal decomposition of the variation of P over the time interval
(T
0
, T
1
) can be written as follows:
(2) ( ) ( )
( ) ( ) ( ) ( )
t d
dt
t dL
L
t P
dt
dt
t d t P
T P T P P
IC
T
T
GC
T
T

c
c
+
c
c
= = A
1
0
1
0
0 1
u
u

where (t) is the mean per capita expenditure and L(t) is the Lorenz curve in period t. We lack
information on the partial derivatives in equation (2) over the entire time interval ( )
1 0
,T T , and
therefore rely on the following approximation of equation (2):
(3) ( ) ( ) ( ) ( ) R L T P T P R dt
dt
dL
T P dt
dt
d
T P P
C I
r L
C G
r
T
T
r L
T
T
r
+ A + A = + + = A

1
0
1
0
u
u
u u

where
( ) ( ) ( ) ( ) ( ) ( ) ( )
( ) ( )
( )
( ) ( ) ( ) ( ) ( ) ( ) ( )
( ) ( )
( )
1 0
1 0
1 0
1 0
, ,
, ,
r r
r
r r
L r
P T L T P T L T P t
P T
T T
P T L T P T L T P t
P T
L L T L T
u
u u
u u u
u u
c
~ =
c
c
~ =
c

approximate the partial derivatives in equation (2), = (T
1
) - (T
0
), L = L(T
1
) - L(T
0
), and R
is a residual term. Note that the decomposition depends on the arbitrary reference period T
r
. Datt
and Ravallion (1992) recommend the use of the initial period (T
r
=T
0
), but other choices are
23
available. One is the ending period (T
r
=T
1
), another is the so-called Shapley decomposition
where the growth and inequality components are assumed to be an average of the approximated
decompositions with T
r
=T
0
and T
r
=T
1
.
Table 10 shows the results for the three decompositions described above, using the main
poverty measures of the FGT class. Poverty decompositions are found to be robust to the
reference period chosen; residual terms are negligible in size, with no exceptions.
Table 10 Growth-inequality decompositions, Madagascar 2001-2005
reference period
2001 2005 Shapley
Headcount (H) 69.728 68.748
Change in H -0.980 -0.980 - 0.980
Growth component 0.506 0.731 0.619
Inequality component -1.711 -1.486 -1.598
Residual 0.225 0.225 0.000
Poverty Gap (PG) 34.906 26.780
Change in PG -8.127 -8.127 -8.127
Residual 0.100 0.100 0.000
Poverty Gap Squared (PG2) 20.923 13.383
Change in PG2 -7.541 -7.541 -7.541
Residual -0.021 -0.021 0.000

Overall the inequality effect is dominant. The contribution of the growth component is low, a
result largely expected because of the substantial stability of mean per capita expenditure during
the period considered. As argued above, however, one has to take into account the timing of the
surveys. The fact that we use 2001, a year immediately preceding a major crisis, and compare it
to 2005, a time by which the recovery from the crisis was just completed, makes the results in
Table 9 difficult to interpret, if not misleading. In particular, the role of the growth component is
likely to be severely under-estimated.
6.4 Sectoral Decomposition of Poverty
Changes in the national poverty level can be decomposed into the relative contributions of
changes in poverty within population sub-groups, and changes in population shares across
sectors. In this section we estimate the relative contributions of these two components by
exploiting the additive decomposability of the FGT class of poverty indices.
Following Ravallion and Huppi (1991) we use the following formula:
(3) ( ) ( ) ( ) ( ) ( ) ( )
1 0 0 1 0 0
1 1
K K
k k k k k k
k k
WITHIN GROUP INTER GROUP
P P T P T n T n T n T P T R
= =

A = + +

where R denotes a residual term.
Table 11 shows the results of the decomposition (3) for selected groups. The main result is that
the within-group effects dominate, regardless of the choice of the poverty measure and the
definition of population sub-groups. For instance, taking the PG decomposition by urban-rural
(top panel in Table 11) we find that the change in PG within rural and urban areas (-8.2) would
have caused a larger reduction in the aggregate PG index than the observed change (-8.1), were
24
it not for the offsetting effect of the population shifts (+0.2), i.e. for the between-area
movements of the population. This finding is consistent with the fact that during the observation
period there was a net inflow towards rural areas, that is, towards areas with higher poverty
incidence and poverty intensity.
The pattern found for the urban-rural decomposition is similar to the decomposition by province
(mid panel in Table 11): it is the change of poverty within provinces that drives the overall
decline in poverty rates in Madagascar. Populations shifts play a marginal role.
The bottom panel in Table 11 tells a slightly different story. While the decline in poverty
continues to be driven by the within-group component, the population movements between
economic sectors play a non negligible role: in the absence of population shifts we would have
observed larger reductions in poverty rates.
Table 11 Sectoral decomposition of poverty, 2001-2005
sub-group partition
change in national
index
(%)
% change in MLD accounted for by changes in
within-group
poverty
intergroup residual
by urban/rural
H -1.0
(100)
-1.2
(120)
0.3
(-30)
-0.1
(10)
PG -8.1
(100)
-8.2
(101)
0.2
(-2)
-0.1
(1)
PG2 -7.6
(100)
-7.6
(100)
0.1
(-1)
-0.1
(1)
by province
H -1.0
(100)
-1.1
(114)
-0.1
(13)
0.3
(-27)
PG -8.1
(100)
-8.1
(100)
0.1
(-1)
-0.1
(1)
PG2 -7.6
(100)
-7.5
(99)
0.1
(-1)
-0.1
(2)
by economic sector: primary/secondary/tertiary
H -1.0
(100)
-3.4
(259)
3.5
(-268)
-1.4
(109)
PG -8.1
(100)
-9.6
(112)
2.3
(-26)
-1.2
(14)
PG2 -7.6
(100)
-8.4
(108)
1.5
(-19)
-0.9
(11)
7 A Model of Household Consumption
In this section we apply multivariate analysis techniques to the EPM data to identify the
determinants of household consumption patterns in Madagascar. Following Razafindravonona
et al. (2001), we estimate distinct rural and urban models of log-consumption for the years 2001
and 2005. Regression results are shown in the appendix. The main findings can be summarized
as follows.
Demographics. Household composition has a major impact on consumption. The dependency
ratio, defined as the number of household members aged below 15 or above 64 divided by the
number of individuals aged 15 to 64, has a large and negative impact on per capita consumption.
25
This finding is robust to the econometric specification of the model; it holds true in urban as
well as in rural areas, for both survey years. Living standards decrease with household size.
9

Education. According to the 2001 estimates, educational attainments of household members are
positively correlated with consumption, albeit with differences by gender, level of education and
location. Estimates for 2005 are more blurred: the coefficients of education variables decrease,
particularly in urban areas.
Occupational status. According to all models estimated, wage-earners and the self employed in
the secondary and tertiary sectors are better off than their peers in agriculture. This finding is
consistent with low estimates of agricultural productivity (Hoftijzer and Stifel 2007). The
negative effect of informal sector employment is particularly strong.
Land. In a country like Madagascar, where a large fraction of the population is employed in the
primary sector, land is perhaps the most important asset. Not surprisingly, land ownership
decreases the risk of poverty, increasing consumption proportionately with the extent of land
holdings.
Geography. Even after controlling for a wide range of socio-economic and demographic
variables, provincial dummy effects are still large in size and significant. Living in provinces
other than Antananarivo implies a higher risk of poverty. According to the 2001 estimates the
households most at risk are those in Toamasina followed by those in Fianarantsoa, Toliara,
Mahajanga and Antsiranana. The ranking hardly changes in 2005.
The non-negligible and significant provincial dummy effects suggest that our model fails to
capture all relevant determinants of poverty. Following World Bank (2002), our strategy is to
explore the roles of (i) infrastructure, (ii) climate shocks and (iii) land tenure patterns.
Unfortunately the EPM surveys do not provide information about these factors. However,
community-level data from a 2001 census of communes can be matched to the EPM data.
Adding variables from the census, we estimate a new model of household consumption with
controls for crop mix, relevant infrastructure, and the natural environment. The estimates in
Tables A1 and A2 indicate that these factors largely explain provincial differences not
attributable to household variables; most provincial dummy effects disappear or lose statistical
significance.
Considering individual variables we find that the major roles are played by public water
provision, specialization in rice production, and remoteness of location, while adverse climate
shocks have a negative but lesser impact.
8 Poverty and Growth Projections
This section presents the results of a simulation projecting poverty measures in Madagascar for
two benchmark years, 2007 and 2010. The simulations are based on forecasts of sectoral value
added and population growth (see Appendix 3).

9
As noted by Lanjouw and Ravallion (1995), the existence of size economies in household consumption
cautions against concluding that larger families tend to be poorer (p. 1415). Further research is needed to
address this issue; the analysis carried out in this paper, based on consumption per capita as the welfare
metric, rules out by construction the existence of economies of scale.
26
Table 12 Impact of growth on poverty in Madagascar
2007 2010
Urban Rural National Urban Rural National
Headcount ratio 49.4 73.1 67.9 42.0 70.1 63.9
PG 18.2 28.8 26.5 15.4 27.2 24.6
PG2 9.1 14.5 13.3 7.7 13.6 12.3
Annualized
change rates (%)
2005-2007 2005-2010
Headcount ratio -2.53 -0.27 -0.58 -4.18 -0.94 -1.44
PG -2.89 -0.17 -0.56 -4.41 -1.21 -1.70
PG2 -1.61 0.00 -0.37 -3.91 -1.27 -1.70
Source: Authors estimates based on EPM 2005 data and World Bank forecasts of sectoral value added and
population growth.

The projections in Table 13 show poverty decreasing at a slow annual rate. At the same time,
the pattern of poverty changes is reversed relative to the 2001-2005 period: poverty decreases
faster in urban than in rural areas. This is due to projections of sectoral GDP growth rates which
indicate more rapid growth in the largely urban secondary and tertiary sectors.
It bears emphasis that the simulation results are contingent on several assumptions. Of particular
importance is the assumption that population shares remain constant across geographic regions
and economic sectors. Other economic factors could affects poverty projections in Table 12.
Similarly, GDP growth rates used in the analysis do not take into account the rise in world
energy prices during the first half of 2006. Andriamihaja and Vecchi (2007) have estimated that
a 17 percent rise in the price of energy products leads to a 1.75 percent average decrease in real
expenditure (2.1 percent for low-income households, 1.5 percent for high-income households).
9 Summary and Final Remarks
The paper has accomplished three main tasks. First, it has documented the changes in poverty
and inequality in the time period between 2001 and 2005. Second, it has examined the factors at
play in determining the evolution of poverty and inequality over time. Third, it has forecasted
poverty on the basis of population and sectoral GDP growth rate projections.
The availability of comparable surveys for the years 2001 and 2005 has made the updating of
the poverty and inequality profiles a relatively straightforward task. The main findings can be
summarized as follows.
The incidence of poverty at the national level has barely changed between the years 2001 and
2005. However, while the headcount ratio among rural households has not changed in a
statistically significant way, it has unambiguously increased among urban households. This
pattern reverses the tendency observed during the second half of the 1990s, when rural poverty
was on the rise and urban poverty was falling precipitously see World Bank (2002).
While the fraction of the population classified as poor has decreased, the absolute count of the
poor has increased by some 2 million people. This leaves a fundamental question unanswered:
has poverty in Madagascar decreased or increased during the first half of the decade? This
finding suggests that population growth rates play a significant role in shaping poverty in
Madagascar and in determining the risk of poverty. Families with low dependency ratios face an
estimated poverty risk 60 percent lower than the average risk. Additional research on the impact
of demographic transition on poverty in Madagascar should be awarded a high priority.
While the evidence on changes in the incidence of poverty is inconclusive, results of the
analysis of the depth and severity of poverty are unambiguous. Both the poverty gap and the
poverty gap squared point to a substantive and statistically robust reduction of poverty at the
27
national level. Moreover, in contrast to what observed with the headcount ratio, there is no
contradiction when we consider total versus average poverty gaps; the absolute shortfall of the
poor in monetary terms decreased by 7.8 percent between 2001 and 2005. With regard to the
pattern of urban and rural poverty we find that rural poverty decreased while the change in
urban poverty was not statistically significant.
Consistent with the trend in poverty is the finding that inequality decreases both nationally and
within urban and rural areas, regardless of the index employed. Even if the growth-inequality
decomposition of poverty leads to the conclusion that inequality dominates the explanation of
the trend in poverty, we believe this result greatly over-emphasizes the role attributed to
inequality. We argue that the political crisis of 2002 is largely responsible for this result.
The analysis of poverty at the provincial level confirms what was previously noted in World
Bank (2002: iv), namely that Madagascar is characterized by a ... high degree of spatial
heterogeneity in poverty levels across the country. We find that provinces not only display
variation in poverty levels but also in the changes of poverty over time. For instance, between
the years 2001 and 2005 the incidence of poverty in Antananarivo increased by 18.7 percent,
while it hardly changed in Toliara (minus 1.7 percent). In Toamasina it decreased by 12.6
percent.
Our second major aim in this paper was to explore the determinants of poverty. The strategy
pursued in the paper capitalized on the regression analysis carried out in World Bank (2002).
We find that even after controlling for a large number of socio-economic and demographic
variables, the provincial dummies remain large in size and statistically significant. The
interpretation of this result is twofold: (i) ceteris paribus, geographical location of households
matters in determining poverty risks, (ii) regression analysis based on the EPM data alone, does
not explain why geographical location for poverty matters so much.
By exploiting data from the Commune Census for 2001, we were able to assess the relative
contribution of three sets of explanatory variables: (i) provincial endowments of infrastructures,
(ii) the structure of the agricultural sector, and (iii) climatic factors. A new regression model,
inclusive of factors (i)-(iii) was estimated separately for urban and rural households. In the rural
model, controlling for (i)-(iii), rendered all provincial dummies, statistically insignificant. A
similar result was obtained for the consumption model estimated for urban households.
Overall, the regression results support the view advanced in a number of recent World Bank
documents, namely that for a country as remote as Madagascar, the lack of and/or poor quality
of the transport and communication infrastructures is a major obstacle to development and
poverty reduction. The above evidence is also consistent with the emphasis placed on the role of
agricultural productivity in lifting people out of poverty. We find that land tenure (as captured
by the ownership and distribution of land across the population) has a large and significant
impact on living standards. Lastly, we find that the occurrence of adverse climatic shocks has a
negative impact on poverty, though their magnitude is substantially lower compared to factors
(i) and (iii).
The third and last aim of the paper was to project poverty to the present day, and to deliver a
forecast of poverty into a not-too distant future. According to our estimates, poverty in 2007
has, largely, remained, at the same level as (and with the same structure of) in 2005. Our
analysis points to a tendency towards improvement, especially in urban areas. Longer-term
poverty forecasts predict that the incidence of poverty will fall from 68.7 percent in 2005 to 63.9
percent in 2010, with a marked pro-urban bias which characterizes the poverty reduction
process.
28
List of References
Andriamihaja, N. and G. Vecchi (2007), An Evaluation of the Welfare Impact of Higher
Energy Prices in Madagascar, in D. Go and J. Page (eds). Africa at a Turning Point? Growth,
Aid External Shocks. Washington: World Bank.
Arndt, C. and K. Simler (2005), Estimating utility-consistent poverty lines, FCND Discussion
Papers 189, International Food Policy Research Institute.
Chakravarti, S., R. Kanbur and D. Mukherjee (2006), Population Growth and Poverty
Measurement, in Social Choice and Welfare, 26, 3.
Chaudhuri, S., J. Jalan and A. Suryahadi (2002), Assessing household vulnerability to poverty
from cross-sectional data: a methodology and estimates from Indonesia, Discussion Paper
0102-52. New York: Columbia University.
Duclos J. and A. Araar (2006), Poverty and Equity. Measurement, Policy, and Estimation with
DAD, Springer, New York.
Datt, G. and M. Ravallion (1992), Growth and Redistribution Components of Changes in
Poverty Measures: A Decomposition with Application to Brazil and India in the 1980s,
Journal of Development Economics, 38.
Hoftijzer, M. and D. Stifel (2007), Exploring The Role of Employment and Earnings in
Poverty Reduction: The Case of Madagascar, photocopy.
Jain, L. and S. Tendulkar (1990), The Role of Growth and Distribution in the Observed
Change in the Head Count Ratio Measure of Poverty: A Decomposition Exercise for India,
India Economic Review, 25, 2.
Jenkins, S.P. (1995), Accounting for inequality trends: Decomposition analyses for the UK,
1971-86, Economica, 62: 29-63.
Kakwani, N. (1993), Poverty and economic growth with application to Cte dIvoire, Review
of Income and Wealth, 39(2): 121-139.
Kakwani, N. (2000), On Measuring Growth and Inequality Components of Poverty with
Application to Thailand, Journal of Quantitative Economics, 16 (1).
Kakwani, N. and H. Son (2004), Economic Growth and Poverty Reduction: Initial Conditions
Matter, International Poverty Centre Working Paper n.2.
Lanjouw, P. and M. Ravallion (1995), Poverty and Household Size, Economic Journal, 105:
1415-1434.
Ligon, E. and L. Schechter (2003), Measuring vulnerability, Economic Journal, 113: C95-
C102.
Mookherjee, D. and A.F. Shorrocks (1982), A decomposition analysis of the trend in UK
income inequality, Economic Journal, 92: 886-902.
Muller, A. (2006) Clarifying Poverty Decomposition, Working Papers in Economics 213,
Gteborg University, Department of Economics
29
Ravallion, M. (1988), Expected poverty under risk-induced welfare variability, Economic
Journal, 98, 393: 1171-82.
Ravallion, M. and B. Bidani (1994), How robust is a poverty profile?," World Bank Economic
Review, 8, 75-102.
Ravallion, M. and M. Huppi (1991), Measuring Changes in Poverty: A Methodological Case
Study of Indonesia During an Adjustment Period, World Bank Economic Review, 5.
Ravallion, M. and M. Lokshin (2006), Testing Poverty Lines, Review of Income and Wealth,
52(3): 399-421.
Shorrocks, A. (1980), The class of additively decomposable inequality measures,
Econometrica, 48: 613-25.
Stifel, D., B. Minten and P. Dorosh (2003), Transaction Costs and Agricultural Productivity:
Implications of Isolation for Rural Poverty in Madagascar, MSSD Discussion Poverty no. 56,
International Food Policy Research Institute.
United Nation Development Programme (2006), Human Development Report.
World Bank (2002), Poverty and Socio Economic Developments: 1993-1999, Report No.
23366-MAG.
30
Appendix 1 Regression Analysis of Household Consumption
In this appendix we show the regression results of consumption models, estimated separately by
urban ad rural households. The dependent variable is the log of per capita household
consumption.
In Table 13 models (1) and (2) are comparable, models (1) and (3) are also comparable, while
models (2) and (3) are not. Estimation of models (1), (2) and (3) uses sampling weights.
Table 13 Regression Estimates of Consumption Models for Rural Households,
Madagascar 2001 and 2005
(1) (2) (3)
2001 rural 2001 rural 2005 rural
hh head age 0.002 0.002 0.001
(0.001) (0.001) (0.001)
1 if male hh head 0.136 0.101 -0.002
(0.058)* (0.054) (0.031)
number of children aged 0-4 -0.172 -0.174 -0.148
(0.026)** (0.026)** (0.015)**
(0.023)** (0.023)** (0.014)**
number of adults male -0.176 -0.174 -0.148
(0.026)** (0.025)** (0.015)**
number of adults female -0.197 -0.209 -0.214
(0.028)** (0.028)** (0.016)**
hh size squared 0.006 0.006 0.005
(0.001)** (0.001)** (0.001)**
dependency ratio -0.596 -0.592 -0.532
(0.095)** (0.094)** (0.055)**
hh head separated or divorces 0.029 0.035 -0.014
(0.064) (0.060) (0.033)
hh head widower -0.021 -0.047 -0.048
(0.068) (0.061) (0.033)
hh head single 0.124 0.092 -0.143
(0.067) (0.066) (0.046)**
hh head with primary 0.096 0.109 0.030
(0.040)* (0.037)** (0.019)
hh head with low secondary 0.370 0.333 0.107
(0.061)** (0.055)** (0.029)**
hh head with upper secondary 0.341 0.333 0.144
(0.076)** (0.069)** (0.049)**
number of male adults with at least primary 0.039 0.021 0.018
(0.024) (0.022) (0.012)
number of female adults with at least primary 0.079 0.065 0.066
(0.022)** (0.020)** (0.010)**
number of members attending school 0.042 0.024 0.018
(0.016)** (0.015) (0.009)*
non-agricultural self-employment 0.300 0.234 0.388
(0.062)** (0.055)** (0.039)**
wage employment (public sector) 0.282 0.288 0.230
(0.098)** (0.092)** (0.085)**
wage employment (enterprise) 0.243 0.107 0.189
(0.067)** (0.058) (0.034)**
wage employment (individual) 0.238 0.098 0.009
(0.090)** (0.083) (0.043)
not employed 0.152 0.212 -0.216
(0.149) (0.178) (0.088)*
informal -0.209 -0.183 -0.139
(0.063)** (0.057)** (0.073)
hh owns no or little land -0.331 -0.189 0.040
(0.065)** (0.058)** (0.033)
hh owns land (0.10-0.49 ha/head) 0.150 0.170 0.122
(0.035)** (0.032)** (0.020)**
31
(1) (2) (3)
2001 rural 2001 rural 2005 rural
(0.048)** (0.045)** (0.025)**
hh own land (more than 1 ha/head) 0.448 0.480 0.439
(0.055)** (0.050)** (0.032)**
Fianarantsoa -0.354 -0.039 -0.166
(0.044)** (0.052) (0.019)**
Toamasina -0.415 -0.088 -0.258
(0.045)** (0.053) (0.025)**
Mahajanga -0.180 0.010 -0.133
(0.047)** (0.057) (0.022)**
Toliara -0.208 0.097 -0.235
(0.051)** (0.058) (0.024)**
Antsiranana -0.139 0.226 -0.234
(0.050)** (0.060)** (0.028)**
drice1s 0.108
(0.032)**
riceirrig 0.002
(0.001)**
shareagr -0.003
(0.001)**
post 0.021
(0.030)
roadnat 0.061
(0.031)
market -0.012
(0.029)
phone -0.061
(0.045)
waterjirama 0.451
(0.056)**
time_quin== 2 -0.153
(0.046)**
(0.040)**
(0.045)**
time_quin==Most Remote -0.007
(0.049)
1 if cyclones during previous 2 years 0.040
(0.017)*
1 if floud during previous 2 years 0.018
(0.015)
1 if drought during previous 2 years -0.036
(0.013)**
1 if phyto disease during previous 2 years -0.017
(0.009)
intercept 14.429 14.237 13.248
(0.137)** (0.144)** (0.092)**

Observations 1982 1963 5907
Adjusted R-squared 0.5116 0.5826 0.3539
Note: * significant at 5%; ** significant at 1%.
Source: Authors estimates on EPM data for 2001 and 2005, and Commune Census data for 2001.

32
Table 14 Regression Estimates of Consumption Models for Urban Households,
Madagascar 2001 and 2005
(1) (2) (3)
2001 urban 2001 urban 2005 urban
hh head age 0.002 0.002 -0.000
[0.001] [0.001]* [0.001]
1 if male hh head 0.087 0.082 0.084
[0.055] [0.054] [0.044]
[0.031]** [0.029]** [0.027]**
[0.027]** [0.027]** [0.021]**
number of adults male -0.205 -0.212 -0.301
[0.026]** [0.024]** [0.024]**
number of adults female -0.232 -0.242 -0.326
[0.027]** [0.025]** [0.026]**
hh size squared 0.005 0.006 0.009
[0.002]** [0.001]** [0.001]**
dependency ratio -0.667 -0.704 -0.694
[0.108]** [0.103]** [0.086]**
hh head separated or divorces -0.032 -0.024 -0.045
[0.060] [0.058] [0.047]
hh head widower 0.023 0.028 -0.056
[0.062] [0.063] [0.053]
hh head single 0.159 0.110 0.020
[0.064]* [0.062] [0.055]
hh head with primary 0.085 0.085 -0.113
[0.041]* [0.039]* [0.031]**
hh head with low secondary 0.425 0.428 -0.077
[0.046]** [0.044]** [0.037]*
hh head with upper secondary 0.260 0.281 0.009
[0.054]** [0.051]** [0.044]
number of male adults with at least primary 0.013 0.021 0.117
[0.020] [0.020] [0.018]**
number of female adults with at least primary 0.091 0.090 0.136
[0.020]** [0.020]** [0.018]**
number of members attending school 0.049 0.053 0.015
[0.015]** [0.014]** [0.013]
non-agricultural self-employment 0.353 0.307 0.434
[0.052]** [0.050]** [0.039]**
wage employment (public sector) 0.246 0.204 0.186
[0.064]** [0.062]** [0.069]**
wage employment (enterprise) 0.317 0.239 0.172
[0.053]** [0.051]** [0.039]**
wage employment (individual) 0.212 0.157 -0.004
[0.056]** [0.055]** [0.059]
not employed 0.283 0.221 -0.114
[0.108]** [0.113] [0.066]
informal -0.227 -0.210 -0.357
[0.036]** [0.036]** [0.046]**
hh owns no or little land -0.240 -0.177 0.025
[0.049]** [0.047]** [0.031]
[0.052] [0.049] [0.036]**
[0.074]* [0.067]* [0.039]**
hh own land (more than 1 ha/head) 0.404 0.364 0.253
[0.065]** [0.067]** [0.053]**
Fianarantsoa -0.359 -0.267 -0.281
[0.037]** [0.043]** [0.029]**
Toamasina -0.423 -0.434 -0.225
[0.038]** [0.043]** [0.032]**
Mahajanga -0.204 -0.120 -0.003
[0.039]** [0.057]* [0.031]
Toliara -0.203 -0.074 -0.200
33
(1) (2) (3)
2001 urban 2001 urban 2005 urban
[0.044]** [0.052] [0.032]**
Antsiranana 0.049 0.065 0.119
[0.038] [0.050] [0.036]**
1 if bank in the community 0.093
[0.042]*
1 if airport in the community -0.205
[0.030]**
1 if harbor river in the community 0.200
[0.045]**
1 fi harbsea 0.028
[0.041]
post -0.271
[0.072]**
roadnat 0.150
[0.050]**
market 0.056
[0.058]
phone 0.174
[0.042]**
waterjirama 0.135
[0.039]**
redz 0.240
[0.044]**
Intercept 14.707 14.464 13.881
[0.108]** [0.138]** [0.092]**

Observations 2705 2704 5851
Adjusted R-squared 0.5671 0.5995 0.4510
Note: * significant at 5%; ** significant at 1%.
Source: Authors estimates on EPM data for 2001 and 2005, and Commune Census data for 2001.
Appendix 2 Sensitivity Analysis of Poverty Estimates to the Choice
of Different Deflators
This Appendix analyses the sensitivity of poverty estimates presented in the text to the choice of
different (spatial) price deflators. The issue arises from the fact that in 2001 the welfare
indicator was deflated at the urban-rural level for each province (faritany). At the time, this was
the finest possible geographical disaggregation level. In 2005, however, the survey was made
representative of each of the newly formed 22 regions (faritra) at the urban-rural level.
Accordingly, deflation in 2001 used 13 different prices (Antananarivo being counted as a
separate area), while in 2005 deflation exploited all the.45 regional prices available (again,
Antananarivo counted as an additional region).
To assess the extent to which the change in the deflation method affects the results obtained in
the text, we first calculated a set of provincial deflators for the year 2005 by aggregating the
regional deflators, and next we used them to re-calculate poverty rates.
Table 16 compares the deflators for 2001, with the new deflators for 2005. The latter were
calculated as population-weighted averages of regional prices.
Table 17 contains the key results of the sensitivity analysis. The main conclusion that we draw
is that the poverty profile is robust to the choice of a regional versus provincial deflation
method. The discrepancy in poverty rates, irrespective of the poverty measure, are on average
lower than 1 percent.
34
Appendix Table 15 Provincial Deflators, Madagascar 2001 and 2005
Province 2001 2005
Antananarivo Urban 0.9132 0.8827
Antananarivo Rural 0.9375 0.8991
Antananarivo Capital 1.0000 1.0000
Fianarantsoa Urban 0.9015 0.8892
Fianarantsoa Rural 0.9208 0.8598
Toamasina Urban 0.9869 0.9030
Toamasina Rural 0.9395 0.8832
Mahajanga Urban 0.8943 0.8929
Mahajanga Rural 0.8398 0.8131
Toliara Urban 0.9250 0.9754
Toliara Rural 0.9386 0.8648
Antsiranana Urban 1.0998 0.9883
Antsiranana Rural 1.0893 0.9974
Source: Authors estimates based on EPM data for 2001 and 2005.

Appendix Table 16 Regional vs. provincial deflators: Sensitivity of poverty estimates

2005
regional deflator
2005
provincial deflator
discrepancy
(%)
province H PG PG2 H PG PG2 H PG PG2
Antananarivo (U) 41.6 13.6 5.8 41.7 13.6 5.8 0.24 0.00 0.00
Antananarivo (R) 64.7 21.9 9.8 64.8 22.0 9.8 0.15 0.46 0.00
Antananarivo 57.7 19.4 8.6 57.8 19.5 8.6 0.17 0.38 0.31
Fianarantsoa (U) 71.6 28.8 14.5 71.4 28.7 14.5 -0.28 -0.35 0.00
Fianarantsoa (R) 78.7 30.9 15.1 79.1 31.0 15.3 0.51 0.32 1.32
Fianarantsoa 77.6 30.6 15.0 77.8 30.6 15.2 0.26 0.12 1.14
Toamasina (U) 55.8 21.4 11.2 57.0 21.6 11.2 2.15 0.93 0.00
Toamasina (R) 75.6 33.1 18.0 75.9 33.1 18.0 0.40 0.00 0.00
Toamasina 71.9 30.9 16.7 72.5 31.0 16.7 0.83 0.30 0.28
Mahajanga (U) 47.0 16.1 7.2 48.4 15.9 7.1 2.98 -1.24 -1.39
Mahajanga (R) 76.6 28.9 13.9 75.5 29.3 14.4 -1.44 1.38 3.60
Mahajanga 70.2 26.2 12.4 69.6 26.4 12.8 -0.85 0.82 3.52
Toliara (U) 64.3 28.3 15.9 65.3 28.4 15.8 1.56 0.35 -0.63
Toliara (R) 77.4 34.0 19.1 77.5 34.0 19.1 0.13 0.00 0.00
Toliara 74.8 32.9 18.4 75.0 32.9 18.4 0.27 -0.05 0.14
Antsiranana (U) 33.8 9.4 3.5 35.3 9.3 3.4 4.44 -1.06 -2.86
Antsiranana (R) 69.8 28.1 14.2 69.6 28.1 14.0 -0.29 0.00 -1.41
Antsiranana 64.2 25.2 12.5 64.3 25.2 12.4 0.16 -0.07 -1.01
Urban 52.0 19.3 9.4 52.5 19.3 9.4 0.96 0.00 0.00
Rural 73.5 28.9 14.5 73.5 29.0 14.6 0.00 0.35 0.69
Madagascar 68.7 26.8 13.4 68.9 26.9 13.5 0.29 0.37 0.75
35
Appendix 3 Sectoral Value Added And Population Growth Rates,
Projections for 2006-2010
Sectors 2006 2007 2008 2009 2010
Primary Sector
Agriculture 2.6 2.3 3.5 3.5 3.5
Farming & Fishing 1.9 3.2 3.2 3.2 3.2
Forestry 1.0 1.0 1.0 1.0 1.5
Other 2.1 2.6 3.2 3.2 3.2
Secondary sector
Agro industry -12.5 6.0 6.0 6.0 6.0
Mining 9.1 10.5 11.5 11.5 11.5
Energy 4.4 5.6 5.6 5.6 5.6
Food industry 0.0 7.3 8.3 8.3 8.3
Chemical 2.2 3.5 3.5 3.5 3.5
Textile 0.8 6.0 6.0 6.0 6.0
Wood 4.1 5.0 5.0 5.0 5.0
Construction 11.6 15.9 16.9 13.0 12.0
Other -6.0 7.0 7.0 7.0 6.0
Tertiary sector
Public Construction 22.5 22.8 18.1 18.0 14.0
Transport 8.7 9.5 11.1 11.0 10.0
Telecom 12.0 10.8 10.8 10.8 10.8
Trade 4.4 7.3 8.9 8.9 8.9
Banks & Insurance 14.2 8.9 7.6 7.6 7.6
Services 5.5 7.9 9.5 9.5 9.5
Public Administration 2.1 2.5 4.1 5.6 5.6

Population 2.8 2.8 2.7 2.7 2.6
Source: Ministre des Finances et de l'Economie, INSTAT, World Bank and IMF projections.
36
Africa Region Working Paper Series
Series # Title Date Author
ARWPS 1 Progress in Public Expenditure Management in
Africa: Evidence from World Bank Surveys

January 1999 C. Kostopoulos
ARWPS 2 Toward Inclusive and Sustainable Development in
the Democratic Republic of the Congo

March 1999 Markus Kostner
ARWPS 3 Business Taxation in a Low-Revenue Economy: A
Study on Uganda in Comparison with Neighboring
Countries

June 1999 Ritva Reinikka
Duanjie Chen
ARWPS 4 Pensions and Social Security in Sub-Saharan
Africa: Issues and Options

October 1999 Luca Barbone
Luis-A. Sanchez B.
ARWPS 5 Forest Taxes, Government Revenues and the
Sustainable Exploitation of Tropical Forests

January 2000 Luca Barbone
Juan Zalduendo
ARWPS 6 The Cost of Doing Business: Firms Experience
with Corruption in Uganda

June 2000 Jacob Svensson
ARWPS 7 On the Recent Trade Performance of Sub-Saharan
African Countries: Cause for Hope or More of the
Same

August 2000 Francis Ng and
Alexander J. Yeats
ARWPS 8 Foreign Direct Investment in Africa: Old Tales and
New Evidence

November 2000 Miria Pigato
ARWPS 9 The Macro Implications of HIV/AIDS in South
Africa: A Preliminary Assessment

November 2000 Channing Arndt
Jeffrey D. Lewis
ARWPS 10 Revisiting Growth and Convergence: Is Africa
Catching Up?

December 2000 C. G. Tsangarides
ARWPS 11 Spending on Safety Nets for the Poor: How Much,
for
How Many? The Case of Malawi

January 2001 William J. Smith
ARWPS 12 Tourism in Africa February 2001 Iain T. Christie
D. E. Crompton

ARWPS 13 Conflict Diamonds

February 2001 Louis Goreux
ARWPS 14 Reform and Opportunity: The Changing Role and
Patterns of Trade in South Africa and SADC

March 2001 Jeffrey D. Lewis
ARWPS 15 The Foreign Direct Investment Environment in
Africa

March 2001 Miria Pigato
ARWPS 16 Choice of Exchange Rate Regimes for Developing
Countries

April 2001 Fahrettin Yagci
ARWPS 18 Rural Infrastructure in Africa: Policy Directions

June 2001 Robert Fishbein
ARWPS 19 Changes in Poverty in Madagascar: 1993-1999 July 2001 S. Paternostro
J. Razafindravonona
David Stifel

ARWPS 20 Information and Communication Technology, August 2001 Miria Pigato
37
Poverty, and Development in sub-Saharan Africa
and South Asia

ARWPS 21 Handling Hierarchy in Decentralized Settings:
Governance Underpinnings of School Performance
in Tikur Inchini, West Shewa Zone, Oromia
Region

September 2001 Navin Girishankar A.
Alemayehu
Yusuf Ahmad
ARWPS 22 Child Malnutrition in Ethiopia: Can Maternal
Knowledge Augment The Role of Income?

October 2001 Luc Christiaensen
Harold Alderman
ARWPS 23 Child Soldiers: Preventing, Demobilizing and
Reintegrating

November 2001 Beth Verhey
ARWPS 24 The Budget and Medium-Term Expenditure
Framework in Uganda

December 2001 David L. Bevan
ARWPS 25 Design and Implementation of Financial
Management Systems: An African Perspective
January 2002 Guenter Heidenhof
H. Grandvoinnet
Daryoush Kianpour
B. Rezaian

ARWPS 26 What Can Africa Expect From Its Traditional
Exports?

February 2002 Francis Ng
Alexander Yeats

ARWPS 27 Free Trade Agreements and the SADC Economies February 2002 Jeffrey D. Lewis
Sherman Robinson
Karen Thierfelder

ARWPS 28 Medium Term Expenditure Frameworks: From
Concept to Practice. Preliminary Lessons from
Africa

February 2002 P. Le Houerou
Robert Taliercio
ARWPS 29 The Changing Distribution of Public Education
Expenditure in Malawi

February 2002 Samer Al-Samarrai
Hassan Zaman
ARWPS 30 Post-Conflict Recovery in Africa: An Agenda for
the Africa Region

April 2002 Serge Michailof
Markus Kostner
Xavier Devictor
ARWPS 31 Efficiency of Public Expenditure Distribution and
Beyond: A report on Ghanas 2000 Public
Expenditure Tracking Survey in the Sectors of
Primary Health and Education

May 2002 Xiao Ye
S. Canagaraja
ARWPS 33 Addressing Gender Issues in Demobilization and
Reintegration Programs

August 2002 N. de Watteville
ARWPS 34 Putting Welfare on the Map in Madagascar August 2002 Johan A. Mistiaen
Berk Soler
T. Razafimanantena
J. Razafindravonona

ARWPS 35 A Review of the Rural Firewood Market Strategy
in West Africa
August 2002 Gerald Foley
P. Kerkhof, D.
Madougou

38
ARWPS 36 Patterns of Governance in Africa September 2002 Brian D. Levy

ARWPS 37 Obstacles and Opportunities for Senegals
International Competitiveness: Case Studies of the
Peanut Oil, Fishing and Textile Industries

September 2002 Stephen Golub
Ahmadou Aly
Mbaye
ARWPS 38 A Macroeconomic Framework for Poverty
Reduction Strategy Papers : With an Application to
Zambia

October 2002 S. Devarajan
Delfin S. Go

ARWPS 39 The Impact of Cash Budgets on Poverty Reduction
in Zambia: A Case Study of the Conflict between
Well Intentioned Macroeconomic Policy and
Service Delivery to the Poor

November 2002 Hinh T. Dinh
Abebe Adugna
Bernard Myers
ARWPS 40 Decentralization in Africa: A Stocktaking Survey November 2002 Stephen N. Ndegwa

ARWPS 41 An Industry Level Analysis of Manufacturing
Productivity in Senegal

December 2002 Professor A. Mbaye
ARWPS 42 Tanzanias Cotton Sector: Constraints and
Challenges in a Global Environment

December 2002 John Baffes
ARWPS 43 Analyzing Financial and Private Sector Linkages
in Africa
January 2003 Abayomi Alawode

ARWPS 44 Modernizing Africas Agro-Food System:
Analytical Framework and Implications for
Operations
February 2003 Steven Jaffee
Ron Kopicki
Patrick Labaste
Iain Christie

ARWPS 45 Public Expenditure Performance in Rwanda March 2003 Hippolyte Fofack
C. Obidegwu
Robert Ngong

ARWPS 46 Senegal Tourism Sector Study March 2003 Elizabeth Crompton
Iain T. Christie
ARWPS 47 Reforming the Cotton Sector in SSA March 2003 Louis Goreux
John Macrae

ARWPS 48 HIV/AIDS, Human Capital, and Economic Growth
Prospects for Mozambique

April 2003 Channing Arndt
ARWPS 49 Rural and Micro Finance Regulation in Ghana:
Implications for Development and Performance of
the Industry

June 2003 William F. Steel
David O. Andah
ARWPS 50 Microfinance Regulation in Benin: Implications of
the PARMEC LAW for Development and
Performance of the Industry

June 2003 K. Ouattara
ARWPS 51 Microfinance Regulation in Tanzania: Implications
for Development and Performance of the Industry

June 2003 Bikki Randhawa
Joselito Gallardo
ARWPS 52 Regional Integration in Central Africa: Key Issues June 2003 Ali Zafar
Keiko Kubota

39
ARWPS 53 Evaluating Banking Supervision in Africa June 2003 Abayomi Alawode

ARWPS 54 Microfinance Institutions Response in Conflict
Environments: Eritrea- Savings and Micro Credit
Program; West Bank and Gaza Palestine for
Credit and Development; Haiti Micro Credit
National, S.A.

June 2003

Marilyn S. Manalo
AWPS 55 Malawis Tobacco Sector: Standing on One Strong
leg is Better than on None

June 2003 Steven Jaffee
AWPS 56 Tanzanias Coffee Sector: Constraints and

June 2003 John Baffes
AWPS 57 The New Southern AfricanCustoms Union
Agreement
June 2003 Robert Kirk
Matthew Stern

AWPS 58a How Far Did Africas First Generation Trade
Reforms Go? An Intermediate Methodology for
Comparative Analysis of Trade Policies
June 2003 Lawrence Hinkle
A. Herrou-Aragon
Keiko Kubota
AWPS 58b How Far Did Africas First Generation Trade
Reforms Go? An Intermediate Methodology for
Comparative Analysis of Trade Policies
June 2003 Lawrence Hinkle
A. Herrou-Aragon
Keiko Kubota

AWPS 59 Rwanda: The Search for Post-Conflict Socio-
Economic Change, 1995-2001

October 2003 C. Obidegwu
AWPS 60 Linking Farmers to Markets: Exporting Malian
Mangoes to Europe
October 2003 Morgane Danielou
Patrick Labaste
J-M. Voisard

AWPS 61 Evolution of Poverty and Welfare in Ghana in the
1990s: Achievements and Challenges
October 2003 S. Canagarajah
Claus C. Prtner

AWPS 62 Reforming The Cotton Sector in Sub-Saharan
Africa: SECOND EDITION

November 2003 Louis Goreux
AWPS 63
(E)
Republic of Madagascar: Tourism Sector Study November 2003 Iain T. Christie
D. E. Crompton

AWPS 63 (F) Rpublique de Madagascar: Etude du Secteur
Tourisme
November 2003 Iain T. Christie
D. E. Crompton

AWPS 64 Migrant Labor Remittances in Africa: Reducing
Obstacles to Development Contributions

Novembre 2003 Cerstin Sander
Samuel M. Maimbo

AWPS 65 Government Revenues and Expenditures in
Guinea-Bissau: Casualty and Cointegration
January 2004 Francisco G.
Carneiro
Joao R. Faria
Boubacar S. Barry

AWPS 66 How will we know Development Results when we
see them? Building a Results-Based Monitoring
and Evaluation System to Give us the Answer

June 2004 Jody Zall Kusek
Ray C. Rist
Elizabeth M. White
AWPS 67 An Analysis of the Trade Regime in Senegal
(2001) and UEMOAs Common External Trade
June 2004 Alberto Herrou-
Arago
40
Policies Keiko Kubota

AWPS 68 Bottom-Up Administrative Reform: Designing
Indicators for a Local Governance Scorecard in
Nigeria

June 2004 Talib Esmail
Nick Manning
Jana Orac
Galia Schechter

AWPS 69 Tanzanias Tea Sector: Constraints and
Challenges
June 2004 John Baffes

AWPS 70 Tanzanias Cashew Sector: Constraints and

June 2004 Donald Mitchell
AWPS 71 An Analysis of Chiles Trade Regime in 1998 and
2001: A Good Practice Trade Policy Benchmark
July 2004 Francesca Castellani
A. Herrou-Arago
Lawrence E. Hinkle

AWPS 72 Regional Trade Integration inEast Africa: Trade
and Revenue Impacts of the Planned East African
Community Customs Union

August 2004 Lucio Castro
Christiane Kraus
Manuel de la Rocha
AWPS 73 Post-Conflict Peace Building in Africa: The
Challenges of Socio-Economic Recovery and
Development

August 2004 Chukwuma
Obidegwu
AWPS 74 An Analysis of the Trade Regime in Bolivia
in2001: A Trade Policy Benchmark for low Income
Countries

August 2004 Francesca Castellani
Alberto Herrou-
Aragon
Lawrence E. Hinkle
AWPS 75 Remittances to Comoros- Volumes, Trends,
Impact and Implications
October 2004 Vincent da Cruz
Wolfgang Fendler
Adam Schwartzman
AWPS 76 Salient Features of Trade Performance in Eastern
and Southern Africa
October 2004 Fahrettin Yagci
Enrique Aldaz-
Carroll

AWPS 77 Implementing Performance-Based Aid in Africa November 2004 Alan Gelb
Brian Ngo
Xiao Ye

AWPS 78 Poverty Reduction Strategy Papers: Do they matter
for children and Young people made vulnerable by
HIV/AIDS?

December 2004 Rene Bonnel
Miriam Temin
Faith Tempest
AWPS 79 Experience in Scaling up Support to Local
Response in Multi-Country Aids Programs (map)
in Africa

December 2004 Jean Delion
Pia Peeters
Ann Klofkorn
Bloome
AWPS 80 What makes FDI work? A Panel Analysis of the
Growth Effect of FDI in Africa
February 2005 Kevin N. Lumbila
AWPS 81 Earnings Differences between Men and Women in
Rwanda
February 2005 Kene Ezemenari
Rui Wu
AWPS 82 The Medium-Term Expenditure Framework: The
Challenge of Budget Integration in SSA countries
April 2005 Chukwuma
Obidegwu

41
AWPS 83 Rules of Origin and SADC: The Case for change
in the Mid Term Review of the Trade Protocol
June 2005 Paul Brenton
Frank Flatters
Paul Kalenga

AWPS 84 Sexual Minorities, Violence and AIDS in Africa

July 2005 Chukwuemeka
Anyamele
Ronald Lwabaayi
Tuu-Van Nguyen,
and Hans
Binswanger

AWPS 85 Poverty Reducing Potential of Smallholder
Agriculture in Zambia: Opportunities and
Constraints
July 2005 Paul B. Siegel
Jeffrey Alwang

AWPS 86 Infrastructure, Productivity and Urban Dynamics
in Cte dIvoire An empirical analysis and policy
implications

July 2005 Zeljko Bogetic
Issa Sanogo

AWPS 87 Poverty in Mozambique: Unraveling Changes and
Determinants

August 2005 Louise Fox
Elena Bardasi,
Katleen V. Broeck
AWPS 88 Operational Challenges: Community Home Based
Care (CHBC) forPLWHA in Multi-Country
HIV/AIDS Programs (MAP) forSub-Saharan
Africa
August 2005 N. Mohammad
Juliet Gikonyo
AWPS 90 Kenya: Exports Prospects and Problems September 2005 Francis Ng
Alexander Yeats

AWPS 91 Uganda: How Good a Trade Policy Benchmark for
Sub-Saharan-Africa
September 2005 Lawrence E. Hinkle
Albero H. Aragon
Ranga Krishnamani
Elke Kreuzwieser

AWPS 92 Community Driven Development in South Africa,
1990-2004
October 2005 David Everatt Lulu
Gwagwa
AWPS 93 The Rise of Ghanas Pineapple Industry from
Successful take off to Sustainable Expansion
November 2005 Morgane Danielou
Christophe Ravry
AWPS 94 South Africa: Sources and Constraints of Long-
Term Growth, 1970-2000
December 2005 Johannes Fedderke
AWPS 95 South Africas Export Performance: Determinants
of Export supply
December 2005 Lawrence Edwards
Phil Alves
AWPS 96 Industry Concentration in South African
Manufacturing: Trends and Consequences, 1972-
96
December 2005 Gbor Szalontai
Johannes Fedderke
AWPS 97 The Urban Transition in Sub-Saharan Africa:
Implications for Economic Growth and Poverty
Reduction
December 2005 Christine Kessides
AWPS 98 Measuring Intergovernmental Fiscal Performance
in South Africa
Issues in Municipal Grant Monitoring
May 2006 Navin Girishankar
David DeGroot
T.V. Pillay
AWPS 99 Nutrition and Its determinants in Southern Ethiopia
- Findings from the Child Growth
Promotion Baseline Survey

July 2006 Jesper Kuhl
Luc Christiaensen
42
AWPS 100 The Impact of Morbidity and Mortality on
Municipal Human Resources and Service Delivery
September 2006 Zara Sarzin
AWPS 101 Rice Markets in Madagascar in Disarray:
Policy Options for Increased Efficiency and Price
Stabilization

September 2006 Bart Minten
Paul Dorosh
Marie-Hlne Dabat,
Olivier Jenn-Treyer,
John Magnay and
Ziva Razafintsalama

AWPS 102 Riz et Pauvrete a Madagascar Septembre 2006 Bart Minten
AWPS 103 ECOWAS- Fiscal Revenue Implications of the
Prospective Economic Partnership Agreement with
the EU
April 2007 Simplice G. Zouhon-
Bi
Lynge Nielsen

AWPS
104(a)
Development of the Cities of Mali
Challenges and Priorities
June 2007 Catherine Farvacque-
V. Alicia Casalis
Mahine Diop
Christian Eghoff

AWPS
104(b)
Developpement des villes Maliennes
Enjeux et Priorites
June 2007 Catherine Farvacque-
V. Alicia Casalis
Mahine Diop
Christian Eghoff

AWPS 105 Assessing Labor Market Conditions In
Madagascar, 2001-2005
June 2007 David Stifel
Faly H.
Rakotomanana
Elena Celada
AWPS 106 An Evaluation of the Welfare Impact of Higher
Energy Prices in Madagascar
June 2007 Noro Andriamihaja
Giovanni Vecchi
AWPS 107 The Impact of The Real Exchange Rate on
Manufacturing Exports in Benin
November 2007 Mireille Linjouom
AWPS 108 Building Sector concerns into Macroeconomic
Financial Programming: Lessons from Senegal and
Uganda
December 2007 Antonio Estache
Rafael Munoz
AWPS 109 An Accelerating Sustainable, Efficient and
Equitable Land Reform: Case Study of the
Qedusizi/Besters Cluster Project
December 2007 Hans P. Binswanger
Roland Henderson
Zweli Mbhele
Kay Muir-Leresche

AWPS 110 Development of the Cites of Ghana
Challenges, Priorities and Tools
January 2008 Catherine Farvacque-
Vitkovic
Madhu Raghunath
Christian Eghoff
Charles Boakye

AWPS 111 Is Africas Recent Growth Robust? January 2008 Jorge Saba Arbache
John Page
AWPS 112 MADAGASCAR: Labor Markets, the Non-Farm
Economy and Household Livelihood Strategies in
Rural Madagascar
March 2008 David Stifel
Stefano Paternostro
AWPS 113 Growth, Inequallity and Poverty in Madagascar,
2001-2005
April 2007 Nicolas Amendola
Giovanni Vecchi

43

WB21847
C:\Documents and Settings\WB21847\My Documents\Working Paper Series\AWS#113madagascar poverty final.Stefano.4.08.doc
18/04/2008 17:06:00

Growth, Inequality and Poverty in Madagascar, 2001-2005 (World Bank - 2008)

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Growth, Inequality and Poverty in Madagascar, 2001-2005 (World Bank - 2008)

Caricato da

Copyright:

Formati disponibili

Growth, Inequality and Poverty in Madagascar, 2001-2005

Africa Region Working Paper Series No. 111

Potrebbero piacerti anche