Role of growth and inequality in explaining changes in poverty in Pakistan.
Anwar, Talat
Changes in the extent of poverty are affected not only by growth in
the mean income but also by changes in the distribution of income. The
effect of these two factors can be separately measured by decomposing
the total change in poverty. In this context, this paper uses new tools
to quantify relative contribution of growth and inequality using the
latest available household survey data. The findings of this paper
suggest that the role of inequality remained important in mitigating the
adverse effects of growth on poverty during the first period, 1998-99 to
2001-02. Alternatively, the role of growth has been fundamental in
reducing absolute poverty in the second period, 2001-02 to 2004-05.
Poverty would have been further reduced, had the distribution not
worsened during this period. The policy implication is that while
pursuit of growth as a strategy is important for poverty reduction in
Pakistan, the contribution of redistribution in favour of the poor
should not be ignored if the effect of growth on poverty reduction is to
be enhanced. Thus, the major challenge is to pursue a poverty reduction
strategy that is based on growth with redistribution.
JEL classification: I300, I320, O150, O400
Keywords: Poverty, Inequality, Economic Growth, Pakistan
1. INTRODUCTION
Pakistan's economy maintained its growth momentum in fiscal
year 2004-05 with GDP registering its fastest growth rate of 9.0 percent
for the last two decades. The economy has grown at an average rate of
almost 6.9 percent per annum during 2003-04 and 2005-06. This fast
sustained pace of expansion has enabled Pakistan to achieve place among
the fastest growing economies of the Asian region. This growth pattern
is strengthened by vigorous performance in industry, agriculture and
services, together with expansion in domestic demand resulting from as a
new investment cycle. Official poverty estimates demonstrate that this
sectoral confluence of growth contributed to a substantial decline of
10.6 percentage points in absolute poverty in Pakistan, from 34.4
percent in 2001-02 to 23.9 percent in 2004-05 [Pakistan (2006)].
Conversely, the trend in inequality as measured by the Gini coefficient
shows that distribution of consumption expenditure has worsened during
this period. It emerges that while rapid economic growth seems to have
reduced the poverty level, it appears to have increased inequality
during this period. However, in determining the role of growth and
inequality in poverty level changes, the available measures, such as the
Gini coefficient, may not be particularly useful. The paper illustrates
how changes in poverty measures can be decomposed into growth and
distributional effects. The growth component of poverty change measures
how much of the change in poverty is due to the variation in mean
expenditure over time, holding the distribution constant, while the
redistribution component evaluates how much of the variation in poverty
is due to a change in the distribution of expenditure, holding the mean
expenditure constant. The objective of the study is to analyse the role
of growth and inequality in explaining increase or decline in poverty in
Pakistan. To assess the contribution of growth and redistribution three
large household survey data sets PIHS 1998-99, PIHS 2001-02 and PSLM
2004-05 have been used.
The organisation of this report is as follows: Section 2 presents a
review of recent studies on poverty and inequality in Pakistan. Section
3 discusses household data sets and outlines the methodology that has
been used to decompose poverty into growth and redistribution effects.
Section 4 presents results of decomposition of poverty. Finally, Section
5 draws some conclusions from the analysis.
2. REVIEW OF POVERTY AND INCOME DISTRIBUTION IN PAKISTAN
A number of attempts have been made to examine the extent of
poverty and inequality in Pakistan. The earlier work is based on grouped
data of Household Income and Expenditure Survey (HIES) conducted by the
Federal Bureau of Statistics, Government of Pakistan. The unit record
data have, however, been made available on computer data files since the
late 1980s which has enabled analysts/institutions to use this primary
source in estimation work. However, most of these studies examined the
issue of poverty and inequality separately. It is, therefore, important
to review the poverty studies first and then present a review of
inequality studies in the next sub-section.
2.1. Review of Poverty
A review of the existing work on poverty shows that a number of
authors/institutions have made attempts to examine this issue in
Pakistan during the last four decades. The earlier work on measurement
of poverty included Naseem (1973, 1979); Alauddin (1975); Mujahid
(1978); Irfan and Amjad (1984); Kruik and Leeuwen (1985); Malik (1988);
Ahmad and Ludlow (1989); Ercelawn (1990); Malik (1994); Anwar (1996,
1998); Amjad and Kemal (1997); Bhatti, et al. (1999); Jafri (1999);
Arif, et al. (2000). These studies define individuals as poor when their
income is not sufficient to obtain the minimum necessities of life such
as food, clothing, housing etc., for the maintenance of physical
efficiency Most of these studies derived absolute poverty lines in terms
of cost of food requirements consistent with 2550 calories per day per
adult equivalent recommended by Planning Commission (1985). These
studies suggest that poverty declined in the 1970s and 1980s after
witnessing a rise in the late 1960s.
Recent work on poverty includes FBS (2001); World Bank (1995,
2002); Arif (2002); Anwar and Qureshi (2002); Anwar, Qureshi, and Ali
(2004); Planning Commission (2003) and Planning Commission/CRPRID
(2006). Planning Commission in 2002-03 defined and recommended poverty
norm as shortfall from minimum 2350 calories per person per day required
for physical functioning and daily activities. Based on this poverty
norm, Planning Commission notified the estimated official poverty line
at Rs 748 per capita per month in 2001-02 prices. The poverty estimate
implied by the above official poverty line suggests that 32 percent of
the population were poor in Pakistan whereas 38.9 percent and 22.6
percent were poor respectively in rural and urban areas in 2001-02. It
is noteworthy that Anwar and Qureshi (2002) using the lower poverty line
of consumption expenditure of Rs 735 per adult per month in 2001-02
prices estimated a headcount of 35.6 percent for the country as a whole.
The lower official poverty headcount of 32 percent with a higher poverty
line of Rs 748 per capita per month in 2001-02 was mainly due to the
fact that some of the households with lower income were dropped from the
sample before computing poverty headcount. Anwar, Qureshi, and Ali
(2004) used this official poverty line of Rs 748 per capita and
estimated a headcount of 38 percent in 2001-02 as opposed to 32 percent
notified by the Planning Commission. Similarly, World Bank (2005) using
official poverty line of Rs 748 per capita also reported 37 percent in
2001-02. The official poverty estimates were never corroborated from
independent sources by any author or institution. Consequently, official
poverty line and the headcount estimate were revised using CPI adjusted
poverty line of Rs 723 in 2001-02 prices which gave 34.5 percent head
count in 2001-02. The revised poverty estimates were in line and
consistent with the findings of Anwar, Qureshi, and Ali (2004) and World
Bank (2005). Table 1 reports the estimates of various studies.
The general consensus emerging from the review of the literature is
that absolute poverty increased during the 1990s. However, the increase
was more rapid in rural compared to the urban areas. In the subsequent
period, rural poverty deteriorated further while urban poverty remained
(1) constant. The rise in absolute poverty in the 1990s was attributed
mainly to low economic growth of an average of 4 percent per annum in
the 1990s declining from a growth trajectory of around 6 percent per
annum in the 1980s.
However, more recently absolute poverty, as measured by the
official methodology, declined substantially by 10.6 percent from 34.5
percent to 23.9 percent between 2001-02 and 2004-05 (see Table 1). The
decline was more pronounced in rural poverty, from 39.3 percent in
2001-02 to 28.1 percent in 2004-05. It is noteworthy that the economy
witnessed an extraordinary growth rate of at 9.0 percent in 2004-05
which seems to have caused a decline in absolute poverty during this
period. It appears that while low economic growth seems to have
increased poverty in the 1990s, the high economic growth seems to have
reduced absolute poverty in the recent period, 2001-02 and 2004-05.
2.2. Review of Inequality
The work on inequality indicates that a large number of attempts
have been to estimate the extent of income inequality in Pakistan during
the last four decades. Various studies on income distribution include
Bergen (1967), Azfar (1973), Khundkar (1973), Naseem (1973), Alauddin
(1975), Chaudhry (1982), Mahmood (1984), Kruik and Leeuwen (1985), Ahmad
and Ludlow (1989), Malik (1992), Malik (1992), Anwar (1998) and Ahmad
(2000). More recently, FBS (2001), World Bank (2002), Anwar (2003, 2005)
and Planning Commission/CRPRID (2006) have estimated Gini coefficients
for the 1990s. Table 2 reports the Gini coefficient estimated by recent
studies during the 1990s. The Gini coefficients reported by FBS (2001),
World Bank (2002) and Planning Commission (2006) are based on the
consumption expenditure, while those reported by Anwar (2005) are based
on household per capita income. According to both FBS (2001) and World
Bank (2002), consumption inequality increased in Pakistan between
1992-93 to 1998-99. Urban inequality followed the same trend. On the
contrary, rural inequality declined as measured by World Bank (2002) but
increased according to FBS (2001) during this period.
However, the above studies are based on consumption expenditure
that has been used as the proxy for income. It is generally held that
consumption expenditure is more equally distributed than income. In this
context, it would be important to review inequality trends based on
income. Anwar (2005) estimated inequality using the methodology that was
consistent throughout the period of the 1990s. The Gini coefficients
based on income were significantly higher than the one based on
consumption implying that income is more unequally distributed than
consumption expenditure among households (Table 2). The author found a
rapid increase in income inequality during the 1990s. The rise was more
rapid in urban compared to the rural areas. Consequently, income
distribution of 1998-99 turned out to be the most unequal income
distribution in the history of Pakistan. In the later period, income
inequality declined in rural areas but continued to worsen in urban
areas between 1998-99 and 2001-02. More recently, Planning Commission
(2006) shows that consumption distribution has worsened between 2001-02
and 2004-05. To sum up, consumption inequality increased between 1992-93
and 1998-99, then declined between 1998-99 and 2001-02 and finally
worsened in 2004-05. Income inequality more or less followed the same
trend during the above mentioned period. It appears that while
inequality increased during the period of slow growth in the 1990s, the
inequality also worsened during the period of rapid growth in 2000s.
It is now clear that these studies examined the issue of poverty
and inequality separately without linking changes in poverty to the
changes in inequality. An increase or decrease in inequality will be all
that we can conclude from it. We cannot deduce to what extent a change
in inequality contributed (2) to a change in poverty over time. This is
because conventional inequality measures including Gini coefficient are
a poor guide to explain changes in mean income and changes in poverty.
It is, therefore, important to find new tools to examine the
contribution of growth in mean income and changes in inequality in order
to quantify relative contribution of these components to poverty
changes. In this perspective, the study is aimed at decomposing the
poverty changes into growth and redistribution effects using the latest
available tools.
3. DATA AND METHODOLOGY
The analysis of decomposition of poverty in this study is based on
household unit record data of three household surveys conducted by
Federal Bureau of Statistics (FBS), Government of Pakistan, Islamabad.
Two household surveys Pakistan Integrated Economic Survey (PIHS) were
carried out in 1998-99 and 2001-02, while the third survey Pakistan
Social and Living Standard Measurement Survey (PSLM) was conducted in
2004-05. These surveys contain information and data on consumption
expenditure on food and non-food items of each household. While income
is generally under reported to the enumerator, the household consumption
expenditure on nondurables is used as an alternative for 'permanent
income' for the decomposition of poverty in this study.
The universe of these surveys consists of all urban and rural areas
of the four provinces of Pakistan defined as such by the 1998 Population
Census. The primary data files contain population weights, which are
designed to obtain the nationally representative estimates of
population. The sample of PIHS, 1998-99 and 2001-02 respectively
consists of 14,679 and 14,705 households whereas sample of PSLM, 2004-05
consists of 14,706 households both rural and urban in all the four
provinces of Pakistan.
3.1. Measuring Poverty
For this study, the monetary value of household consumption
expenditure in Pakistani rupee is chosen as a welfare measure. The
methodology used by the Government of Pakistan in the derivation of the
official poverty estimates has been used in this study. The official
method prefers consumption expenditure on non-durables over income to
estimate poverty. Inevitably, consumption is a preferred measure of
wellbeing in developing countries for various reasons. First,
consumption is a better indicator of the person's welfare because
it is more closely related to the well-being than income. Second,
consumption can be measured more precisely than income due to the
widespread practice of tax evasion in developing countries. Finally,
consumption is less volatile compared to income, and can be a better
indicator of a household's actual standard of living.
Following the official method of measuring poverty, the measure of
well-being used in this study is corrected for spatial and temporal
prices. The study has adjusted household consumption expenditure for
economies of scale by using 1 for adult and 0.8 for children 0-18 years.
The regional price index has been used to take an account of regional
differences between rural and urban areas. In general, the poverty line
is a cutoff point and individuals with consumptions below this value are
considered as poor. The official poverty line adjusted for inflation has
been used to estimate the incidence of poverty in Pakistan over time.
The official poverty line is defined by the government as the cost of
buying a diet of 2350 calories per capita per day plus non-food
expenditures to satisfy subsistence needs. The Planning Commission
notified the estimated official poverty line at Rs 673 per capita per
month in 1998-99 prices which has been adjusted for inflation by the
consumer price index (CPI) for the survey period between 1998-99 and
2001-02 and between 2001-02 and 2004-05.
In poverty literature, a number of measures of poverty have been
proposed. Among these scores of poverty measures the following three
measures are commonly used and these measures belong to a class of
poverty measures popularised by Foster, Greer, and Thorbecke (1984).
[P.sub.[alpha]] = 1/n [q.summation over (i=1)]
[[(Z-[y.sub.i]/Z].sup.[alpha] ... ... ... ... ... (1)
These measures have clear advantages for evaluating policies which
aim at reaching the poorest. Note that if [alpha] = 0, the FGT index,
[P.sub.[alpha]] = Headcount measure, if [alpha] = 1, [P.sub.[alpha]] =
Poverty gap index or quotient and if [alpha] = 2, [P.sub.[alpha]] is the
mean of squared proportionate poverty gaps and indicates greater
severity of poverty among the poorest. The study uses the three poverty
measures--poverty headcount, poverty gap and poverty severity to
decompose the changes in these measures into growth and distribution
components.
3.2. Poverty Decomposition Methodology
To decompose the changes in poverty into growth effect and
inequality effect, the Datt and Ravallion (1992) methodology has been
used. For this purpose, the focus is on poverty measures which can be
fully characterised in terms of the poverty line, the mean income of the
distribution, and the Lorenz curve representing the structure of
relative income inequalities. The poverty measure [P.sub.t] at date t is
written as
[P.sub.t]== (z/[[mu].sub.t], [L.sub.t]) ... ... ... ... ... ... ...
(2)
Where P is a poverty measure written as a function of the ratio of
the mean consumption [[micro].sub.t], to the poverty line z and the
parameters of the Lorenz curve [L.sub.t] at t date. Homogeneity in z and
p is a common property of poverty measures. The level of poverty may
change due to a change in the mean income [[mu].sub.t] relative to the
poverty line, or due to a change in relative inequalities [L.sub.t].
Lorenz curves may be presumed to follow a particular parametric form and
fit to the data. The parametric specification of the Lorenz curve is
given in Appendix.
For any two dates 0 and 1, the growth component of a change in the
poverty measure is defined as the change in poverty due to a change in
the mean from [[mu].sub.0] to [[mu].sub.1] while holding the Lorenz
curve constant at reference level [L.sub.r]. The redistribution
component is defined as the change in poverty due to a change in the
Lorenz curve, while holding the mean income constant at the reference
level [[micro].sub.r]. Hence, a change in poverty over dates t and t+n
(say) can be decomposed as follows:
[P.sub.t+n] - [P.sub.t] = G(t,t+n;r) + D(t,t+n;r) + R(t,t+n;r) ...
... ... (3)
Thus, total change in poverty
=Growth effect + Redistribution effect + Residual
In the above formulation the growth and redistribution components
are given by
G(t,t+n;r) = P(z]/ [[mu].sub.tn], [L.sub.r]) - P(z/[[mu].sub.t],
[L.sub.r])
G(t,t+n;r) = P(z]/ [[mu].sub.r], [L.sub.t+n]) - P(z/[[mu].sub.r],
[L.sub.t])
Where R(t,t+n;r) in (3) stand for the residual. The residual in (3)
exists whenever the poverty measure is not additively separable between
P and L, specifically, whenever the marginal effects on the poverty
index of changes in the mean (Lorenz curve) depend on the precise Lorenz
curve (mean). In general, the residual does not vanish. Nor can it be
apportioned between the growth and redistribution components, as some
recent attempts at poverty decomposition have done. In general, the
residual would not vanish. It can vanish only if the mean income or the
Lorenz curve remains unchanged over the decomposition period. This is
very unlikely for most of the empirical works.
The remainder of the study is divided into two interrelated
sections. The first part investigates the changes in poverty and
inequality, while the second section of the study decomposes changes in
poverty into growth and distribution effects of poverty.
4. CHANGES IN POVERTY: 1998-99 TO 2004-05
To estimate the poverty in 1998-99, official poverty line in
1998-99 prices notified by the Planning Commission has been used.
Poverty estimates for 2001-02 and 2004-05 have been computed by
adjusting the official poverty for inflation using CPI during this
period. The direction of change in poverty is then examined by looking
at differences in poverty estimates during this period. Table 3 reports
estimates of poverty in Pakistan for 1998-99 and 2001-02. The results
show that poverty in Pakistan initially increased from 30.6 percent in
1998-99 to 34.4 percent in 2001-02 and then declined rapidly to 23.9
percent in 2004-05. The direction of change in poverty at the regional
level shows that in absolute terms rural poverty fluctuated more than
the urban poverty, but this partly reflects the higher base level of
rural poverty. However, the relative decline was much larger in urban
areas during the above period.
While rural poverty increased substantially from 34.7 percent (3)
in 1998-99 to 39.3 percent in 2001-02 and then declined to 28.1 in
2004-05, the urban poverty increased marginally in 2001-02 and then
declined substantially from 22.7 percent to 14.9 percent during the
above period. Both poverty gap, [P.sub.1] and poverty severity measures
FGT [P.sub.2] indicate similar trends during this period. The results
relating to inequality are quite contrary to those relating to poverty.
In contrast to poverty, inequality in Pakistan initially declined
between 1998-99 and 2001-02 and then increased between 2001-02 and
2004-05.
These results are quite consistent with the macroeconomic trends in
the country during this period. While low economic growth due to drought
seems to have increased rural poverty during the first period, the
exceptional growth seems to have resulted in a rapid decline in poverty
in the country during the second period. In contrast to this, low growth
seems to have resulted in lower level of inequality during the first
period, whereas 'high growth seems to have resulted in higher level
of inequality during the second period.
5. DECOMPOSITION OF POVERTY, 1998-99 AND 2004-05
Table 4 presents the contributions of growth and redistribution to
changes in poverty using the headcount, poverty gap and FGT [P.sub.2],
poverty severity measures. The table shows that during the first period,
1998-99 to 2001-02, growth component is positive implying that growth
component contributed to the rise in poverty over this period. On the
other hand, negative redistribution suggests that had the inequality not
declined, the increase in poverty would have been much higher. By
components in terms of poverty headcount, growth accounted for 5.6
percentage points for the rise in poverty, while distributional shift
accounted for 2.05 percentage points for lessening the negative effect
of growth on poverty (see Figure 1). Consequently, the total rise in
poverty was 3.83 percent in 2001-02. Notably, the adverse growth
component significantly dominates the favourable redistribution
component. This implies that a decline in per capita household
consumption due to drought, particularly in rural areas, contributed to
an increase in poverty during this period. This is also supported by the
changes at regional levels as the adverse growth effect was higher in
rural compared with urban areas. Similarly, a favourable redistribution
effect was also higher in rural compared to urban areas, otherwise the
rise in rural poverty would have been higher during the period.
In the second period, growth component is negative and
redistribution component is positive in all regions implying that though
growth contributed to a decline in poverty, the redistribution component
dampened the effects of growth on poverty reduction over this period.
While growth accounted for 12.48 percentage points in poverty reduction,
the distributional shift adversely accounted for 1.42 percentage points
and reduced the positive effects of growth on poverty. This implies that
poverty would have been further reduced in 2004-05, had the government
taken adequate measures to cheek deterioration of distribution.. At
regional level, growth and redistribution components were different in
relative terms. The growth component was more dominant in rural compared
to urban areas. Growth accounted for 14.29 percentage points reduction
in poverty in rural areas, whereas it accounted for 8.06 percentage
points reduction in urban poverty. On the other hand, distributional
shift accounted for adversely 2.2 percentage points in rural areas
compared to 1.18 percentage points in urban areas. Thus, poverty would
have been further reduced in urban areas, had the distribution not
worsened over the period.
The residual in the decomposition is relatively small for Pakistan
compared with other countries particularly with India and Brazil
computed by Datt and Ravallion (1992). The residual is the difference
between the growth (redistribution) component measured at the terminal
and initial Lorenz curve (mean consumption) respectively. If the mean
consumption or the Lorenz curve remains unchanged over the decomposition
period, then the residual disappears.
Table 4 also reports the decomposition components of growth and
redistribution over the whole period, 1998-99 to 2004-05. The growth
effects remained dominant over the period as a whole. Redistribution
also contributed to decline in poverty but decline was pronounced in
urban areas. The growth component accounted for 6.82 percentage points
in poverty reduction, whereas the distributional shift accounted for
0.63 percentage points for the country as a whole. Consequently, total
reduction in poverty was 6.73 percent between 1998-99 and 2004-05. One
important thing also emerged from the results. The growth component
remained dominant not only in two different periods but also over the
period as a whole, though the shift in distribution in opposite
direction was significant in the recent period with dominance of
redistribution component in rural areas.
[FIGURE 1 OMITTED]
Decomposition results relating to poverty gap ([P.sub.1]) and
poverty severity measure FGT ([P.sub.2]), follow the same pattern, if
examined separately during the two periods. However, during the period
as a whole, 1998-99 to 2004-05, these two measures show an adverse
redistribution effect in rural areas which was not captured by
conventional inequality measures such as Gini coefficient. For example,
Gini coefficient shows a decline in rural inequality from 0.2521 in
1998-99 to 0.2519 in 2004-05 (see Table 3). In contrast, both poverty
gap and FGT [P.sub.2], poverty severity measures demonstrate an adverse
effect of inequality on poverty in rural areas indicating that
conventional inequality measures may be a poor guide to the way
distribution shift can affect poverty measures (see Table 4).
[FIGURE 2 OMITTED]
5.1. Growth Incidence Curve
The growth incidence curve (GIC) is a useful tool to analyse the
impact of economic growth over a wide range of distribution. The growth
incidence curve shows the growth rate in income or consumption between
two points in time at each distribution percentile. It would, therefore,
be interesting to look at the growth incidence curve to evaluate the
impact of economic growth over the two periods. The household
consumption expenditure on non-durables has been used to examine the
growth incidence curve, while CPI has been used to convert the nominal
consumption expenditure into real values. Figures 2 and 3 show the
growth incidence curve for both urban and rural regions for the first
period, 1998-99 to 2001-02. The CICs, broadly, are downward sloping
which shows a decreasing level of consumption expenditure over this
period. However, consumption declined at faster rates in higher
percentile groups compared to lower percentile groups resulting in lower
inequality over this period.
[FIGURE 3 OMITTED]
Further, GIC for both urban and rural areas lies mostly below zero
implying that absolute poverty has increased over this period with
reference to some conceivable poverty lines including the official.
However, the growth rate was positive for the poorest about 5 percent of
the population. But against a very low poverty line, the poorest 5
percent of population did not suffer any deterioration between 1998-99
and 2001-02.
[FIGURE 4 OMITTED]
In contrast to this growth incidence curve in the second period,
the period between 2001-02 to 2004-05 shows entirely different trends in
urban as well as rural areas. Growth in consumption of the richest
quintile was the highest and the lowest for the poorest quintile,
particularly in urban areas. This suggests an increased gap between the
rich and the poor particularly in urban areas over the period (see
Figures 4 and 5). Nevertheless, the growth in consumption of the lower
deciles was sufficient to reduce absolute poverty. It appears that while
growth seems to have contributed to a decline in absolute poverty in
rural and urban regions, it seems to have increased the gap between the
rich and the poor over the period. The GIC for rural areas needs
particular attention as it lies below zero in the range of 1 percent of
population implying that while absolute poverty has fallen over this
period for some conceivable poverty lines, such as the official poverty
line, the living standard of the poorest 1 percent had not improved. If
one draws a very low poverty line in stringent terms, absolute poverty
increased amongst the poorest 1.0 of population between 2001-02 and
2004-05 which the poverty headcount measures did not capture.
[FIGURE 5 OMITTED]
5.2. Comparison with Earlier Studies
The decomposition of changes of poverty into growth and
redistribution components has not received adequate attention in
Pakistan. The World Bank (2002) is the exception that made an attempt to
decompose poverty using Datt and Ravallion's (1992) methodology for
1990-91 to 1998-99. According to World Bank (2002), decline in urban
poverty was entirely due to the growth component and redistribution had
had a negative effect. On the contrary, reduction in inequality in the
absence of growth in consumption resulted in a small reduction in rural
poverty during 1990-91 to 1998-99.
The finding of the World Bank study for the recent period is
different from that for the earlier period of the 1990s (2002) in
contrast to which growth contributed to a significant decline in poverty
in both rural and urban areas over the whole period, 1998-99 to 2004-05.
The growth components remained dominant both in urban and rural areas.
The decline in poverty was mainly attributable to consumption while
redistribution had adverse effect on poverty over the whole period.
However, adverse distributional changes became more important in the
recent period, 2001-02 to 2004-05 compared to the earlier era of low
economic growth (1990-91 to 199899). This is in line with the view that
rapid economic growth in recent time has benefited the middle and upper
income classes more than the poorest segments of the population.
6. CONCLUSION
The study analysed the decomposition of changes in Pakistan's
poverty profile covering rural and urban regions during 1999-98 to
2001-02 and 2001-02 to 2004-05. The first period, 1999-98 to 2001-02
relates to a low growth period mainly due to drought in the country. The
second period, 2001-02 to 2004-05 relates to a period dominated by a
growth-oriented poverty reduction strategy. The main conclusions that
emerge from the analysis may be stated as follows.
During the first period, 1998-99 to 2001-02, the dominant growth
component contributed adversely to the rise in poverty over this period.
This is a low growth period characterised by drought that contributed to
increase in poverty, particularly in rural areas. However, there are
agriculture and industry linkages that seem to have affected the growth
of consumption expenditure adversely leading to a rise in poverty in
urban areas over this period. On the other hand, the redistribution
component affected the poverty situation favourably otherwise the rise
in poverty would have been much higher over the period. Thus, the role
of inequality remained important in mitigating the adverse effects of
growth on poverty between 1998-99 and 2001-02.
In contrast to this, the dominant growth component contributed
significantly to decline in poverty, whereas the redistribution effects
in the opposite direction depressed the effects of growth on poverty
reduction during the second period. This suggests that poverty would
have been further reduced in 2004-05, if the government had taken
adequate measures to improve distribution. While the growth component
remained dominant across rural and urban areas, the adverse
redistribution component remained relatively large in rural areas. In
sum, the role of growth has been more important in explaining changes in
poverty in the recent period.
Over the period as a whole, from 1998-99 to 2004-05, while the
effects of growth remained dominant, the redistribution component seems
to have benefited only the urban areas. On the other hand,
redistribution seems to have adversely affected the poor in rural areas.
The policy implication of this empirical finding is that while the
pursuit of growth as a strategy is important for poverty reduction in
Pakistan, the contribution of redistribution in favour of the poor
should not be ignored in order to enhance the effect of growth on
poverty reduction. In this context, the major challenge is to pursue a
poverty reduction strategy that is based on growth with redistribution.
To meet this challenge, the government can enforce tax regimes and
pursue expenditure policies that redistribute incomes from the rich to
the poor while backward regions and provinces get preference in
development programmes.
Analysis based on Growth Incidence Curve also supported the
findings of the decomposition exercise performed in this study. The
Growth Incidence Curve highlighted the role of inequality in the first
period and that of growth in the second period in explaining the changes
in absolute poverty. The use of GIC unveiled the rise in poverty amongst
the poorest 1.0 of population in rural areas between 2001-02 and 2004-05
which the conventional poverty measures failed to register. The rise in
poverty amongst poorest 1.0 of population indicates their social
exclusion from the process of growth. Targeted policy measures such as
income or food support are required to protect this segment of
population.
Appendix
A BRIEF NOTE ON PARAMETRIC SPECIFICATION OF THE LORENZ CURVE
The Lorenz curve is a method for representing the distribution of
income. It is created by plotting cumulative income shares against
cumulative population shares and forms the foundation of several
inequality measures including the popular Gini coefficient. Lorenz
curves may be constructed from grouped data using interpolation
techniques or may be presumed to follow a particular parametric form and
fit to tabulated data.
A Lorenz curve may be defined as
[eta] = f([eta])
Where
[pi] is the cumulative population share of persons earning income
equal to or below income level x.
[eta] is the cumulative income share of population subgroup [pi].
A Lorenz curve must have the following properties:
D[eta]/d[pi]>0, [D.sup.2][eta]/d[[pi].sup.2]>0,
[eta](0)=0, [eta](1)=1
and is defined on the domain 0 [less than or equal to] [pi] [less
than or equal to] 1
Author's Note: Comments on an earlier draft of the paper by Dr
G. M. Arif, Pakistan Institute of Development Economics, lslamabad and
Safdar Pervaz, Asian Development Bank, Islamabad are gratefully
acknowledged. The views expressed are those of the author and do not
necessarily reflect those of the State Bank of Pakistan, Asian
Development Bank, and Canadian International Development Agency.
REFERENCES
Abroad, E. and S. Ludlow (1989) Poverty, Inequality and Growth in
Pakistan. The Pakistan Development Review 28:4, 813-850.
Ahmad, Mahboob (2000) Estimation and Distribution of Income in
Pakistan, Using Micro Data. The Pakistan Development Review 39:4,
807-824.
Alauddin, Talat (1975) Mass Poverty in Pakistan: A Further Study.
The Pakistan Development Review 14:4, 577-602.
Amjad, R. and A. R. Kemal (1997) Macroeconomic Policies and their
Impact on Poverty Alleviation in Pakistan. The Pakistan Development
Review 36:1, 39-68.
Amjad, R. and M. Irfan (1984) Poverty in Rural Pakistan. In
ILO-ARTEP (1984) (eds.) Impact of Return Migration on Domestic
Employment in Pakistan: A Preliminary Analysis.
Anwar, Talat (1996) Structural Adjustment and Poverty: The Case of
Pakistan. The Pakistan Development Review 35:4, 911-926.
Anwar, Talat (1998) Absolute Poverty in Pakistan: Evidence and
Alleviating Strategy. Pakistan Academy for Rural Development, Peshawar.
Journal of Rural Development and Administration 2.
Anwar, Talat (2003) Trends in Income Inequality in Pakistan between
1998-99 and 2001. The Pakistan Development Review 42:4, 809-821.
Anwar, Talat (2005) Long-term Changes in Income Distribution in
Pakistan: Evidence Based on Consistent Series of Estimates. CRPRID,
Islamabad. (Discussion Paper No. 3).
Anwar, Talat and Sarfraz Khan Qureshi (2002) Trends in Absolute
Poverty in Pakistan: 1990-2001. The Pakistan Development Review 41:4,
859-878.
Anwar, Talat, Sarfraz Khan Qureshi, and Hammad Ali (2004)
Landlessness and Rural Poverty in Pakistan. The Pakistan Development
Review 43:4, 855-874.
Arif, G. M. (2002) Measuring Poverty in Pakistan: A Critical Review
of Recent Poverty Lines. In Human Condition Report 2002. Centre for
Research on Poverty Reduction and Income Distribution (CRPRID),
Islamabad.
Arif, G. M., Hina Nazli, and Rashida Haq (2000) Rural
Non-agriculture Employment and Poverty in Pakistan. The Pakistan
Development Review 39:4, 1089-1110.
Asian Development Bank (2002) Poverty in Pakistan, Issues, Causes
and Institutional Responses. Pakistan Resident Mission, Islamabad.
Azfar, J. (1973) The Distribution of Income in Pakistan: 1966-67.
Pakistan Economic and Social Review 11, 40-66.
Bergan, A. (1967) Personal Income Distribution and Personal Savings
in Pakistan, 1963-64. The Pakistan Development Review 7:2, 160-212.
Bhatti, M. A., R. Haq, and T. Javed (1999) A Sectoral Analysis of
Poverty in Pakistan. The Pakistan Development Review 38:4, 859-872.
Chaudhry, M. G. (1982) Green Revolution and Redistribution of Rural
Incomes: Pakistan's Experience. The Pakistan Development Review
21:3, 173-205.
Cheema, I. A. (2005) A Poverty Profile in Pakistan. CRPRID,
Islamabad.
Datt, G. and M. Ravallion (1992) Growth and Redistribution
Components of Changes in Poverty Measures. A Decomposition with
Applications to Brazil and India in the 1980s. Journal of Development
Economics 38, 275-295.
Ercelawn, Aly (1990) Absolute Poverty in Pakistan. Applied
Economics Research Centre, Karachi. (Mimeographed).
Federal Bureau of Statistics (2001) Poverty ill the 1990s.
Islamabad: Statistics Division.
Federal Bureau of Statistics (Various Issues) Pakistan Integrated
Economic Survey (PIHS), 1998-99 and 2001-02, and Pakistan Social and
Living Standard Measurement Survey (PSLSM), 2004-05. Islamabad.
Foster, James, J. Greer, and E. Thorbecke (1984) A Class of
Decomposable Poverty Measures. Econometrica 52:761-765.
Irfan and Amjad (1984) Impact of Return Migration on Domestic
Employment in Pakistan: A Preliminary Analysis. ILO-ARTEP.
Jafri, S. M. Y. (1999) Assessing Poverty in Pakistan. In A Poverty
Profile of Poverty in Pakistan. Mahbub ul Haq Centre for Human
Development, Islamabad.
Khandkar, R. (1973) Distribution of Income and Wealth in Pakistan.
Pakistan Economic and Social Review 11:1, 1-39.
Kruik and Leeuwen (1985) Changes in Poverty in Pakistan. The
Pakistan Development Review 24.
Mahmood, Zafar (1984) Income Inequality in Pakistan: An Analysis of
Existing Evidence. The Pakistan Development Review 23:2-3, 365-376.
Malik, Mohammad Hussain (1988) Some New Evidence of Poverty in
Pakistan. The Pakistan Development Review 27:4, 509-16.
Malik, Shahnawaz (1992) Rural Poverty in Pakistan. Unpublished
Ph.D. Thesis submitted to the University of Sussex.
Malik, Sohail J. (1992) Chapter on Poverty in Pakistan. In M.
Lipton and J. Van deer Gaag (eds.) Including the Poor. Washington, DC:
The World Bank.
Malik, Sohalil, J. (1994) Poverty in Pakistan 1984-85, 1987-88 and
1990-91. IFPRI, Washington, DC. (Mimeographed.)
Mujahid, G. B. (1978) Note on Measurement of Poverty and Income
Inequalities in Pakistan: Some Observations of Methodology. The Pakistan
Development Review 17:3, 365-377.
Naseem, S. M. (1973) Mass Poverty in Pakistan: Some Preliminary
Findings. The Pakistan Development Review 12: 4, 312-360.
Naseem, S. M. (1979) Underdevelopment, Poverty and Inequality in
Pakistan. Islamabad: Vanguard.
Pakistan, Government of (2006) Economic Survey, 2004-05. Islamabad:
Ministry of Finance.
Planning Commission (1985) Energy and Protein Requirements. Report
of a joint FAO/WHO Ad-hoc Experts Committee, Geneva, Islamabad.
Planning Commission (2003) Poverty in Pakistan Based on the
Officially Notified National Poverty Line, Reported in Full-poverty
Reduction Strategy Paper. Finance Division, Islamabad.
Planning Commission/CRPRID (2006) Poverty chapter published in
Economic Survey, 2004-05. Ministry of Finance, Islamabad.
World Bank (2002) Poverty in Pakistan: Vulnerabilities, Social
Gaps, and Rural Dynamics. Washington, DC.
World Bank (2005) Pakistan Poverty Update, 2005. Washington, DC.
(1) anwar and Qureshi (2002), Anwar, Qureshi, and Ali (2004) and
Cheema (2005) have also arrived more or less at the same conclusion.
(2) While Bhatti, et al. (1999) examined the relative contribution
of various sectors to aggregate poverty, the author did not decompose
poverty into inequality and growth components.
(3) For 1998-99, poverty estimates are based on Rs 670 per capita
month. However, if Rs 673 per capita per month is used as poverty line,
it gives 31 percent below poverty line in 1998-99 as opposed to 30.6
percent in 1998-99 which was notified officially.
Talat Anwar <
[email protected]> was Senior Joint
Director, Research Department, State Bank of Pakistan, Karachi and
currently working as Economic Advisor, Canadian International
Development Agency, Programme Support Unit, Islamabad.
Table 1
GDP Growth Rates and Headcount Measure for Pakistan-1990-91 to 2004-05
World Bank Planning
(2002) Commission
FBS (2001) Urban Rs 767 (2003)
Rs 682 Rural Rs 680 Rs 748
in 1998-99 in 1998-99 in 2001-02
Poverty Prices Prices Prices
Lines GDP Headcount Measure
Growth (% Overall below Poverty Line)
Years Rates %
Overall
1992-93 2.1 26.6 25.7 -- --
1993-94 4.4 29.3 28.6 -- --
1998-99 4.2 32.2 32.6 30.6
2001-02 3.1 -- -- 32.1
2004-05 8.6 -- -- --
Rural
1992-93 -- 29.9 27.7 --
1993-94 -- 34.7 33.4 --
1998-99 -- 36.3 35.4 34.6
2001-02 -- -- -- 38.9
2004-05 -- -- -- --
Urban
1992-93 -- 20.7 20.8 --
1993-94 -- 16.3 17.2 --
1998-99 -- 22.4 24.2 20.9
2001-02 -- -- -- 22.6
2004-05 -- -- -- --
Anwar and Planning
Qureshi Commission/
(2003) CRPRID (2006)
Rs 735 Rs 723
in 2001-02 in 2001-02
Poverty Prices Prices
Lines
Years
1992-93 -- 25.5
1993-94 -- 28.2
1998-99 30.4 31.1
2001-02 35.6 34.5
2004-05 -- 23.9
1992-93 -- 27.6
1993-94 -- 33.5
1998-99 32.1 35.1
2001-02 41.0 39.3 *
2004-05 -- 28.1 *
1992-93 -- 19.9
1993-94 -- 15.4
1998-99 26.39 21.4
2001-02 26.47 22.7 *
2004-05 -- 14.9 *
Source: Various studies cited above.
* Planning Commission /CRPRID (2006), based on inflation (CPI)
adjusted official poverty line of Rs 723 in 2001-02 and Rs 878.64
in 2004-05.
(1) Anwar and Qureshi (2002), Anwar, Qureshi, and Ali (2004) and
Cheema (2005) have also arrived more or less at the same
conclusion.
Table 2
Gini Coefficient for Pakistan by Regions-1990-91 to 2004-05
Planning
FBS World Bank Commission/ Anwar
Years (2001) (2002) CRPRID (2006) (2005) *
Overall
1992-93 0.2680 0.276 0.3937
1993-94 0.2709 0.276 0.3864
1998-99 0.3019 0.296 0.4187
2001-02 -- -- .2752 0.4129
2004-05 -- -- .2976
Rural
1992-93 0.2389 0.252 0.3668
1993-94 0.2345 0.246 0.3647
1998-99 0.2521 0.251 0.3796
2001-02 .2367 0.3762
2004-05 .2519
Urban
1992-93 0.3170 0.316 0.3970
1993-94 0.3070 0.302 0.3685
1998-99 0.3596 0.353 0.4510
2001-02 .3227 0.4615
2004-05 .3388
Source: Various studies cited above.
* Based on household per capita income.
Table 3
Trends in Poverty Incidence, Intensity, Severity and Gini
Coefficient, 1998-99 to 2004-05 in Pakistan
Headcount FGT Poverty Gap
([P.sub.O]) Index ([P.sub.1])
Regions 1998-99 2001-02 2004-05 1998-99 2001-02
Pakistan
Overall 30.6 * 34.4 23.9 6.5 * 6.9
Rural 34.7 * 39.2 28.1 7.4 * 8.0
Urban 20.9 * 22.7 14.9 4.2 * 4.5
Pakistan Gini Coefficient
Overall 0.3019 0.2752 0.2976 -- --
Rural 0.2521 0.2367 0.2519 -- --
Urban 0.3596 0.3227 0.3388 -- --
FGT
Poverty
Gap
Index FGT Index
([P.sub.1]) ([P.sub.2])
Regions 2004-05 1998-99 2001-02 2004-05
Pakistan
Overall 4.8 2.0 * 2.1 1.5
Rural 5.6 2.3 * 2.4 1.8
Urban 2.9 1.3 * 1.3 0.8
Pakistan
Overall -- -- -- --
Rural -- -- -- --
Urban -- -- -- --
Source: Calculations are based on primary data of PIHS 1998-99,
2001-02 and 2004-05, Federal Bureau of Statistics, Government of
Pakistan, Islamabad.
Note: All poverty indices are expressed as percentages. * Based on
poverty line of Rs 670 per capita month.
Table 4
Decomposition of Poverty for Pakistan by Regions between
2001-02 to 2004-05 mid 1998-99 to 2001-02
Total
Change in
Growth Redistribution Residual Poverty
Index
Headcount Index (H)
1998-99 to 2001-02
Pakistan 5.66 -2.05 0.22 3.83
Urban 4.58 -1.82 -0.99 1.77
Rural 6.12 -2.23 0.7 4.59
2001-02 to 2004-05
Pakistan -12.48 1.42 0.5 -10.56
Urban -8.06 1.18 -0.91 -7.79
Rural -14.29 2.2 0.93 -11.16
1998-99 to 2004-05
Pakistan -6.82 -0.63 -0.72 -6.73
Urban -3.48 -0.64 1.9 -6.02
Rural -8.17 -0.03 -1.63 -6.57
Poverty Gap index (PG)
1998-99 to 2001-02
Pakistan 1.59 -1.00 0.05 0.59
Urban 1.14 -0.73 0.12 0.41
Rural 1.79 -1.14 0.03 0.65
2001-02 to 2004-05
Pakistan -3.04 0.95 0.14 -2.09
Urban -2.05 0.46 0.02 -1.58
Rural -3.45 1.32 0.05 -2.13
1998-99 to 2004-05
Pakistan -1.45 -0.05 0.19 -1.50
Urban -0.91 -0.26 0.14 -1.17
Rural -1.67 0.19 0.08 -1.48
FGT Index ([P.sub.2])
1998-99 to 2001-02
Pakistan 0.60 -0.44 0.07 0.16
Urban 0.40 -0.30 0.05 0.10
Rural 0.68 -0.50 0.08 0.18
2001-02 to 2004-05
Pakistan -1.02 0.51 0.12 -0.51
Urban -0.66 0.26 0.06 -0.40
Rural -1.17 0.68 0.05 -0.49
1998-99 to 2004-05
Pakistan -0.42 0.07 0.19 -0.35
Urban -0.26 -0.05 0.11 -0.30
Rural -0.49 0.18 0.13 -0.31
Source: Author's computation from PIHS, 1998-99, 2001-04 and
PSLM, 2004-05.
