The News

Monday, July 10, 2006

Poverty in Pakistan
John Wall

Poverty is an ethical concept, not a statistical one. Inherent in the term “poverty”,
when applied to human beings, is the notion of a life situation that should not exist.
It is not only lack of roti, kapra aur makan—food, cloth and shelter. Amartya Sen
aptly sums up many dimensions of poverty as lack of “capability”—capability to
overcome violence, hunger, ignorance, illness, physical hardship, injustice and
voicelessness. The World Bank has argued that poverty often lies in the absence of
opportunity, empowerment and security, and not just the absence of food on the
table.

Still, there is a hunger to have a statistic that sums up poverty, something handy
both for analysis and for comparison across groups and among time periods. For this
reason there are a variety of statistical measures of poverty. None of them do a very
good job of capturing the multi-dimensional concepts of poverty discussed above. All
require severe conceptual compromises to make them comparable either across
groups or time. Nevertheless, they provide whatever is available to monitor poverty
reduction in a consistent manner.

Comparing poverty in the same country at different periods of time raises many
difficulties. One is that, often, household income and expenditure surveys change
their methods (wording of questions, sampling method, interview technique) that
make them difficult to compare. If the poverty line used changes between periods,
the results are incomparable. An inescapable problem is that even with no
differences in the surveys themselves and in the poverty line, adjusting for prices
changes between periods to make poverty lines comparable bristles with difficulties.
This is particularly so in Pakistan, where there are serious flaws with the two price
indices available.

One is the consumer-price index (CPI), which deals with many commodities of
consumption but covers only urban areas; it does not capture price changes in rural
areas where the bulk of the poor live. The other, the Survey Based Index (SBI) of
prices, has the advantage of being collected at the same time from the same
households as the consumption data. Unfortunately, it is difficult to estimate price

1
The News
Monday, July 10, 2006

changes for all goods that households consume because prices of non-food items are
not available in the survey. These are the two choices. Pakistan should improve its
prices indices with a sense of urgency for many reasons.

A comparable survey using the same poverty line in 1998-99, 2000-01 and 2004-05,
adjusted by both the CPI and SPI revealed the following: According to both
measures, poverty headcount had been rising throughout the 1990s and peaked in
2000-01, a bad drought year. It then fell sharply in 2004-05, a very good
agricultural crop year. Under the CPI, poverty headcount dropped by 10.6 per cent,
under the SBI it dropped five per cent. Those two estimates probably capture the
extremes. A less flawed price index might well find a third estimate in between these
extremes.

How can poverty drop so sharply in just four years? How can the same data yield
such different results due to different estimates of the same thing — price changes?

The answer is that incomes of a very large portion of the population are just above
and just below the official poverty line. Compared to 2000-01, the consumption
distribution has improved substantially in 2004-05, meaning almost all families are
better off. The fact that there is an enormous clustering of population around the
poverty line means that even small changes in consumption or income can affect
poverty headcount ratios dramatically.

This is the case in Pakistan, where the improvement in consumption distribution in

2004-05 has led to a sizeable decline in poverty. It also explains why a difference in
the estimated rate of price inflation between 2000-01 and 2004-05 of about eight
per cent (depending on whether we use CPI or SBI) makes such a huge difference in
the estimated fall in poverty; i.e., over 10 per cent in one case and five per cent in
the other.

This clustering of Pakistan’s population just above and just below the poverty line
also implies that families are quite vulnerable to falling into poverty with the slightest
run of bad luck. A drought or bad agricultural year, an illness of a breadwinner, rises
in prices of basic commodities not compensated by rises in income—all of these can
cause families to fall into poverty.

2
The News
Monday, July 10, 2006

Sustaining a declining poverty trend requires a concerted effort to improve the

capabilities of the poor and vulnerable. It also requires well-designed programmes
that help to mitigate the vulnerabilities induced by economic downturns or natural
disasters. Pakistan’s poverty-reduction strategy is consistent with these objectives—
broad-based rapid growth along with improvements in social services and putting in
place a social safety net. The challenge lies in carefully designing and implementing
the policies to achieve the desired outcomes.

The 2004-05 poverty data show that rapid growth did reduce poverty quite sharply
— whether by five or 10 per cent is somewhat academic. It also points to the urgent
need for enhancing capabilities through better social mobilisation, education, public
health and rural infrastructure. Finally, strengthening the social-welfare support
system that is a notable impulse of Pakistan’s Islamic heritage will help go a long
way in reducing poverty and vulnerability in all its dimensions. Last but not least,
making data publicly available will not only promote transparency but also allow
everyone the opportunity to assess the progress that is being achieved.

The writer is the World Bank’s country director for Pakistan