Journal Issue
Share
Article

Pakistan: Selected Issues and Statistical Appendix

Author(s):
International Monetary Fund
Published Date:
January 2001
Share
  • ShareShare
Show Summary Details

IV. A SURVEY OF POVERTY IN PAKISTAN80

A. Introduction

171. Despite respectable real GDP growth rates for the past 50 years, poverty has continued to be an endemic problem in Pakistan. Even periods of high growth, as in the l960s and 1980s, have seen either rising or slowly declining poverty trends. Moreover, recent studies have found that poverty was on the rise in the 1990s. This paper attempts to shed some light on the reasons explaining Pakistan’s limited success in tackling poverty, by surveying some existing literature. The primary sources include a set of studies presented at the meeting of the Pakistan Society of Development Economists, “Pakistan Poverty Assessment” (1995) by the World Bank, and “A Profile of Poverty in Pakistan” by Human Development Center and UNDP.

172. The Section is organized as follows. In discussing poverty the first question is what exactly do we mean by poverty, so the next subsection considers the definition of poverty. Subsequent subsections are devoted to poverty trends in Pakistan and broader indicators of poverty. The following subsection considers the nature of poverty in Pakistan, in particular the incidence of poverty in different sections of the population and the economy.

B. Definition of Poverty

173. In the traditional sense poverty refers to lack of means to meet one’s needs. Absolute poverty was defined as inability to meet the physiological needs for survival. This led to measures of poverty based on daily calorie intake and food production. Later, the definition of poverty was expanded to include other “basic needs” such as shelter, clothing, access to water, health, and education, leading to measures of poverty based on basic social services and human development indicators. In recent years the definition of poverty has been further expanded to include concepts of vulnerability to risks and socio/political access. Since the measures of vulnerability and access are still in the early stages of development, this paper only considers measures based on income levels and human development indicators.

174. Income is the most obvious and easily observable variable for measuring poverty status. When asking if someone is poor, we are really asking whether the person has enough income to acquire what he/she “needs”. The decision of what a person “needs” is quite subjective and the root cause of lack of uniformity in poverty measures. The question of how one determines “needs” and transforms those needs into monetary units has significant implications for the level and trends in poverty. The ability of a person to afford a certain bundle of goods can be measured either by his/her income or expenditure. Income and expenditure would, of course, be the same if there was no access to credit and no savings. Since expenditure is more difficult to assess, most studies use income statistics.

175. Poverty line is the threshold income/expenditure level - the minimum income required to acquire what an individual or a household “needs”. There are many different ways of determining the poverty line. For example, the World Bank arbitrarily uses a dollar a day as the poverty line in determining global poverty. Since the assessment of basic needs of the poor is subjective, different authors have used different poverty lines. Another method of determining the poverty line employed by Ahmad (1993) and Gazdar et al. (1994) is to consider a basket of basic needs. Ah’ (1995) uses the linear expenditure system methodology to determine the basic needs of the poor. Some authors have also used calories based measures, i.e., the estimated cost of food consistent with a benchmark daily calorie requirement.81 Since trends in poverty are quite sensitive to the choice of the poverty line, it is difficult to compare results of studies across different time periods.

176. Once a poverty line has been chosen, the next step is to select a measure of poverty. The most obvious measure is to look at the head count ratio which gives the percentage of individuals (or households) which fall below the poverty line. The disadvantage of using this measure is that it does not capture the inequality among the poor. This led to the use of the poverty gap measure which measures both incidence of poverty and the income shortfall. These two measures are a subset of a class of poverty measures introduced by Foster, Greer, and Thorbecke (FGT; 1984).

The FGT class of poverty measures may be defined as follows:

Pa is the level of poverty

N is the population size

Q is the number of poor

Z is the poverty line

Yi is the per capita household income

Note that if α = 0, then this is just the head count index H. If α = 1, then it becomes the poverty gap index. Even the poverty gap index suffers from insensitivity to inequality among the poor. If we consider Pα with α = 2, this measure not only captures the shortfall of income but also the distribution of income within the poor. The main advantage of using the FGT Indices is that they are additively decomposable in the sense that the total poverty is a weighted sum of subgroup poverty levels (this will be useful when we consider the different subgroups of the population). Except for the head count index, we can interpret these measures only in relation to other known values to get a sense of the direction of the index. For this reason we will primarily use the head count index, which has a more direct interpretation.

C. Trends in Poverty

177. It is clear from the discussion on poverty lines that trends in poverty are very sensitive to the choice of methodology employed. This section reviews the poverty studies on Pakistan and their broad findings. To make analysis of a trend easier the studies will be divided into four different periods, the 1960s, 1970s, 1980s, and 1990s.

178. There appears to be a consensus amongst various studies of poverty in Pakistan that poverty rose in the 1960s, and declined in the 1970s and early 1980s (Table IV-1). In particular, it seems that in the l960s, poverty rose in rural areas and declined in urban areas. Both the 1970s and the early 1980s saw a decline in poverty in both urban and rural areas. There does not appear to be a consensus regarding the trend in poverty for the late 1980s and the 1990s. In discussing trends in poverty, few studies have attempted to explain the trends in poverty or tried to corroborate their findings with the existing macroeconomic conditions.

Table IV-1.Pakistan: The Existing Evidence on Trend in poverty
StudyOverallRuralUrbanData PointsPoverty Line/Unit
Poverty in the Sixties
1Naseem (1973)DeclineDecline1963/64, 1966/67,1968/69,1969/70Per Capita Expenditure, % of Households
2Allaudin (1975)DeclineDeclinePer Capita Income, % of Households
3Naseem (1977)Increase95%, 92%, 90% of 2100 calories
4Mujahid (1978)Increase“ except 1968/69Per Capita Expenditure, % of Households
5Man and Amjad (1984)Increase“ except 1968/692550 calories
6MH Malik (1988)IncreaseIncreaseDecline“ except 1968/69Per Capita Exp (2500 cal + non food), % of Population
Poverty in the seventies
1Man and Amjad (1984)Decline1969/70, 1978/792550 calories, % of Population
2MH Malik (1988)DeclineDeclineDeclineSame as Malik (1988) above
3Kruijik and Leewin (1985)DeclineDeclineDeclineMonthly of Rs. 700 at 1979 prices, % of Population
4Ali (1997)IncreaseUtility Function based concept of Poverty
5Amjad and Kemal (1997)DeclineDeclineDeclineSame as Malik (1988) above
Poverty in the Eighties and Thereafter
1MH Malik (1988)DeclineDeclineDecline1978/79,1984/85Same as Malik (1988) above
2Shirazi (1995)IncreaseIncreaseIncrease1987/88, 1990/91Basket of Basic Needs, % of Population
3SJ Malik (1994)DeclineDeclineDecline1984/85, 1987/882550 calories, % of Population
IncreaseIncreaseIncrease1987/88,1990/91
4Gazdar et.al (1994)DeclineDeclineDecline1984/85, 1987/88Basket of Basic Needs, % of Population
DeclineDeclineDecline1987/88,1990/91
5Ali (1997)Increase1984/85, 1987/88,1990/91Utility Function based concept of Poverty
6Amjad and Kemal (1997)DeclineDeclineDecline1978/79, 1984/85,1987/88Same as Malik (1988) above
IncreaseIncreaseIncrease1990/91,1992/93
7Jafri (1999)IncreaseIncreaseIncrease1986/87,1987/88Basic Needs based on Expenditure
DeclineDeclineDecline1990/91
IncreaseDeclineIncrease1992/93
IncreaseIncreaseDecline1993/94
Source: Ali and Tahir (1999)
Source: Ali and Tahir (1999)

179. One set of studies for the 1960s is based on a poverty line arbitrarily fixed in terms of a given per capita expenditure and income. The pioneering work on poverty in Pakistan was by Naseem (1973), using data from Household Income and Expenditure Surveys (HIES) from the 1960s. He used the income poverty line of PRs. 300 in rural areas and Rs. 375 in urban areas. This study showed that between 1963/64 and 1969/70, rural and urban poverty declined. Allaudin (1975) used expenditure data and used the poverty lines at Rs. 250 for rural areas and Rs. 300 for urban areas. Both of these studies show that poverty declined in both urban and rural areas in 1960s. The only caveat to this conclusion is that the use of per capita data instead of household level data requires adjustment for household composition. Mujahid (1978), corrected for this methodological error and found that rural poverty had actually increased over that period.

180. Another set of studies for the 1960s relates the poverty line to the recommended diet of 2550 calories per day per adult equivalent. These studies include those by Naseem (1977), Irfan and Amjad (1984), and Malik (1988). All these studies found that poverty increased in rural areas and the one by Malik (1988) found that poverty declined in urban areas.

181. Poverty trends in 1970s are difficult to establish because there were no Household Income and Expenditure Surveys (HIES) between 1971/72 and 1978/79. The four studies for the period are Amjad and Kemal (1997), Irfan and Amjad (1984), Kruijik and van Leewin (1985), and Malik (1988). These studies unanimously find that there was a decline in poverty in the 1970s. The only exception was Ali (1997) who found that overall poverty increased in the l970s. This may be due to the fact that he uses a utility function based concept of poverty which leads to a higher poverty line.

182. Poverty seems to have declined in the early the 1980s but there is some uncertainty about the trend in the late the 1980s. The two main studies for the period 1984/85 to 1990/91 are Malik (1994) and Gazdar et al (1994). Malik uses a calorie based poverty line approach and finds that while poverty declined between 1984/85 and 1987/88, it marginally rose between 1987/88 and 1990/91. Gazdar uses a basic needs approach and finds the poverty declined from 1984/85 to 1990/91. For the 1980s the complete HIES tapes were available for 1984/85,1987/88 and 1990/91 surveys, which allowed calculations of the poverty gap index and FGT index. Use of these indicators yield similar results as the head count index.

183. Results on poverty trends in the 1990s differ across studies. For the periods after 1990/91 the three main studies are Jafri (1999), Qureshi and Arif (1999), and Amjad and Kemal (1997). Amjad and Kemal (1997) found that poverty rose between 1987/88 and 1992/93. They used a similar methodology to that of Malik (1996). Using HIES 1993/94 primary data set, Jafri (1999) finds a gradual decline in poverty in both urban and rural areas. Using data from the 1998–99 Pakistan Socio-Economic Survey (PSES), Qureshi and Arif (1999) estimate that basic needs based poverty in Pakistan stands at 35.2 percent with 39.8 percent rural poverty and 31.7 percent urban poverty. This is considerably higher than most of the estimates for the 1980s, and they concluded that poverty had risen from 1990/91.

184. There have been few attempts to determine comparable poverty levels over time. While a comparable time series of poverty is necessary in order to do any trend analysis or statistical work, most studies tend to simply calculate poverty rates for particular periods to determine the sensitivity of their choice of poverty line, and to compare their results with earlier findings. Two not able exceptions are Malik (1988) and Amjad and Kemal (1997). These studies use a combination of basic need and calorie in-take approach to poverty instead of a purely calorie intake based analysis. Since food expenditure account for only 50 percent of total expenditure, even for the poor, and since it is difficult to find poverty estimates for a purely calorie intake based poverty figures for recent years, a basic needs approach to poverty is appropriate. Malik (1988) defined the poverty line with reference to a calorie requirement of 2550 for the adult, and the revealed expenditure pattern of the poor between food and non-food expenditures (considered the average ratio of food to non-food consumption for the poor). Taking the 1984–85 poverty line as the benchmark, earlier poverty lines were estimated by deflating the current poverty line using the current CPIs for those years. Amjad and Kemal (op. cit.), inflated the 1984/85 poverty line by the current CPI figures to arrive at new poverty levels up to 1992/93. Qureshi and Arif (1999) used a similar methodology to arrive at poverty estimates for the period l993/94 and 1998/99. Inc Chart IV-1, the poverty level estimates of Amjad and Kemal (op. cit.) and Qureshi and Arif (op. cit.) were spliced to time series for the period 1963/64-1998/99.

Chart IV-1Pakistan: Trends in Poverty,1963/64-1998/99

185. In broad strokes, the studies seem to indicate that poverty declined from beginning of the 1970s to late 1980s but has been on a rising trend since then. Although many of the studies are not strictly comparable, the conclusions seem to be that: (a) poverty rates were higher in 1969/70 than in 1963/64; (b) between 1969/70 and 1979 poverty declined in both rural and urban areas; (c) this decline continued until 1986/87; and (d) since 1992/93 there has been a gradual rise in poverty. The preliminary data sets from recent surveys also indicate that poverty is on the rise.

D. Broader Poverty Measures

186. As the definition of poverty has grown in scope, more indicators of poverty have been suggested instead of just looking at the income, or consumption. There is a need for indicators to measure aspects of poverty such as inequality, economic opportunity, health, nutritional status, vulnerability, discrimination based on gender, ethnicity, or race, and empowerment/voice.

Income inequality

187. Although the head count index is the most widely used measure of poverty, it is insensitive to changes in extent of poverty. Income inequality measures attempt to measure the extent of poverty. Why do we care about income inequality? The first reason is that rising inequality would explain why rampant poverty may persist despite robust economic growth. It suggests that policies are not benefiting the poorest sections of the population. Rising inequality can also make problem of poverty even more acute.

188. The most common income based measures of inequality are the Gini coefficient82, percent shares of income (lowest 10 percent and highest 10 percent), and the FGT Index (described earlier). Gini coefficients for various years when the HIESs were conducted are presented in Table IV-2. In the 1960’s there seems to have been a decline in income equality in both rural and urban areas. Generally, in 1970s there seems to be an overall worsening of income inequality, followed by a reversal of the trend in 1980s. The trend is more uncertain for the l990s but there seems to be a gradual rise.

Table IV-2Pakistan: Income Distribution
YearGini Coefficient
TotalRuralUrban
1963/640.3550.3480.368
1966/670.3510.3140.388
1968/690.3280.2930.370
1969/700.330.2950.361
1970/710.3260.2730.359
1971/720.3440.3090.381
1978/790.3750.3190.380
1984/850.4280.3450.379
1985/860.3550.3300.354
1986/870.3460.3120.357
1987/880.3480.3070.366
1990/910.4070.4100.390
1992/930.3900.3670.384
Source: HIES, Various Year (Jaffri, 1999)
Source: HIES, Various Year (Jaffri, 1999)

189. The measures mentioned above are based on income distribution but since Pakistan is primarily an agriculture-based economy, another variable that can be used to gauge the trend in inequality is the distribution of land. Table IV-3, replicated from Jafri and Khattak (1995), shows a worsening of the inequality in distribution of land. Farm holdings of greater than 50 acres accounted for 23.8 percent of the land in 1990 as opposed to only 8.3 percent in 1981. Income inequality can be used to illustrate an important point regarding use of socioeconomic indicators to measure different aspects of poverty.

Table IV-3.Pakistan: Distribution of Land (1981 and 1990)
Farm Size

(acres)
Number of FarmsPercentage of
(In thousands)FarmsArea
198119901981199019811990
<51384.12404.134.147.47.111.2
5 to 2523022322.356.745.852.149.2
26 to 50358.3237.58.84.732.515.8
Over 5013.9106.90.32.18.423.8
Source: Agriculture Census Organization, Agriculture Census 1981 and 1990
Source: Agriculture Census Organization, Agriculture Census 1981 and 1990

190. Trends in poverty are very sensitive to choice of indicators. For example although the distribution of land statistics suggest that inequality worsened between 1981 and 1990. The Gini coefficients from the intervening period 1984/85 to 1987/88 showed that income inequality has actually improved.

Access to credit

191. In measuring the economic opportunity aspect of poverty, one variable that stands out is the access to and use of credit, in particular the use of credit by the poor and women. To measure the access of poor to credit we can look at the actual use of credit by the poor and also the real interest rate differences between formal and informal sectors. In a recent study, Malik and Nazli (1999), used data collected by International Food and Policy Research Institute (IFPRI) from a sub-sample of households in 1985 Rural Credit Survey of Pakistan to analyze credit use in rural Pakistan. They found that using head count measures about a fifth of the rural households in Pakistan could be classified as poor or below the poverty fine. However, when expenditures met through credit are netted out, about half the households drop below the poverty line. The study showed that shopkeeper credit was an important factor in meeting consumption needs. They also found evidence that households which use credit have significantly higher values of farm output in each expenditure quintile. Further, both formal and informal sources are used to finance farm expenditures, but in general, the proportion of expenditure met through the formal channels is much lower than those from the informal sources.

192. Higher credit access and use of credit for farm production are likely to be linked to poverty alleviation. However, credit access is a difficult measure to use since data on credit use are sparse and as shown in Table IV-4 (based on 1990 data compiled by IFPRI), much of the credit for the poorer households is through informal sources. For the lowest two quintiles nearly 90 percent of the loans are from the informal sector. Since the informal sources are fragmented and thin, it is difficult to determine what percentage of the individuals have access or actually take loans. Although credit indicators such as credit to private sector or credit through agriculture development banks may not reflect trends in poverty directly, but higher credit use may result in more employment opportunities for the poor, thus leading to poverty alleviation.

Table IV-4.Pakistan: Distribution of Loans
Per Capita

Expenditure Quintiles
FormalInformal
Lowest6.993.1
Second10.589.5
Third35.564.5
Fourth27.572.6
Highest44.455.5
All31.768.3
Source: IFPRI (1990).
Source: IFPRI (1990).

Illiteracy rate

193. Illiteracy rate is another important indicator of economic opportunity and human development. Pakistan has one of the highest illiteracy rates in the developing world. In 1997, the illiteracy rate was estimated to be 45 percent for males and 75 percent for women. Although this is an improvement from 1980s, the high rates and the stark differences between male and female rates are a cause for concern. Education has been shown to have an impact on many socio-economic variables, including income and productivity. In the case of Pakistan, Bhutt (1984) shows that for farmers, five or more years of education led to increased farm productivity, reduced use of farm labor, and increased use of yield augmenting inputs. Azhar (1988) also reports a significant relationship between number of years of schooling and increases in farm output due to increased efficiency. Education can also have an impact on health and child mortality. An educated parent would be more informed about the treatment and precautions for common diseases, and is more likely to provide better healthcare to the child. Consider the following example, despite extensive public health campaigns, Rukanuddin and Hasan (1992) found that, 21 percent of the women surveyed, reported that they had reduced the amount of fluid given to children during diarrhea episode, which could make diarrhea fatal.

194. In discussing poverty, it is sometimes difficult to distinguish between the causes and effects of poverty, particularly when poverty is measured using human development indicators. On the one hand, low levels of education can cause poverty through lower productivity and job opportunities. On the other hand, poverty can lead to lower literacy levels since parents do not have enough resources to send their children to school or they depend on the child’s income to meet the household expenses. Arif et al (1999) using a poverty dummy in addition to an income variable found that poverty exerts a negative influence on a child’s probability of attending school in addition to the effect of the low household income. But regardless of whether it causes or is an effect of poverty, higher literacy levels indicate lower poverty.

195. Illiteracy rates can also capture certain aspects of gender bias in poverty. Lower literacy rates restrict women’s access to employment, training opportunities, and available social services. According to the latest statistics, the female labor participation rate is 13 percent as compared to the 74 percent male participation rate. The female labor force participation rate is extremely low compared to other countries. Since most women are uneducated, their employment opportunities are restricted primarily to agriculture. A 1987/88 Labor Force Survey83 shows that 73 percent of the female labor force works in agriculture, 13 percent in manufacturing, and 11 percent in services. As female primary school attendance rates are low (25 percent), the prospects for a significant improvement in female literacy rates in the near term appear limited. The low primary school attendance rates and labor participation rates may also be a result of social and religious norms.

196. Illiteracy is a direct result of low primary school attendance. There are many reasons for the low primary school attendance. The first problem is the limited access to schools and teachers for much of rural Pakistan. For those who have access, variables such as household income, the father’s education and tenure of status as land owner, cost of books, and village literacy level seem to play a role in school attendance decision.

Health and nutrition

197. Health and nutrition are essential components of human development. Poverty usually results in deprivation of food and essential medical services leading to deterioration of the health indicators. So a decrease in poverty should affect this category of variables positively. The commonly used health and nutritional indicators are infant mortality, life expectancy at birth, maternal mortality, and child malnutrition. Variables for access to health include doctors per 1000 and access to safe water. Considering the data in Table IV-5, Pakistan seems to have made some progress in health indicators, both in terms of the condition of health and access to health, suggesting that poverty may indeed have declined.

Table IV-5.Pakistan: Main Health Indicators
Variable1966–701971–751975–801981–851986–901991–951996–98
Infant Mortality per 100014514214012711110595
Life Expectancy (All)49.450.655.156.259.059.761.3
Life Expectancy (M)49.650.654.655.658.259.660.8
Life Expectancy (F)49.850.555.756.960.060.862.6
Physicians Per 10000.20.30.40.50.5
Access to Safe Water (Urban)77757784848085
Access to Safe Water (Rural)452228284056
Access to Sanitation (R)455824
Access to Sanitation (U)4856564875
Source: World Development Indicators by the World Bank
Source: World Development Indicators by the World Bank

198. Another outcome of poverty is child malnutrition.84 Despite an increase in per capita food availability the National Health Survey shows that prevalence of malnutrition has not changed over the last 20 years. Recent estimates place the number of malnourished children at 8 million. The three common indicators used to measure malnutrition are percent of children underweight (low weight for age), stunted (low height for age), and wasted (low weight for height). The National Nutritional Survey (1988) found that about 52 percent of the children were underweight, 42 percent were stunted and 11 percent had low weight for height ratio. The National Health Survey (1996) observed that the proportion of children stunted declined to 36 percent but the proportion of children wasted increased to 14 percent. According to the Human Development Report (1999), between 1990–97, 25 percent of the infants had low birth weights. Nearly 45 percent of the women in rural areas and 37 percent of the women in urban areas are anemic. The poor nutritional status results in increased vulnerability to infectious diseases and water borne diseases. Variables which have been shown to have a positive impact on malnutrition are the mother’s education, per capita calorie intake, income, and community factors such as access to safe drinking water, sanitation, disposal of solid waste and awareness of health issues.

E. Distribution of Poverty

199. To acquire an accurate picture of the nature of poverty in Pakistan, it is necessary to consider the incidence of poverty in different sections of the population and the economy. In looking at subsets of the poor, it may be possible to deduce where poverty alleviation policies may have the most impact or are most needed. Earlier in the paper the only division considered was rural versus urban poverty. In this subsection we will consider several other partitions of the population, based on geography, industry, and household characteristics.

200. It is not enough to look at just the poverty in each subgroup but it is also necessary to see how much each group contributes to overall poverty. We want a measure of whether a specific category has more than or less than its share of the poor. The share of poverty (SPi) for population subgroup i is define as:

si is the share of subgroup in total population,

pi is the poverty rate in the sub-group

P is the poverty rate in the population

To measure the relative contribution to poverty we will use a location index (LI) defined as follows:

The location index gives the poverty share as percent of population share. If poverty were distributed evenly across population subgroups, then the population share should be equal to the contribution to poverty, i.e, LIi is 100. If this index has a value of greater than 100, it indicates that poverty is more severe in this subgroup compared to other subgroups.

Geographical distribution

201. We will first consider if there are any differences in poverty across geographical units. One reason to look at geographical distribution is to see if particular areas of the country need more attention than others. There may also be some other factors such as land productivity or environmental conditions, which may explain persistent poverty in some areas compared to others.

202. According to the head count index, the highest poverty amongst the provinces appears to be in Northwest Frontier Province (NWFP) with an over-all head count index of40 percent (Table IV-6). It is also clear from the concentration index that NWFP has a greater share of poor compared to other provinces. In particular, the urban areas in NWFP seem to have the most severe poverty compared to its size. Only 7 percent of the total urban population of Pakistan lives in NWFP but it constitutes 9.3 percent of the urban poor. Sindh appears to be better off in terms of poverty with low head count index for both urban and rural areas. Surprisingly Balochistan shows the lowest poverty rate in all areas. It should be noted that the statistics for the smaller provinces, NWFP and Balochistan, should be taken with a grain of salt. The data show inconsistencies across different surveys conducted at the same time. The HIES 1990/91 showed that Balochistan had a poverty rate of 22 percent while the PHIS 1991 (Pakistan Household Integrated Survey) yielded a reverse trend with Balochistan having the highest poverty rate at 41 percent (Pakistan Poverty Assessment, World Bank, 1998).

Table IV-6.Pakistan: Geographic Distribution of Poverty
Geographic UnitShare of

Population

(In percent)
Poverty

within Group

(In percent)
Share of

Poverty

(In percent)
Location

Index
Pakistan100.034.0
Punjab59.835.963.1105.6
Sindh22.527.618.381.2
NWFP13.540.015.9117.6
Balochistan4.022.02.664.7
PakistanUrban29.828.024.582.4
PunjabUrban16.929.459.5105.0
SindhUrban10.324.129.786.1
NWFPUrban2.137.09.3132.1
BalochistanUrban0.626.71.995.4
PakistanRural70.236.976.2108.5
PunjabRural42.938.563.8104.3
SindhRural12.530.814.983.5
NWFPRural11.440.617.9110.0
BalochistanRural3.420.92.756.6
Source: World Bank (1995) “Pakistan Poverty Assessment” based on HIES 1990/1991.
Source: World Bank (1995) “Pakistan Poverty Assessment” based on HIES 1990/1991.

203. Considering the bigger provinces, Punjab seems to have a higher poverty rate than Sindh, particularly in the rural areas. There is a higher concentration of poor in Punjab than in Sindh, as indicated by the location index difference of 105 to 81. Also, poverty seems to be higher in the rural areas in comparison to urban areas, with total rural poverty in 1990/91 at 28 percent compared to 37 percent in rural areas.

Household characteristics

204. Household characteristics are important factors in the incidence of poverty. In the following paragraphs we will look at partitions based on household size, number of earners, and head of household characteristics, including sex, age, and education. The purpose of considering this breakdown is to see if there are any particular demographic groups which show higher poverty compared to others. In this section the poverty measure is based on the distribution of expenditure for the year 1993/94. We will be using the head count measure and the location index to compare poverty across subgroups.

205. The data clearly indicate that poverty is most concentrated in the largest subgroup, which comprises households with 7 or more individuals (Table IV-7). Household size is an important determinant of poverty because the greater number of dependents means that there will be few per capita resources available. In particular, the poverty rate is 34.3 percent which is significantly higher than the 8.2 percent for household with 1–4 individuals. Due to the size of the subgroup and high subgroup poverty, 66 percent of the poor are from households with 7 or more individuals. The size of the household may be offset by a higher number of earners so we also consider the breakdown of poverty according to number of earners. There does not seem to be much variation in poverty rates related to the number of earners for each household. Most household have one earner and on average the poverty rate seems to be around 28 percent.

Table IV-7.Pakistan: Household Attributes
House Hold SizeShare of

Population

(In percent)
Poverty

with in Group

(In percent)
Share

of Poverty

(In percent)
Location

Index
1 to 425.58.28.734.0
5 to 628.121.725.390.0
>746.434.366.0142.3
Overall100.024.1
Number of Earners
08.125.47.289.5
151.329.653.5104.3
222.628.822.9101.4
310.925.19.688.4
>47.027.26.795.8
Overall99.928.4
Head of the Household Characteristics
Sex
Male93.128.29399.9
Female6.928.57101
Overall28.2
Age
>2911.424.49.886.3
30-3925.531.928.8112.8
40-4927.229.128102.9
>5035.926.333.493
Overall28.3
Education
No Formal Education59.035.874.7126.6
Kindergaten2.431.82.7112.5
Primary12.926.111.992.3
Middle7.118.54.665.4
Matriculation9.013.24.246.7
Intermediate3.67.91.027.9
B.A/B.Sc3.56.00.721.2
M.A/M.Sc./LL.B1.91.00.13.5
MBBS/Eng0.42.20.07.8
M.Phil/Ph.D etc0.10.00.00.0
Overall99.928.3
Source: A Profile of Poverty in Pakistan.
Source: A Profile of Poverty in Pakistan.

206. There also does not seem to be much difference in poverty rates across sex of the head of the household. About 7 percent of the households in the survey were headed by females. Across age groups the highest poverty is in the in 30–39 group with poverty rates of32 percent. The lowest poverty is seen in younger group (<29) and the oldest group (> 50).

207. There are no surprises in the distribution of poverty across educational subgroups. The households with the lowest levels of education have the highest incidence of poverty. About 60 percent of the households in Pakistan are headed by individuals with no formal education. The poverty rate in this group is 36 percent and it accounts for 74 percent of the poor in Pakistan. Poverty rates decline considerably for individuals with education levels higher than matriculation.

Economic activity

208. Productivity and real wages differ across sectors, as does the incidence of poverty. In what follows, we will consider partitions of the economy according to economic activity by sector, for Pakistan as a whole, and by type of wage earner.

209. Reflecting its central role in the economy, agriculture (including forestry and fishery) is the most important sector in terms of employment. With low productivity and real wages, the incidence of poverty in this sector is the second-highest economy-wide (Table IV-8). Poverty is only more severe in the construction sector, which employs a large work force of unskilled casual labor. Of all the poor, 14.4 percent are from this sector although it constitutes only 9.3 percent of the population.

Table IV-8.Pakistan: Distribution of Poverty by Economic Activity
ActivityShare of

Population

(In percent)
Poverty

within Group

(In percent)
Share of

Poverty

(In percent)
Location

Index
Agriculture/Forestry/Fishery28.733.734.1118.7
Mining0.212.30.143.3
Manufacturing8.425.97.791.2
Electricity/Gas/Water1.118.40.764.8
Construction9.344.014.4154.9
Trade12.620.18.970.8
Restaurants/Hotels6.325.65.790.1
Transport1.43.40.212.0
Social Services15.122.812.180.3
Undefined16.827.416.296.5
Overall99.928.499.9100.0
Source: Profile of Poverty in Pakistan, HIES 1993/94
Source: Profile of Poverty in Pakistan, HIES 1993/94

210. In Table IV-9, the population is divided into employment categories: agricultural workers, wage earners outside agriculture, self-employed outside agriculture, and a residual “other”. The agricultural worker are further categorized based upon access to land, and the wage earners are divided into “white collar”, skilled/semi-skilled and casual/manual. The reason for partitioning the population in this way is to gauge how access to capital (physical or human) affects poverty. We would expect that those groups with the least access to capital would show the worst poverty rates.

Table IV-9.Pakistan: Employment Profile of Household Heads and Incidence of Poverty (PIHS 1991)
RuralShare of

Population

(In percent)
Poverty

within Group

(In percent)
Share of

Poverty

(In percent)
Location

Index
Agriculture63.635.164.7101.7
Owner cultivator36.630.232.087.5
Tenant13.643.817.3127.0
Agricutural laborer7.056.011.4162.3
Other Agriculture6.421.03.960.9
Wage earners in other sectors17.929.815.586.4
Self-emp outside agricutlure15.236.316.0105.2
Other3.441.14.1119.1
Overall100.134.5
RuralShare of

Population

(In percent)
Poverty

within Group

(In percent)
Share of

Poverty

(In percent)
Location

Index
Wage earners43.828.439.690.4
By job type:
White collar14.222.110.070.4
Skilled/semi-skilled20.128.118.089.6
Casual/manual9.538.311.6122.1
Self-employed36.334.039.3108.3
Other19.933.721.4107.5
Overall100.031.4
Source: Poverty in Pakistan: Measurement, Trends, and Patterns, 1994
Source: Poverty in Pakistan: Measurement, Trends, and Patterns, 1994

211. In the urban areas 43.8 percent of the household are headed by wage earners with a poverty rate of 28.4 percent. White collar workers constitute 14.2 percent of the total urban population and they have the lowest poverty rate in urban areas. Casual/Manual laborers appear to be the worse off with a poverty rate of 38.3 percent and a location index of 122. Poverty also seems to be high amongst self-employed and the residual category.

212. Rural areas are primarily dominated by the agricultural sector with 63.6 percent of the rural households headed by agricultural wage earners. Since access to land plays an important role in the poverty status, the agricultural workers are divided according to land ownership. As expected poverty seems to be high in sections with least access to land, i.e. the tenant farmers, and agricultural laborers. Poverty is most severe amongst the agricultural laborers with a poverty rate of 56 percent.

F. Conclusions

213. The broad conclusion seems to be that although some progress has been made in human development indices in Pakistan, overall poverty has been on the rise. Since the early 1990s we have seen a general rise in poverty, exacerbated by slowing growth, falling public social expenditure, and high population growth. Even in times of robust economic growth, Pakistan experienced high poverty due to high income inequality. This suggests that rapid economic growth by itself is not sufficient to reduce poverty, instead a more direct and comprehensive approach in needed to address the problem of poverty.

214. There is a need for increased social expenditure, particularly in education sector. As was shown in this paper education has a positive impact on many quality of life indices including economic opportunity, productivity, and health. Education is a long term strategy and will yield high returns through a more skilled pool of human capital. But in the short run more direct efforts are required to stem poverty.

215. Recently, there has been a shift of policy in Pakistan, which has brought the issue of poverty to the forefront of economic strategy. The new government has promised an integrated approach to attacking poverty, through social safety nets, small infrastructure projects in poor rural and urban areas, and establishment of a micro-credit bank. The large funding commitment for these projects promised by the new government is a good start in trying to bolster the long neglected social sector in Pakistan. Over the longer term, sustainable high rates of economic growth would be needed to meaningfully improve the living standards of the poor in Pakistan.

References

Prepared by Farhan Hameed (MED).

The Gini coefficient ranges between zero for perfect equality and one for perfect inequality. It is the ratio of the area between the 45 degree line and the Lorenz Curve to the area uner the 45 degree line, where the Lorenz curve is a plot of the proportion of total income held by each percentile of the population, ranked in order of income.

Source: World Bank, “Pakistan: Country Gender Profile”

This section is primarily based on Qureshi et al (1999)

Other Resources Citing This Publication