Poverty, Household Food Security, and Nutrition in Rural Pakistan

The pathways from economic and social policies to improved food security and nutrition for the poor often are not well understood. Yet each day governments decide on policies that ultimately affect their well-being. How households increase their incomes, acquire food, improve health, or cope with insecurity are important concerns that need to be examined in order to devise policies to help eradicate poverty. Nearly all attempts to study these issues have used snapshot approaches--those that look at one point in time. These approaches are limited in that they do not reveal anything about the actual dynamics of poverty, food security, and their consequences for nutrition and health.

In Poverty, Household Food Security, and Nutrition in Rural Pakistan, Research Report 96, Harold Alderman and Marito Garcia address these concerns by looking at longitudinal data for a three-year period, 1986-89, and analyzing fluctuations in incomes, consumption, savings, nutrition and health-seeking behavior of 800 households in five districts in rural Pakistan (Faisalabad and Attock in Punjab province, Badin in Sind, Dir in North-West Frontier Province, and Mastung/Kalat in Baluchistan). The report examines income sources and wage formation in rural Pakistan and investigates the level and distribution of income in poor households. It contributes to analysis of the temporal dimensions of poverty and thus adds to the literature on coping strategies of households. Although the three-year panel of data analyzed is too short to model fully the dynamics of poverty, it is sufficient to indicate the fluidity of the economic environment that households in Pakistan face. The report also traces the efficiency by which household incomes are converted to better nutritional well-being and the influence of other intervening factors such as health and education.

Income Sources and Wage Formation in Rural Areas

Although the households surveyed were all located in rural areas, their sources of livelihood were not strictly agricultural. Crop earnings represented less than 45 percent of all earnings, whereas nonfarm wages and earnings from own enterprises were 41 percent of all incomes. A diversity of sources of income other than crops and livestock were found, including artisan work, village crafts, public conveyances, and various trading activities. Moreover, many households received substantial remittances from household members working in large Pakistani cities such as Karachi or abroad (the Middle East, for example). This diversity of income sources held true for all districts studied. Only 14 percent of households in the entire sample earned less than 20 percent of household income from activities outside of crop cultivation or livestock tending.

These findings reinforce the conclusion that rural development is not totally congruent with agricultural development. This does not negate, however, the findings of other studies indicating that agricultural development strongly influences the demand for services and nonfarm production. To be sure, rural nonfarm income is often derived from self-employment in business and activities related to the production of inputs or processing of agricultural output. Nevertheless, strategies for rural development must involve a broader set of policies than development of agriculture per se. These strategies must include development of education and infrastructure as well as broadening of credit availability to rural nonfarm enterprises.

Although it is occasionally argued that technological change and farm fragmentation have increased the number of households dependent on agricultural wage labor, little evidence for this is found in either the IFPRI panel survey or nationally representative surveys. Moreover, the pay for labor in Pakistani agriculture is similar to that received for urban unskilled labor and higher than that in most other Asian and African countries (Figure 1). For example, the average agricultural wage laborer in the IFPRI sample could buy 11-17 kilograms of wheat for a day's work, while unskilled construction workers in Karachi or Lahore could buy 15-18 kilograms of flour with their wages.

In general, food is relatively cheap throughout the year in Pakistan. This reflects the extensive subsidy and stabilization policies pursued by the government. At the margin, there is little scope for further increases in grain consumption through such policies.

Income Inequality and Poverty in Rural Areas

Income inequality was moderately high in rural areas. An overall Gini coefficient (measure of income inequality) of 0.40 is calculated for the populations covered in this report. A decomposition of the Gini coefficient indicates that of the five sources of income--agriculture, nonfarm, livestock, rent, and transfers--agricultural income accounts for the largest share of overall income inequality. On the other hand, livestock and nonfarm income sources help decrease income inequality. These findings indicate that policies that seek to promote livestock development and to attract nonfarm investments in rural areas are likely to promote better distribution of incomes in Pakistan.

The rich data set allows the authors to construct nine different definitions of poverty--including per capita expenditures, per capita income, per capita income plus transfers, expenditure per adult equivalency, per capita food expenditures, per capita calorie consumption, landless, per capita predicted income, and the share of food in total expenditures.

Analysis of these alternative indicators shows that there are enormous pitfalls in defining poverty using a single dimension. The repeated observations of households in this research (12 visits in three years) clearly show that it is difficult to determine unambiguously which are the poorest households in a community: some households move in and out of poverty over time; some are excluded when using one indicator but included using another. This implies that instead of attempting to identify the poverty levels of different groups, perhaps one should identify characteristics of the poor--by looking through the statistical correlates of being poor, for example--in order to design programs that take into account the constraints.

Income Fluctuations and Household Coping Strategies

Fluctuations in income, even over the relatively short three-year period, were considerable. Weather, illness, and decline in remittances from abroad were among the reasons. A moderate share of income fluctuation is explained in the report by district variables and a far greater share by village-level variables. Consumption risks in these households were only partially mitigated by sharing through family networks. Income risks were also reduced by the diversification of income sources.

Households in rural Pakistan clearly use savings and credit markets to stabilize consumption. For example, although incomes are seasonal, there are no statistically significant seasonal patterns of food consumption in any of the sample districts.

Households also save or draw on their savings as a result of transitory annual income fluctuations. On average, 70 percent of an increase after a windfall is saved. Similarly, in response to a temporary decrease in income, households borrow or reduce assets to cover about 70 percent of a shortfall. Even low-income households manage to save half of their transitory income increases, although they save only 10 percent of the overall income. Remittances are mostly saved; households put half of marginal remittances from abroad into financial savings (debt reduction, bonds, and bank accounts) and an additional 30 percent into physical property.

Household Food Security

Over the three-year period, the calorie supply of the sample households was moderately high--2,400 calories per person per day, on average--compared with households in many parts of South Asia. No evidence of seasonal differences in calories consumed is detected, although the composition of grains consumed in one region changed with the seasons. Households generally coped with seasonal lows by using savings, including storage of grains.

Households opted to use credit from informal sources--such as relatives and friends--to maintain a fairly constant consumption level. In these households, the effects from fluctuations in crop production are mitigated by the great diversity in income sources.

Calorie income elasticities in the report range from 0.12 to 0.39, on average, and from 0.14 to 0.46 for the poorest income quintile. Food expenditure elasticities, however, are about 1.5-2.0 times the calorie elasticities, which suggests that households are likely to choose a higher-quality and more diversified diet over quantity as income increases.

The results indicate that underconsumption is not likely to disappear in the normal course of economic growth. In these households, the report finds that it would take a 30 percent increase in incomes to achieve a 10 percent rise in calorie intakes. Other concomitant policies need to accompany increases in incomes in order to attain food security. Improved public education, particularly for women, is found to be a critical determinant. A household with the same level of income but with adult women with some education consumes about 150 calories per person per day more than similar households where women do not have this education.

Translating Income and Food Security to Improved Nutrition

Despite the relatively high level of calorie availability in Pakistan, the proportion of malnourished children (indicated by the prevalence of underweight) has remained high--more than 40 percent in the study area (Figure 2). The results in this report indicate that converting income into higher calories and consequently into growth is hampered by the negative effects of infection in these children. The prevalence of illness and diarrhea is particularly high. The main message, then, is that the interaction between diet and infection is particularly critical in an environment where disease is widespread.

The nutritional status modeling in this report indicates that critical community services--including primary health-care services, sanitation, village water supply, and public drainage systems--are necessary to stem the spread of infectious diseases (Figure 3). Public health programs that reduce illness, such as immunization or those that encourage prenatal care, are important instruments for influencing nutrition. However, mere physical presence of these services in a community is not enough: quality of services is equally important.

In this report, education, particularly of females, is found to be strongly correlated with better nutrition in children. In fact, the impact of education is much stronger than that of increasing incomes. Educating women to at least the primary level is likely to be nearly three times more effective than increasing household incomes by 10 percent.

The implications for policy formation are clear. To reduce malnutrition in the country, improvements in per capita food supply and food security need to be matched by a systematic reduction in the prevalence of infectious diseases, particularly for the preschool-age population. This can be achieved by increasing public investments in health and education.