Differences in consumption aggregates and poverty estimates in South Asia

In a paper published in the latest edition of Asian Development Review (Vol.38. No.1), Islam, Newhouse and Yanex-Pagans study how differences in the construction of consumption aggregate in South Asian countries contribute to total error (arising out of nonsampling error and the error in the process of determining the international poverty line) in international extreme poverty measurement. Methodology and questionnaire for household survey differ among the countries, contributing to total error of the subsequent international extreme poverty line obtained for each country. They examine how sampling and survey design; spatial deflation to account for cost-of-living differences and intertemporal deflation; and construction of nominal consumption aggregate contribute to total error. 

Some factors that affect total error are incomplete coverage of the country’s population, errors in measuring consumption data, errors in calculating the poverty line, use of the consumer price index (CPI) to deflate prices in a manner that may not be consistent with the consumption patterns of the poor, and geographic differences in prices.

On sampling and survey design, the paper examines (i) sampling design, (ii) monetary welfare measure, (iii) food consumption questionnaire and data collection methods, (iv) self-production and meals outside home, (v) nonfood durables, (vi) durables, (vii) housing expenditures, and (viii) health and education expenditures. 

Among these, there exists significant differences in the way food consumption data are collected, especially the number of food items in the consumption questionnaire. Inclusion of more food items tend to increase levels of reported consumption, leading to lower reported poverty rate. While Pakistan has the lowest number of food items (69) listed in the survey, Sri Lanka has the highest (227). Bangladesh has 141, Bhutan 130, India 143, Maldives 92, and Nepal 74. For nonfood items, Afghanistan has the lowest number of items (38) and Maldives has the highest number of items (483). India has 338 and Nepal 95. 

Consumption data are collected either using diary method (households record all consumption data over a certain period in a notebook) and/or recall method (households list what they consumed over a specific past reference period). Length of consumption recall is also different. Lowering the recall period have increased reported consumption by households, which resulted in poverty rates falling by half in India. The authors note that “this simple change in the method of collecting data “lifted” 175 million Indians out of poverty”. 

The South Asian countries also have different questionnaires for the value of consumption of self-produced food and meals from outside home. For instance, Bangladesh, Pakistan and Sri Lanka do not account for food expenditures on meals outside the household as a part of their consumption aggregate. Note that consumption of food eaten outside the home is shown to raise extreme poverty rate. Similarly, Maldives and Pakistan do not account for consumer durables, and in Afghanistan and Nepal it is imputed. In the case of housing, all countries except Maldives include actual rent for urban and rural areas. They also include imputed rent except for (India and Maldives). All countries include health and education expenditure (except for Nepal in the case of health expenditure). 

Spatial deflation is used to adjust cost-of-living differences, which lowers the possibility of overestimation of poverty in rural areas and underestimation in urban areas (since an urban household needs to spend more to maintain the same standard of living as that of a rural household). Use of appropriate regional price indexes is important in this regard, but this could be challenging to construct. All South Asian countries do spatial deflation when they calculate their national poverty estimates, but when calculating international extreme poverty rates, the World Bank spatially deflates consumption aggregates in Bhutan and Nepal only. Bhutan uses survey-based price index to deflate prices, but Nepal and Bangladesh use an implicit spatial price index to deflate prices. The authors show that using spatial deflation is important in the case of international extreme poverty estimates, especially given the fact that without spatial deflation urban areas have less poverty and rural areas have more poverty. They recommend collecting regional price data for different durable and nondurable goods, and services. They also suggest collecting rental cost of housing at regional level so that imputation for rental rate of owner-occupied housing could be done suitably. 

A third source of total error is the way standardized consumption aggregates are computed compared to the national consumption aggregates. Here, standardized consumption aggregate refers to the one obtained from the standardized consumption datasets created by the World Bank. It basically reclassifies expenditure items into the various categories used in the International Comparison Program (ICP). However, note that the method of data collection and questionnaire design affect standardization. The authors show slightly different average per capita consumption using the standardized and the national consumption aggregates. Not much difference, but in the case of India there is notable difference because of imputed rents for home owners. Also, standardization decreases housing expenditure in Nepal, Bhutan and Sri Lanka. 

The authors shows that standardization of consumption aggregate also changes share of particular item on total consumption expenditure. In Nepal, while food items account for 57% of national consumption expenditure, the standardized value is a bit less at 53%. Standardization reduces the share of food expenditure in consumption aggregate in all South Asian countries except Sri Lanka. 

The standardization of consumption aggregates increases poverty rate in Bangladesh, Bhutan Pakistan, and Sri Lanka. However, they authors show that it decreases the international extreme poverty rate in India, Maldives and Nepal.