Why Differences in Household Expenditure Estimates Matter

Abstract

Government estimates on average household spending can be obtained from two sources: sample surveys, such as the Family Income and Expenditure Survey (FIES), and national accounts data. The typical estimate is the household final consumption expenditure (HFCE) estimate from national accounts which is more timely, as it is released quarterly, than survey estimates, which are triennial. Estimates between sample surveys and national accounts have wide discrepancies across time, with their divergence generally growing. However, this scenario is not unique to the Philippines (Karshenas 2003). In India, the Committee on Private Final Consumption Expenditure of the Central Statistics Office (2015) revealed that the discrepancy in Indian estimates had increased from 5 percent in 1972–1973 to 45 percent in 2011–2012 (also see Minhas 1988; Sundaram and Tendulkar 2003; and Deaton and Kozel 2005). Often, survey-based estimates are lower (Ravallion 2003; Deaton 2005). Such discrepancies have consequences to policy. For instance, poverty can be overestimated if survey-based estimates are biased downward. As a result, resources for poverty reduction will go to those who do not need them. Meanwhile, underreporting from wealthy households in surveys also underestimate income inequality. This Policy Note describes this issue in detail, looking at how estimates are derived and other related issues.