Estimating Wealth Effects Without Expenditure Data: Evidence From Rural Ethiopia

Abstract

In economic studies, household economic status is usually proxied by measures of consumption or income. In recent years, several studies have advanced an asset-based index as an alternative measure of wealth status. In most studies, the asset-based wealth index is constructed with a standard list of assets comprising household ownership of consumer durables, the characteristics of the household's dwelling and sometimes household landownership.

Although a standard list may be useful when comparing households across countries or urban and rural residents in one country, the assets included may not always be relevant for studies focussing on rural areas or a particular rural area only. This paper addresses the question what assets should be included in the wealth index to the best reflect long-term economic status in rural Ethiopia. We use data from the Ethiopian Rural Household Survey (ERHS) collected in 1994 and 1995 to construct several asset-based wealth indices. We find not all standard assets are relevant locally and signs and heights of factor loadings differ substantially between localities, supporting the case of specified (local) asset listings. The specified asset index performs best compared to other asset indices when considering the distribution of food security across wealth quintiles and is at least as good a predictor of food security as per capita consumption measure for the same households.


Ethiopian Journal of Economics Vol. 15 (1) 2006: pp. 35-54