Are Unit Values Reliable Proxies for Prices ? Implications of Better Price Data for Household Consumption Measurement in a Low-Income Context

Household Consumption and Expenditure Surveys are key to consumption-based monetary poverty measurement. In the absence of market price surveys that are linked to Household Consumption and Expenditure Surveys, unit values are used as proxies for market prices in estimating nominal consumption aggregates, price deflators, poverty lines, and poverty statistics. This practice relies on the Hicksian separability assumption: within-commodity group relative prices are constant across space and the price of a single good is an accurate proxy for the commodity group price. To test, for the first time in a low-income context, whether Hicksian separability holds, this paper uses the price data collected for an extensive list of food items, including several variety/quality-differentiated products for specific items, in a national market survey that was conducted in Malawi in sync with the Household Consumption and Expenditure Survey that is the source of official poverty statistics. The analysis demonstrates that Hicksian separability fails to hold across space and time and that unit values are biased proxies for prices. Integrating the Household Consumption and Expenditure Survey and market survey data based on location and timing of fieldwork permits an assessment of consumption and poverty estimation based on market prices versus unit values. Relative to unit values, using market prices leads to higher food and overall consumption expenditures—both in nominal and real terms—while generating higher poverty lines and higher food and overall poverty rates. Compared to their counterparts based on unit values, spatially-disaggregated poverty estimates based on market prices exhibit a stronger correlation with nightlights —an objective proxy for living standards.