Price Adjustments and Poverty Measurement

Measuring poverty entails making interpersonal welfare comparisons, that should account fordifferences in prices faced by households, both over time and across space. This paper investigates the impact ofseemingly minor differences in the practical implementation of price adjustments, by developing an analytical frameworkthat is consistent with standard consumer theory and mindful of the data limitations faced by practitioners. The mainresult is at odds with common sense: even when multiple price indexes are available, say a food and a nonfoodConsumer Price Index, it turns out that using a single price index, the total Consumer Price Index, to adjust theconsumption aggregate is recommended. The practice of adjusting the components of the consumption aggregateseparately, using matching deflators—food expenditure with the food index and nonfood expenditure with the nonfoodindex—can lead to a systematic bias in the welfare measure, and consequently in poverty and inequality measures. Thedirection of the bias can be easily predicted based on the price level and household consumption patterns. On theinterplay between spatial and temporal deflation, the findings show that temporal deflation should be carried outbefore implementing adjustments to spatial cost-of-living differences. The paper illustrates these findings using theIslamic Republic of Iran’s 2019 Household Income and Expenditure survey: the bias in the headcount poverty ratedue to incorrect deflation is substantive (5–10 percent for estimates at the national level, 15–20 percent in urban andrural areas, and more than 30 percent for district-level headcount rates). Higher-order Foster-Greer-Thorbecke poverty measures are even more affected.