Rags and Riches: Relative Prices, Non-Homothetic Preferences, and Inequality in India

Abstract

It is well known that consumption patterns change with income. Relative price changes would therefore affect rich and poor consumers differently. Yet, the standard price indices are not income-specific, and hence, they cannot account for such differences. In this paper, we study consumption inequality in India, while fully allowing for non-homotheticity. We show that the relative price changes during most of the period from 1993 to 2012 were pro-poor, in the sense that they favored the poor relative to the rich. As a result, we also find that conventional measures significantly overstate the rise in real consumption inequality during this period.The main lesson from our study is the importance of accounting for non-homotheticity when measuring inequality. The price index literature has, as of yet, paid relatively little attention to this. In our application, however, it turns out that the allowance for non-homotheticity is quantitatively much more important than much discussed adjustments, such as those for substitution in consumption.

Measuring Welfare and Inequality with Incomplete Price Information

“…This finding contrasts starkly with the changes in real income calculated using food and fuel Paasche and Lespeyres indices which slightly magnify the already substantial convergence seen in nominal incomes. This result also stands in contrast to Almås and Kjelsrud (2017) who estimate non-homothetic price indices using a Quadratic AIDS demand system that does not impose quasi-separability but requires well-measured price changes for the full consumption basket to implement. 38 They find that inflation was pro-poor over the period 1993-2005. Why are our price index estimates lower for richer households?…”

“…In particular, the rich spent a large and increasing share of their budget on durables such as manufactures and on services. 39 These are exactly the categories for which unobserved quality differences make price data unreliable and so are omitted in Deaton's CPI approach which only covers well-measured food and fuels, and are crudely captured, if at all, by the government non-food CPI in the Almås and Kjelsrud (2017) approach. As touched upon in the introduction, lower inflation in these specific categories is consistent with the fact that the Indian trade 38 See footnote 6 for a detailed description of how Almås and Kjelsrud (2017) the ratio of the Indian food and nonfood CPI to navigate the lack of well-measured price data for categories beyond food and fuels.…”

“…Looking across the income distribution, our estimates bring to light that price inflation has been far from uniform, with significantly lower inflation rates for richer households-something that is not apparent from calculating standard price indices even when using income-groupand district-specific expenditure weights, or from estimating non-homothetic price indices using a Quadratic AIDS demand system and goods with observable price data as in Almås and Kjelsrud (2017). 6 Thus, while estimates based on standard approaches designed for settings with complete price data suggest that India saw significant convergence between poor and rich households over this period, we find that this convergence entirely disappears once we account for the differential inflation across income groups revealed by our approach.…”

The representative agent bias in cost of living indices

“…The nationally representative consumer expenditure survey of 2004-05 is used which samples about 120,000 households across rural and urban India. Following Almås and Kjelsrud (2017), we classify all expenditure into 11 categories. Tables 1 and 2 list these categories and also display across the urban and rural sectors, the mean budget shares as well measures of dispersionboth evaluated by the democratic density.…”

To adjust or not to adjust? Spatial price variation and the measurement of poverty*

“…For simplicity, I do not distinguish between different types of publicly provided goods and services. In the empirical exercise, however, I use measures of availability for several types of government facilities.12 Price levels are similarly likely to change with income because people's consumption patterns typically change, meaning that there is no such thing as a unique cost of living relevant for all income groups(Muellbauer, 1974;Pendakur, 2002;Almås and Kjelsrud, 2017).C 2022 The Author. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of Föreningen för utgivande av the SJE.…”