Paper, Experimental CPI for lower and higher income households (U.S. Bureau of Labor Statistics)

This paper examines CPI indexes for subsets of the target population defined by the bottom and top of the income distribution and compares results with the target population. We use data from the Consumer Expenditure Surveys (CE) to construct biennial and monthly market basket shares for groups of respondents based on their reported income, in order to calculate CPIs using modified Laspeyres and Tornqvist formulas respectively. From 2003 to 2018, we find the Laspeyres index for the lowest income quartile population rose faster than the index for all urban consumers. The Laspeyres index for the highest income quartile population rose slower than the index for all urban consumers. Chained CPI indexes for the income quartile populations rose slower than their Laspeyres counterparts. The measure of consumer substitution was lowest for the lowest income quartile population; the difference between the Laspeyres and Tornqvist index for the lowest income quartile population was less than half the difference for all urban consumers.