Estimates of State Price Levels for Consumption Goods and Services: a first brush (PDF)

This paper develops exploratory estimates of the spatial price differences for consumption goods and services at the U.S. state level. Spatial (place-to-place) price differences are important to regional and other sub-national accounting frameworks as they make possible comparisons of economic data that are adjusted for geographic differences in price levels. In international comparisons, these adjustments are termed purchasing power parities (PPP); when divided by exchange rates they are called national price levels. In areas with a common currency like the Euro, the exchange rates are the same and the PPP becomes the price level. Just as there are differences in price levels between European Union member countries, there are significant differences in the purchasing power of a currency across diverse areas of the United States, for example between metropolitan New York compared to rural South Dakota. I use the term Spatial Price Indexes (SPIs) to label these sub-national estimates of PPPs. The SPIs can be used to adjust consumption-related statistics, such as per capita incomes, expenditures and output, providing users with a more accurate picture of regional economic differences at one point in time. The SPIs are built up in this paper from two main data sets. The first is the principal source of consumer price information in the United States, the Bureau of Labor Statistics Consumer Price Index (CPI) for 38 metropolitan and urban areas, which is of course a time-to-time index. Aten (2006) presented spatial price index estimates for 2003 and 2004 for these 38 areas, which cover 87% of the population but only about 15% of U.S. counties. In addition, some states are not covered at all by the CPI. The second source of information is the county level rent surveys from the U.S. Census Bureau. The estimates presented here are generated using a multi-stage approach that bridges the results in the areas sampled by the CPI price surveys to the remaining non-sampled areas using the Census rent information.

Bettina H. Aten

Published