A procedure for estimating price indices in localities within one state, North Carolina, is presented and implemented. The paper makes a number of improvements over previous work, including estimating the association between local prices and national prices, and testing the potential capitalization of local taxes and public output into local prices. Intrastate income variation is reduced when the price index estimates are applied to nominal per capita incomes.
I.The Review of Regional Studies
II. PREVIOUS WORKThere is now a large body of evidence which suggests substantial variation in prices between localities in the nation. First is evidence from direct surveys of prices in different locations in the country. One part of this evidence comes from the, now discontinued, surveys of family budgets conducted by the Bureau of Labor Statistics (U. S. Bureau of Labor Statistics, 1982). The final report of these budgets for autumn 1981 showed a 12 percent difference in budgets between metropolitan and nonmetropolitan areas of the country for a four person family with an intermediate standard of living. The largest difference existed for rent (27 percent), and the smallest difference was found for clothing, transportation, and food at home (each with a 2 percent difference between metropolitan and nonmetropolitan areas). Furthermore, similar differences were found between metropolitan and nonmetropolitan areas in regions of the country.Other direct evidence of locational price differences comes from the American Chamber of Commerce Researchers Association (ACCRA). Each quarter ACCRA collects price data for 61 products and services in 289 locations across the country. ACCRA forms a price index by taking a weighted average of these products and services. 1 For example, the first quarter, 1995 report showed a 166 percent difference between the highest cost of living location (Manhattan, New York City) and the lowest cost of living location (Kennett, Missouri). Again, the greatest difference existed for housing, at 526 percent (American Chamber of Commerce Researchers Association, 1995).The last set of direct evidence comes from housing costs collected by the U.S. Department of Housing and Urban Development (HUD). HUD collects rent data as part of its housing subsidy programs. The rents are for units of a certain size and certain quality. Within North Carolina, for example, the 1993 data showed a 92 percent difference between the highest rent ($545/month) and lowest rent ($284/month) for a two bedroom apartment (U. S. Government Printing Office, 1993).Other evidence comes from studies which estimate differences in price indices between locations. Most of these studies have focused on price differences between states or metropolitan areas (