Model Identification for Estimating Missing Values in Space-Time Data Series: Monthly Inflation in the US Urban System, 1977–1990

Abstract

Many applications- as well as theoretically-oriented scholars who study economic phenomena are interested in features of inflation. Inflation in the United States has been indexed in a variety of different ways, including with the consumer price index (CPI). Because of time and resource constraints, this index is computed for selected urban areas either on a monthly, a bimonthly, or a semiannual basis. Because of representativeness and accuracy concerns, the manner in which these statistics are computed changed in 1987. The general problem to be addressed in this paper pertains to the incomplete time series that are due to (1) bimonthly collection of data for some urban places, and (2) redefinition of the original series for all urban places. The usefulness of completing series is found in studies that, say, try to relate temporal inflation trends with spatial variations in cost of living (see, for example, cost of living data compiled by the American Chamber of Commerce Researchers Association, 1989).