| Abstract: | ABSTRACT. Kraybill and Dorfman (1992) propose a model of intermediate and export demand which uses ordinary least-squares and linear systems techniques to produce a state-space representation of the time element of output change. Their model produces dynamic multipliers which trace the temporal path of regional growth, and has many advantages over previously employed time series methods. This study extends their methodology to accommodate structural shifts and outliers found in the least-squares relationship between industry and export output by using a recently-introduced technique–multiprocess mixture estimation. An application of the Kraybill-Dorfman method and the extensions proposed here to monthly time series data on Ohio employment is used to illustrate these issues. |
