摘要：During the past two decades, gross state product (GSP) has become a major indicator of state economic activity. GSP has great conceptual appeal with its definition being "the gross market value of the goods and services attributable to labor and property located in a state", (Renshaw, et.al., 1988, p. 30). Until the late 1980s, GSP was not an official indicator released by a government agency. Rather, it was constructed by approximation techniques traceable to the work of Kendrick and Jaycox (1965). Since there was no official standard of computation, analysts were on their own to compute GSP accounts for their locale as they saw fit. The method proposed here for creating preliminary GSP estimates uses predictions from single equation models that relate official GSP estimates to either the synthetic GSP estimates of the corresponding industry or to selected components of the synthetic GSP estimates. Four prediction models for official GSP will be examined, one pertaining to current dollar GSP and the remaining three pertaining to constant dollar GSP. This division of attention between current and constant dollar magnitudes is made to allow for the uncertain future of reporting lags in industry GDP and deflator data.
