Decomposing Technical Change with Panel Data: An Application to the Public Sector.

The estimation of technical change in panel data models is considered using two competing models: the time trend model and the general index model. The authors' measure of technical change in both models is based on the input requirement function. Technical change is further decomposed into pure and nonneutral components. Empirically, focus is on measuring technical change in 366 Swedish Social Insurance Offices during the time period 1974-87 on the basis of a translog input (labor) requirement function with four outputs (services). Copyright 1995 by The editors of the Scandinavian Journal of Economics.