Public sector productivity: the sensitivity of productivity indexes to alternative output measures and index number formulas

Abstract

Focusing on a specific government agency, the Internal Revenue Service, and using publicly available data for the years 1956–1982, this article investigates two issues. First, because government agencies produce services which, in general, are not sold in markets, the appropriate measure and weighting of outputs is not obvious. Thus, it is of interest to know how sensitive an index of labor productivity is to the way in which output is measured. Second, in computing labor productivity, an index number formula must be selected. Dispersion among the index numbers is not uncommon, making the choice among formulas an important issue. Results indicate that the choice of both the output measure and index formula are important.