An Empirical Study of Production Inefficiency in the Presence of Errors-in-the-Variables.

The concept of a frontier production has been attacked on the grounds that mismeasurement of output makes it impossible to separate efficient from inefficient firms, i.e., what looks like inefficiency may actually be mismeasurement of output. In this paper, we illustrate one method for estimating a frontier production relation when output is poorly measured--leading to errors-in-the-variables. The technique, based on Goldbergers factor analysis model, is meant to avoid not only spurious findings of inefficiency but also an overestimate of scale economies. Our empirical example involves a military application U.S. Naval Bases. In this example, our taking account of the errors-in-variable problem does not decrease the indicator of average inefficiency. It does, however, substantially reduce the measured economies of scale. Author