Investment and Value: a Neoclassical Benchmark

Abstract

Do investment models fit firm-level data? - which model fits best? To answer this question we estimate alternative models using Compustat data. We find that both a version of the Hayashi (1982) and a model with decreasing returns to scale in production fit firm-level data very well. Our estimates suggest that there is substantial measurement error in Q. This measurement error implies that the investment-cash-flow regressions that have received so much attention are ineffectual to discriminate among alternative models. In fact, the models that we estimate generate empirically plausible cash-flow and lagged-investment effects even though they were not designed to produce them.