The Impact of Capital Measurement Error Correction on Firm-Level Production Function Estimation

Abstract

Based on a large panel of Czech manufacturing firms, we estimate firm-level production functions in 2003–2007 using the Levinsohn and Petrin (2003) and Wooldridge (2009) approaches, correcting for the measurement error in capital. We show that measurement error plays a significant role in the size of the estimated capital coefficient. The capital coefficient estimate approximately doubles (depending on the particular industry) when we control for capital measurement error. Consequently, while the majority of industries exhibit constant or (in)significantly decreasing returns to scale when the standard methods are used, increasing returns cannot be rejected in some industries when the estimation is corrected for capital measurement error.