Exploring ways to estimate endogenous productivity
Files
First author draft
Date
DOI
Authors
Guillard, Charlotte
Olivari, Jocelyn
Jaumandreu, J.
Version
OA Version
Citation
Charlotte Guillard, J Jaumandreu, Jocelyn Olivari. "Exploring Ways to Estimate Endogenous Productivity."
Abstract
This paper explores methods to assess the impact on firm productivity of the investment in
innovation activities (endogenous productivity). It uses 23 years of firm-level data generated by the Spanish ESEE survey (1990-2012). We first apply traditional approaches to the measurement of productivity such as Solow Residual and Multilateral Index. We then replicate the estimation of the model in Doraszelski and Jaumandreu (2013) using more data now available. We briefly compare both approaches and discuss about the importance of treating inputs and productivity as endogenous. We then discuss ways to apply the model for endogenous productivity when there are no firm-level output price indices available, a limitation of many data bases. Including the demand of the firm in the estimation allows us to obtain a "composite" of productivity, demand elasticity, and demand heterogeneity. This unobservable, often called "revenue productivity", is the estimate of productivity used by most scholarly studies. We find that this composite does not behave as productivity and, in particular, neither is greater for firms that perform R&D nor its distribution shows stochastic dominance. Its persistence and returns also give misleading results. Our findings highlight the importance of producing more complete databases, especially if policy implications are to be drawn. They also suggest caution in interpreting the results based on revenue productivity.