Estimating the Innovation Function from Patent Numbers_ GMM on Count Panel Data

SUMMARY
The purpose of this paper is to estimate the patent equation, an empirical counterpart to the 'knowledgeproduction
function'. Innovation output is measured through the number of European patent applications
and the input by research capital, in a panel of French manufacturing firms. Estimating the innovation
function raises specific issues related to count data. Using the framework of models with multiplicative
errors, we explore and test for various specifications: correlated fixed effects, serial correlations, and
weak exogeneity. We also present a first extension to lagged dependent variables. O 1997 by John Wiley &
Sons, Ltd.
J. Appl. Econ., 12, 243-263 (1997)
No. of Figures: 0. No. of Tables: 7. No. of References: 19.

1. INTRODUCTION
The relationship between R&D and productivity has been widely studied, raising some intriguing
puzzles. First, most of the empirical studies fail to find a significant and reasonable effect of R&D
on productivity. Hall and Mairesse (1995), comparing France and the USA, show that the
returns to R&D are very imprecisely estimated and parameter estimates are often small and even
negative as soon as endogeneity of regressors is taken into account. Second, it is very difficult to
identify the forces that drive technical progress. Pakes and Griliches (1985) and Griliches, Hall,
and Pakes (1988) have shown that it is empirically difficult to separate out, within the R&Dproductivity
relationship, shifts due to the innovation process from shifts coming from the
demand for innovations.
In this paper we focus on the relationship between investment in R&D and patents, used as a
measure of incremental knowledge. More precisely, we raise two issues.