Follow us
Publications Commentary Research People Events News Resources and Videos About IFS
Home Publications GMM estimation with persistent panel data: an application to production functions

GMM estimation with persistent panel data: an application to production functions

Richard Blundell and Stephen Bond
IFS Working Paper W99/04
We consider the estimation of Cobb-Douglas production functions using panel data covering a large sample of companies observed for a small number of time periods. Standard GMM estimators, which eliminate unobserved firm-specific eects by taking first differences, have been found to produce unsatisfactory results in this context (Mairesse and Hall, 1996). We attribute this to weak instruments: the series on rm sales, capital and employment are highly persistent, so that lagged levels are only weakly correlated with subsequent first differences. As shown in Blundell and Bond (1998), this can result in large finite-sample biases when using the standard first-differenced GMM estimator. Blundell and Bond (1998) also show that these biases can be dramatically reduced by exploiting reasonable stationarity restrictions on the initial conditions process. This yields an extended GMM estimator in which lagged first-differences of the series are also used as instruments for the levels equations (cf. Arellano and Bover, 1995). Using data for a panel of R&D-performing US manufacturing companies, similar to that in Mairesse and Hall (1996), we show that the instruments available for the production function in first differences are indeed weak. We find that the additional instruments used in our extended GMM estimator appear to be both valid and informative in this context; this estimator yields much more reasonable parameter estimates. We also stress the importance of allowing for an autoregressive component in the productivity shocks.

More on this topic

Cemmap Working Paper CWP09/22
We compare two approaches to using information about the signs of structural shocks at specific dates within a structural vector autoregression (SVAR): imposing ‘narrative restrictions’ (NR) on the shock signs in an otherwise set-identified SVAR; and casting the information about the shock ...
Cemmap Working Paper CWP07/22
Economists are obsessed with rankings of institutions, journals, or scholars according to the value of some feature of interest.
Cemmap Working Paper CWP05/22
We designed a coaching program that focused on one aspect of teacher quality—teacher-child interactions—that researchers in education and psychology have argued is critical for child development and learning.
Cemmap Working Paper CWP04/22
It is often desired to rank different populations according to the value of some feature of each population. For example, it may be desired to rank neighbourhoods according to some measure of intergenerational mobility or countries according to some measure of academic achievement.
Cemmap Working Paper CWP44/21
We study a dynamic ordered logit model for panel data with fixed effects.