Output, wages and the demand for labour : Some evidence from panel data

Abstract

Using a sample of small and medium sized firms, we investigate the relationship between wages and employment. Our data reveal a stable cross-section correlation between wages and productivity, consistent with the neoclassical demand for labour theory in general and the Cobb Douglas assumptions in particular. These results do not indicate anything about the direction of causation. A positive correlation between wages and productivity can arise from capital-labour substitution as wages change but other explanations are also plausible. Intervening variables are for example a particular concern in the analysis of panel data. In this paper, we adopt the generalized random effects framework put forward by Chamberlain. A series of exogeneity tests gives some support to the neoclassical notion that at the micro level, wages affect employment and productivity but not vice versa. The evidence presented in this paper is rather weak, however, and our data do not reject a restriction to a purely static relationship. In this specification, parameter estimates are not neoclassical. The wage-elasticity estimates obtained from the neoclassical cost-minimization model are of order 0.2 - 0.4, which is quite reasonable. Our results give support to the hypothesis that measurement errors have biased some of the earlier elasticity estimates from panel data towards 1.