Differences in Growth Rates and Kaldor's Laws

Abstract

where p is the rate of growth of productivity, q is the rate of growth of output, e is the rate of growth of employment, and a, (3, -y and 8 are parameters. Kaldor regressed e on q to support his conclusion that the slow growth of industrial productivity was due to a shortage of labour which prevented exploitation of economies of scale. He took the growth rate q of manufacturing output as an independent variable and regressed the growth rate of productivity p and of employment e on q, concluding that each percentage addition to the growth of output requires a 0.5 per cent increase in the growth rate of employment, in terms of man-hours, and is associated with a 0.5 per cent increase in the growth of productivity. The argument depends on a positive relationship between productivity and employment, and Rowthorn (1975a) examined this by regressing p directly on e rather than indirectly by regressing p or e on q as shown in (1) and (2). He concluded that the positive association depended entirely on the presence of one extreme observation, and that if Japan were excluded from the sample the association between p and e disappeared. The main purpose of this paper is to argue that both sets of results are subject to simultaneous equations bias and that mis-specification of estimated equations leads to incorrect estimates and wrong conclusions.