The cyclical behaviour of wages
In 1938 Dunlop challenged the assertion in the General Theory that wages moved countercyclically. The resulting debate on the cyclical movement of wages deserves study as an episode in the history of economic thought. This is done in chapter 2 which reviews the theoretical issues and chapter 3 which reviews the empirical work. To understand this history requires some analysis of the meaning and significance of the debate. At one level the debate can be interpreted as the search for a 'stylised fact'. This is apparently an empirical question and part of the thesis will be concerned to use data for various countries, time-periods, cycle phases, and industries to examine whether there is any systematic cyclical pattern in wage movements. The conclusion of the empirical analysis is that there is no such empirical regularity. At a second level the debate was theoretical. The empirical observation that wages moved procyc1ica1ly was thought to falsify a prevailing theory. What is interesting about this debate is the light it sheds on the response of economists to apparent falsification. A third level of the debate is the issue of inference. Keynes tended to treat theory as prior, attacking 'pseudo natural science procedures'. Keynes was not opposed in principle to statistical work informing theory: although in practice he did not attempt the empirical investigation into cyclical wages for which he called. Thus from a different methodological standpoint Burns and Mitchell criticise the theorist who 'often stops before his work is finished'. Current econometrics would emphasise the need for identifying assumptions before estimates could be used to test hypotheses. In this framework, the implications for theory of any reduced form regularity would be ambiguous in the absence on non-data based identifying assumptions. This thesis uses the history of the debate and the empirical analysis to illustrate these themes of observation, theory and inference.