The 'Phillips Curve' relationship between wage changes and unemployment arose from a study of the forces determining the wage rate and the rate of employment of labour in the British economy from 1857 to 1913. Since Phillips wrote in 1958, the literature on labour market adjustment has increasingly focussed on models in which the labour market clears but in which there is imperfect information and slow adjustment.

In Discussion Paper No. 23, Research Fellow Tim Hatton examines a simple model which captures these new elements but at the same time bears a resemblance to the Phillips curve. 'New classical' macroeconomics emphasizes the difference between actual and expected wages as a source of economic fluctuations. In keeping with this, Hatton constructs a 'surprise' labour supply function in which the rate of employment depends on the unanticipated change in wages - the deviation of actual from expected wages - and on past values of the employment rate. The labour market is assumed to clear continuously, and so unlike the traditional Phillips curve literature, Hatton's model includes an equation for labour demand, depending negatively on the wage and also on current and lagged values of other variables. In order to derive the expected wage in the supply equation, Hatton uses the rational expectations approach. The expected wage is therefore assumed to be formed in a manner consistent with the model itself.

Hatton notes that these assumptions give rise to various restrictions, which can provide tests of the model. One approach is to a take fairly general model and test for the importance of variables which theory suggests should be excluded. He finds that the results of this test are not very decisive, but they do suggest that the strict classical version of the model is not strongly supported. These findings do not take one very far, however, since dropping any one of the model's several hypotheses might lead to the rejection of the model. Hatton uses a second type of test to determine which hypothesis might be leading to rejection. This involves testing for specific values of parameters within and between supply and demand equations. The results suggest that the restrictions implied by the rational expectations postulate are not strongly rejected while those implied by the surprise supply function are. Hatton finds that on testing another variant of the model the current change in the wage rather than the unanticipated change or 'surprise' is important for labour supply.

The period before 1914 is often thought of as 'classical' in the sense that the economy adjusted smoothly and rapidly, unimpeded by institutional rigidities. Nevertheless, as Hatton notes, his results support the traditional interpretation of Phillips, rather than the 'new classical' interpretation of the labour market before the first World War.