Labour Markets
Reinterpreting Phillips's Curve

Empirical work on the aggregate British labour market from 1857 to 1938 is dominated by the famous pioneering work of Phillips and Lipsey. However, models of equilibrium unemployment which have developed from the original literature on the Phillips curve have not been generally applied to the period before 1945. In a recent Discussion Paper, Research Associate Tim Hatton examines some simple equilibrium models of the aggregate wage and unemployment rate for various phases of British economic history from 1857 to 1938.

Studies which have concentrated on the pre-1945 and especially the pre-1914 period have followed directly in the tradition of Phillips. Proportionate changes in money wages are treated as a response to labour market disequilibrium and/or other variables. An alternative interpretation of such a relation is as a labour supply function in which the supply of workers for employment (and hence the level of measured unemployment) depends on the deviation of actual from expected or perceived wages and/or prices. Such models may be based on search theory or on inter- temporal labour/leisure substitution. For a proper market model one also needs a labour demand function - a feature largely ignored in the early literature on the Phillips curve.

Hatton specifies a basic model in the context of which a number of historical issues may be raised. He uses a modern 'classical' specification of labour supply in which supply depends on the deviation between the local wage and workers' estimates of the economy-wide average wage. Thus, in aggregate, labour market fluctuations will depend on the current 'surprise' component of the average wage. The rest of the model consists of a standard labour demand curve and an assumption that the market clears each year.

In previous work Hatton reported a version of this model with a simple 'nominal wage surprise' supply function, for the period 1857-1913. In that study the restrictions implied by rational expectations were tested and, on the whole, not rejected while those implied by the surprise supply function were rejected. In Discussion Paper no.18, Hatton takes a more flexible approach and tests several variants of the model. In particular, Hatton examines whether the wage 'surprise' generated by a simple forecasting equation outperforms the current change in wages as featured in the traditional Phillips curve. He finds that in general wage and price change terms tend to dominate wage and price surprise terms in the model. A simple specification including only nominal wage changes cannot be rejected against a variety of more general specifications.
Some variants of this model (particularly when prices are excluded from the labour supply equation) perform well for the period up to 1913. When the period is broken into three at 1874 and 1894 the functions change from one period to the next but these changes are not statistically significant. But when the 1857-1913 period as a whole is compared with 1921-38, significant differences do emerge. In particular, Hatton finds that the interwar period does not support the classical interpretation as well as does the prewar period.

Hatton considers whether these findings are simply an artefact of data or a reflection that the interwar and prewar labour markets did operate differently. One obvious possibility is the extension of national collective bargaining (which was virtually absent before the War) via national representative bodies of unions and employers, and in unorganised industries, wage boards made the difference. This, coupled with structural readjustment imposed on the economy in the 1920s and 1930s could have led to labour market disequilibrium.

An alternative view, inspired by Benjamin and Kochin, is that the introduction of national insurance benefits and contributions served to mask the operation of the equilibrium labour market by shifting supply and demand curves for labour. Hatton investigated whether the introduction of social insurance, involving the right to unemployment benefit for employees and the contribution to the insurance fund by employers, could account for the shift after the first World War. He found little evidence that this could explain the break in supply and demand functions. Insofar as insurance did have direct effects, it appears to have been on the demand for labour rather than its supply.


The British Labour Market in Different Economic Eras: 1857-1938
T J Hatton

Discussion Paper no.18, May 1984 (HR)