1930s Unemployment
No single explanation

Throughout the interwar period two benefit systems operated in parallel in the United Kingdom, the national unemployment insurance initiated in 1911 and made available to all workers in 1921, and the vestiges of the nineteenth century Poor Law. Neither had been designed as a universal system of payments as of right, yet a series of Acts abolished most of the conditions and limitations on benefits. This has prompted arguments by Benjamin and Kochin among others that unemployment increased during this period in large part because people became 'work-shy', as unemployment benefits rose in real terms or relative to wages. According to this view the increase in unemployment arose from a deficiency of supply rather than of aggregate demand as was commonly supposed.

A second controversy has been sparked by the claim that excessive growth in real wages created unemployment during the interwar period. Mrs Thatcher and Mr Lawson have recently advanced a similar argument, claiming that 'excessive' real wage growth has created unemployment in the 1980s. In Discussion Paper No. 105, Research Fellow Michael Beenstock and Peter Warburton suggest that these two arguments are related. If the social security system encouraged people to join the dole queues, the supply of labour was reduced. This in turn put upward pressure on real wages and so reduced the level of employment and output that firms found profitable.

Beenstock and Warburton explore these issues empirically and find evidence for both propositions, using an econometric model of the interwar labour market in Britain. They find that during the interwar period not only did real wages affect unemployment in interwar Britain, but that real wages behaved as they did in part because unemployment benefit became more generous.

Beenstock and Warburton argue that the Benjamin and Kochin argument cannot be properly evaluated by regressing interwar unemployment rates on GDP and 'replacement ratios' (the ratio of net income from employment to unemployment benefit). Instead, a structural model of the interwar labour market is needed. The authors specify labour demand as varying inversely with the 'own product' real wage (the ratio of wage rates to output prices), as well as other variables such as the volume of world trade, the economy's capital stock and the money supply. Labour supply in Beenstock and Warburton's model depends on disposable real wages and other variables such as the unemployment rate (representing the 'discouraged worker' effect). Money wage growth is modelled as an 'augmented Phillips curve' and depends on inflation expectations, an underlying rate of real wage growth and the difference between the actual and 'equilibrium' rates of unemployment. The model is completed by an equation which specifies the 'equilibrium' unemployment rate, which is taken to depend on the real level of benefits. The equilibrium unemployment rate is of course unobservable and so the final equation cannot be estimated: Beenstock and Warburton therefore substitute this equation in the money wage equation to obtain a composite equation. The labour supply, labour demand and wage change equations are then solved as a complete system.

In CEPR Discussion Paper No. 66, Research Fellow Tim Hatton argued that existing market-clearing models do not adequately characterize the interwar period. Beenstock and Warburton argue that their model does possess a market-clearing pedigree and that simulations of the estimated model over the period 1922-38 track the observed data quite impressively. Beenstock and Warburton conduct a series of simulation exercises designed to shed light on the reasons for the dramatic interwar surge in unemployment and its subsequent decline. It is impossible, they conclude, to attribute the rise in unemployment to a single cause. They first simulate what the level of unemployment would have been had the level of benefit remained constant, and find that increases in benefit levels were responsible for 2.7 percentage points of unemployment at its 1932 peak of 15.1%. In addition, the 1930 decision to shift onto the administrators of benefit the burden of proving whether each claimant was 'genuinely seeking work' accounted for a further 2.8 percentage points of the 1932 unemployment rate. The simulations also indicate that the level of aggregate demand had an independent influence on the behaviour of employment. Simulating a modest growth rate of 2% per annum in place of the collapse of world trade which actually occurred removed a further 2.4 percentage points from the 1932 level of unemployment.

It is often argued that the disparity between the high unemployment of the 1930s and the low unemployment of the late 1940s is a reflection of the success of the Keynesian revolution. Beenstock and Warburton therefore conduct a final simulation, assuming no changes in benefit levels, no slump in the volume of world trade or in the wholesale prices of UK goods, and a smooth growth of the real money supply. The UK economy emerges from this simulation with an underlying unemployment rate of around 3% in the late 1930s, not greatly different from the rate observed 10 years later.


The Market for Labour in Interwar Britain
Michael Beenstock and Peter Warburton

Discussion Paper No. 105, April 1986 (IM)