Economic stagnation and high unemployment in the 1980s have led many politicians and policy-makers to look back fifty years to the Great Depression in an attempt to draw parallels, test hypotheses and develop policies to cope with some of today's economic problems. CEPR Research Fellow Tim_Hatton convened a one-day workshop on 'Labour Markets in an Underemployed Economy: The Experience of the Interwar Years', bringing together a group of economists, historians and civil servants to discuss some of the historical problems of the British economy and to propose areas of research which might inform current debate on the causes of and solutions to mass unemployment. The Economic and Social Research Council provided a grant towards the cost of the meeting.

CEPR Programme Director Barry Supple chaired the morning session, in which three papers were discussed. In the first, Barry_Eichengreen, Harvard University and CEPR, presented preliminary results of his analysis of microeconomic data on unemployment rates and benefits collected in the late 1920s for the New_Survey_of_London_Life_and_Labour. He pointed out that aggregate data on unemployment cannot provide detailed information on how the experience of unemployment and the benefits of unemployment insurance affected individual families, but that cross-section panel data can. There are, of course, problems with the use of such panel data - London is not representative of the UK, the income of families may be mis- reported, definitions of 'unemployment' may vary among households depending on the way in which short-time work or casual employment is reported. Nevertheless, with a sample of 1300 working class households, Eichengreen felt that some consistent patterns were likely to emerge. In his regression model, he found that unemployment was positively related to age, family size and wage, but that it was negatively related to insurance benefit entitlement. This suggested that some recent and controversial interpretations of macroeconomic data which identify a positive link between the real level of unemployment insurance benefit and the proportion of the workforce unemployed are not obviously corroborated at the microeconomic level. The juvenile labour market, however, behaved rather differently, with high wages associated with low unemployment probabilities. Eichengreen concluded that his disaggregated microeconomic approach indicated a need to recognize wide economic and behavioural differences among different groups of workers.

Several participants questioned the reliability of the cross-section panel data. To what degree was it right to impute unemployment benefit income to those unemployed individuals who did not declare a benefit income on the survey forms? It was suggested that the level of disqualification from benefit experienced by the workforce was likely to be higher than Eichengreen had allowed. The microeconomic approach received a good deal of support, however, and further work with the small amounts of interwar panel data available was encouraged.

In the second paper, Stuart_Riddle, Cambridge University, looked at old people in the workforce in the interwar years and how their job prospects were affected by government action. A recurrent belief of governments has been that employment prospects for young persons can be improved if the elderly are removed from the workforce, but there is no clear evidence from any period that this in fact occurs. It was, nonetheless, an important driving force behind the introduction of pensions for sixty-five year olds in the mid-1920s.

Riddle showed that older workers suffered both a greater incidence of unemployment and longer periods of unemployment than did young workers, and that they became overly concentrated in declining, low-wage industrial sectors. This was partly due to the policy of some employers to reward loyal workers by giving them low-wage, low-productivity 'pension posts' in old age. Demand for old workers fell in the interwar period due to technical change. This fall in demand may have exceeded the fall in the supply of old workers brought about by the extension of the pension scheme. Did this cause the growing plight of older workers or was it a function of a general decrease in aggregate demand? Riddle stressed the role of interwar demographic trends which increased the number of elderly in the population at a time when employment opportunities were declining, and he noted that similar trends are affecting Britain today.

Steve_Broadberry, University College Cardiff, concluded the morning session with a paper on nominal and real wage adjustment in interwar Britain. Then, as now, inflexible wages were often cited as a major cause of mass unemployment and a barrier to economic recovery. Broadberry developed a two-equation model of wage and price determination to test wage adjustment to nominal shocks, such as the return to the gold standard in 1925, and real shocks such as the American great depression. Nominal wages decreased substantially between 1921 and 1923. This was difficult to model adequately, and seemed to be more closely related to the fall in trade union density than to the operation of sliding-scale wage agreements. Taking the period 1924-38, he found that there was some nominal wage rigidity in the short run. In the long run the real wage was flexible with respect to real shocks, being determined jointly by the terms of trade and labour productivity. When these conclusions are set in the longer-run perspective of changes since 1870, they point to an historically high degree of wage flexibility in the interwar period. The high level of interwar unemployment could not be simply explained by labour market rigidities.

The model adopted by Broadberry was questioned by Mark Casson and James Foreman-Peck, who doubted that labour productivity was exogenous or that the terms of trade was the best measure of the impact of the foreign sector - import prices were suggested as an alternative. Unemployment was not a significant determinant of nominal wages in Broadberry's model, and he used this to deny the existence of a Phillips curve between the wars. Jon Stern challenged this view, arguing that potentially large changes in the supply of labour in this period would make the unemployment/vacancies ratio a more appropriate explanatory variable.

In the afternoon, CEPR Programme Director Roderick Floud chaired discussion of two papers on female labour force behaviour between the wars. In the first paper, Gillian D'Arcy pointed out that most contemporary and modern discussion of women's employment has been predicated upon a belief that female participation rates should rise as political and social emancipation progresses. The reasons for such a relationship are not clear. The fall in family size and the development of light manufacturing and clerical work in the 1920s and 30s increased opportunities for female employment, but the participation rates rose only slowly. D'Arcy suggested that social convention, women's domestic ideology and a desire to re-establish pre-war social conditions were all barriers to greater economic participation by women. Tim_Hatton, Essex University and CEPR, followed this with a paper reporting a preliminary analysis of female participation based on cross- section evidence of 144 urban areas drawn from the 1931 census. There were specification problems which generated some unexpected results, but his estimates suggested a positive relationship between the wage rate and participation. The most important explanatory variable, however, was local history, or more precisely, the town's female participation rate twenty years earlier. Hatton's paper confirmed the conclusions of D'Arcy that local patterns of female employment in the 19th century continued to dominate women's attitudes to the labour market in the interwar years; the reduction of these local and regional disparities has taken place mainly since 1951.

In the discussion of these two papers, Sean Glynn suggested that the high female participation rates achieved during World War Two and maintained through the 1950s point to the ability of economic forces, especially high wages, to overcome social conventions. Heather Joshi felt the picture was more complex. Since the Second World War, women workers have been concentrated in part-time jobs, which allows them both to earn money and maintain a domestic role; the 1930s offered many fewer part-time jobs. Such dissimilarities between the structure of the labour market in the 1930s and the 1980s are as important to recognize as are the parallels.

What conclusions emerged from the workshop - what can we learn from the 1930s? All five papers touched on the difficulty of finding adequate data on interwar and indeed more modern labour markets. Macroeconomic data give little scope for a disaggregated analysis of different types of labour, and the small number of annual observations is a constraint on econometric work; microeconomic data are difficult to obtain, patchy and possibly unrepresentative, and they may say little about change over time. More use might be made of quarterly data, of unpublished unemployment insurance records, and of panel data. There was a need to channel new information found in archives to economists who could analyse it formally. This is particularly important as investigation of subsections of the labour market, delineated by age or sex or location, requires specialized data often not available in published sources.

The papers suggested that while labour market responses to purely economic forces in interwar Britain were fairly rapid, there were important constraints imposed by historical experience and social attitudes, and these took several decades to change. The interwar period is the only example we have of long-run mass unemployment in the modern period. If high unemployment were to persist through the 1980s, the experience of the interwar period could contribute much to our understanding of effects on family income, wealth, health and poverty.