1930s London
New Microdata

Traditionally, the interwar years in the United Kingdom have been portrayed as a period of "depression, gloom and failure,' dominated by the persistence of high levels of unemployment and its consequent hardship.

There are strong disagreements among economists however on the correct interpretation of the interwar labour market. According to the "Keynesian' interpretation, interwar unemployment was largely involuntary, a consequence of labour market disequilibrium and the insufficiency of aggregate demand. According to the "Equilibrium' interpretation, interwar unemployment was a voluntary phenomenon resulting from the generous provision of unemployment benefits. If British unemployment can be characterized as a voluntary response to the dole, then the presumed association between joblessness and hardship may not exist. The debate hinges on the proper interpretation of the available statistical evidence. The equilibrium view, for example, implies a positive correlation between unemployment and the ratio of benefits to wages. Yet the available data are sparse, and analyses of the interwar period typically make use of no more than 19 annual observations. Tests of relationships based on so small a sample are not powerful and are inevitably sensitive to the inclusion of particular observations.

In Discussion Paper No. 51 Susan Freiwald and Research Fellow Barry Eichengreen argue that new sources of data are needed to clarify the operation of the interwar British labour market. They note that labour economists increasingly test hypotheses about individual behaviour at the individual level, using "microdata' sets, which take the individual or household as the unit of observation. Our understanding of British economic history would be enriched by comparable microeconomic studies of the interwar labour market.

"Microdata' based on official sources, such as the census, are unfortunately much less readily available in the UK than in other countries. For some time to come, microdata for interwar Britain will have to be drawn from other, non-official sources. Freiwald and Eichengreen draw attention to the local surveys of British wage earners carried out by academic or charitable bodies during the interwar period. These surveys, however, were usually conducted for specific purposes related to the interests of the individual or charitable organization in charge, and they pose problems for the historian who wishes to utilize them for other purposes.

The Discussion Paper describes one such survey conducted by the London School of Economics, the "New Survey of London Life and Labour', which gathered information on the Greater London working class in the years 1929-1931. Freiwald and Eichengreen use the survey to construct a historical data set which gives a picture of unemployment in interwar London and can be used to analyse the influence of unemployment insurance on labour market behaviour.

Problems arise in the analysis of data from such surveys. The New Survey does not extend beyond London, includes no observations on middle-class wage earners, and provides at best limited information on employment by occupation and industry. The authors describe how these limitations may be circumvented. The absence of information on the middle class creates a sample which is non-random and unrepresentative of the population as a whole. Nevertheless, econometric techniques can be used to correct the estimates for the absence of high-income workers in the sample. The geographical limitations and the paucity of occupational data in the survey can be overcome by using it in conjunction with surveys of other regions.

Such survey data can illuminate a variety of historical issues, in addition to the debate over the causes of interwar unemployment. Freiwald and Eichengreen argue that studies of female labour-force participation, fertility behaviour or housing expenditure could benefit from the use of such historical materials.


From Survey to Sample: Labor Market Data for Interwar London
Barry Eichengreen and Susan Freiwald

Discussion Paper No. 51, February 1985 (HR)