In recent debates over the impact of unemployment benefits, Benjamin and Kochin suggested that, with higher and easily available benefits, workers were induced to remain longer on the unemployment register, either searching more widely or enjoying more leisure. They estimated that benefit-induced unemployment accounted for one-third of the total between 1922 and 1930. Their "search-turnover" approach to the labour market implies an inverse relation between vacancies and unemployment. The higher the unemployment benefits, the more choosy workers will be, both in searching out job offers and accepting them once offered. An increase in benefits relative to wages will lead to higher vacancies at every level of unemployment (or higher unemployment at each level of vacancies). Deviations of actual from expected wages and changes in the structural imbalance of labour supply and demand might also be expected to shift the vacancies- unemployment relation.

CEPR Research Fellow Tim Hatton tests the Benjamin-Kochin theory by examining the relationship between the flow of workers off the unemployment register and the levels of vacancies and unemployment during the 1920s. Using data on the stock of vacancies outstanding, which are available only for the 1920s, he estimates the average duration of unemployment at 7-8 weeks and the average duration of a vacancy at less than one week. His analysis suggests that labour turnover was surprisingly high and that there were frequent short spells of unemployment. This might be thought to enhance the applicability of the search-turnover model.

But Hatton's econometric estimates for the 1920's do not reveal the strong vacancies-unemployment relation predicted by the search-turnover theory. Nor is the impact of the benefit-to-wage ratio on vacancies for a given unemployment level what would have been predicted by the theory. He finds no support for the Benjamin and Kochin view at all, insofar as it rests on an implicit search model of the labour market.

How does Hatton interpret the evidence from the 1920's? Given there were 8.5 wholly unemployed workers to every vacancy, workers could not be choosy about accepting job offers. Genuine new vacancies were taken up almost immediately, leaving only a few slow-moving vacancies at the labour exchanges. Workers were simply queuing at the exchanges, and a rise in the unemployment rate merely lengthened the queue - it did not cause vacancies to be taken up more quickly. Similarly, though unemployment insurance might be expected to shorten the queue of active job seekers, it did not retard the movement into vacancies of those in the queue. Hatton thus concludes that the search-turnover view of labour market frictions is not an appropriate framework for the 1920s and, if benefits did cause higher unemployment, then this must have operated through some other mechanism.