well-established tenet of the theory of unemployment is that unemploymentbenefits tend to increase the unemployment rate. This prediction is basedon two central pieces of labour economics, namely the job-search theory andthe efficiency-wage theory. In the job-search theory, unemployment benefitscause an increase in the duration of unemployment by reducing the incentive ofunemployed workers to search for and accept a job; in the efficiency-wagetheory, they reduce the cost of being unemployed and induce workers to ask forhigher wages, so the demand for labour decreases. Since the unemploymentrate is a function of the duration of unemployment and its incidence, unem-ployment therefore increases with both the level and the duration of benefits.Despite the strength of this theoretical prediction, the empirical evidencethat supports it is not very strong. Those studies which have detected an effectof benefits on unemployment duration have found it to be rather small. Othershave failed to detect any such effect at all. On average, estimates have impliedthat a 10 percentage point increase in the replacement rate (i.e. the ratio ofunemployment benefits to earnings from employment) would be associatedwith an increase of about 1-1½ weeks only in unemployment duration (Lancasterand Nickell, 1980; Meyer, 1990). This suggests that the impact of benefits onunemployment probably depends on a set of factors wider than the one usuallyconsidered in labour economics. The aim of this article is to investigate theconditions under which the provision of unemployment benefits would leavethe rate of unemployment unchanged.The main argument brought forward in the following analysis hinges onthe distinction between