Theoretical predictions about the effect of unemployment on school leaving are ambiguous. Unemployment might affect the decision to leave school in a number of different ways. (i) Current youth unemployment. A higher probability of current unemployment for those who choose to leave school would tend to discourage early school leaving, first by reducing the expected gains from job-search (a 'discouraged worker' effect) and second by reducing the opportunity cost of additional education. (ii) Expected future unemployment. The theoretical predictions here are less clear-cut. An increase in the probability of future unemployment that affected both those with and without additional education would tend to reduce the rate of return to education, and hence education demand. On the other hand, those with more education may be less affected by rises in unemployment and this may encourage further schooling. These two opposing effects have been formalised as a shift and a tilt in the unemployment probability-schooling relationship (Kodde, I988). Fig. I suggests that in the United Kingdom both effects could have accompanied the transition from the late 1970S to the I98o?. The full effects of uncertainty on education investment in a two-period model depend on the correlation between the uncertainty in the two periods (Levhari and Weiss, 1974). How current unemployment rates are used to form expectations of future unemployment is clearly an important issue here. Also, some individuals may be able to undertake job search before committing themselves to leave school, and whilst affected by the risk of future unemployment may be comparatively unaffected by the risk of current unemployment. (iii) Effects ofparental unemployment. Parents' unemployment in general reduces family income; their children's educational demand may fall as a consequence for three reasons. First, the consumption demand for education will be lower. Second, a reduction in the first period income reduces the investment demand for education, assuming decreasing absolute risk-aversion (Kodde, I 988).