Examination of cross-section data on non-contractual construction workers in Egypt reveals strong attachment to the sector despite extreme demand instability. Also present are statistically significant wage differentials between construction trades that cannot be attributed to differential costs of skill acquisition. These observations suggest that employers might be compensating their non-contractual workforce for recurrent unemployment so that they can rely on a steady supply of qualified workers. We rely on a static labor supply model and investigate the theoretical consequences of rationing, turnover, and randomness in employment and unemployment durations. Assuming that the risk- averse worker would demand an expected utility which is at least as high as the reservation utility provided in the unconstrained sector, we derive structural expressions that quantify the anticipated and unanticipated components of compensation for employment risk. To test the model we use data from two rounds of the labor force surveys conducted in 1988. The first data set allows us to construct trade-specific measures of risk based on information on individual spells of employment and unemployment. The second allows us to estimate a restricted version of our model based on quarterly information on aggregate unemployment experiences of the workers. We find strong evidence that workers are compensated for the anticipated component and weak evidence that they are compensated for the unanticipated component. The estimated magnitudes reveal that employers provide only partial compensation against recurrent unemployment.
