اثرات بیمه بیکاری بر درآمد پس از بیکاری

There is surprisingly little research into the effects of unemployment insurance (UI) on postunemployment wage outcomes. Moreover, the few existing studies are sufficiently varied in their approach and conclusions that experienced observers have reached very different interpretations of their implications. We provide new estimates of the effect of UI on subsequent earnings, using data on workers displaced in the period 1983–1990. Our objective is to provide a systematic evaluation of the approaches used in the existing literature. We find some limited evidence of a favorable impact of UI on earnings, but only when we compare recipients with nonrecipients. Even in this case, our point estimates lie well below those reported in earlier studies that pointed to beneficial UI effects.

“At the present time one can find no compelling evidence in support of the proposition that UI increases wages because of better job matches and increased job stability.” ( Cox and Oaxaca, 1990; p. 236) “In the smaller number of studies that examine the predicted beneficial effects of UI, economists have found support for their partial equilibrium predictions. Better insured workers appear to find higher-wage jobs than those found by less-insured workers.” ( Burtless, 1990; p. 102) Given the universal finding that unemployment insurance (UI) extends the length of unemployment spells, the available evidence would seem to suggest that the incentive structure of the UI system is primarily negative, and that there is little reason to defend the system other than on equity grounds. This is perhaps even more the case today than in the past, as the most recent estimates of the duration effects of UI can be construed as pointing to rather strong disincentive effects. The reason for the seeming indictment is that there has been little research regarding the benefits of the UI system on the job-search process of unemployed workers, and in particular on its effects on postunemployment wage outcomes. As suggested by the above quotations, the limited number of studies of the effect of UI on postunemployment wages are varied enough in their approach and conclusions that experienced observers can reach very different interpretations as to their findings. In this paper, we provide new estimates of the effect of UI on subsequent earnings for a uniform sample of no-fault job losers who were not recalled by their previous employers. Our goal is to offer a unified approach within which the approaches taken in the extant literature can be evaluated. In our analysis, we look for broad patterns in the data in order to establish the circumstances (if any) where UI is associated with wage gains or better job matches. To anticipate our findings, we report that there is some modest evidence in support of UI increasing the postunemployment wages of recipients. This finding only emerges when we compare the earnings of recipients with nonrecipients, with due account of biases that can occur in constructing the sample for this comparison. The evidence is not strong, however, and is even weaker when we restrict our attention to observations on the immediate job following displacement. Additionally, we find no evidence that UI leads to greater job stability, but some evidence that receipt of UI tends to lower the variability of wage changes.

In marked contrast with the analysis of UI effects on unemployment, there has been remarkably little research into the wage effects of subsidized search. This is surprising, given the difficulty of discriminating between the alternative theoretical models on the basis of duration data alone and in view of the argument that UI enables workers to sort themselves into better jobs. Not only is the extant evidence on wage gains sparse, but it has also raised more questions than it has answered. The present paper has sought to provide a clarification of the issues that have arisen by examining a fairly large number of specifications. Unlike much of the previous literature, we find little evidence of beneficial effects of UI on wages using samples consisting of claimants only. However, limited evidence of broad positive effects are uncovered from a comparison of recipients with nonrecipients, controlling in an admittedly crude way for sample construction problems attendant upon the distinction between UI eligibility and receipt. The point estimates of UI effects from this comparison are nontrivial, but they are considerably smaller than those reported in earlier studies that found support for a UI wage effect. The evidence of the effects is not strong in a statistical sense, however, as the coefficient estimates themselves are at best marginally statistically significant. In sum, it is difficult for us to argue that our results are clearly inconsistent with either of the quotes with which we began the paper. For an individual with the priors of someone like Burtless, our estimates are not likely to lead to a change in the belief that UI is probably beneficial in terms of wages. However, our estimates are of a lower order of magnitude than reported in studies conditioning such beliefs. And for individuals with Cox and Oaxaca's priors, it is possible to point to the weak statistical basis of our support to continue to conclude that there is still no `compelling evidence' that UI improves earnings. The resolution of these disparate beliefs may ultimately depend on additional evidence, either from alternative datasets or more fundamentally from experimental designs that randomly vary actual benefits.