افزایش بهره وری از بیمه بیکاری

This paper argues that unemployment insurance increases labor productivity by encouraging workers to seek higher productivity jobs, and by encouraging firms to create those jobs. We use a quantitative model to investigate whether this effect is comparable in magnitude to the standard moral hazard effects of unemployment insurance. Our model economy captures the behavior of the U.S. labor market for high school graduates quite well. With unemployment insurance more generous than the current U.S. level, unemployment would increase by a magnitude similar to the micro-estimates; but because the composition of jobs also changes, total output and welfare would increase as well.

This paper argues that unemployment insurance increases labor productivity by encouraging workers to seek higher productivity jobs, and by encouraging firms to create those jobs. We use a quantitative model to investigate whether this effect is comparable in magnitude to the standard moral hazard effects of unemployment insurance (UI). Our results indicate that a decrease in the generosity of UI from its current U.S. level would not only decrease welfare but also reduce the level of output. Most analyses of unemployment insurance focus on its consumption-smoothing and risk-sharing roles. For example, Gruber (1997) finds that workers who receive more generous unemployment benefits experience a smaller drop in consumption following the loss of a job. Standard approaches to unemployment insurance compare this benefit with the adverse moral hazard effects, and compute the optimal amount of UI by equating marginal costs and benefits (Shavell and Weiss, 1979; Hopenhayn and Nicolini, 1997). While this tradeoff is likely to be important, UI could also affect the type of jobs that workers look for and accept.1 According to the theory we develop, in an economy without any UI, workers avoid the risk of unemployment by applying to low productivity jobs that are easier to obtain. Firms offer implicit insurance to workers by opening jobs with low unemployment risk, and charging an insurance premium in the form of lower wages. The resulting composition of jobs is inefficient and can be improved by a moderate level of unemployment insurance, which encourages workers to take on more risk, and increases both welfare and the level of output (see also Acemoglu and Shimer, 1999a). Although this effect is qualitatively reasonable, a major goal of this paper is to show that it is likely to be quantitatively important as well. In a realistic environment, it must outweigh two significant forces. First, unemployment insurance will encourage workers to reduce their search effort, lowering employment and output. Second, workers can self-insure by saving, considerably reducing the need for unemployment insurance. To address these issues, we consider a quantitative dynamic economy. Workers are risk averse, with constant relative risk aversion (CRRA). They optimally choose their consumption, labor supply, and search effort while unemployed. Unfortunately, this model cannot be solved analytically,2 because the workers’ optimal policy (their consumption, labor supply, and search rules) depends on their wealth level, which is itself determined by the optimal policy. We therefore undertake a calibration exercise, anchoring our model to plausible preferences and to the unemployment rate and unemployment insurance system faced by U.S. workers with a high school degree. With a coefficient of relative risk aversion of four, a replacement rate a little below 50% for six months, and productivity differences between good and bad jobs on the order of 30%, our model generates levels of unemployment, the consumption drop upon job loss, and low-frequency income variability similar to those found in U.S. data. Moderate increases in the replacement rate or the duration of unemployment benefits lead to increases in unemployment duration similar to those observed in the U.S. economy. For example, Meyer (1989) finds that a 10% increase in unemployment benefits raises unemployment duration by about a week, while Ehrenberg and Oaxaca (1976) and Atkinson and Micklewright (1991) estimate a slightly smaller response. At the same time, the policy change raises average wages by about 1.2%, which is a substantial effect, but still somewhat less than the gains reported by Ehrenberg and Oaxaca (1976). The overall effect of the policy change is to raise output and welfare by a little over . The result that economies with moderate UI have higher output and welfare those without social insurance is very robust. Increasing the value of leisure, reducing risk aversion, reducing the wage gap between different types of jobs, and allowing on-the-job search does not alter this conclusion, although it does sometimes affect whether the output-maximizing replacement rate is above or below U.S. levels. Our paper is related to a number of previous studies. As mentioned above, we build on Acemoglu and Shimer (1999a). Other papers have also pointed out beneficial effects of UI, inter alia Diamond (1981), Acemoglu (1998), and Marimon and Zilibotti (1999). For example, Acemoglu shows that UI may improve welfare by encouraging workers to wait for higher capital jobs that pay higher wages because of holdup problems. Marimon and Zilibotti emphasize matching between workers and firms according to comparative advantage, and show that UI encourages workers to wait for jobs better suited to them. All these papers consider risk-neutral agents, however, so unemployment benefits are simply a subsidy to search. Our approach differs in explicitly modeling risk aversion and precautionary saving, and in contrasting the beneficial effects of UI with its conventional costs. We are not aware of any other study that has compared these costs and benefits in such a realistic setting. Other studies, including Shavell and Weiss (1979), Hansen and Imrohoroglu (1992), Atkeson and Lucas (1995), and Hopenhayn and Nicolini (1997) analyze optimal unemployment insurance with asymmetric information, but do not model labor market search. Costain (1996), Valdivia (1996), and Gomes et al. (1997) examine labor market behavior in a quantitative general equilibrium framework, but do not look at the productivity gains from unemployment insurance. Our emphasis on the importance of uninsured risk may appear to contradict the findings of Aiyagari (1994), Huggett (1993), and Krusell and Smith (1998). These papers show that wealth heterogeneity and risk are unlikely to have an important effect on the behavior of aggregate macroeconomic variables. They consider general equilibrium models with an endogenous interest rate. If all workers have the same rate of time preference and there are no exogenous borrowing constraints, the interest rate will be quite close to the discount rate. The precautionary savings motive then induces workers to build up a large buffer stock of assets, insulating them from risk. In contrast, we focus on the behavior of low skill workers (e.g., high school graduates) rather than the whole economy. We use a partial equilibrium model, with an exogenous interest rate, significantly less than the discount rate.3 Workers maintain low asset levels, which provides an accurate characterization of the behavior of preretirement, low skill workers (see the numbers given in footnote 11). Since self-insurance is imperfect, agents change their behavior to avoid risk. Another significant difference between our model and the previous literature, is that in these other papers, agents only affect macroeconomic outcomes through their savings decisions.4 Since poor agents own very little of the aggregate capital stock, they only have a limited effect on aggregate outcomes. In contrast, in our economy poor agents’ behavior determines the composition of jobs, and therefore has a large effect on aggregate income. In the next section, we start with a static model that illustrates the main tradeoffs that we focus on. We outline our full dynamic model and characterize the solution in Section 3. The model is calibrated to features of the U.S. labor market for high school graduates and the U.S. unemployment insurance system in Section 4. We show that the model performs well and matches a variety of salient patterns of the labor market that we did not use in our choice of parameters. We also find in this section that moderate increases in UI generosity starting from U.S. levels could increase output, productivity, and welfare. In Section 5, we show that our general results are robust to introducing on-the-job search, to reducing the wage gap between different types of jobs, and to lower levels of risk aversion. Section 6 concludes.

Conventional wisdom views unemployment insurance as a serious distortion that we have to live with in order to smooth income risk and consumption variability. In this paper, we have argued that moderate unemployment insurance may actually improve the allocation of resources. Unemployment insurance enables workers to pursue riskier options, including jobs that are harder to get, but possibly also more productive. As a result, moderate UI may raise output by improving the composition of jobs. In practice, however, this effect of UI may be outweighed by the traditional moral hazard cost: insured workers search less hard for jobs, and therefore spend more time unemployed. Moreover, one might conjecture that the importance of uninsured risks is limited by workers’ ability to self-insure by building up buffer stocks of assets. To investigate the quantitative importance of these opposing forces, we constructed a dynamic model in which workers make search effort, savings, and job application decisions. Although the decisions in question are complex, our model is sparse, enabling us to calibrate the few parameters to U.S. data, in particular to the labor market for high school graduates. The model performs well not only along the dimensions in which it is calibrated, but in a number of other dimensions as well, including the standard deviation of annual income, the decline in consumption following job loss, and the responsiveness of unemployment duration and wages to the magnitude and duration of unemployment benefits. Armed with the confidence that this model captures some of the tradeoffs faced by workers in real labor markets, we investigated the implications of different UI policies on unemployment, output, labor productivity, and welfare. We find that reducing UI from its current U.S. level would reduce both risk sharing and total consumption. Conversely, moderate increases in UI raise output and improve risk sharing. As this is a calibration exercise, our results are necessarily sensitive to parameter choices. To address this issue, we verified the robustness of our results to several key parameter changes. A complementary strategy would be to look for direct evidence that more generous unemployment insurance programs encourage the creation of more specialized and higher productivity jobs. At a more general level, we believe that more work needs to be done to understand the role and optimal design of UI programs. Labor market reform is a key issue in Europe. Social insurance programs, especially UI programs, are likely to be modified during the next decade. Similar reforms are underway in the U.S. If we are correct that social insurance programs have a beneficial effect not only on welfare but also on output and productivity, the reforms may have very different consequences than currently envisioned. Finally, we have assumed that UI is provided by the government. Why unemployment insurance is almost always publicly provided, in contrast to most other insurance contracts, remains an important, unresolved question.22 The answer will likely be relevant to the optimal design of unemployment insurance programs.