Working Papers

The Value of Unemployment Insurance: Liquidity vs. Insurance Value [Current Version] [Cleveland Fed WP]

with Kaixin Liu

This paper argues that the value of unemployment insurance (UI) can be decomposed into a liquidity component and an insurance component. While the liquidity component captures the value of relieving the cost to access liquidity during unemployment, the insurance component captures the value of protecting the worker against a potential permanent future income loss. We develop a novel sufficient statistics method to identify each component that requires only the labor supply responses to changes in the potential duration of UI and severance payment and implement it using Spanish administrative data. We find that shutting down the liquidity component decreases the value of UI by 66 percent. However, the relevance of the liquidity channel is highly heterogeneous across different groups of workers. Poorer and wealthier value similarly the liquidity UI provides, but poorer workers value UI more because the insurance channel is crucial for them. On the other hand, wealthier workers and workers with more cash-on-hand value UI equally, but the wealthier value its liquidity, while those with more liquidity care about its insurance value. Finally, we show that while extending the potential duration of Spain's UI would increase welfare, starting a zero-interest rate public loan program of the same cost would be more welfare-improving, even under large default risks. 

The Hedgehog’s Curse: Knowledge Specialization and Displacement Loss [Current Version] [Cleveland Fed WP]

with Roberto Pinheiro and Hans Holter

This paper studies the impact of knowledge specialization on earnings losses following displacement. We develop a novel measure of the specialization of human capital, based on how concentrated the knowledge used in an occupation is. Combining our measure with individual labor histories from the NLSY 79-97 and Norway’s LEED, we show that workers with more specialized human capital suffer larger earnings losses following exogenous displacement. A one standard deviation increase in pre-displacement knowledge specialization increases the earnings losses post-displacement by 3 to 4 pp per year in the US, and by 1.5 to 2 pp per year in Norway. In the US, the negative effect of higher pre-displacement knowledge specialization on post-displacement earnings is driven by the negative impact of knowledge specialization on well-paid outside opportunities. By contrast, this association between outside opportunities and knowledge specialization plays no role in post-displacement earnings losses in Norway, where the negative effect of specialization is in part explained by its association with the routine content and the offshoring probability of the occupation.

Estimating Duration Dependence on Re-employment Wages When Reservation Wages Are Binding [Current Version]

with Kaixin Liu

This paper documents a novel finding indicating that re-employment wages are elastic to the level of unemployment insurance (i.e., a binding reservation wage) and adapts the IV estimator for duration dependence in \textcite{schmieder_effect_2016} to account for this fact. Using administrative data from Spain, we find that unemployed workers lower their re-employment wages by 3 percent immediately after the exhaustion of unemployment insurance (UI) benefits. Workers' characteristics and permanent unobserved heterogeneity cannot explain this. To estimate duration dependence, we extend the IV framework proposed by \textcite{schmieder_effect_2016}, whose estimator of duration dependence is proportional to the response of wages to an extension of the potential duration of UI, to account for the response of reservation wages. We find that while extending the potential duration of UI has an insignificant effect on re-employment wages, duration dependence is strongly negative. We estimate that the degree of duration dependence in Spain is approximately 0.8 percent per month in daily wages. Workers' inability to find full-time jobs as the duration of non-employment increases is an important mechanism behind this effect, since the duration dependence of hourly wages is 0.25 percent per month. Failing to account for the fact that reservation wages are binding would underestimate the magnitude of duration dependence by 15 to 20 percent.

Skill Complementarity and the Shale Boom [Current Version]

Fear of automation has spiked in the last decade. This paper aims to understand whether machines can replace workers in the labor market and if so, which types of workers are the ones that can be substituted with technology. By exploiting a natural experiment with an instrumental variables strategy that takes advantage of the increase in the relative demand for low-skill labor as a consequence of the shale boom in the US, I find mild evidence of substitutability between low-skill labor and capital in the manufacturing sector. The structure of my data allows me to assess how is best defined low skill labor in this scenario. I find that the strongest conclusions are reached when the definition of high-skill labor includes only those with a college degree or above that level of education. 

Work in Progress

Fighting Income Inequality with International Trade [Coming Soon]

with Nicholas Kozeniauskas and Roman Merga

How does international trade affect wage and income distribution across workers? We use detailed employer-employee data from Spain that goes from 1987 to 2004 to answer this question. Our results show that, in contrast with recent economic development in international trade theory, international trade reduces wage and income inequality. Using a new instrumental variable approach to disentangle the effects that trade openness has over the distribution of income and wages, we document 4 effects that international trade has on the labor market. First, an increase in local trade exposure reduces average wages and income; second, it lowers wage and income inequality; third exposure to trade openness generates a reallocation of workers towards small firms and low-skilled jobs; fourth there is a reduction in firm size, when the size of the firm is measured according to their amount of workers. These results are contrary to the findings and predictions of recent models in the literature and to some of the previous empirical findings. The policy implications of these results to mitigate inequality are potentially different than the ones that arise from the standard models, while they raise the question about the mechanism behind and its relevance in other countries.


Social distancing merely stabilized COVID‐19 in the US [STAT] (2020)

with Aaron B. Wagner, Elaine L. Hill, Sean E. Ryan, Ziteng Sun, Grace Deng, Sourbh Bhadane, Peter Wu, Dongmei Li, Ajay Anand, Jayadev Acharya, and David S. Matteson.

Social distancing measures have been imposed across the US in order to stem the spread of COVID‐19. We quantify the reduction in doubling rate, by state, that is associated with this intervention. Using the earlier of K‐12 school closures and restaurant closures, by state, to define the start of the intervention, and considering daily confirmed cases through April 23rd, 2020, we find that social distancing is associated with a statistically‐significant (p < 0.01) reduction in the doubling rate for all states except for Nebraska, North Dakota, and South Dakota, when controlling for false discovery, with the doubling rate averaged across the states falling from 0.302 (0.285, 0.320) days‐1 to 0.010 (‐0.007, 0.028) days‐1. However, we do not find that social distancing has made the spread subcritical. Instead, social distancing has merely stabilized the spread of the disease. We provide an illustration of our findings for each state, including estimates of the effective reproduction number, R , both with and without social distancing. We also discuss the policy implications of our findings.