Research

Selection in the Disability Insurance Market
This paper tests for selection in disability insurance (DI) and quantifies the associated welfare loss. DI pays income benefits to individuals with disabilities that are not necessarily work-related. The market for optional private DI is small despite its important role in supplementing public DI programs. The DI market is speculated to be adversely selected, but empirical evidence is scant as most DI programs have mandatory coverage. This paper estimates demand and cost curves of an employer-provided DI program that allows the choice of coverage. I use rich administrative data on workers' take-up, premiums, claims, sick leave use, and disability diagnoses, and exploit variation in premiums that depends on the performance of the DI program investment fund. I find that selection in this market is advantageous. As premiums increased, covered workers claimed less DI. This can be explained by high-earning workers being more likely to have coverage even though they are less likely to claim DI compared to low-earning workers.

Prevalence of Disabilities and the Capacity to Return to Work
While there is a large literature on the labor response to public disability insurance (DI) and return-to-work programs, much less is known about the period prior to DI application. During this period, workers are more attached to their jobs, suggesting that interventions to support employment among workers with a disability may be more fruitful compared to interventions done after workers have started receiving public DI benefits. For interventions to work, it is important to understand the employment capacity of workers with a disability. This paper studies the prevalence of disabilities among Wisconsin state public employees, their use of employer-provided DI benefits, and their capacity to return to work after claiming short-term DI. Generally, low-earning workers are less likely to be covered by employer-provided DI although they tend to need and claim DI at a higher rate. Next, the paper examines the use of short-term DI and worker's trajectories before and after claiming short-term DI. 66 percent of claimants return to work, 21 percent transition to long-term DI, while the rest quit their jobs or retire. Short-term DI claimants are able to maintain their earnings and hours worked up to the year of claim. Afterwards, their employment drops even among those who return to work. Moreover, workers who eventually claim short-term DI use a lot of sick leave even in the years prior to claiming. 

The Effects of Opioids on Labor Market Outcomes and Use of Social Security Disability Insurance
Prescription opioids are widely used to treat pain. They can be beneficial by helping a person with a medical condition to manage pain and to be gainfully employed rather than seeking Social Security Disability Insurance (SSDI). On the other hand, opioids are addictive; their use may lead to substance abuse and an exit from the labor market. This paper investigates the effects of prescription opioids on labor market outcomes and the use of SSDI. Estimating the effects of prescription opioids is challenging because those prescribed opioids are likely to be different from those who were not, for example, in terms of their health conditions. To obtain causal estimates, I use marketing payments from opioid manufacturers and distributors to physicians as an instrument to predict opioid prescribing. The underlying idea is that, holding other factors constant (e.g.  health conditions), doctors who receive marketing dollars are more likely to prescribe opioids than those who do not. I find that a higher opioid prescription rate has negative effects on labor market outcomes by reducing labor force participation and increasing unemployment. Greater opioid prescribing also increases the use of SSDI in terms of applications, awards, and claims. I also find that the effect of opioids is greater in more urban counties. Lastly, a higher opioid prescription rate reduces full-time employment, but increases part-time work.