Session Recap: Health Insurance and the Labor Market
November 10, 2014 02:30 PM
By Sana Ahmad, Rutgers University
Health and employment often go hand-in-hand with a person’s health status, contributing to their ability to work and employment potentially impacting health both in positive and potentially negative way. With such a simultaneous relationship between health and employment, policies in both realms are bound to influence the other variable. Additionally, with Employer Sponsored Insurance accounting for more than half of private insurance coverage, health insurance policies have a tremendous implication for and impact on labor. Further, through their income eligibility requirements, social programs like Medcaid can also impact, or be impacted by, employment. The two presenters in this panel look at the relationship between labor market elements, such as unemployment and workers compensation, and aspects of the healthcare system.
The discussants for this session were Nicole Nestoriak, Bureau of Labor Statistics, and Jamie Rubenstein Taber, U.S. Census Bureau.
The Effects of Provider Choice Policies on Workers Compensation Costs and Outcomes, by David Neumark, University of California, Irvine and Bogdan Savych, Workers Compensation Research Institute
Savych indicated that the authors’ primary interest in this paper was to assess how states’ provider choice policies affect workers compensation. Workers compensation is a short term disability program, for which employers are required to provide or at least be self-insured. When injured, insurance pays all cost of medical care and pays out a percentage income that is lost. Workers compensation represents 25 of the overall medical spending in the US. Provider choice policies refer to the ability to choose a provider for medical care. There are multiple policy approaches towards provider choice in workers comp: 1) employers choose the provider, 2) employers give workers choice of a network of providers, or 3) workers can choose providers or have the freedom to change providers. Provider choice is important because the physicians are not only responsible for providing care but also for determining when it is safe for employee to return to work.
People make arguments for benefits of both employers and employees being able to make decisions regarding provider. However, there is no research on looking at state policies and spending outcomes for workers compensation. The analysis in this paper focuses on 3-way classification of states provider choice policies: 1) Employer choice, 2) worker choice from network, or 3) worker choice or worker ability to readily change provider.
The authors use reduced form estimates as empirical method and predict average effect of policy on. Since there are no changes in provider choice regulations within examined states overtime, Neumark and Savych cannot control for all other aspects of workers compensation through state fixed effects. Instead, they control for various elements of workers compensation including generosity of indemnity benefits as well as for medical policies. Using worker compensation claims from 2007-2010 for 33 states, they measured effect of policy on medical and indemnity costs.
Estimates from medical cost differences not statistically significant, thus there is no evidence that increasing employee freedom to choose providers affects medical costs. The researchers then estimated if medical cost if broken down into quintiles, and results showed that provider choice policies don’t make a difference. Since one aspect that potentially affects medical cost is the type of injury, they showed quantile regression by specific types of injury showing that for inflammatory conditions there are lower medical costs when workers have more power to choose providers: medical costs and for back and neck sprains there is significant increase in medical costs when workers have more choice. For indemnity costs, there is no evidence for cost difference, on average, but when looking at different medical conditions, indemnity costs are much higher for the 95th quantile when workers have more choice of provider.
The discussants generally had a positive response to the research but pointed out that workers compensation is essentially a different program in each state where the variability is not accounted for in the analysis. Additionally, the authors’ focus on one aspect of workers compensation policy, provider choice, but there are many other elements of the policy which might be playing a role.
In terms of policy implications, there is no empirical evidence in the study supporting employer control of provider choice. Contrarily, the authors provided some evidence that states with policies giving workers control over provider choice have better outcomes in terms of medical and indemnity costs. Thus, policies framing or affecting workers compensation should keep this in view when allocating the decision making power over choosing medical providers.
The Impact of Unemployment on Medicaid Enrollment by Cheryl A. Camillo, University of Maryland, Baltimore County
Camillo began her discussion by providing two purposes for her paper: 1) to forecast Medicaid participation (new and continuing enrollment) and 2) to examine opportunities to identify and enroll eligible but un-enrolled individuals. Whereas health insurance rates have dropped and Medicaid enrollment has increased significantly, tens of millions of American remain uninsured. Based on Urban Institute’s early results using data from the Health Reform Monitoring Survey, 13.9 percent of people are eligible but uninsured. Additionally, unemployment rates remained above pre-recession values.
This research examined participation rates in the post welfare reform era (2002-2010) during which Medicaid eligibility and enrollment policies and procedures for children and non-disabled adults changed significantly. Additionally, during this time period, there were dramatic increases in unemployment rates.
Camillo did a preliminary study using cross-section of changes design. She used OLS to estimate effect of unemployment on Medicaid participation rates. The model showed that a on percent increase in unemployment leads to a statistically significant 0.15 percent increase in Medicaid enrollment. An additional finding was that Medicaid enrollment grew at a much faster rate in states that covered all non-disabled (i.e. childless) adults.
Using pooled OLS with clustered errors, she found a statistically significant positive relationship between unemployment rates and Medicaid participation rates among working age individuals. Camillo reported that a one percent increase in unemployment leads to 0.1 percent increase in Medicaid participation. The author conducted various sensitivity analyses, which did not impact model fit.
One potential limitation of the study that the discussants pointed out is use of administrative data. Additionally, they also highlighted that OLS might not be the best form to model the relationship the author is studying.
Regarding policy implications of this research, the Camillo suggested utilizing a policy solution to help unemployed individuals enroll in Medicaid. She concluded that state welfare agencies should factor unemployment rate into projections of Medicaid enrollment. She also suggested consideration of designating state unemployment agencies as express lane eligibility agencies.