Affordable Care Act(ACA), enacted back in 2010, has great impact on health benefits, taxes, and employment. Since health insurance subsidies that the act provides to some people will be phased out as their income rises, thus creating an implicit tax on additional earnings. Some provisions may imply higher effective tax rates on earnings from labor, thus reduce workers’ willingness of supplying labor. From the view point of health benefits, massive expanding public health insurance can decrease the employment rate, as people previously get insurance through their employers, now they have access towards public health insurance with less requirements to be eligible. ACA is tightly connected with labor economics in many ways, which is why it is a good topic to focus on.
The studies that I choose on ACA are as follow:
- Labor Market Effects of the Affordable Care Act: Evidence from a Tax Notch, written by Kavan Kucko, Kevin Rinz, and Benjamin Solow. Authors focus on “coverage gap” for states refusing to raise their medicaid income eligibility cutoffs to 138 percent of the FPL. Since ACA provides refundable premium tax credit, when lacking access to subsidized health insurance, civilians of these states will have strong incentive to a large increase in post-tax income at the FPL level, in order to increase their income and obtain the premium tax credit. Authors used tax return to study the relationship between labor income and the implementation of ACA.
- Effects of the Affordable Care Act on Part-Time Employment: Early Evidence, written by Marcus Dillender, Carolyn Heinrich, and Susan Houseman. ACA mandates employers with minimum number of 50 full-time employees to offer equivalent affordable health insurance to those who work 30 or more hours per week; otherwise there will be a penalty. To avoid such penalty, employers attempt to reduce weekly working hours below 30 hours, by using temporarily workers, small or independent contractors. Authors study the effect of ACA on part time employment by using monthly CPS data, since mandatory health insurance for full-time employees is a financial loss to some extent for employers.
- The Effect of Public Insurance Coverage for Childless Adults on Labor Supply, written by Laura Dague, Thomas Deleire, and Lindsey Leininger. Authors provide causal estimates of medicare expansion on childless adults. By regressing discontinuity and propensity score with difference-in-difference methods to look at the enrollment cap, authors discover that enrolling in public health insurance lead to a statistically significant employment reduction up to 9 quarters later minimal.
- Public Health Insurance, Labor Supply, and Employment Lock, written by Craig Garthwaite, Tal Gross, and Matthew J. Notowidigdo. By looking at the largest medicaid disenrollment in the history of the United States, TennCare, authors try to predict the effect of expanding ACA on labor supply. They suggest a “employment lock”, that workers seeking for jobs are mainly looking for health benefits through jobs. With implemented ACA, authors expect a fall in employment rate in the future.
The last study, Public Health Insurance, Labor Supply, and Employment Lock, is the one I mainly focused on. Previously, public health insurance mainly focused on disabled, low-income parents, and those who are over 65 years old. ACA mainly targets childless adults, who used to be not eligible for public health program. Similarly, TennCare shares same demographic features as ACA. Following by the disenrollment of TennCare starting in 2005, within two years, the number of enrollee fell from 1,337,000 to 1,170,000. Authors discover a immediate increase in labor supply and job searching behavior after the disenrollment, thus providing a good predictor of effects of ACA: there will be a decrease in employment rate and labor supply after the implementation of ACA, as expanding public medical care provides benefits that people used to get through their employers with private insurance.
Authors first compare the TennCare with ACA, showing similarity between two programs, so it would be plausible to predict effects of ACA by TennCare. They use data from the Current Population Survey(CPS) on health insurance coverage and labor market outcomes, from 2000 to 2007. Then restricting people from age 21 to 64, and looking at whether they are employed or not, what type of insurance they have, and how many hours they work per week.
The key method to study the causal effect of the disenrollment is a state by year difference-in-difference regression: [image: ], with variable “yst” representing outcome of state s and in year t, such as the share of the population with public health insurance coverage. This model includes state fixed effects(α) and year fixed effects(δ). Authors give out a comparison of Tennessee and other southern states, showing their similarity in summary statistics, so Tennessee can become the “treated group”, with disenrollment being the “treatment”, and other southern states being the “untreated”. β is the difference-in-difference estimate of the effect of the disenrollment. By comparing outcomes in Tennessee before and after the disenrollment to outcomes in other southern states, assuming outcomes in Tennessee would not be different with other states without the disenrollment.
To further implement this regression, also concerning other potential national policy changes and recessions, authors use triple difference regression model: [image: ], this is based on the difference-in-difference model above, and is used to control for any unobservable common shocks that may affect childless adults in Tennessee and other Southern States. This model includes a new demographic group fixed effects (γ), and all two-way interactions between each fixed effects. Again, β is the triple-difference estimate of the effect of TennCare disenrollment on childless adults relative to other adults, and the key assumption is that inside Tennessee, the two demographic groups would have be similar without the disenrollment. A “triple-difference” analysis is used primarily to eliminate potential bias.
For results, authors first provide a comparison of share publicly insured between Tennessee and other southern states, and a triple difference comparison between childless adults in Tennessee and other adults. In 2006 following the disenrollment, there is a massive decline in share publicly insured, about 4 percent, both for Tennessee residents and childless adults, which other southern states and citizens with children do not experience. For the effect of TennCare disenrollment on employment, there is a 4.6 percent decrease in public insurance coverage for Tennessee residents, a 7.3 percent decrease for Tennessee residents with no children, and at the same time employment rate raise by 2.5 percent and 4.6 percent respectively. This effect concentrates on people working over 20 hours a week, implying an increase in childless adults seeking private health insurance through their employers.
From labor supply’s perspective, authors anticipate an increase in labor supply would lead to decrease in wages, as they find out that log wage and residualized log wage decreased by 2.3 percent and 1.8 percent respectively, which support the idea that labor supply has increased after the disenrollment. To check robustness, since disenrollment mainly focus on childless adults, people over 65 years old who are almost fully enrolled in medicaid programs, should have little impact on insurance coverage and labor supply. They find a 0.1 percent increase in insurance coverage and 0.2 percent increase in labor supply, which is quantitatively and statistically insignificant.
Lastly, authors even provide Google Trend for phrase “job openings” searched, during disenrollment period, the searches reaches its peak. Data from Local Area Unemployment Statistics and Behavioral Risk Factor Surveillance System also suggest an increase in share employed by month and a decrease in share insured by month. By looking at TennCare, authors predict that there will be 4.2 million people switching from private insurance to public insurance after ACA is implemented. There will be a decline in employment of 530,000 to 940,000 people, which is a decline in employment rate of 0.3 to 0.6 percent from ACA itself.
This study is the most credible, due to its well constructed model, and concrete evidences. A triple difference model is used to complement difference-in-difference model, there is robustness check, and authors use results from various aspects: including share insured publicly and privately, wage changes, employment rates, and even Google Trends. Although in study: Public Health Insurance, Labor Supply, and Employment Lock: Effects or Data Artifacts? Written by Elena Gouskova, author critiques the inaccuracy of data, as during the time of study, CPS underwent a series of changes to its data, including calculation weights, sampling scales, etc. Using the same diff-in-diff method and adjusted data, Elena comes up with statistically insignificant effect on labor supply. Author also claims that the increase in employment rate is due to the decline in total population in Tennessee rather than the increase in workers employed. However, this study cannot explain the reason behind the Google Trend for increase in job seeking behavior. It could not be simply a coincidence that people suddenly start to look for jobs after the disenrollment.
My critiques about this study is that, it does not include any demand side factor in the labor market, as it only mentions labor supply. TennCare and ACA can be similar enough, but is Tennessee representative enough to predict effects of ACA in the entire nation? I would also like to see some analysis on crowdout effect influencing the insurance market, does it has anything to do with employment rate? How would private vs public insurance affect labor market? ACA is one enormous policy to study, as it covers too many aspects of the society, there is no perfect way to study this policy, all we can do is to consider as many aspects as possible to avoid potential bias.
- Public Health Insurance, Labor Supply, and Employment Lock Craig Garthwaite, Tal Gross, Matthew J. Notowidigdo, NBER Working Paper No. 19220, Issued in July 2013 https://www.nber.org/papers/w19220
- Labor Market Effects of the Affordable Care Act: Evidence from a Tax Notch Kavan Kucko, Kevin Rinz, Benjamin Solow, United States Census Bureau, Issued in July 13, 2017 https://www.census.gov/library/working-papers/2017/adrm/carra-wp-2017-07.html
- The Effect of Public Insurance Coverage for Childless Adults on Labor Supply Laura Dague, Thomas DeLeire, Lindsey Leininger, NBER Working Paper No. 20111, Issued in May 2014 https://www.nber.org/papers/w20111
- Effects of the Affordable Care Act on Part-Time Employment: Early Evidence Marcus Dillender, Carolyn Heinrich, Susan Houseman, Labour Economics 43(2016): 151-158, Issued in June, 2016 https://research.upjohn.org/up_workingpapers/258/
- Public Health Insurance, Labor Supply, and Employment Lock: Effects or Data Artifacts? Elena Gouskova, November 10, 2015, Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2688582