Content uploaded by Devika Hazra
Author content
All content in this area was uploaded by Devika Hazra on Sep 22, 2018
Content may be subject to copyright.
1
WHO DOES THE AFFORDABLE CARE ACT HELP AND WHO
DOES IT FAIL?
Zhen Cui*
Devika Hazra†
California State University, Los Angeles
December 2017
Abstract
This study uses the 2007 and 2013 Annual Social and Economic Supplement of the Current
Population Survey to examine the effect of the Affordable Care Act (ACA) on health insurance
coverage among adults in the U.S. It finds that the ACA has improved coverage for men, youth,
minorities, and low-income and less-educated individuals. However, those who are self-employed
or do not work full-time have been negatively impacted.
* California State University - Los Angeles, Email: zcui@calstatela.edu
† California State University - Los Angeles, Email: dhazra@calstatela.edu
2
I. Introduction
The Affordable Care Act (ACA) has been a contention for policymakers since its inception.
While evidence on actual health outcomes is still sparse, a few studies have shown a decrease in
the number of the uninsured. A study done by Sommers, Buchmueller, Decker, Carey, and Kronick
(2013) found sizable coverage gains for adults aged 19–25. The gains continued to grow
throughout 2011, with the largest gains seen in unmarried adults, non-students, and men.
According to Sommers, Gunja, and Finegold (2015), low-income adults within the states that
expanded Medicaid reported significant gains in insurance coverage and access compared with
adults within the states that did not expand Medicaid. Moreover, Sommers, Maylone, Blendon,
Orav, and Epstein (2017) assessed changes in health care use and self-reported health after three
years of the ACA’s coverage expansion, using survey data collected from low-income adults
through the end of 2016 in two states (Arkansas and Kentucky) that expanded coverage, and Texas
that did not expand coverage. By the end of 2016, the uninsurance rate in the two expansion states
had dropped by more than 20 percentage points relative to the non-expansion state. Finally, Figure
1 shows the percentage of adults aged 18–64 who were uninsured or had private or public coverage
at the time of interview in the U.S. between 1997–2016.
Figure 1 Percentages of adults aged 18–64 who were
uninsured or had private or public coverage at the time of interview:
United States, 1997–2016
After the main ACA provisions went into effect in 2014, racial disparities in coverage
declined slightly as the percentage of adults who were uninsured decreased by 7.1 percentage
3
points for Hispanics, 5.1 percentage points for Blacks, and 3 percentage points for Whites
(Buchmueller et. al. 2016). McMorrow, Long, Kenny, and Anderson (2015) found significant
improvements in insurance coverage for all racial and ethnic groups between the second and third
quarters of 2013 as well as 2014, which translated into reductions in absolute disparities in the
uninsurance rates for Blacks and Hispanics in both expansion and non-expansion states.
Furthermore, Chen, Vargas-Bustamante, Mortensen, and Ortega (2016) demonstrated that racial
and ethnic disparities in access had been reduced significantly during the initial years of the ACA
implementation.
Despite coverage improvement for men, single adults, and minorities, the ACA has a
negative effect on other subpopulations. For example, Blumberg, Corlette, and Lucia (2014)
roughly estimated that the number of uninsured self-employed individuals and entrepreneurs
would relatively increase by more than 11 percent following the ACA.
Against the above background, the present study applies a probit regression separately to
the 2007 and 2013 Annual Social and Economic Supplement of the Current Population Survey
(ASEC CPS). It then compares the effects of main demographic traits and socioeconomic factors
on health insurance coverage in the U.S. for adults across all ages before and after the signing of
the ACA. We find that the coverage has improved for men, youth, minorities, and low-income and
less-educated individuals, while adversely impacting those who are self-employed or do not work
full-time. These findings are largely consistent with the existing studies. Thus, this study makes
two major contributions to the literature. First, it adds more empirical evidence regarding the
impact of the ACA on health insurance coverage for the adult population in the U.S. Second, it
provides practical policy and business insights by identifying specific groups which policymakers
and practicing managers should focus on extending the insurance coverage to in the post-ACA era.
The rest of the paper proceeds as follows. Section 2 describes the data and the econometric
model. Section 3 explains the regression results. Section 4 presents the concluding remarks and
makes policy and business recommendations.
II. Empirical Strategy
The ASEC CPS is a rich dataset that has detailed information on employment,
demographics, and health insurance. The data used in this study are extracted from its 2007 and
2013 series, marking the three years before and after the signing of the ACA. After removing
missing values and restricting respondents to civilians aged 18–79, the number of individual
observations in our final sample is 247,943 (125,851 for 2007 and 122,092 for 2013).
The dataset has a constructed variable that indicates whether a respondent had any health
insurance coverage (private or public) in the previous year. This study considers a respondent
covered by health insurance if the answer is yes, and not covered by health insurance if the answer
is no. Per this definition, about 82% of our sample had health insurance (83% for 2007 and 82%
for 2013). Clearly, not everyone was covered even three years after the ACA was signed into law.
This highlights the relevance and importance of studying the determinants of health insurance
coverage in the post-ACA era. All statistics reported in this study are appropriately weighted.
Existing studies on the determinants of health insurance coverage typically employ a binary
response model (i.e., logit or probit regression). For example, Gius (2010) used the 2008 National
Health Interview Survey and adopted a logit regression to examine the determinants of health
insurance coverage for young adults. Cantiello, Fottler, Oetjen, and Zhang (2015) used the 2005
and 2008 Medical Expenditure Panel Survey to investigate the factors that influence young adults’
4
decision to have private health insurance. They incorporated structural equation modeling into a
standard logit regression.
Following those studies, we apply the probit regression below to the 2007 and 2013 data
separately. A comparison of results from these two years would enable us to examine the effect of
the ACA on adult health insurance coverage in the U.S.
coveredi = α + β1 agei + β2 malei + β3 log_incomei + β4 fulltimei + β5 self_empi
+ β6 highschooli + β7 collegei + β8 graduatei + β9 whitei + ui,
where coveredi is a dummy variable that takes 1 if individual i has health insurance coverage and
0 otherwise, agei denotes the individual’s age, malei is a dummy variable that takes 1 if the
individual is male and 0 otherwise, log_incomei denotes the individual’s logged annual pre-tax
wage and salary income,
1
fulltimei is a dummy variable that takes 1 if the individual has a full-time
job and 0 otherwise, self_empi is a dummy variable that takes 1 if the individual is self-employed
and 0 otherwise, highschooli is a dummy variable that takes 1 if the individual has completed 12
years of schooling or received a high school diploma and 0 otherwise, collegei is a dummy variable
that takes 1 if the individual has some college education or received an Associate’s or a Bachelor’s
degree and 0 otherwise, graduatei is a dummy variable that takes 1 if the individual has a Master’s
degree or above and 0 otherwise, whitei is a dummy variable that takes 1 if the individual is White
and 0 otherwise, and ui denotes the standard classical error term. Therefore, our analysis takes into
account both basic demographic traits (i.e., age, gender, and race) and various socioeconomic
factors (i.e., income, employment, and educational attainment). Table 1 describes the summary
statistics of all the variables.
Table 1 Summary statistics
2007
2013
VARIABLES
MEAN
STANDARD
DEVIATION
MIN
MAX
MEAN
STANDARD
DEVIATION
MIN
MAX
covered
0.83
0.38
0
1
0.82
0.39
0
1
age
44.14
16.03
18
79
45.05
16.40
18
79
male
0.49
0.50
0
1
0.49
0.50
0
1
log_income
5.99
5.80
-2.30
13.16
5.55
5.91
-2.30
13.80
fulltime
0.57
0.50
0
1
0.51
0.50
0
1
self_emp
0.08
0.26
0
1
0.07
0.25
0
1
highschool
0.33
0.47
0
1
0.31
0.46
0
1
college
0.46
0.50
0
1
0.49
0.50
0
1
graduate
0.09
0.29
0
1
0.10
0.31
0
1
white
0.81
0.39
0
1
0.79
0.41
0
1
III. Results
Table 2 shows the average marginal effects estimated using the probit regression model for
the 2007 and 2013 data, respectively. The estimates suggest that all marginal effects are highly
1
The income amounts have been adjusted for inflation using the Consumer Price Index (1999 =
100). Also, striving for the maximum number of observations, we use log (inflation-adjusted income + 0.1)
to account for income value of zero.
5
statistically significant. The age and gender gap in health insurance coverage has shrunk. A one
year decrease in age reduced the average probability of having health insurance by 0.54 percentage
point in 2007, but by 0.50 percentage point in 2013. The probability of adult males having health
insurance was 3.32 percentage points lower than that of adult females in 2007; this number dropped
to 2.99 in 2013.
Table 2 Average marginal effects of probit model, 2007 vs. 2013
VARIABLES
2007
2013
age
0.0054***
0.0050***
(7.30e-05)
(7.28e-05)
male
-0.0332***
-0.0299***
(0.0025)
(0.0025)
log_income
0.0031***
0.0007**
(0.0003)
(0.0003)
fulltime
0.0040
0.0330***
(0.0034)
(0.0034)
self_emp
-0.0859***
-0.1130***
(0.0044)
(0.0047)
highschool
0.0884***
0.0801***
(0.0033)
(0.0037)
college
0.1850***
0.1780***
(0.0033)
(0.0036)
graduate
0.2890***
0.2740***
(0.0069)
(0.0064)
white
0.0391***
0.0368***
(0.0029)
(0.0029)
number of observations
125,851
122,092
Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
The race and income gap in health insurance coverage has also shrunk. For example,
Whites were 3.91 percentage points more likely to have health insurance than non-Whites in 2007,
but were 3.68 percentage points more likely in 2013. A 10% decrease in income reduced the
average probability of having health insurance by 0.031 percentage point in 2007, but by 0.007
percentage point in 2013.
In addition, the education gap in health insurance coverage has shrunk. High school
graduates were 8.84 percentage points more likely to have health insurance than those with below
high school education in 2007; this number fell to 8.01 in 2013. For college-educated individuals,
the probability of having health insurance was 18.5 percentage points higher than those who did
not go to college in 2007, but were 17.8 percentage points higher in 2013. Individuals with
graduate degrees or higher were 28.9 percentage points more likely to have health insurance than
those who did not have graduate degrees in 2007; this number fell by 1.5 in 2013.
Unfortunately, health insurance coverage has deteriorated for individuals who are self-
employed or do not have full-time jobs. The self-employed were 8.59 percentage points less likely
to have health insurance than their non-self-employed counterparts in 2007, but were 11.3
percentage points less likely in 2013. For people without full-time jobs, they were 0.4 percentage
point less likely to have health insurance than full-time job holders in 2007, but were 3.3
percentage points less likely in 2013. These two marginal effect changes are the biggest among all
the factors examined.
6
Lastly, despite the changes observed between 2007 and 2013, the magnitudes of most
changes are fairly small. Also, men, minorities, and young people are still less likely to have health
insurance, so are low-income and less-educated individuals as well as those who are self-employed
or do not have full-time jobs.
IV. Conclusion
This study examines the determinants of adult health insurance coverage in the U.S.
between 2007 (pre-ACA) and 2013 (post-ACA). We find that after the signing of the ACA, health
insurance coverage has improved among males, youth, minorities, and low-income and less-
educated individuals, but deteriorated for people who are self-employed or do not work full-time.
While the first set of findings are encouraging, the observed changes are still rather small.
Moreover, the ACA did not expand health insurance coverage as fast and drastically as expected.
Perhaps more public outreach and education are needed to magnify the positive impact. Also,
future research should re-examine this topic when more recent data become available.
The second set of findings are troubling. In light of these findings, we make the following
recommendations to policymakers and practicing managers. For policymakers, we suggest that
future policies offer companies more incentives to provide their part-time employees with health
insurance. Also, policymakers should pay more attention to people who are self-employed by
encouraging them to obtain coverage and making health insurance more affordable to them. This
group of people will grow rapidly in size in the near future as technology advancement makes
freelancing and entrepreneurship more accessible to the general public.
For practicing managers, as internet-based technology progresses further, freelancing will
become the future of work either by workers’ choice or due to business contingency. Currently,
freelancers are considered as self-employed and are responsible for their medical and dental
insurance entirely on their own. While this group of people are largely neglected by policymakers,
one wonders if there is a solution in the private sector. For example, businesses could offer to
partially contribute to a freelancer’s health insurance if the freelancer agrees to pay into the
company’s account reserved for covering the benefits of their regular employees. Given our
findings, this approach not only demonstrates a company’s commitment to social responsibility
but potentially makes itself more attractive to certain freelancers.
Author Contact Information
Cui, Zhen: zcui@calstatela.edu; Hazra, Devika: dhazra@calstatela.edu.
References
Blumberg, L.J., Corlette, S., and Lucia, K. 2014. The Affordable Care Act: Improving incentives
for entrepreneurship and self-employment. Public Policy & Aging Report, 24(4), 162-7.
Buchmueller, T. C., Levinson, Z. M., Levy, H. G., and Wolfe, B. L. 2016. Effect of the Affordable
Care Act on racial and ethnic disparities in health insurance coverage. American Journal
of Public Health, 106, 1416-21.
Cantiello, J., Fottler, M. D., Oetjen, D., and Zhang, N. J. 2015. The impact of demographic and
perceptual variables on a young adult’s decision to be covered by private health insurance.
BMC Health Services Research, 15, 1-15.
7
Chen, J., Vargas-Bustamante, A., Mortensen, K., and Ortega, A. N. 2016. Racial and ethnic
disparities in health care access and utilization under the Affordable Care Act. Medical
Care, 54(2), 140-6.
Gius, P. M. 2010. An analysis of the health insurance coverage of young adults. International
Journal of Applied Economics, 7(1), 1-17.
McMorrow, S., Long, S. K., Kenney, G.M., and Anderson, N. 2015. Uninsurance disparities have
narrowed for Black and Hispanic adults under the Affordable Care Act. Health Affairs,
34(10), 1774-8.
Sommers, B. D., Buchmueller, T., Decker, S. L., Carey, C., and Kronick, K. 2013. The Affordable
Care Act has led to significant gains in health insurance and access to care for young
adults. Health Affairs, 32(1), 165-74.
Sommers, B. D., Gunja, M. Z., Finegold, K., and Musco, T. 2015. Changes in self-reported
insurance coverage, access to care, and health under the Affordable Care Act. Journal of
the American Medical Association, 314(4), 366-74.
Sommers, B. D., Maylone, B., Blendon, R. J., Orav, E. J., and Epstein, A. M. 2017. Three-year
impacts of the Affordable Care Act: Improved medical care and health among low-income
adults. Health Affairs, 36, 1119-28.