Public and private health insurance premiums: how do they affect the health insurance status of low-income childless adults?The Patient Protection and Affordable Care Act (ACA) will substantially increase public health insurance eligibility and alter the costs of insurance coverage. Using Current Population Survey (CPS) data from the period 2000 2008, we examine the effects of public and private health insurance premiums on the insurance status of low-income childless adults, a population substantially affected by the A CA. Results show higher public premiums to be associated with a decrease in the probability of having public insurance and an increase in the probability of being uninsured, while increased private premiums decrease the probability of having private insurance. Eligibility for premium assistance programs and increased subsidy levels are associated with lower rates of uninsurance. The magnitudes of the effects are quite modest and provide important implications for insurance expansions j or childless adults under the A CA.
Low-income childless adults are among the most likely groups of Americans to be uninsured. In 2008, childless adults accounted for 58% of the nonelderly uninsured. Among childless adults below the federal poverty level, 47% are uninsured (Hoffman et al. 2009). One reason for low coverage among this population is that many do not have access to employer-sponsored insurance and cannot afford coverage through the individual market. Additionally, few nonelderly childless adults qualify for public health insurance programs unless they are disabled or pregnant (Artiga and Schwartz 2010). Other factors leading to high uninsurance levels among childless adults include: disenrollment from their parents' plan at a certain age, a belief that they are healthy and not in need of medical care, and a reliance on the health care safety net.
These high levels of uninsurance lead to incomplete risk pooling within the health insurance market. Evidence suggests that lack of health insurance results in worse health outcomes because the uninsured tend to receive less preventive care, to be diagnosed at more advanced disease stages, to receive less comprehensive care, and to have higher mortality rates (Coleman et al. 2002). Lack of insurance also places individuals and their families at risk of financial catastrophe, as medical bills have been found to be a contributing factor in many personal bankruptcies (Dranove and Millenson 2006).
This study examines state-level childless adult health insurance expansions to analyze the effect of both public and private premium levels on the insurance status of low-income childless adults. State-level expansions have taken a variety of forms. States can use Section 1115 or other waivers to provide coverage through a public insurance program, or subsidize private health insurance through premium assistance programs. Given the flexibility states have in designing and implementing their expansion programs, many childless adult programs require more cost-sharing than the typical Medicaid program, often in the form of premiums. While premiums may limit the extent of private insurance crowd-out and help states constrain program costs, they can also potentially reduce participation rates since low-income enrollees may be unable to afford the premiums. Some state programs have relied on the use of premium subsidies to aid in the purchase of private health insurance, including both group and individual insurance, to promote coverage through the private market. For example, premium assistance programs may subsidize privately offered health insurance products through an employer as well as individual plans among those without access to employer-sponsored insurance. While each state is unique in the structure and subsidy levels of its premium assistance programs, premium subsidies are typically provided on a sliding scale by income.
Recent passage of the Patient Protection and Affordable Care Act (ACA) will substantially increase eligibility for publicly funded health insurance and alter the costs of private coverage among the low-income childless adult population. Under the ACA childless adults with incomes up to 133% of the federal poverty level (FPL) will be eligible for public coverage through Medicaid without a premium, and those with incomes between 133% and 400% FPL will be eligible for subsidies to purchase private insurance through insurance exchanges. For those receiving subsidies, premium contributions will be limited to a percentage of income on a sliding scale from 4% to 9.5%. Given this policy change, it is important to understand how premium levels will affect insurance status in the population targeted by the ACA. The purpose of this paper is to provide an analysis of the effect of both public and private premium levels on the insurance status of low-income childless adults. This is a timely analysis with important implications given recent passage and the impending implementation of the ACA.
A number of earlier studies have found that new or increased premiums resulted in lower enrollment in public insurance programs (Artiga and O'Malley 2005; Kenney et al. 2007; Ku and Coughlin 1999). Similar studies have examined policy changes in individual states, consistently finding that program participation depends heavily on premium levels. In Oregon, premium increases resulted in large enrollment reductions, with approximately one-third of enrollees citing premium costs as the reason for disenrollment (Carlson et al. 2005; McConnell and Wallace 2004). Even without increases in premiums, low-income individuals may be unable to afford continued premium payments leading to disenrollment (Office of Health Care Statistics 2004). On the other hand, premium reductions have been shown to lead to increased program participation (Long and Marquis 2002).
While these studies examine the effect of public premiums on insurance rates, very few control for the price of private health insurance. Economic theory would suggest it is the relative price that affects behavior and the literature confirms that private health insurance premiums impact overall health insurance status, and hence, are likely to affect the ability of states to charge premiums and still achieve intended goals (Chernew, Cutler, and Keenan 2005; Cunningham, Hadley, and Reschovsky 2002). In an effort to increase or maintain private health insurance coverage rates, state premium assistance programs provide subsidies to help lower enrollee premiums. The literature suggests that while insurance subsidies increase participation, rather high levels are needed to encourage significant participation rates (Chernew, Frick, and McLaughlin 1997; Marquis and Long 1995; Thomas 1994; Thorpe et al. 1992).
Health insurance expansions are designed with the intention of increasing insurance coverage among those currently uninsured. However, once individuals qualify for public health insurance they may drop their private insurance and participate in the public program. While there is little agreement in the literature on the exact magnitude of this "crowd-out" there is broad agreement that it does occur (Gruber and Simon 2010; Aizer and Grogger 2003; Blumberg, Dubay, and Norton 2000; Cutler and Gruber 1996; Dubay and Kenney 1996, 1997; Ham and Shore-Sheppard 2005; Hudson, Selden, and Banthin 2005; Lo Sasso and Buchmueller 2004; Shore-Sheppard et al. 2008; Thorpe and Florence 1998; Yazici and Kaestner 2000). In an effort to limit the extent of crowd-out, many state health insurance expansions include anti crowd-out provisions, such as higher cost-sharing requirements often in the form of premiums.
Few studies have examined how public health insurance premiums affect insurance status while controlling for the cost of private insurance coverage, although two studies examined the role of both public and private premiums on children's health insurance coverage. These studies' authors found that increases in public (private) premiums reduced enrollment in public (private) programs, while increasing the likelihood of both private (public) coverage and uninsurance. The authors suggested that states that imposed or increased insurance premiums for near-poor children would succeed in discouraging the crowd-out of private insurance, but at the expense of higher uninsurance rates among children (Hadley et al. 2006/2007; Kenney, Hadley, and Blavin 2006/2007).
Data and Methods
Our analysis is based on a standard economic model of the demand for health insurance. Health insurance decisions are based on the costs of available coverage choices, income, health, and personal characteristics (Hadley and Reschovsky 2002; Marquis and Holmer 1996). Childless adult health insurance expansions alter the choice set available to those eligible. Adults eligible for public coverage can choose to take up such coverage or remain either privately insured or uninsured. Similarly those eligible for private coverage subsidies can choose to enroll or remain in their current insurance situation. The key variable of interest in the analysis is the premium required to enroll in the public or private insurance programs. The empirical model assumes that the choice an individual makes depends on the costs of each of the alternative insurance options. The probability of each type of insurance is estimated while controlling for the cost of each alternative option and individual characteristics likely to affect the choice of coverage. It is hypothesized that for those eligible for public insurance, the lower the public premium the greater the likelihood of public insurance, holding constant the price of private insurance. Additionally, as states' premium subsidies reduce the price of private health insurance coverage, it is expected that increased subsidies will reduce the likelihood of being uninsured.
The primary data for the analysis come from the March supplements to the Current Population Survey (CPS) from the 2000-2008 period. The March supplement to the CPS provides a large database with detailed demographic, employment, and health insurance information on individuals in the United States. Most importantly, the CPS includes information on family structure, household income, and state identifiers, making it possible to impute eligibility and premium levels for health insurance expansion programs given each state's eligibility criteria. Public insurance premiums were obtained for each state and year using state program websites and premium schedules in effect at the time of the survey. To estimate private health insurance premiums, data from the publicly available Medical Expenditure Panel Survey-Insurance Component (MEPS-IC) summary tables were abstracted at the firm size/state/year level.
Table 1 displays the income eligibility criteria and average premium levels among the 16 states and the District of Columbia included in the analysis of states with childless adult public health insurance expansions. Table 2 displays the income eligibility criteria, as well as the average unadjusted private premium, average subsidy level, and adjusted private premium among the six states included in our premium assistance programs analysis. We excluded individuals eligible for both public health insurance and a premium assistance program from the premium assistance analysis since public health insurance eligibility is unrelated to the subsidy. These insurance expansions provide a quasi-experiment in that childless adults are exposed to changes in eligibility and some to changes in premiums while others are not.
The dependent variable used in the analysis is self-reported health insurance status. In the CPS, individuals are asked to report all types of health insurance coverage they had in the previous year, leading some individuals to report multiple types of coverage. We used a hierarchy to assign one type of insurance coverage to each individual: those reporting any form of private coverage, followed by those reporting public coverage, followed by those with no insurance. Those with access to other forms of public insurance, such as pregnant women and those reporting coverage through TRICARE, are excluded because these individuals face a different insurance choice set. Additionally, students under age 23 are excluded due to the potential eligibility through their parents' private insurance. The definition of a family within the CPS data is not necessarily the same as the definition used by states when determining insurance eligibility. To determine insurance expansion eligibility, household members were placed into health insurance units (HIUs), allowing for the grouping of individuals according to their insurance eligibility rather than family relation or household membership. The study sample consists of adults ages 19 to 64 who live in HIUs without children and who meet the eligibility criteria for their state health insurance program. Sample sizes are shown in Table 3.
Eligibility and Premium Determination
The key independent variables in the analysis are the public and private health insurance premiums faced by childless adults eligible for health insurance expansions. To be considered eligible, individuals must meet both income and categorical requirements. Income eligibility is determined by comparing HIU income relative to the federal poverty level (FPL) to the maximum income allowed by the insurance program in each individual's state at the time of the interview. The eligibility determination also takes into account family structure, age, and in some cases, firm size and employment status. Among those eligible for public insurance, it is determined whether they face a premium and, for those facing a premium, the annual premium amount. We set the public premium to zero for childless adults eligible for public programs without premium requirements. We used the Consumer Price Index (CPI) to adjust all premiums to 2008 dollars.
To estimate private health insurance premiums, we applied individual plan private premium levels by firm size in each state and year from the MEPS-IC to observations in the CPS by work status and firm size for adults living in the household. To merge the MEPS-IC data with the CPS data set, we created an HIU level firm size hierarchy variable, as previously used in the literature (Kenney, Hadley, and Blavin 2006/2007). The hierarchy applied to all HIUs with at least one non-self-employed worker as follows: any adult in the HIU working for a firm with 1,000 employees or more, with 100 to 999 employees, with 25 to 99 employees, with 10 to 24 employees, and with fewer than 10 employees. The underlying assumption of this approach is that if only one member of a married couple is not self-employed, the couple will take up coverage available to that worker; if both members of a married couple in an HIU are employed, the couple will take up coverage from the larger employer. Adults in an HIU without a worker or with only self-employed workers were assigned the premium level faced by employees in firms with one to nine employees. Premiums faced by those in the nongroup market, essentially a group of one, should be closest to premiums among those faced in the small-group market (one to nine employees). Additionally, small-group and nongroup market premiums should be correlated due to the underlying geographic variation in provider prices and benefit mandates within states. A key assumption in our private premium imputations is that individuals face the full price of the private health insurance premium, either directly or through reduced wages, an assumption commonly supported in the literature (Miller 2004; Olson 2002; Sheiner 1999). We made additional adjustments for private premium levels among those eligible for premium assistance programs using state rules in place at the time of the survey. For example, if an individual were determined to be eligible for a premium subsidy of 30%, the adjusted annual private premium would be calculated by reducing the premium by the corresponding amount. In instances where MEPS-IC data were unavailable, premium levels were imputed by calculating the average premium from the years preceding and following the year with missing data. As with public premiums, private insurance premiums were adjusted to 2008 dollars using the CPI.
Multivariate Analysis and Identification
We used two separate analyses to examine the effect of premiums on insurance status. We employed a multinomial logistic regression model to analyze the effects of both public and private health insurance premiums among childless adults eligible for public health insurance expansions with and without a premium. This analysis allows for the assessment of a multinomial dependent variable of insurance status, indicating either public insurance, private insurance, or uninsurance. The key independent variables of interest in this model include both the imputed annual public health insurance premium and the imputed annual private health insurance premium.
Analysis of the impact of premium assistance programs and the associated subsidies on the insurance status of eligible childless adults was examined in two parts. First, using a difference-in-difference multivariate logistic regression model, we examined the effect of eligibility on health insurance status. With difference-in-difference modeling, changes in insurance status from a control group are subtracted from changes in insurance status among those eligible for premium assistance programs. The control group consists of childless adults in expansion states who have incomes up to 400% of the poverty level and are not eligible for any health insurance expansions. Use of difference-in-difference modeling provides a within-state control for other factors that may have changed in the absence of insurance expansions. With the difference-in-difference model, the expansion effect is secured through the interaction of the time difference (expansion program operational) and the group difference (expansion eligible individual). As Ai and Norton (2003) and Norton, Wang, and Ai (2004) have noted, marginal effects and their standard errors cannot be obtained directly from the coefficients of interaction terms in nonlinear models. To obtain the correct marginal effects we used the margins command in Stata 11, taking into account the interaction term between the variables (Karaca-Mandic, Norton, and Dowd 2011).
Second, we used a multivariate logistic regression model to investigate the effect of premium subsidy levels on insurance status while controlling for the full private health insurance premium among childless adults eligible for a premium assistance program. The sample for this analysis is comprised of childless adults eligible for premium assistance programs. The dependent variable in this specification is a dichotomous measure of insurance status of the eligible childless adult (insured/uninsured). In this specification, the key independent variables of interest are the annual unadjusted private health insurance premium and the annual premium assistance subsidy. Unlike the public insurance model, a public insurance premium is not included since this population is not eligible for public coverage.
Each of the multivariate models includes a number of variables to control for demographic and socioeconomic characteristics. Childless adult characteristics include: age, gender, race/ethnicity, marital status, HIU income (CPI adjusted), citizenship status, health status, highest educational attainment level in the HIU, and HIU work status (any adult working full time/full year, any adult working full time/part year, any adult working part time). County-level unemployment rates are also included because local labor market characteristics are likely to affect both access to private health insurance and enrollment in public insurance programs (Cawley and Simon 2005). State and year fixed effects are included in the model to capture permanent time-invariant differences in state characteristics and overall secular trends that may affect insurance status. All of the models were estimated using Stata version 11.1. Standard errors were computed with the delta method, and adjusted for clustering on state and year.
Identification of the public premium effect on health insurance status is derived from within-state variation due to varying premium requirements by HIU income, and over time due to changes in program premium levels. Between 2000 and 2008, the average annual inflation-adjusted public insurance premium increased from $134 to $235. Identification of the private insurance premium effect on health insurance status is derived from a combination of within-state variation in premiums and changes over time in private insurance premiums. Between 2000 and 2008, the average annual inflation-adjusted private insurance premium increased from $3,109 to $3,858. Identification of the premium assistance subsidy on health insurance status is derived from varying subsidy levels over time and within states among individuals eligible for premium assistance programs. For example, between 2000 and 2008, the average annual inflation-adjusted subsidy level increased from $2,133 to $3,682.
Table 3 provides summary statistics for the sample of eligible childless adults pooled across states and over the 2000 2008 period. The first column includes childless adults eligible for public health insurance expansions, while the second column includes only those eligible for premium assistance programs. Childless adults eligible for public insurance expansions were more likely to be insured, more likely to have lower household income, less likely to be in a household with a full-time worker, and less likely to be married than those eligible for premium assistance programs. Among the publicly eligible group, the average annual public health insurance premium was $173, while the average annual private health insurance premium was $4,250. The average unadjusted private health insurance premium among those eligible for premium subsidies was $4,570, while the average subsidy was $3,161.
Table 4 presents the multinomial logistic regression results for childless adults eligible for public health insurance expansions. The marginal probabilities indicate that higher public premiums reduce the likelihood of being publicly insured while increasing the probability of being uninsured. Specifically, a $1,000 increase in public premiums was associated with a 14.2-percentage-point reduction in the probability of public insurance and an 8.2-percentage- point increase in the probability of being uninsured. The results also indicate that higher private premiums are associated with a reduced likelihood of being privately insured. Specifically, a $1,000 increase in private premiums was associated with a 3.3-percentage-point reduction in the probability of private insurance. The covariates included in the model were in the expected direction. For example, people who were black, Hispanic, in poorer health, and unmarried were less likely to have private health insurance and more likely to be publicly insured, while citizens and those with higher incomes and education levels were less likely to be uninsured.
The results of the premium assistance models are presented in Tables 5 and 6. Results from the difference-in-difference analysis indicate that childless adults eligible for a premium assistance program were 4.7 percentage points less likely to be uninsured (Table 5). Covariates in the model were in the expected direction, for example those with higher household income, higher levels of education, citizens, and females were less likely to be uninsured. Table 6 reports the logistic regression results for the model examining the effect of premium assistance subsidies on insurance status. The marginal probabilities indicate that higher unadjusted private insurance premiums were associated with an increase in the likelihood of being uninsured, while increased subsidies were associated with a reduction in the likelihood of being uninsured. Specifically, a $1,000 increase in private premiums was associated with a 10.4-percentage-point increase in the likelihood of being uninsured, while a $1,000 increase in the subsidy level was associated with a 3.4-percentage-point reduction in the likelihood of being uninsured. As with the previous model, the covariates were in the expected direction.
To examine the sensitivity of our results to the imputation of private health insurance premiums, we estimated a number of alternative models. First, we used a weighted average private premium, as previously used in the literature (Kenney, Hadley, and Blavin 2006/2007). The private premium was calculated with data from the MEPS-IC at the firm size/state/year level using the following formula:
Private Premium = (insurance offer rate * employee share of premium) + ((1 - insurance offer rate) * full premium).
The employer-sponsored insurance offer rate was set to zero among households without a worker or with only self-employed workers, thus assigning these individuals the full premium amount. This specification also allowed us to test our results to the varying health insurance offer rates by firm size. Second, given the uncertainties around premium levels in the nongroup market, we adjusted premiums upward by 10% in one specification and downward by 10% in another among households without a worker or with only self-employed workers. The results from our analyses using alternative private health insurance premiums, demonstrates the robustness of our findings to the imputation of private health insurance premiums, in particular to the assignment of private premiums among households without a worker or with only self-employed workers.
Alternative models were used to examine the robustness of our findings to the treatment of individuals with nongroup private health insurance. To account for the possibility that some public health insurance programs could be mistaken by survey respondents for private health insurance plans (Lo Sasso and Buchmueller 2004), we examined models in which nongroup private coverage is classified as public insurance. We also examined the sensitivity of our results when excluding individuals with nongroup private insurance. We found that our results are not sensitive to the treatment of those with nongroup private health insurance.
A number of additional models were estimated to examine the robustness of our findings to various specifications. To test the robustness of the results by the treatment of respondents reporting more than one type of coverage, public coverage was placed at the top of the hierarchy, followed by private insurance and the uninsured category. To address the possibility that states requiring premiums for public health insurance expansion programs differ systematically from those without premium requirements in ways not accounted for by state fixed effects, we examined models including only the subset of states with premium requirements. Given the variation in program design and its potential impact on take-up, we examined an alternative model excluding programs with limited benefits, and a model excluding programs with enrollment caps. Additionally, we explored the sensitivity of our results to the Massachusetts individual health insurance mandate by examining an alternative model excluding Massachusetts. Additional analyses were performed excluding noncitizens, given their potential exclusion from state expansion programs. Overall, the various sensitivity checks showed the same pattern of results, implying a generally robust relationship between premium levels and health insurance status.
While the results from this study are robust to many different model specifications, the generalization of the results may be limited because only childless adults eligible for health insurance expansions in 19 states and the District of Columbia are included in the analysis. There are a number of limitations regarding our premium level calculations. First, our reliance on imputed values for public and private health insurance premiums may introduce measurement error into our analysis. A second limitation is the use of individual private health insurance premiums from the MEPS-IC regardless of marital status. However, we examined our models excluding married individuals and our findings were robust. An additional limitation regarding private insurance premiums is the unavailability of individual market, nongroup private health insurance premiums for the years included in our analysis. However, as discussed, we examined several alternative specifications varying the private health insurance premiums among households without a worker or with only self-employed workers. An additional limitation of our analysis is the use of pooled cross-sectional data rather than panel data; thus we were unable to follow eligible childless adults over time and observe changes in health insurance premiums.
Discussion and Policy Implications
We have examined the effect of public and private health insurance premiums on the insurance status of low-income childless adults. We find that higher public insurance premiums are associated with a decreased probability of having public insurance and an increased likelihood of being uninsured. Additionally, among those eligible for public insurance, higher private health insurance premiums are associated with a reduction in the probability of having private insurance. The findings from this analysis are consistent with previous research analyzing the effect of public and private health insurance premiums on insurance status (Hadley et al. 2006/2007; Kenney, Hadley, and Blavin 2006/2007). This study extends upon the previous literature by examining the effects among childless adults, a population seldom examined but one that will be affected significantly by the ACA. Additionally, this study analyzed the effect of private health insurance subsidies on the insurance status of those eligible for premium assistance programs, finding that both program eligibility and increased subsidy levels are associated with a reduction in the likelihood of being uninsured.
Although we find that higher public insurance premiums are associated with a reduction in the likelihood of public insurance and an increased likelihood of uninsurance, the results are quite modest. For example, among those eligible for public insurance, the average annual public insurance premium was $173; a $100 premium increase was associated with a mere 1.4-percentage-point reduction in public insurance and a .8-percentage-point increase in uninsurance. These findings suggest that Medicaid expansions with no premium requirements for childless adults with incomes up to 133% of the FPL under the ACA have the potential to increase public insurance take-up. However, the resulting change in uninsurance rates may be small, leaving a large portion of individuals uninsured.
Previous studies have also found modest increases in insurance coverage following the expansion of health coverage to similar adult populations (Aizer and Grogger 2003; Busch and Duchovny 2005; Guy 2010; Kronick and Gilmer 2002). The magnitude of the effects may be small for several reasons; the availability of charity care has been shown to reduce the demand for health insurance and increase the likelihood of being uninsured, especially among the low-income population (Herring 2005; Rask and Rask 2000). Additionally, information and administrative costs, along with the perceived stigma and reputation of public insurance, have been shown to be important barriers to enrollment in public insurance programs (Aizer 2007; Ketsche et al. 2007).
Our analysis also finds that premium assistance program eligibility and increased subsidy levels reduce the likelihood of uninsurance. However, we find modest effects of subsidy levels on the insurance status of those eligible for such programs. Similar to previous research, our results suggest that rather high subsidy levels are needed to encourage substantial take-up (Chernew, Frick, and McLaughlin 1997; Marquis and Long 1995; Thomas 1994; Thorpe et al. 1992). For example, in our analysis the average annual private insurance subsidy was $3,161; a $1,000 subsidy increase was associated with a modest 3.4-percentage-point reduction in uninsurance. The ACA includes premium subsidies for individuals with incomes between 133% and 400% FPL to purchase private insurance through insurance exchanges; our results suggest these subsidies will have to be rather large to encourage significant take-up of private health insurance.
Overall, this study provides interesting insight into the effect of both public and private health insurance premiums on the eve of health care reform. Under reform, insurance coverage will be significantly expanded among the low-income childless adult population, and the costs of coverage will be dramatically altered. In addition to expanding public insurance eligibility and private insurance subsidies, the individual health insurance mandate under the ACA may also play an important role in increasing coverage. Our results indicate that expanded public health insurance without premium requirements through Medicaid and private health insurance subsidies under the ACA should reduce the uninsurance rates among low-income childless adults. However, our results suggest that the reductions in uninsurance may be quite modest. An additional provision of the ACA, the individual mandate, gives added incentive, which may aid in increasing insurance coverage rates. The combination of financial incentives, as well as behavioral responses and the desire to comply with the mandate, may lead to further reductions in uninsurance rates, as suggested in Massachusetts (Buettgens, Garrett, and Holaham 2010).
Reversing the historical exclusion of childless adults from Medicaid and providing premium credits for the purchase of private health insurance are important steps in reducing the number of low-income uninsured Americans, but there is much work left to be done. As evidenced in our study sample, despite being eligible for public coverage or private insurance subsidies, many individuals remain uninsured. It is clear that targeted policy and outreach efforts will be needed to identify hard-to-reach groups and address their particular barriers to health insurance take-up. Along with increased public insurance eligibility and private health insurance subsidies, these efforts can potentially lead to increased participation rates and ultimately reduce the number of uninsured Americans.
Ai, C., and E. C. Norton. 2003. Interaction Terms in Logit and Probit Models. Economies Letters 80(1):123-129.
Aizer, A. 2007. Public Health Insurance, Program Take-up, and Child Health. Review of Economics and Statistics 89(3):400-415.
Aizer, A., and J. Grogger. 2003. Parental Medicaid Expansions and Health Insurance Coverage. NBER Working Paper No. 9907. Cambridge, Mass.: National Bureau of Economic Research (NBER).
Artiga, S., and M. O'Malley. 2005. Increasing Premiums and Cost Sharing in Medicaid and SCHIP: Recent State Experiences. Washington, D.C.: Kaiser Commission on Medicaid and the Uninsured.
Artiga, S., and K. Schwartz. 2010. Expanding Health Coverage for Low-Income Adults: Filling the Gaps in Medicaid Eligibility. Washington, D.C.: Kaiser Commission on Medicaid and the Uninsured.
Blumberg, L. J., L. Dubay, and S. A. Norton. 2000. Did the Medicaid Expansions for Children Displace Private Insurance? An Analysis Using the SIPP. Journal of Health Economics 19(1):33-60.
Buettgens, M., B. Garrett, and J. Holaham. 2010. Why the Individual Mandate Matters: Timely Analysis of Immediate Health Policy Issues. Washington, D.C.: The Urban Institute.
Busch, S. H., and N. Duchovny. 2005. Family Coverage Expansions: Impact on Insurance Coverage and Health Care Utilization of Parents. Journal of Health Economics 24(5):876-890.
Carlson, M. J., B. Wright, L. K. Gallia, J. Smith, and J. Miller. 2005. The Impact of Program Changes on Enrollment, Access, and Utilization in the Oregon Health Plan Standard Population. Salem, Ore.: Office of Oregon Health Policy and Research.
Cawley, J., and K. I. Simon. 2005. Health Insurance Coverage and the Macroeconomy. Journal of Health Economics 24(2):299-315.
Chernew, M., D. M. Cutler, and P. S. Keenan. 2005. Increasing Health Insurance Costs and the Decline in Insurance Coverage. Health Services Research 40(4):1021-1039.
Chernew, M., K. Frick, and C. G. McLaughlin. 1997. The Demand for Health Insurance Coverage by Low-Income Workers: Can Reduced Premiums Achieve Full Coverage? Health Services Research 32(4):453-470.
Coleman, M. S., A. L. Kellerman, R. M. Andersen, J. Z. Ayanian, R. J. Blendon, S. P. Davis, et al. 2002. Care Without Coverage: Too Little, Too Late. Washington, D.C.: National Academies Press.
Cunningham, P. J., J. Hadley, and J. Reschovsky. 2002. The Effects of SCHIP on Children's Health Insurance Coverage: Early Evidence from the Community Tracking Study. Medical Care Research and Review 59(4):359-383.
Cutler, D. M., and J. Gruber. 1996. Does Public Insurance Crowd Out Private Insurance? Quarterly Journal of Economics 111(2):391-430.
Dranove, D., and M. L. Millenson. 2006. Medical Bankruptcy: Myth Versus Fact. Health Affairs 25(2):w74-w83.
Dubay, L., and G. Kenney. 1997. Peer Review: Did Medicaid Expansions for Pregnant Women Crowd Out Private Coverage? Health Affairs 16(1):185-193.
Dubay, L. 1996. The Effects of Medicaid Expansions on Insurance Coverage of Children. The Future of Children 6(1): 152-161.
Gruber, J., and K. Simon. 2008. Crowd-Out 10 Years Later: Have Recent Public Insurance Expansions Crowded Out Private Health Insurance? Journal of Health Economics 27(2):201-217.
Guy, G. P. 2010. The Effects of Cost Sharing on Access to Care among Childless Adults. Health Services Research 45(6): 1720-1739.
Hadley, J., and J. D. Reschovsky. 2002. Small Firms' Demand for Health Insurance: The Decision to Offer Insurance. Inquiry 39(2): 118-137.
Hadley, J., J. D. Reschovsky, P. Cunningham, G. Kenney, and L. Dubay. 2006/2007. Insurance Premiums and Insurance Coverage of Near-Poor Children. Inquiry 43(4):362-377.
Ham, J. C., and L. Shore-Sheppard. 2005. The Effect of Medicaid Expansions for Low-Income Children on Medicaid Participation and Private Insurance Coverage: Evidence from the SIPP. Journal of Public Economics 89(1):57-83.
Herring, B. 2005. The Effect of the Availability of Charity Care to the Uninsured on the Demand for Private Health Insurance. Journal of Health Economics 24(2):225-252.
Hoffman, C., K. Schwartz, J. Howard, J. Tolbert, A. Cook, and E. Lawton. 2009. The Uninsured, A Primer: Key Facts about Americans Without Health Insurance. Menlo Park, Calif.: Henry J. Kaiser Family Foundation, Kaiser Commission on Medicaid and the Uninsured.
Hudson, J. L., T. M. Selden, and J. S. Banthin. 2005. The Impact of SCHIP on Insurance Coverage of Children. Inquiry 42(3):232-254.
Karaca-Mandic, P., E. C. Norton, and B. Dowd. 2012. Interaction Terms in Nonlinear Models. Health Services Research 47(1):255-274.
Kenney, G., R. A. Allison, J. Costich, J. Matron, and J. McFeeters. 2007. The Effects of Premium Increases on Enrollment in SCHIP Programs: Findings from Three States. Inquiry 43(4):378-392.
Kenney, G., J. Hadley, and F. Blavin. 2006/2007. Effects of Public Premiums on Children's Health Insurance Coverage: Evidence from 1999 to 2003. Inquiry 43(4):345-361.
Ketsche, P., E. K. Adams, K. Minyard, and R. Kellenberg. 2007. The Stigma of Public Programs: Does a Separate SCHIP Program Reduce It? Journal of Policy Analysis and Management 26(4):775-790.
Kronick, R., and T. Gilmer. 2002. Insuring Low-Income Adults: Does Public Coverage Crowd Out Private? Health Affairs 21 (1):225-239.
Ku, L., and T. Coughlin. 1999. Sliding-Scale Premium Health Insurance Programs: Four States' Experiences. Inquiry 36(4):471-480.
Long, S. H., and M. S. Marquis. 2002. Participation in a Public Insurance Program: Subsidies, Crowd-Out, and Adverse Selection. Inquiry 39(3):243-257.
Lo Sasso, A. T., and T. C. Buchmueller. 2004. The Effect of the State Children's Health Insurance Program on Health Insurance Coverage. Journal of Health Economics 23(5): 1059-1082.
Marquis, M. S., and M. R. Holmer. 1996. Alternative Models of Choice Under Uncertainty and Demand for Health Insurance. Review of Economics and Statistics 78(3):421-427.
Marquis, M. S., and S. H. Long. 1995. Worker Demand for Health Insurance in the Non-group Market. Journal of Health Economics 14(1):47-63.
McConnell, J., and N. Wallace. 2004. Impact of Premium Changes in the Oregon Health Plan. Office for Oregon Health Policy and Research.
Miller, R. D. 2004. Estimating the Compensating Differential for Employer-Provided Health Insurance. International Journal of Health Care Finance and Economics 4(1):27-41.
Norton, E., H. Wang, and C. Ai. 2004. Computing Interaction Effects and Standard Errors in Logit and Probit Models. Stata Journal 4(2): 154-167.
Office of Health Care Statistics. 2004. Utah Primary Care Network Disenrollment Report: July-August 2003. Salt Lake City, Utah: Utah Department of Health.
Olson, C. A. 2002. Do Workers Accept Lower Wages in Exchange for Health Benefits? Journal of Labor Economics 20(2):91-114.
Rask, K. N., and K. J. Rask. 2000. Public Insurance Substituting for Private Insurance: New Evidence Regarding Public Hospitals, Uncompensated Care Funds, and Medicaid. Journal of Health Economics 19(1):1-31.
Sheiner, L. 1999. Health Care Costs, Wages, and Aging: Finance and Economics Discussion Series. Board of Governors of the Federal Reserve System (U.S.).
Shore-Sheppard, L. D., B. H. Hanratty, A. Krueger, C. Rouse, and J. Rudd. 2008. Stemming the Tide? The Effect of Expanding Medicaid Eligibility on Health Insurance Coverage. The B.E. Journal of Economic Analysis & Policy 8(2):Article 6.
Thomas, K. 1994. Are Subsidies Enough to Encourage the Uninsured to Purchase Health Insurance? An Analysis of Underlying Behavior. Inquiry 31 (4):415-424.
Thorpe, K., and C. Florence. 1998. Health Insurance among Children: The Role of Expanded Medicaid Coverage. Inquiry 35(4):369-379.
Thorpe, K. E., A. Hendricks, D. Garnick, K. Donelan, and J. P. Newhouse. 1992. Reducing the Number of Uninsured by Subsidizing Employment-Based Health Insurance: Results from a Pilot Study. Journal of the American Medical Association 267(7):945-948.
Yazici, E. Y., and R. Kaestner. 2000. Medicaid Expansions and the Crowding Out of Private Health Insurance among Children. Inquiry 37(1):23-32.
Gery P. Guy Jr., Ph.D., M.P.H., is a contract researcher, and E. Kathleen Adams, Ph.D., is a full professor, both in the Department of Health Policy and Management, Rollins School of Public Health, Emory University. Adam Atherly, Ph.D., is an associate professor and chair of the Department of Health Systems, Management and Policy, Colorado School of Public Health, University of Colorado. Address correspondence to Dr. Guy at Department of Health Policy and Management, Rollins School of Public Health. Emory University, 1518 Clifton Road, N.E., Atlanta, GA 30322. Email: email@example.com
Table 1. Summary of childless adult public health insurance expansions: income eligibility and average annual premiums, 2000-2008 Average Average Income public private State program eligibility premium ($) premium ($) Arizona 0-100% FPL 0 4,195 (322) Delaware 0-100% FPL 0 4,891 (627) District of Columbia 0-200% FPL 0 4,810 (345) Hawaii 0-100% FPL 0 3,870 (461) Indiana 0-200% FPL 447 (420) 4,563 (236) Iowa 0-200% FPL 391 (378) 4,179 (154) Maine 0-100% FPL 0 5,053 (375) Maryland 0-116% FPL 0 4,490 (201) Massachusetts 0-400% FPL 0 3,298 (178) (a) Michigan 0-35% FPL 0 4,799 (216) Minnesota 0-200% FPL 281 (304) 4,178 (377) New York 0-100% FPL 0 4,711 (414) Oregon 0-100% FPL 150 (83) 209 (30) (a) Pennsylvania 0-200% FPL 420 (6) 4,581 (344) Utah 0-150% FPL 48 (13) 3,873 (337) Vermont 0-300% FPL 405 (409) 4,625 (499) Washington 0-200% FPL 487 (461) 4,028 (401) Notes: Eligibility and premiums determined using the Centers for Medicare and Medicaid (CMS) fact sheets, State Coverage Initiatives, and program websites. Premium amounts are adjusted using the Consumer Price Index (CPI) to 2008 dollars, and rounded to the nearest dollar. Standard deviations are in parentheses. FPL=federal poverty level. (a) Individuals are also eligible for premium assistance subsidies. Table 2. Summary of childless adult premium assistance programs: income eligibility, premiums, and subsidy levels, 2000-2008 Average unsubsidized Income private State program eligibility premium ($) Idaho 0-185% FPL 3,753 (259) Maine 100-300% FPL 4,963 (403) Massachusetts 0-400% FPL 4,947 (1,524) New Mexico 0-200% FPL 4,373 (325) Oklahoma 0-200% FPL 4,776 (389) Oregon 100-185% FPL 4,081 (552) Average subsidized Average private State program subsidy ($) premium ($) Idaho 1,248 (32) 2,505 (258) Maine 1,031 (425) 3,499 (1,524) Massachusetts 3,499 (1,524) 1,448 (1,178) New Mexico 3,787 (750) 587 (641) Oklahoma 3,670 (857) 1,109 (832) Oregon 3,347 (725) 734 (651) Notes: Eligibility and premiums determined using CMS fact sheets, State Coverage Initiatives, and program websites. Premium amounts are adjusted using the CPI to 2008 dollars, and rounded to the nearest dollar. Standard deviations are in parentheses. FPL=federal poverty level. Table 3. Descriptive statistics of childless adult sample pooled across states and years, 2000-2008 Eligible Eligible for public for premium Variable coverage assistance Insurance status Percent insured 57.82 52.85 Percent with public coverage 20.70 -- Percent with private coverage 37.12 -- Percent uninsured 42.18 47.15 Public premium ($) 173 -- Private premium ($) 4,250 4,570 (a) Subsidy ($) -- 3,161 Individual characteristics Age in years 36.60 37.09 Percent female 48.72 44.99 Percent black 18.45 4.78 Percent other race 11.12 9.65 Percent Hispanic 15.56 23.89 Percent citizens 86.57 85.04 Percent reporting excellent/very 51.44 56.89 good health status Percent reporting fair/poor health 20.24 13.58 status Percent married 15.37 20.82 Household characteristics Percent full-time, full-year worker 16.83 38.27 Percent full-time, part-year worker 16.17 22.57 Percent part-time worker 20.84 23.53 Percent college graduate 21.02 23.20 Percent some college 17.79 19.28 Percent high school graduate 38.6 37.31 Income ($) 6,661 14,224 Number of observations 22,521 3,616 Source: Tabulations of the 2000-2008 Current Population Survey (CPS) March supplement. (a) Unadjusted premium. Table 4. The marginal effects of premium levels on insurance status of childless adults eligible for public health insurance expansion programs Marginal effect Private insurance Public insurance Public premium (a) .0601 (.0248) ** -.1421 (.0327) *** Private premium (a) -.0326 (.0144) ** .0214 (.0114) * Age .0005 (.0003) -.0004 (.0002) * Female .0401 (.0057) *** .0418 (.0052) *** Black -.0826 (.0057) *** .0469 (.0084) *** Other race -.0347 (.0151) ** .0092 (.0153) Hispanic -.1365 (.0122) *** .0463 (.0103) *** Citizen .0766 (.0128) *** .0814 (.0099) *** Health status: excellent/ .0922 (.0076) *** -.0835 (.0070) *** very good Health status: fair/poor -.0581 (.0095) *** .1249 (.0064) *** Household: full-time, .1232 (.0117) *** -.2462 (.0122) *** full-year worker Household: full-time, .0452 (.0098) *** -.1550 (.0103) *** part-year worker Household: part-time .0610 (.0099) *** -.0959 (.0088) *** worker Married .0693 (.0126) *** -.0613 (0114) *** Household: high school .0703 (.0107) *** -.0564 (.0060) *** graduate Household: some college .1395 (.0116) *** -.0870 (.0073) *** Household: college .2073 (.0120) *** -.1442 (.0079) *** graduate Household income (a) .0050 (.0010) *** .0098 (.0012) *** Unemployment rate -.3954 (.2038) * .2572 (.1780) Marginal effect Uninsured Public premium (a) .0820 (.0270) *** Private premium (a) .0111 (.0160) Age -.0001 (.0003) Female -.0819 (.0064) *** Black .0357 (.0099) *** Other race .0255 (.0154) * Hispanic .0902 (.0143) *** Citizen -.1580 (.0138) *** Health status: excellent/ -.0087 (.0079) very good Health status: fair/poor -.0668 (.0104) *** Household: full-time, .1230 (.0142) *** full-year worker Household: full-time, .1099 (.0121) *** part-year worker Household: part-time .0349 (.0110) *** worker Married -.0080 (.0125) Household: high school -.0138 (.0104) graduate Household: some college -.0525 (.0120) *** Household: college -.0631 (.0113) *** graduate Household income (a) -.0148 (.0011) *** Unemployment rate .1381 (.2504) Notes: Regression model also includes state and year dummies. Standard errors, in parentheses, are corrected for clustering by state/year. (a) In thousands of dollars. *** Significant at p<.01; ** significant at p<05; * significant at p<.10. Table 5. The marginal effect of premium assistance program eligibility on the insurance status of childless adults Marginal effect Uninsured Expansion individual .0737 (.0119) *** Expansion program operational .1141 (.0088) *** Expansion effect -.0468 (.0218) ** Age -.0020 (.0002) *** Female -.0593 (.0029) *** Black .0206 (.0065) *** Other race .0360 (.0110) *** Hispanic .0520 (.0076) *** Citizen -.1278 (.0075) *** Health status: excellent/very good -.0226 (.0041) *** Health status: fair/poor -.0417 (.0061) *** Household: full-time, full-year worker .0149 (.0075) ** Household: full-time, part-year worker .0764 (.0074) *** Household: part-time worker .0512 (.0073) *** Married .0451 (.0051) *** Household: high school graduate -.0440 (.0056) *** Household: some college -.0752 (.0060) *** Household: college graduate -.1175 (.0063) *** Household income (a) -.0070 (.0002) *** Unemployment rate .3249 (.1299) ** Notes: Regression model also includes state and year dummies. Standard errors, in parentheses, are corrected for clustering by state/year. (a) In thousands of dollars. ***Significant at p<.01; ** significant at p<.05; * significant at p<.10. Table 6. The marginal effect of subsidy levels on the insurance status of childless adults eligible for premium assistance programs Marginal effect Uninsured Unadjusted private premium (a) .1044 (.0355) *** Subsidy (a) -.0344 (.0172) ** Age -.0005 (.0009) Female -.1032 (.0132) *** Black .0428 (.0326) Other race .0702 (.0590) Hispanic .0615 (.0247) ** Citizen -.1473 (.0297) *** Health status: excellent/very good -.0576 (.0184) *** Health status: fair/poor -.1062 (.0261) *** Household: full-time, full-year worker .0814 (.0409) ** Household: full-time, part-year worker .1088 (.0400) *** Household: part-time worker .0753 (.0410) * Married -.0306 (.0289) Household: high school graduate -.0504 (.0230) ** Household: some college -.0854 (.0243) *** Household: college graduate -.1291 (.0335) *** Household income (a) -.0051 (.0016) *** Unemployment rate .5487 (.6132) Notes: Regression model also includes state and year dummies. Standard errors, in parentheses, are corrected for clustering by state/year. (a) In thousands of dollars. *** Significant at p<.01; ** significant at p<.05; * significant at p<.10.