Arellano, M., and S. Bond. 1991. “Some Tests of Specification for Panel Data: Mote Carlo Evidence and an Application to Employment Equations.” Review of Economic Studies 58(2):277–97.CrossrefGoogle Scholar
Congressional Budget Office. 2005. “Changes in Participation in Means-Tested Programs.” Economic and Budget Issue Brief.Google Scholar
Decker, S. L., and F. W. Selck. 2012. “The Effect of the Original Introduction of Medicaid on Welfare Participation and Female Labor Supply.” Review of Economics of the Household 10:541–556.Web of ScienceCrossrefGoogle Scholar
Gruber, J. 2000. “Medicaid.” In Means-Tested Transfer Programs in the United States, edited by R. A. Moffitt, 15–78. Chicago, IL: National Bureau of Economic Research, University of Chicago Press.Google Scholar
Haider, S. J., and D. S. Loughran. 2008. “The Effect of the Social Security Earnings Test on Male Labor Supply: New Evidence from Survey and Administrative Data.” Journal of Human Resources 43(1):57–87.CrossrefWeb of ScienceGoogle Scholar
Ham, J. C., and L. D. Shore-Sheppard. 2005. “Did Expanding Medicaid Affect Welfare Participation?” Industrial and Labor Relations Review 58(3):452–70.Google Scholar
Hamersma, S., and M. H. Kim. 2013a. “Participation and Crowd Out: Assessing the Effects of Parental Medicaid Expansions.” Journal of Health Economics 32(1):160–71.PubMedWeb of ScienceCrossrefGoogle Scholar
Hamersma, S., and M. H. Kim. 2013b. “The Role of Public Health Insurance during Employment Transitions.” Working Paper.Google Scholar
Hepner, M., and R. Reed. 2004. “The Effect of Welfare on Work and Marriage: A View from the States.” Cato Journal 24(3):349–37.Google Scholar
Kaiser Commission on Medicaid and the Uninsured. June 2002. “Transitional Medical Assistance (TMA): Medicaid Issue Update.” Available from Kaiser Family Foundation, http://www.kff.org
Liebman, J., and R. Zeckhauser. 2004. “Schmeduling.” Working Paper, Kennedy School of Government, Harvard University.Google Scholar
Meyer, B. D. 2002 “Labor Supply at the Extensive and Intensive Margins: The EITC, Welfare, and Hours Worked.” AEA Papers and Proceedings (May, 2002).Google Scholar
Pohl, V. 2011. “Medicaid and the Labor Supply of Single Mothers: Implications for Health Care Reform.” Working Paper.Google Scholar
Romich, J. 2005a. “Difficult Calculations: Low-Income Workers and Marginal Tax Rates.” Social Service Review 80(1):27–66.Google Scholar
Romich, J. 2005b. “Do High Effective Marginal Tax Rates Reduce Employment and Earnings Among Low-Income Workers?” Working Paper.Google Scholar
Stock, J., and M. Yogo 2005. “Testing for Weak Instruments in Linear IV Regression.” In Identification and Inference for Econometric Models: Essays in Honor of Thomas J. Rothenberg, edited by J. Stock and D. Andrews, 80–108. New York, NY: Cambridge University Press.Google Scholar
Welfare Rules Databook. State TANF Policies as of July XXXX” for XXXX =1999 through 2006. Urban Institute publications available at www.urban.org.
About the article
Published Online: 2013-08-27
There is a federally mandated Transitional Medical Assistance (TMA) program that allows workers to continue receiving Medicaid for 6–12 months (and longer in a few states) after their incomes have exceeded the eligibility threshold. Implementation varies across states, but unfortunately there are no national data showing how many families receive TMA or the composition of those families as states are not required to separately report TMA enrollment (Kaiser Commission on Medicaid and the Uninsured, 2002). TMA coverage allows some limited transition time for a family that has lost eligibility to maintain coverage and motivates my use of lags later in the analysis. The Food Stamp program also has a small notch due to its gross income standard.
In an independently conceived project, a working paper by Pohl (2011) examines a similar question with different data and methodology and finds that simulated increases in eligibility increase full-time relative to part-time work, consistent with my findings here. The Medicaid thresholds used in Pohl (2011) are identical to those used here, as he requested and obtained them directly from me.
There were some major expansions for children’s coverage and pregnant women’s coverage in years prior to 1996 (and since), but expansions in parental coverage have occurred only in more recent years since Medicaid eligibility was formally separated from cash assistance at the federal level. Yelowitz (1995) exploits the pre-1996 variation in child thresholds to examine labor supply and welfare participation among single mothers.
The information on caseloads in this paragraph is drawn from a brief from the Congressional Budget Office (2005).
For example, Hamersma and Kim (2013a) examine the effects of eligibility expansions on health coverage and crowd-out and Gruber and Yelowitz (1999) examine the role of Medicaid asset limits on savings behavior. A detailed review of other Medicaid research, as well as additional information about Medicaid’s legislative history, can be found in Gruber (2000).
The key issue is that Yelowitz (1995) examined the effects of Medicaid thresholds using a variable called “GAIN%” defined as the percent difference between the Medicaid threshold and the AFDC (welfare) threshold. This effectively forced the Medicaid and AFDC thresholds to have effects that were equal in magnitude and opposite in sign. Ham and Shore-Sheppard (2005) test whether this assumption holds in the data and reject the assumption. They instead allow AFDC and Medicaid thresholds to enter separately.
Moffitt and Wolfe (1992) utilize the Survey of Income and Program Participation, which is a panel data set, but they use the panel information primarily to assemble variables related to past health care for use in a cross-sectional analysis.
One paper that focuses on the extensive margin (Ham and Shore-Sheppard, 2005) does briefly mention estimates on the intensive margin as well. In both cases, the authors find no effect of earnings limits.
Hamersma and Kim (2009) note that only 28% of unmarried working mothers with income below the federal poverty line had employer-provided coverage during 1996–2003. Further work analyzing the probability of transitioning from Medicaid to private coverage is currently in progress (see, for example, Hamersma and Kim 2013b).
The SIPP does identify those receiving disability benefits (SSI), which may help measure medical care needs. However, SSI receipt provides automatic eligibility for Medicaid regardless of the state’s earnings thresholds. Disability also likely influences employment in ways beyond the incentives created by relevant programs. For these reasons, I do not include women receiving SSI in my sample.
The average Medicaid threshold (with each state weighted equally) is $948, and the median annual increase (for those state-years with an increase) is $45, or about 5%. Fewer than 25% of all annual changes are more than $75; fully 25% of all changes are less than $30, which in many cases reflects a threshold tied to the rate of inflation or poverty line. There are a large number of changes overall: across 561 state-year changes (51 * 11 year-changes), there are 226 increases in the threshold and only 9 decreases. While it is easiest to discuss these changes in annual terms, note that my analysis takes advantage of these changes at the month-to-month level.
The SIPP data do not separately identify the state for those living in ME, VT, ND, SD, and WY, but instead group them together. Since Medicaid thresholds vary by state, these states are dropped from the SIPP sample. I drop OR and MN from both data sets due to their implementation of state health insurance programs in the early 1990s that complicate the role of Medicaid.
Seam bias is the estimation problem that results from individuals commonly reporting their information in four-month blocks, such that transitions (ex. between jobs) appear to happen disproportionately on the “seam” between surveys. While nothing in this study examines or relies directly on spell lengths, the use of just the most recent month from each interview is a common method of dealing with this measurement problem.
For clarity in terms of comparing earnings to specific Medicaid thresholds each month, I do not adjust earnings or thresholds for inflation. Such an adjustment would just scale both measures and presumably have little effect on my estimates.
Identifying this bunching may be difficult since the shape of the underlying distribution near the threshold may differ across states; for instance, states with a very low threshold are likely on a more upward-sloping portion of the earnings distribution in the region of the threshold, so that the bunching may be less apparent than in a flatter distribution. However, an examination of earnings distributions and thresholds revealed that the vast majority of thresholds are on the upward-sloping part of the earnings distribution, so this issue should not create any serious problems. Additional kernel densities with subgroups of the data (defined by location of threshold relative to median of the earnings distribution) are available upon request.
In an earlier iteration of this work, I performed the same analysis using data from the Current Population Survey Merged Outgoing Rotation Groups, in which sample sizes are larger. Using a sample that extended from 1999 to 2003, I found that the histogram was similarly inconclusive. I also performed the basic regression analysis, using state fixed effects (as the data set is not longitudinal in the same way) and found a small but statistically significant effect of the Medicaid threshold on earnings. Because the SIPP analysis handles unobserved heterogeneity much better, I do not report the other results here, but they are available upon request.
Mothers without earnings are not included, since the focus here is on the intensive margin; results for the extensive margin are presented later in the paper. Box-Cox estimates support the use of the logged dependent variable with levels on the right-hand side (resulting in an estimate of a partial elasticity).
It also does not fully incorporate the incentives of higher earning poor workers, who may reduce earnings upon an increase in the Medicaid threshold in order to become eligible (once it requires only a small reduction to do so). If there are many of these workers, it will (appropriately) decrease the estimated value of γ. This estimate is thus a “net effect” and not simply the effect on workers below the threshold. The earnings growth model estimated later in the paper addresses this issue.
The use of a sample of working women brings to mind the issue of selection into work; this issue is examined in Section 5.
Note that other anti-poverty programs such as Food Stamps and the smaller WIC and National School Lunch programs all follow national eligibility rules tied to the Federal Poverty Guidelines, so my cross-state identification strategy will not inadvertently pick up effects of these programs.
When characterizing thresholds for children’s coverage, I use the highest threshold that applies to them whether it is officially funded by SCHIP or children’s Medicaid. This is usually SCHIP, and so I refer to the children’s threshold as the “SCHIP threshold” even though in a few states it reflects the Medicaid program itself rather than an SCHIP-funded program.
I provide one additional examination of the relationship between Medicaid thresholds and earnings levels by exploring the hypothesis that education level may affect the salience of Medicaid thresholds to labor market decisions. In particular, one might expect that more- educated women have more access to other forms of health insurance or care, while less-educated women may be more sensitive to the parameters of public assistance. If I allow a different effect of the Medicaid threshold on women with and without a diploma, I find no net effect for those with a diploma and a very small but statistically significant positive effect for those without a diploma. The coefficient suggests that these women’s earnings rise by 0.5% for a change of $100 in the Medicaid threshold. While interesting, this effect remains very small (about $10/month at the mean) and thus I do not further explore it here.
This is the case for the majority of states post-1997, but only reaches about 60% even by 2003. Alternative versions of the figure that illustrate the opposite case are available from the author upon request. Both figures illustrate a notched budget constraint that predicts bunching at the threshold.
The argument for a lack of substitution effect (which would generally predict this behavior to be theoretically impossible) breaks down here because both the initial and resulting choices of hours are at corner solutions.
One might argue that these workers do not “choose” their earnings, in the sense that their wage is not their choice. However, in the range of workers with positive earnings meaningfully below the Medicaid threshold, low earnings are at least partly driven by few hours worked. (Full-time work would put people near or above the Medicaid threshold in most states, even at low wages.)
This section implicitly assumes that wages are not changing over the period of the adjustment; if they are, then the change in hours will be an underestimate of the effects of the change in the threshold.
I include changes in the number of children (reflecting the addition of new children or aging-out of older children), and changes in education. Beyond these, I also include race (allowing earnings growth over time to differ by race), urban status (allowing different earnings growth patterns in urban settings), age, and education level (accounting for the stylized fact that more highly educated people tend to have faster earnings growth). The selection of these variables reflects insights from several papers, including Berger (1984), Brunello and Comi (2004), and Wheeler (2006).
The first stage for each of the lagged instruments generates an F-test value of at least 100 (depending on the specification), well above the thresholds for weak instruments indicated in Stock and Yogo (2005).
It is impossible to be more than $1,000 below the threshold in many states, as the thresholds themselves are less than $1,000 per month (see Table 1). To provide more context for the shape of the marginal effects, we can look at the effect of increasing the Medicaid threshold by $100 on earnings growth at additional points. I find that it varies from over 30% (for those within a couple hundred dollars of the threshold) to closer to 5% for those in the realm of $800 below the threshold. The interested reader can insert any threshold levels and changes in the formulas given in the text to obtain specific predictions at different points in the distribution.
Looking across the whole distribution of distance from the Medicaid threshold, a change of $100 in the Medicaid threshold never increases predicted hours worked by more than an hour a month.