The Economic Impact of the Affordable Care Act on Arkansas

by Carter C. Price, Evan Saltzman

This Article

RAND Health Quarterly, 2013; 3(1):1

Abstract

The Affordable Care Act (ACA) will increase coverage through the expansion of Medicaid and the creation of a Health Insurance Exchange with subsidies. RAND researchers analyzed the ACA’s economic impact on the state of Arkansas and found that by 2016, about 400,000 people will be newly insured, net federal payments to the state will amount to $430 million annually, and the total gross domestic product will see a net increase of $550 million.

For more information, see RAND RR-157-ACHI at https://www.rand.org/pubs/research_reports/RR157.html

Full Text

Introduction

The Affordable Care Act (ACA) includes two provisions that will push billions of dollars into state economies: the expansion of Medicaid to cover the poorest segment of the population (those under 138 percent of the Federal Poverty Level) and subsidies for low- and medium-income people to buy health insurance (those between 100 percent and 400 percent of the Federal Poverty Level). As one of the poorest states in the union, Arkansas will disproportionately benefit from the expansion of Medicaid and the Exchange subsidies.

These provisions are paid for by cuts to Medicare spending and new taxes. Additionally, while the federal government pays for all of the Medicaid expansion for the first few years, by 2020, the state will be responsible for 10 percent of the costs for those newly eligible for Medicaid. People who are currently Medicaid eligible but not enrolled may choose to enroll because of the penalties for being uninsured. The costs for these people will be borne at the current Federal Medical Assistance Percentage (FMAP) instead of the more generous rate under the ACA. For Arkansas, nearly 30 percent of these costs would be borne by the state. Thus, Arkansas will also see increases in the required spending, particularly in the later years.

The net effect of this shift in spending will vary substantially among states because the distribution of this new spending will not be evenly distributed across the states. This article contains an assessment of the economic impact of the ACA on Arkansas using the RAND COMPARE model. This assessment focuses on those impacts that can be quantified using available data and literature.

Methods

We used the RAND COMPARE microsimulation to model the coverage and spending impact of the ACA on Arkansas. We then applied the Regional Input-Output Modeling System (RIMS II) multipliers from the Bureau of Economic Analysis (BEA) to determine the broader economic effects of the changes in spending. We also calculated the mortality and tax implications of the ACA using relevant studies from the academic literature. All of these measures have been scaled to the county level to provide additional visibility to the local effects of the ACA.

Assumptions

All of the analysis in this document is based in the year 2016. We expect 2016 to be close to the “steady state” for the effects of the ACA because many of the law's major components do not begin until 2014 and it will take a couple years for the population to fully react. The model assumes Arkansas expands Medicaid availability to everyone up to 138 percent of the Federal Poverty Level (except for recent and undocumented immigrants). In the status quo, not everyone enrolls in Medicaid either because they lack eligibility information or because there is stigma attached to doing so. We assume this stigma is also held by the newly eligible, although the Medicaid take up rate will not be the same for the newly enrolled because of changes in the law. We also assume the insurance Exchange has split the risk pools for individuals and firms. The model includes the individual mandate and assumes perfect enforcement, meaning that everyone subject to the penalty will pay. The employer mandate is included as well, but does not include some of the possible behavioral reactions to the employer mandate. For example, employers may reduce their employees' hours to avoid both offering coverage and the employer mandate penalties.

The economic analysis includes money flows from the individual subsidies, but not the employer subsidies because these last for only two years and are substantially smaller than the individual subsidies. We include Medicaid costs for both the expansion and for the so-called “woodwork” effect (people currently eligible for Medicaid but not enrolled). The costs for the woodwork effect will be borne using the current FMAP, while the newly eligible will be fully covered until after 2016. We also consider the flows out of the state included in the Congressional Budget Office's (CBO) analysis (CBO, 2012). The COMPARE model is used to estimate the effects of the individual mandate, the employer mandate, the private health insurance tax in section 9010, and the costs associated with reinsurance. We calculate the size of the Medicare cuts by scaling CBO's estimates for size of the cuts by Arkansas's share of national Medicare spending. All other taxes, fees, and other flows to the federal government in the ACA that were included in CBO's estimates are distributed proportionately based on Arkansas's and each county's share of the population. The economic analysis does not include the states' administrative costs associated with the Exchange or the expansion of Medicaid. This may not have a large net effect on the economy because the money will be primarily collected and spent in the state, but this could be very dependent on the implementation.

COMPARE Model

COMPARE models the decisions made by various economic actors—including individuals, families, and firms—using a utility maximization framework. Premiums are derived from aggregate choices of the population and decisions can be changed in light of the new premium information. Equilibrium is reached when none of the economic actors want to change their decisions. The model is calibrated to accurately reproduce the decisions made in the current system. We present a basic overview of the COMPARE model, but a more thorough description of the microsimulation can be found in Eibner et al. (2010).

COMPARE uses a synthetic population constructed from several data sources, along with synthetic families and firms, as the basis for the model. Records from the 2008 Survey of Income and Program Participation (SIPP) are used to assign demographic characteristics to the synthetic population (U.S. Census Bureau, 2012). To get medical spending values for each synthetic person, the SIPP records are matched to records in the Medical Expenditure Panel Survey (MEPS) Household Component from 2002 and 2003 based on demographic profiles (Agency for Healthcare Research and Quality, 2010). The synthetic people form synthetic families based on the family relation indicators in the SIPP. Each person and family has an income, family poverty level, and health care costs. We construct a utility for each individual based on the premium, out-of-pocket expenditures, out-of-pocket spending risk, and a general utility of health care. Families will make decisions to maximize the aggregate utility of the family.

In the COMPARE model, workers are assigned to synthetic firms. These synthetic firms are created from the 2010 Kaiser/Health Research and Educational Trust (HRET) and people are matched to firms based on the firm size, sector, and region (Kaiser/HRET, 2010). Firms choose to offer health insurance to their employees based on the aggregated utilities of their employees, the employer penalty, premium subsidies, and the tax advantage of employer-sponsored insurance. Factors such as the employers' contribution to their employees' health insurance are based on the distribution from the Kaiser/HRET survey.

Plan enrollments are determined from the decisions of individuals and families in light of premiums and their insurance options. Premiums are created endogenously (meaning that they are derived from within the model) through the plan enrollment. Government spending values are calculated based on individuals' choices, premiums, and—for Medicaid—individuals' health spending.

Unfortunately, the synthetic population created from the SIPP, Kaiser/HRET, and MEPS datasets is too small to be representative of state-level populations, particularly for a relatively small state such as Arkansas. Consequently, selecting the records in the synthetic population who reside in Arkansas might result in a misleading analysis. As an alternative approach, COMPARE reweights the records in the SIPP to reflect the specific demographics of Arkansas, including age, gender, race, income, insurance type (if insured), and employer's firm size (if employed). COMPARE employs the Iterative Proportional Fitting (IPF) procedure to ensure that each demographic category has the right number of individuals; for instance, if the benchmark state-level data suggest that there are 50,000 females between the ages of 18 and 24 in Arkansas, the IPF procedure reweights female records between the ages of 18 and 24 in the synthetic population such that the sum of their weights equals 50,000. Benchmark data for Arkansas were obtained directly from state publications where available (Arkansas Department of Human Services, 2011); data from the Integrated Public Use Microdata Series, Statistics of U.S. Businesses, and the Kaiser Family Foundation were employed to supplement data obtained directly from the state. In addition to reweighting records in the synthetic population, COMPARE also adjusts the health expenditures of individuals to reflect health spending in Arkansas.

After running COMPARE for the entire state of Arkansas to assess the impact of the ACA in the state, we reweighted records once again to obtain county-level results. Records were reweighted based on joint age and income data from the Five-Year 2006–2010 American Community Survey. From the reweighted population, we derived estimates at the county-level for insurance type enrollment and expected additional government spending for newly insured individuals.

Economic Impact

The change in government spending includes the money entering the state for the Medicaid coverage and subsidies, and the money leaving the state for the various taxes, fees, and Medicare spending cuts. The Medicaid and subsidy money coming from the federal government is an output calculated by the COMPARE model. The money leaving the state for the individual mandate, employer mandate, health insurance premium taxes, and reinsurance costs are also calculated using the COMPARE model. The remaining changes in spending are calculated outside of the model. The Medicare cuts represent money no longer being sent to the state and are calculated by taking CBO's estimate for the 2016 Medicare cuts and scaling them based on Arkansas's share of Medicare spending. Similarly, other taxes, fees, etc., are calculated by taking Arkansas's per capita proportion of CBO's reported values. These cuts include the Medicare and Medicaid Disproportionate Share Hospital (DSH) payments made to hospitals that have a higher than average share of low-income patients. This decrease shifts the cost for uncompensated care from the federal government onto states, localities, and hospitals.

An inflow of federal funds under the ACA provides income that will be spent in the broader economy. Thus, for every dollar spent by the government, there is a multiplier effect that will have a broader impact on the state Gross Domestic Product (GDP). The BEA has compiled estimates of these multipliers for a variety of different industries including health care. The data also include estimates for the employment impact of this additional spending. We apply Arkansas's RIMS II multipliers for ambulatory health care services and hospitals to the COMPARE model's output to determine the implications of the increased spending on total gross output and employment.

Other Measures

In addition to the changes in Arkansas's gross output and employment, we calculated the expected decrease in mortality from the Medicaid expansion and the reduction in uncompensated care costs due to the increase in enrollment.

The expansion of Medicaid has the potential to provide substantial benefits to the population under 138 percent of the poverty level. This segment of the population is disproportionately uninsured and therefore has limited access to health care services. A recent study by Sommers, Baicker, and Epstein (2012) indicated that expansion of Medicaid would lead to a substantial decrease in mortality. They found that states that had expanded Medicaid since 2000 saw a decrease in adjusted mortality of 19.6 per 100,000 for the entire population. They also isolated the reduction for the newly covered population and estimated a decrease in mortality of 2,840 per 500,000 per year (or 568 per 100,000). These values will vary substantially depending on the underlying health of the newly insured population. There are other effects of Medicaid expansions that we did not quantify because of ambiguity in the literature or insufficient data. Gruber (2003) summarizes much of the literature on these effects and notes that beyond mortality, the effects of insurance expansions on outcomes are variable and inconsistent.

In addition to benefiting the new enrollees, providing insurance to this segment of the population should reduce the burden of uncompensated care. Hadley et al. (2008) found that uncompensated care from the uninsured population cost $56 billion in 2008 (which would amount to nearly $80 billion in 2016 with medical inflation). They also found that state and local governments cover 30 percent of this total. Thus, the states are bearing substantial costs to treat the uninsured despite the fact that they are not formally enrolled on a public insurance program. In 2010, Arkansas had nearly $340 million in uncompensated care (Arkansas Hospital Association, 2012). To assess Arkansas's reduction in uncompensated care costs, we determined the reduction in spending by the uninsured population scaled to Arkansas's total uncompensated care costs and calculated the state's share of these costs.

Results

A summary of the state-level enrollment results can be seen in Table 1, and the state-level economic change is listed below. In addition to the statewide analysis, we produced county-level estimates for each of Arkansas's counties.

Table 1

Comparison of Nonelderly Enrollment

Measure

2016 Baseline

2016 ACA

Enrolled in Medicaid (<65)

675,000

865,000

Enrolled in Nongroup Exchange (<65)

NA

288,000

Total Enrollment (<65)

1,894,000

2,296,000

Uninsured (<65)

571,000

170,000

Table 2

Economic Changes

Measure

2016 ACA

Increase in Medicaid Spending

$750,000,000

Increase in Individual Exchange Subsidies

$850,000,000

Impact on Taxes, Medicare cuts, etc.

($1,180,000,000)

Net Change

$430,000,000

Economic Impact of Net Change

$550,000,000

Jobs Created by Spending Change

6,200

Increase in State and Local Tax Revenue

$19,000,000

Decrease in State Spending on Uncompensated Care

($67,000,000)

Increase in State Spending due to “Woodwork” Effect

$3,000,000

All of the numbers presented in this section are from 2016 results unless otherwise specified. The COMPARE model estimates that the number of nonelderly people with health insurance will be nearly 2.3 million with the ACA, compared to 1.9 million without. The model estimates that 190,000 of the additional 400,000 people with insurance will be on Medicaid and the remaining growth in coverage will be in the nongroup market through the Exchange. There is little net change in the coverage through employer-sponsored insurance.

As discussed above, the analysis of Sommers et al. can be applied to estimate the decrease in mortality caused by an increase in Medicaid enrollment. With an additional 190,000 people on Medicaid in Arkansas with the ACA, Arkansas might expect as many as 1,100 fewer deaths per year. There has not been similar analysis to assess the impact of the increase in nongroup coverage, but it may have a comparable effect, in which case the increase in coverage on the Exchange would result in as many as 1,200 fewer deaths per year. Thus, the state may see a reduction in annual deaths as high as 2,300 due to the expansion in insurance coverage. These estimates assume the newly insured in Arkansas will be similar to the newly covered population studied by Sommers et al. and the actual reduction in mortality will depend on the cohort's relative health.

The federal government provides most of the funding required to expand coverage to these 400,000 additional people through additional Medicaid spending and subsidies for the Exchange, but the state will lose some federal money through cuts in Medicare reimbursement, a reduction in DSH payments, and additional taxes on insurance plans. The net effect of these changes will be an increase of roughly $430 million. This total comes from $750 million from federal Medicaid spending plus $850 million in federal subsidies for the Exchange, minus $1,180 million due to the various cuts (these numbers may not add to totals because of rounding). Because of the multiplier effect of this net increase in federal spending, the total impact on the state's GDP will be a gain of around $550 million. Additionally, employment would rise by about 6,200. In the county-level analysis, nearly every county will see increased federal spending and economic activity as a result of the ACA, which would raise total state and local tax revenue by about $19 million. Based on the increased enrollment and spending distributions for Arkansas, the state could see a decrease of about $67 million dollars in uncompensated spending costs in 2016. The economic findings are consistent with a 2003 study by Miller and Pickett that assessed the economic impact of a hypothetical increase in Medicaid spending in Arkansas.

Beyond 2016, Arkansas's share of Medicaid spending will gradually increase to 10 percent of the costs for the newly eligible. If the state maintains a balanced budget, $75 million annually will need to come from increased taxes and fees, decreasing the economic impact proportionately. If the state does not expand Medicaid, the direct net loss in payments would be $670 million dollars for the state in the out years. Thus, while the expansion of Medicaid is not without costs in later years, there is a net positive impact for the state's economy.

Conclusions

CBO estimates that the ACA results in an increase in net revenue to the federal government, but because Arkansas is a relatively poor state, there will be a net flow of money into the state. Specifically, the increased federal spending on subsidies and Medicaid for the poor will outweigh the decreased Medicare spending and increases in taxes and fees. Only relatively poor states such as West Virginia and Mississippi can expect to see a favorable balance of transfer payments and an economic impact similar to that in Arkansas. Most other states will likely see an increase in the transfer payments to the federal government. For these states, additional analysis will be required to assess if the net payments will lead to a negative economic impact or if health improvements negate the spending declines.

While this analysis shows that the ACA will provide a net economic benefit to Arkansas because of the net increase in federal spending in the state, these results are forecasts for 2016 and contain a degree of uncertainty. Should the Arkansas legislature not expand Medicaid to everyone under 138 percent of the federal poverty level, the number of people with coverage will be substantially smaller and the federal funds being directed to the state will also be lower, making the economic impact proportionately lower and potentially negative.

Additionally, these forecasts are based on current economic conditions and behaviors. If the state promotes Medicaid and Exchange enrollment through robust outreach beyond current efforts, the take-up rates could be higher than estimated by the COMPARE model and the economic impact could be greater. This is a particularly salient point because surveys of employers and individuals have found that many people do not understand how the ACA affects them. For example, a 2012 national survey of employers by Deloitte found that a majority of employers understood portions of the law, such as the individual penalty, but not other parts, such as the Exchanges (Deloitte Center for Health Solutions & Deloitte Consulting, 2012). A similar 2012 survey of individuals by researchers from Stanford found very little understanding of the components of the law (Gross et al., undated). If fewer individuals understand the options available to them than anticipated, the take-up rates and hence the economic impact would likely be lower than our estimates. On the other hand, if efforts to promote an understanding of the ACA's pertinent components to target populations are more successful, enrollments could be higher—increasing the federal subsidies, Medicaid payments, and associated economic growth.

References

Agency for Healthcare Research and Quality, Medical Expenditure Panel Survey, page last revised April 23, 2010. As of December 21, 2012:
http://meps.ahrq.gov/mepsweb/data_stats/download_data_files.jsp

Arkansas Department of Human Services, Arkansas Medicaid Program Overview, Little Rock, Ark.: 2011. Available as of December 21, 2012:
http://humanservices.arkansas.gov/reportDocuments/Medicaid%20Program%20Overview.pdf

Arkansas Hospital Association, “Uncompensated Hospital Care Cost Fact Sheet,” Arkansas Hospitals, Summer 2012.

CBO—see Congressional Budget Office.

Congressional Budget Office, letter to House Speaker John Boehner providing an estimate for H.R. 6079, repeal of Obamacare, July 24, 2012. As of December 21, 2012:
http://www.cbo.gov/publication/43471

Deloitte Center for Health Solutions & Deloitte Consulting, 2012 Deloitte Survey of U.S. Employers: Opinions about the U.S. Health Care System and Plans for Employee Health Benefits, June 2012. As of December 21, 2012:
http://www.deloitte.com/assets/Dcom-UnitedStates/Local%20Assets/Documents/us_dchs_employee_survey_072512.pdf

Eibner, Christine, Federico Girosi, Carter Price, Amado Cordova, Peter Hussey, Alice Beckman, and Elizabeth McGlynn, Establishing State Health Insurance Exchanges: Implications for Health Insurance Enrollment, Spending, and Small Businesses, Santa Monica, Calif.: RAND Corporation, TR-825-DOL, 2010. As of December 20, 2012:
http://www.rand.org/pubs/technical_reports/TR825.html

Gross, Wendy, Tobias Stark, Jon Krosnick, Josh Pasek, Gaurav Sood, Trevor Tompson, Jennifer Agiesta, and Dennis Junius, Americans' Attitudes Toward the Affordable Care Act: Would Better Public Understanding Increase or Decrease Favorability? Robert Wood Johnson Foundation, GfK, Stanford University, and the Associated Press, undated. As of December 21, 2012:
http://comm.stanford.edu/faculty/krosnick/docs/2012/Health%20Care%202012%20-%20Knowledge%20and%20Favorability.pdf

Gruber, Jonathan, “Medicaid,” Means-Tested Transfer Programs in the United States, University of Chicago Press, 2003, pp. 15–78. As of December 21, 2012:
http://www.nber.org/chapters/c10254.pdf

Hadley, Jack, John Holahan, Teresa Coughlin, and Dawn Miller, “Covering the Uninsured in 2008: Current Costs Sources of Payment, and Incremental Costs,” Health Affairs, August 25, 2008. As of December 20, 2012:
http://content.healthaffairs.org/content/27/5/w399.full

Kaiser Family Foundation and Health Research and Educational Trust, Employer Health Benefits: 2010 Annual Survey, 2010. As of December 20, 2012:
http://ehbs.kff.org/2010.html

Miller, Wayne, and John Pickett, Economic and Fiscal Impact of Additional $100 Million in State Funding for Medicaid Programs, University of Arkansas, Division of Agriculture, March 24, 2003. As of December 21, 2012:
http://www.arcommunities.org/community/economic/economicimpact/medicaid.htm

Sommers, Benjamin, Katherine Baicker, and Arnold Epstein, “Mortality and Access to Care Among Adults After State Medicaid Expansions,” New England Journal of Medicine, Vol. 367, September 13, 2012. As of December 21, 2012:
http://www.nejm.org/doi/full/10.1056/NEJMsa1202099#t=article

U.S. Census Bureau, Survey of Income and Program Participation, webpage last revised June 6, 2012. As of December 21, 2012:
http://www.census.gov/sipp/

The research described in this report was sponsored by the Arkansas Center for Health Improvement and was conducted within RAND Health, a division of the RAND Corporation.

RAND Health Quarterly is produced by the RAND Corporation. ISSN 2162-8254.