The impact of Medicaid expansion on state expenditures through the COVID‐19 era

Abstract

Objective

To investigate the impact of Medicaid expansion on state expenditures through the end of 2022.

Data Sources

We used data from the National Association of State Budget Officers (NASBO)'s State Expenditure Report, Kaiser Family Foundation (KFF)'s Medicaid expansion tracker, US Bureau of Labor Statistics data (BLS), US Bureau of Economic Analysis data (BEA), and Pandemic Response Accountability Committee Oversight (PRAC).

Study Design

We investigated spending per capita (by state population) across seven budget categories, including Medicaid spending, and four spending sources. We performed a difference‐in‐differences (DiD) analysis that compared within‐state changes in spending over time in expansion and nonexpansion states to estimate the effect of Medicaid expansion on state budgets. We adjusted for annual state unemployment rate, annual state per capita personal income, and state spending of Coronavirus Relief Funds (CRF) from 2020 to 2022 and included state and year fixed effects.

Data Collection/Extraction Methods

We linked annual state‐level data on state‐reported fiscal year expenditures from NASBO with state‐level characteristics from BLS and BEA data and with CRF state spending from PRAC.

Principal Findings

Medicaid expansion was associated with an average increase of 21% (95% confidence interval [CI]: 16%–25%) in per capita Medicaid spending after Medicaid expansion among states that expanded prior to 2020. After inclusion of an interaction term to separate between the coronavirus disease (COVID) era (2020–2022) and the prior period following expansion (2015–2019), we found that although Medicaid expansion led to an average increase of 33% (95% CI: 21%–45%) in federal funding of state expenditures in the post‐COVID years, it was not significantly associated with increased state spending.

Conclusions

There was no evidence of crowding out of other state expenditure categories or a substantial impact on total state spending, even in the COVID‐19 era. Increased federal expenditures may have shielded states from substantial budgetary impacts.

What is known on this topic

• COVID‐19 policies led to increased Medicaid enrollment across states, with Medicaid enrollment rising approximately 29% in expansion states from 2020 to 2022.

• Both the Affordable Care Act (ACA) rollout and COVID‐19 led to changes in the Federal Medical Assistance Percentage for adults covered under Medicaid.

• Through 2019, there was no evidence that state spending due to Medicaid expansion crowded out other state expenditure categories, and some studies found evidence of net savings from Medicaid expansion.

What this study adds

• The impact of Medicaid expansion on Medicaid spending was substantially greater in the post‐COVID era (2020–2022) compared with the pre‐COVID era following Medicaid expansion (2015–2019).

• Even with this increased spending in expansion states, there was no evidence of crowding out of other state expenditure categories.

• Increased federal expenditures may have shielded states from substantial budgetary impacts.

1. INTRODUCTION

Under the Affordable Care Act (ACA), states began expanding Medicaid to cover adults with income up to 138% of the federal poverty level. ¹ Overall, 27 states (including the District of Columbia) expanded coverage in 2014 or had previously expanded to some extent before this. ¹ Numerous states also expanded in the following years and have continued to do so, with North Carolina most recently announcing its planned Medicaid expansion in the fall of 2023. ¹ , ²

Opponents of Medicaid expansion have argued that it could increase state spending on Medicaid, which could crowd out other areas of state spending or lead to greater overall spending. ³ However, several studies have shown net savings as a result of Medicaid expansion in numerous states and national estimates find no evidence of increased state spending due to Medicaid expansion. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸

There are multiple potential reasons that could explain why Medicaid expansion has not been linked to increased state spending. First, under the ACA, states receive a higher Federal Medical Assistance Percentage (FMAP) for adults covered through the expansion compared with nonexpansion Medicaid beneficiaries, meaning that states pay a lower percentage of costs for expansion enrollees compared with traditional Medicaid enrollees. In addition, expansion states are able to use this higher FMAP to cover some expansion enrollees that may have originally been covered under traditional Medicaid where states would be paying a higher percentage of their fees, such as women who enrolled as part of the expansion and then became pregnant. ⁴ Finally, it is possible that states saw savings in other areas of state expenditures that could now be covered through Medicaid, such as corrections (via inmates' health care), state‐funded mental health and substance use services, and health care for the uninsured. ⁴ , ⁵ , ⁶ For instance, some states used federal Medicaid funds to cover behavioral health programs rather than state general funds, and thus saw savings in their state general funds. ⁵

However, there have been numerous changes since 2019, when the last national study was conducted, ⁴ that could change the relationship between Medicaid expansion and state expenditures. First, when Medicaid expansion began in 2014, the federal government paid 100% of the costs for the expansion population through the FMAP. This changed in 2017, when states were required to begin paying 5%, and incrementally rose to the full 10% in 2020 that states pay currently for the expansion population. ⁵ In addition, the Families First Coronavirus Response Act (FFCRA) temporarily prevented disenrollment in Medicaid in conjunction with the public health emergency, thus resulting in rising numbers of Medicaid enrollees across states. ⁹ , ¹⁰ From February 2020 to December 2022, Medicaid enrollment rose approximately 29% in expansion states. ⁹ Moreover, the FFCRA also resulted in a temporary 6.2 percentage point increase in the FMAP for traditional Medicaid, ¹¹ and the American Rescue Plan Act (ARPA) allowed for states that newly expanded Medicaid to receive an additional 5 percentage point increase in the FMAP for traditional Medicaid beyond this as well. ¹² In addition, the Coronavirus Aid, Relief and Economic Security (CARES) Act allocated $150 billion in Coronavirus Relief Funds (CRF) to state and local governments, ¹³ which could have affected state spending. There is a need for current estimates of the impact of Medicaid expansion on state budgets given recent substantial policy changes.

In this study, we investigate the impact of Medicaid expansion on state expenditures through the end of 2022. We compare spending in Medicaid expansion and nonexpansion states across several categories of state and federal funding. We then conduct a difference‐in‐differences analysis to assess the impact of Medicaid expansion on Medicaid and other areas of funding and further assess if the relationship changed in the COVID‐19 era.

2. METHODS

2.1. Data Sources

Our primary data for this study were from NASBO's State Expenditure Report, which measures annual state expenditures. NASBO is a “professional membership organization for state budget and finance officers” that collects data from states. ¹⁴ Its state expenditure report consists of state‐reported fiscal year expenditures that are broken down both into categories of spending (elementary and secondary education, higher education, Temporary Assistance for Needy Families (TANF), other cash assistance, Medicaid, corrections, transportation, all other), and source of spending (state general funds, federal funds, other state funds, bonds). ¹⁵ We aggregated the categories of TANF and other cash assistance together into public assistance and aggregated state general funds and other state funds into overall state funds, as in previous research. ³ Our analytical sample included expenditures from 2009 to 2022 across all 50 states. We obtained month and year of Medicaid expansion from KFF's Medicaid expansion tracker. ¹

We then merged these data to two additional data sources to obtain state‐level characteristics for each year. We included the state unemployment rate from BLS ¹⁶ and state per capita personal income from BEA ¹⁷ for 2009–2022. Then, to account for additional state expenditures during the COVID‐19 era, we merged state spending of CRF from March 2020 to December 31, 2022, from PRAC data. ¹³ In sensitivity analyses, we also merged in state Medicaid enrollment data from KFF. ¹⁸ , ¹⁹

2.2. Key variables

The first outcome we examined was Medicaid spending per capita (state Medicaid spending/state population). We also assessed all other per capita state budget spending across seven state spending categories: Medicaid, elementary and secondary (K‐12) education, higher education, public assistance, corrections, transportation, and all other spending, and across spending sources: federal, state funds (state general funds + other state funds), and bonds. The key independent variable we examined was Medicaid expansion status, a binary indicator for each year following a state's ACA Medicaid expansion. We adjusted for state‐level factors by year of analysis including state unemployment rate, personal income per capita, and federal allocations for CRF.

2.3. Statistical analysis

First, we descriptively compared mean expenditures across the seven state spending categories and four spending sources in expansion states and nonexpansion states. We did this for 2022 and for all years pre‐Medicaid expansion (2009–2013), post‐Medicaid expansion (2015–2019), and in the COVID‐19 era (2020–2022) across expansion and nonexpansion states (see Table S1 for mean expenditures across all time periods).

Second, we compared trends in Medicaid spending from 2009 to 2022 among states that expanded Medicaid in 2014 compared with nonexpansion states. This analysis did not include states that expanded Medicaid after 2014 to graphically isolate a single expansion year. We also depicted trends for all other spending categories in Figure S1.

We then performed a difference‐in‐differences analysis that compared within‐state changes in spending over time in expansion and nonexpansion states to estimate the effect of Medicaid expansion on state budgets. We adjusted for annual state unemployment rate, annual state per capita personal income, and state spending of CRF from 2020 to 2022, as well as state and year fixed effects. Please see Appendix 1 of the Supplemental Materials for the main regression equation. Given that Medicaid expansion began in 2014 and data from 2009 to 2022 were analyzed, the pre‐period prior to Medicaid expansion was a minimum of 5 years and the postperiod was a maximum of 8 years. See Table S2 for log‐transformed regression coefficients and robust standard errors, clustered at the state level, from these analyses. While our primary analyses assess per capita spending with log transformations to reflect percent changes in per capita spending (Table S2), we also include per capita dollar regressions (real dollars as of 2022) in Table S3.

In addition, we evaluated whether there was a differential effect of Medicaid expansion on spending in the COVID era when Medicaid enrollment numbers rose dramatically. To do this, we estimated difference‐in‐differences regressions with an interaction term of the treatment (Medicaid expansion) and an indicator for the COVID‐era years (2020–2022). A full table of regression coefficients and standard errors are shown in Table S4 with log‐transformation and in Table S5 with per capita real dollars as of 2022. In Table S4, which reflects our primary DiD model with the interaction term, we performed a sensitivity analysis where we adjusted for annually reported December Medicaid enrollment from 2009 to 2022, obtained from KFF. ¹⁸ , ¹⁹

In our difference‐in‐differences analyses, both with and without the interaction term, we excluded five states that expanded Medicaid from 2020 to 2022 (Nebraska, Idaho, Utah, Oklahoma, and Missouri) to isolate COVID‐era differences within states that expanded Medicaid prior to the 2020. By removing these states, we thus did not have to account for Oklahoma's and Missouri's increased FMAP rates for their traditional Medicaid population due to ARPA. We also conducted a secondary analysis of the primary difference‐in‐differences analysis (without the COVID‐era interaction) including all states, which is also shown in Tables S2 and S3.

The underlying assumption of the difference‐in‐differences analyses is that Medicaid expansion states would have followed similar trends in Medicaid spending (and other spending) as nonexpansion states had Medicaid expansion not occurred. Although this assumption cannot be directly tested, we examined if there were differences in trends prior to Medicaid expansion between expansion and nonexpansion states that might be indicative of a violation of this assumption with event study analyses. In the event study analyses, we compared differences over time in Medicaid expansion and nonexpansion states each year relative to the implementation of the expansion, adjusted for state per capita income, unemployment rate, and CRF funds (see Figure S2a–m for event study model results).

We also conducted a robustness check that employs the Callaway Sant'Anna (CS) DiD estimator to produce alternative estimates of the average treatment effect on the treated due to the staggered implementation of Medicaid expansion and heterogeneous treatment effects over time. ²⁰ , ²¹ See Tables S6 and S7 and Figure S3 for regression results and the event study plot, respectively.

3. RESULTS

In 2022, expenditures were higher in all categories of spending in expansion states compared with nonexpansion states, with the exception of higher education spending (Figure 1). Differences in spending per capita between expansion and nonexpansion states in 2022 were statistically significant (p‐value <0.05) within the spending categories of Medicaid, K‐12 education, and all other spending, and within spending sources of federal funds and state general funds. Average per capita Medicaid spending in Medicaid expansion states was $2676 compared with $1606 in nonexpansion states. Expansion states also spent on average $3897 per capita from federal funds compared with $2624 in nonexpansion states and spent on average $6174 per capita from state funds compared with $4331 in nonexpansion states. From 2020 to 2022, expansion states also had significantly higher total spending—spending on average $9482 compared with $6571 in nonexpansion states (shown in Table S1).

FIGURE 1.

Mean state expenditures per capita across Medicaid expansion and nonexpansion states, 2022. Mean state expenditures in each spending category and spending source were calculated among Medicaid expansion and nonexpansion states. The analysis excluded the five states that expanded Medicaid between 2020 and 2022 (Nebraska, Idaho, Utah, Oklahoma, and Missouri). *Differences between expansion and nonexpansion states that are statistically significant with a p‐value <0.05. Source: Authors' analysis of data from the National Association of State Budget Officers State Expenditure Report, 2009–2022.

Following Medicaid expansion in 2014, the gap in state Medicaid spending per capita grew between expansion and nonexpansion states, and this gap continued to widen over time (Figure 2). In 2013, before the expansion, the average per capita Medicaid expenditure in 2014 expansion states was $1384 (95% CI: $1232–$1535) compared with $1148 (95% CI: $1007–$1288) in nonexpansion states. After this, expansion states saw a much faster rise in Medicaid spending per capita, with an average expenditure of $2073 (95% CI: $1898–$2247) in 2019 and then $2400 (95% CI: $2276–$2523) in 2020–2022 compared to $1345 (95% CI: $1182–$1509) in 2019 and $1502 (95% CI: $1396–$1609) in 2020–2022 in nonexpansion states, respectively. For trends in other spending categories and sources comparing nonexpansion states and expansion states, see Figure S1.

FIGURE 2.

State annual Medicaid expenditures per capita per year, in 2014 Medicaid expansion states and nonexpansion states, 2009–2022. The mean state Medicaid spending from 2009 to 2022 among states that expanded Medicaid in 2014 compared with states that had not expanded Medicaid as of 2022. Source: Authors' analysis of data from the National Association of State Budget Officers State Expenditure Report, 2009–2022.

Medicaid expansion led to an average 21% (95% CI: 16%–25%) increase in per capita Medicaid spending in the up to 8 years after Medicaid expansion among states that expanded prior to 2020 (Figure 3). In addition, Medicaid expansion was associated with a 10% decline (95% CI: −17% to −4%) in state per capita higher education spending and a 16% increase (95% CI: 2% to 29%) in other spending. However, Medicaid expansion was not associated with any changes in funding for public assistance, K‐12 education, corrections, or transportation. Although Medicaid expansion led to a 25% average increase (95% CI: 17% to 33%) in federal funding of state expenditures, it was not significantly associated with an increase in state funds spending.

FIGURE 3.

Changes in spending per capita resulting from Medicaid expansion in states that expanded Medicaid pre‐2020, 2009–2022. The difference‐in‐differences coefficients reflect differences in within‐state changes in log per capita spending in expansion states compared with nonexpansion states from 2009 to 2022, adjusting for annual state unemployment rate, annual state per capita personal income, and state spending of Coronavirus Relief Funds from 2020 to 2022, as well as state and year fixed effects. The error bars reflect the 95% confidence intervals for the coefficients clustered at the state level. The analysis excluded the five states that expanded Medicaid between 2020 and 2022 (Nebraska, Idaho, Utah, Oklahoma, and Missouri). Public assistance spending and bonds spending both represented small percentages of state spending and were excluded from this figure due to wide confidence intervals and large numbers of 0 observations but can be seen in the Tables S2–S5. Source: Authors' analysis of data from the National Association of State Budget Officers State Expenditure Report, US Bureau of Labor Statistics data, and US Bureau of Economic Analysis data, 2009–2022 and Pandemic Response Accountability Committee Oversight data, 2020–2022.

In the post‐COVID years, Medicaid expansion was associated with a 27% increase (95% CI: 21%–33%) in per capita Medicaid spending (2020–2022) compared with 19% (95% CI: 15%–23%) in the pre‐COVID years (2015–2019) (Figure 4). Medicaid expansion led to a 23% (95% CI: 15%–32%) per capita increase in federal spending in the pre‐COVID years and was associated with 33% (95% CI: 21%–45%) per capita greater federal spending in the post‐COVID years. However, Medicaid expansion was not associated with differences in state funding or other sources of funding in the post‐COVID years. We explored increased enrollment as a potential mechanism in Table S4, finding that controlling for enrollment does not substantially modify the larger post‐COVID impact of Medicaid expansion on Medicaid spending.

FIGURE 4.

Changes in spending per capita resulting from Medicaid expansion in pre‐COVID years (2015–2019) and post‐COVID Years (2020–2022) in states that expanded Medicaid pre‐2020, 2009–2022. The difference‐in‐differences coefficients reflect differences in within‐state changes in log per capita spending in expansion compared with nonexpansion states pre‐COVID (2015–2019) and post‐COVID (2020–2022), adjusting for annual state unemployment rate, annual state per capita personal income, and state spending of Coronavirus Relief Funds from 2020 to 2022, as well as state and year fixed effects. The error bars reflect the 95% confidence intervals for the coefficients clustered at the state level. The analysis excluded the five states that expanded Medicaid between 2020 and 2022 (Nebraska, Idaho, Utah, Oklahoma, and Missouri). Public assistance spending and bonds spending both represented small percentages of state spending and were excluded from this figure due to wide confidence intervals and large numbers of 0 observations but can be seen in Tables S2–S5. Source: Authors' analysis of data from the National Association of State Budget Officers State Expenditure Report, US Bureau of Labor Statistics data, and US Bureau of Economic Analysis data, 2009–2022 and Pandemic Response Accountability Committee Oversight data, 2020–2022.

The event study specifications were next analyzed to assess differences in trends prior to Medicaid expansion. Figure S2a shows this specification first for Medicaid spending per capita and demonstrates parallel pre‐trends prior to expansion. Following the expansion, there was a clear jump in Medicaid expenditures per capita, with spending continuing to grow several years post‐expansion. Findings were broadly similar in magnitude using the CS estimator, although the CS estimator produced more precise estimates in some instances such that the Medicaid expansion coefficient was statistically significant (Tables S6 and S7 and Figure S3). For example, Medicaid expansion was associated with a 6% increase in state general funds in the adjusted CS model including expansion states that expanded prior to 2020 (95% CI: 3%–8%). This would represent an approximate 6% increase over the pre‐expansion mean of $2542 in expansion states (see Table S1 for pre‐expansion mean state expenditures).

4. DISCUSSION

This paper examined the relationship between state spending and Medicaid expansion in the pre‐ to post‐COVID eras. We found that while Medicaid expansion was associated with greater increases in per capita Medicaid and federal spending in the post‐COVID era, there was still no evidence of substantial increases in state general funds expenditures. Further, there was no evidence that Medicaid expansion crowded out state investment in other areas of state expenditures.

Our findings are consistent with the estimates from Sommers and Gruber's work that used data through 2015 as well as their National Bureau of Economic Research update through 2018 following changes to the FMAP after 2016. ³ , ⁷ We show that, despite large increases in Medicaid spending per capita in the COVID era, Medicaid expansion continues to not be associated with substantial increases in state funds expenditures. The evidence suggests that increased Medicaid spending was largely driven by increased federal spending. We estimated a nearly $1000 per capita significant difference in expenditures from federal sources between expansion states and nonexpansion states during the COVID era. In turn, we find no evidence that Medicaid expansion has crowded out spending in other state budget categories, such as education or transportation.

There are numerous reasons why there might have been higher spending in the COVID era. First, inflation could have contributed to higher Medicaid and federal expenditures, as the NASBO's State Expenditure Report does not account for inflation; however, since our primary models log‐transform expenditures and include year fixed effects, inflation was likely not the primary reason for increased spending during the COVID era. ²² In addition, the large growth in Medicaid enrollment during the COVID era given the continuous coverage clause may have contributed to this growth in spending. However, there were also large increases in enrollment in nonexpansion states. ⁹ We find that there was still a larger Medicaid expansion effect on Medicaid spending during the COVID era even after adjusting for Medicaid enrollment. In addition, we find that CRF funds explained a portion of spending differences between expansion and nonexpansion states in the COVID era, but that there were still significant increases in Medicaid spending when adjusting for this spending. It is possible that other pandemic‐era funding through federal legislation helped to contribute to a larger part of total increased spending, such as the Coronavirus Response and Relief Supplemental Appropriations Act, which allocated additional funds to states in multiple areas such as child care and transit and additional business loans. ²³ In addition, states have allocated $173 billion funds through ARPA as of December 2022 via two rounds of funding, ²⁴ which likely contributed to overall state spending across expansion and nonexpansion states.

Furthermore, differences in utilization could have resulted in increased Medicaid spending in expansion states compared with nonexpansion states during the COVID era. Recent literature suggests that Medicaid utilization may have differed between expansion states and nonexpansion states during the pandemic. Although nonexpansion states experienced slightly higher percentage growth in Medicaid enrollees from February 2020 to December 2022, ⁹ it is possible that these enrollees incurred fewer expenses due to lower utilization during the pandemic. For instance, a study in a non‐Medicaid expansion state found that those who enrolled in Medicaid during the pandemic had lower utilization than those who enrolled before the pandemic. ²⁵ Moreover, some enrollees may not have been aware that their coverage was automatically renewed during the public health emergency. ²⁶ Further research is needed to understand whether utilization differed between expansion and nonexpansion states during the COVID era.

Several factors may have contributed to why higher Medicaid expenditures in the COVID era did not drive substantially greater state funds expenditures. First, under the FFCRA, the federal match for traditional Medicaid temporarily increased during the COVID‐19 pandemic, thus providing more federal dollars to states for their nonexpansion Medicaid population. ¹¹ In addition, federal funding for pandemic relief through the CARES Act as well as supplemental legislation may have helped to provide additional federal funding for states. Moreover, it may be that the expansion population is less high need than the traditional Medicaid population and uses less health care, thus limiting the impact of the 10% of Medicaid spending contributed by states for the expansion population. This supports our findings that state general funds did increase, but results were nonsignificant. Furthermore, this federal match is still higher than that of the traditional Medicaid match and thus puts less of a burden on states. In addition, previous research has found that many states that expanded Medicaid were able to do so without adding to their administrative costs per enrollee, particularly in larger states with the greatest increases to their eligible population, ²⁷ thus suggesting that expanding the number of people eligible for Medicaid does not add significant administrative financial burden to states. Our findings suggest that Medicaid spending did not crowd out other areas of state spending, consistent with previous research, despite substantial increases in Medicaid enrollment resulting from the continuous coverage provision.

Medicaid expansion was associated with greater increases in Medicaid spending in the COVID era than it was prior to 2020, mostly driven by federal funds. However, the relationship between Medicaid expansion and state expenditures is likely continuing to evolve. Due to the end of the public health emergency, the increased COVID‐era FMAP and continuous coverage provision have also ended. Further research will be needed to assess how the unwinding of COVID‐era policies impacts state budgets. On the one hand, enrollment in Medicaid will diminish due to the ending of the continuous coverage provision. KFF estimates that between 8 and 24 million individuals will lose Medicaid coverage. ¹⁰ On the other hand, decreases in the FMAP will mean that states will need to contribute a higher share of their own Medicaid expenditures.

4.1. Limitations

This study is subject to limitations. First, the NASBO dataset is self‐reported by state governments rather than based on administrative records detailing the actual spending. The data are not officially reviewed and may therefore contain inaccuracies. However, as stated in previous work, the data are often used by policymakers to analyze state budgets, which may point to their validity. ³ Second, it is possible that other policies enacted following Medicaid expansion, or around the same time, resulted in differential trends among Medicaid expansion and nonexpansion states. We attempted to mitigate this limitation by adjusting regressions for time‐varying characteristics, including state income per capita, unemployment rates, and CRF allocated to states, in addition to state and year fixed effects. Given that we exclude states that expanded Medicaid from 2020 to 2022 to isolate COVID‐era effects, we are not able to take into account changes to the FMAP for newly expanding states due to ARPA. Moreover, we do not take into account other COVID‐era legislation besides CRF via the CARES Act that provides states with additional funding, such as ARPA. It is also possible that between‐state migration could bias our estimates if states experienced difference patterns in migration, though previous work has not found evidence that Medicaid expansion is associated with migration across states. ²⁸ , ²⁹ Finally, changes in expenditures could reflect changes in health care utilization that were also occurring during the COVID‐19 pandemic. If these utilization changes were correlated with state Medicaid expansion status, they may influence our estimates.

4.2. Conclusions

We studied the effects of Medicaid expansion on state budgets before and during the COVID‐19 era, when Medicaid enrollment increased greatly. We found no evidence that Medicaid spending crowded out other state expenditures despite increased Medicaid expenditures. Increased federal expenditures likely shielded states from substantial budgetary impacts.

Supporting information

Data S1. Supporting Information.

Markell J, Meiselbach MK. The impact of Medicaid expansion on state expenditures through the COVID‐19 era. Health Serv Res. 2024;59(5):e14331. doi: 10.1111/1475-6773.14331