The Effect of the Massachusetts Health Care Cost Growth Benchmark on Hospital Revenues and Other Spending Outcomes: Associations with Hospital Revenue, Hospital Prices, and Insurance Premiums

As part of Chapter 224 of the Acts of 2012, Massachusetts introduced a health care cost growth benchmark that serves as a target for constraining health care spending (Commonwealth of Massachusetts, 2012). The cost growth benchmark is set annually by the Massachusetts Health Policy Commission (HPC) and monitored by the HPC and the Massachusetts Center for Health Information and Analysis. Over time, the cost growth benchmark has ranged from 3.1 percent to 3.6 percent. Each year, the HPC monitors accountability of payers and providers through reports and hearings and assesses the effects of proposed mergers and acquisitions on health care spending growth. Furthermore, providers and payers with spending growth in excess of the cost growth benchmark may be subject to performance improvement plans and can face fines of up to $500,000. However, Massachusetts has rarely exercised its enforcement authority.

As of early 2026, spending growth in Massachusetts had stayed within the cost growth benchmark in four of the 11 years for which assessment data were available. The frequency with which spending growth exceeded the cost growth benchmark suggests that the policy is not working as intended. However, it is possible that the policy constrained health care spending growth relative to what would have occurred without the cost growth benchmark.

In this analysis, we assessed whether the Massachusetts cost growth benchmark reduced growth in spending-related outcomes using statistical techniques to predict what these outcomes would have been without the cost growth benchmark. We focused on three types of outcomes: hospital revenue, hospital prices, and premiums. Revenue data came from the Centers for Medicare & Medicaid Services (CMS) Hospital Cost Reporting and Information System, hospital price data came from the RAND Hospital Price Transparency Study (Rounds 4 and 5), and data about insurance premiums came from the CMS Center for Consumer Information and Insurance Oversight (CCIIO). While both the CMS Hospital Cost Reporting and Information System and the CMS CCIIO data contain close to a universe of submitters, the RAND Hospital Price Transparency data are a convenience sample of employer claims that is not representative of the full Massachusetts market. We had access to data from 2010 through 2023 for revenue outcomes, 2016 through 2022 for prices, and 2015 through 2023 for premiums.

Methodological Considerations

In a previous analysis, we considered the effects of health care cost growth benchmarks in eight other states that have similar policies in place: California, Connecticut, Delaware, New Jersey, Oregon, Rhode Island, Vermont, and Washington (Eibner et al., 2026). In that study, we found scant evidence to suggest that health care cost growth benchmarks were associated with reductions in hospital revenue, hospital prices, or insurance premiums. However, we did not consider the effects in Massachusetts in that article because of unique challenges inherent to the state. The most formidable of these challenges is that the Massachusetts cost growth benchmark took effect in 2013, immediately before the 2014 implementation of the Affordable Care Act (ACA). Because the ACA expanded insurance coverage and set minimum benefit requirements, ACA implementation could have affected health care spending separately from the Massachusetts cost growth benchmark. Moreover, because Massachusetts enacted an ACA-like insurance expansion in 2006 (Chapter 58 of the Acts of 2006), the effects of the ACA in Massachusetts differed substantially from effects in other states (Commonwealth of Massachusetts, 2006). These challenges affect the validity of standard statistical methods, such as difference-in-differences approaches that compare outcome trends in Massachusetts with outcome trends in comparison states. A further challenge is that our data sources for premiums and prices exist for only post–cost growth benchmark implementation years, limiting our ability to use methods that require preintervention and postintervention comparisons for these outcomes.

For these reasons, we chose to consider the effects in Massachusetts separately from the effects in the other eight states with health care cost growth benchmarks. Furthermore, we used four alternative strategies to estimate the effects of the Massachusetts cost growth benchmark:

1. We used entropy balancing (EB) and difference-in-differences (DD) models to compare hospital revenue outcomes in Massachusetts before and after the cost growth benchmark with hospital revenue outcomes among hospitals in states that did not have cost growth benchmarks and did not expand Medicaid. We focused on nonexpansion states because they were less affected by the ACA's insurance reforms than other states. To address the fact that these states are quite different from Massachusetts, we used EB to weight nonexpansion state hospitals to resemble Massachusetts hospitals, including weighting on preperiod trends in the outcome variable. Although differences in unmodeled covariates likely remain, DD adjusts for time-invariant differences, such as fixed differences in income between Massachusetts and other states.
2. We used EB and DD to compare hospital revenue outcomes in Massachusetts with hospital revenue outcomes among hospitals in counties with high pre-ACA insurance levels (90-percent insured or more) in states without cost growth benchmarks. We chose high-insurance counties because these counties were likely less affected by the ACA than counties with lower insured rates. Because relatively few counties outside Massachusetts had high insured rates prior to the ACA, this approach had a smaller sample size than the first approach. As in the first approach, we used EB weights to make comparison hospitals as similar as possible to Massachusetts hospitals.
3. We used an interrupted time series (ITS) approach to assess whether the 2013 implementation of the Massachusetts cost growth benchmark led to a shift in hospital revenue outcomes through either a discrete change in revenue levels or a shift in revenue growth trends. This approach did not require a comparison group but did not allow us to separate the effects of the Massachusetts cost growth benchmark from other factors that may have influenced revenue trends in the state.
4. We developed a model to predict hospital revenue, hospital price, and premium outcomes in non-Massachusetts hospitals and counties and used this model to estimate what these outcomes would have been in Massachusetts without the cost growth benchmark. This was the only approach that enabled us to analyze prices and premiums because data for these outcomes are available only for after the Massachusetts cost growth benchmark was implemented. However, a drawback of this approach was that it required us to assume that the observed predictor variables can fully explain variation in outcomes, which may not be accurate.

All four of our analytic approaches had important drawbacks, forcing us to regard the findings as exploratory and associative. The approaches did not always produce similar findings for the same outcome. When these contradictions occurred, we summarized the evidence across the approaches, but these discrepancies further weakened our confidence in the results. Nevertheless, these analyses provided suggestive evidence on the effects of the Massachusetts cost growth benchmark.

Findings

Table 1 shows the findings, strengths, and weaknesses of each approach. For hospital inpatient revenue per discharge, we found statistically significant evidence of an effect of the Massachusetts cost growth benchmark in Approach 4 and marginally significant evidence of an effect in Approach 2. However, the effects were null in Approaches 1 and 3, suggesting that this finding was not robust. For hospital outpatient revenue per discharge equivalent, we found statistically significant evidence of an effect of the cost growth benchmark in Approach 4 and marginally significant evidence of an effect in Approach 3, but null findings from Approaches 1 and 2. We found no effects on health insurance premiums on average, although effects were significant in some years for Approach 4. Using Approach 4, we found a statistically significant negative effect on both inpatient and outpatient hospital prices. However, given that this finding is based on a single approach that likely cannot fully adjust for the differences between Massachusetts and comparison states, we still regard this finding as exploratory.

Conclusions

Our analysis suggests that the Massachusetts health care cost growth benchmark may have reduced hospital revenue and prices. However, the revenue results were not consistent across models. For both inpatient and outpatient revenue, we obtained a negative and statistically significant finding from one approach, a negative and marginally significant finding from a second approach, and null findings from two other approaches. For prices, we were able to estimate the models using only Approach 4, and we found large effects; for example, the reduction in inpatient prices was $11,165, which is roughly one-third lower than what would have been expected without the cost growth benchmark. However, there were statistical challenges for using Approach 4 for the hospital price outcomes, which reduces our confidence in the results. For health insurance premiums, our estimated effects were not statistically significant.

Although other states have health care cost growth benchmarks, the Massachusetts cost growth benchmark is unique and important to study because of its long tenure and because Massachusetts is one of the few states that has enforcement mechanisms for its benchmark. However, it is very difficult to evaluate the effects of the Massachusetts cost growth benchmark—even with sophisticated statistical methods—and these methodological limitations constrain confidence in our results. The cost growth benchmark may have had some effect on hospital revenue and prices in Massachusetts, but the evidence for this is weak. Policymakers aiming to reduce health care spending growth could consider policies for which there is more-compelling evidence, such as stronger enforcement of cost growth benchmarks, combining cost growth benchmarks with complementary policies (such as value-based payments and price caps), and adopting more targeted approaches that were successful in other states.