Do Chargemaster Prices Matter? An Examination of Acute Care Hospital Profitability

Abstract

Background:

It remains widely debated whether chargemaster price markups are tied to hospital profitability.

Objective:

To evaluate the effect of chargemaster markups on hospital profitability in the presence of unobserved hospital specific (time-invariant) confounders, and cross-sectional dependence due to latent (common) policy shocks.

Design:

We use interactive fixed effects methods to address concerns of unobserved hospital specific (time-invariant) confounders, and cross-sectional dependence.

Setting:

U.S. acute care hospitals, 1996 through 2017 (i.e., 22 years).

Participants:

Using primarily Medicare cost report data, we construct an unbalanced panel of 3,499 acute care hospitals per year, or a total of 76,972 hospital-year observations.

Measurements:

chargemaster markups (above cost), profits per hospital inpatient discharge.

Results:

Between 1996 and 2017, chargemaster markups increased (on average) by 155% and the standard deviation of the chargemaster markup distribution increased by 324% – indicating growing variability in the average markup strategies pursued by hospitals. Our preferred model specification implies that a unit increase of the hospital chargemaster markup is associated with a $261 (p<0.01; 95% CI: $232-$291) increase in profits per hospital inpatient discharge. These results are robust to a wide set of model specifications, the use of alternative profitability measurements, and the use of an alternative instrumental variable identification strategy. Additional subsample analysis that controls for a rich set of hospital quality measures and system affiliation information also yield similar results.

Conclusions:

We show that higher chargemaster markups are associated with higher hospital profitability. Additional research is needed to understand how chargemaster pricing impact health outcomes and healthcare disparities.

Introduction

Chargemaster rates are list prices that providers assign to each of their medical services and are commonly charged in full to uninsured patients.^1–6 Recent work has further found that chargemaster rates are also commonly billed to insured patients receiving care at out-of-network facilities, or to patients who receive care at an in-network facility by an out-of-network provider.^7,8 These practices highlight that chargemaster rates present an important risk for both uninsured and insured individuals, and as such, understanding the setting of these chargemaster rates is of primary policy relevance. While prior work has helped identify important correlations between chargemaster price markups and hospital profitability,⁹ and noted evidence suggestive of strategic chargemaster setting in the state of California, ¹⁰ it still remains debated whether chargemaster prices are set strategically more broadly across the U.S. and whether they in fact matter for hospital profitability.^11,12

In this study, we draw on a rich panel dataset of U.S. acute care hospitals for the years 1996 – 2017 in order to document trends for chargemaster markups and profitability. We further utilize recent techniques for dealing with long, cross-sectionally dependent panel data in order to estimate the chargemaster-to-cost markup effect on hospital profitability. In particular, we employ the estimation technique proposed by Bai (2009),¹³ which allows us to obtain estimates conditional on our rich set of covariate data, hospital specific (time invariant) unobservables (using traditional hospital fixed effects), as well as time-varying unobserved (common) policy shocks to which hospitals may respond heterogeneously (using interactive fixed effects). We further show that our main findings are robust to the inclusion of rich time-varying quality controls, the use of alternative measures of hospital profitability, and to the use of an alternative instrumental variable identification strategy. Given prior work that has noted potential heterogeneities in the profit objectives of different hospital types (for-profit, nonprofit and publicly owned) we also examine whether heterogeneities exist pertaining to the relationship between chargemaster markups and hospital profits across these ownership types.^14,15

Conceptual Model

To describe how chargemaster markups (the charge to cost ratio) affect hospital profits, we use a directed acyclic graphical model. This model is presented within Figure 1A, and it explicitly delineates all the relationships (given by directed lines) between the model variables (given by the square nodes).¹⁶ In particular, we see here that patient, hospital, and market, specific characteristics may all influence both the outcome of interest (hospital profits) and chargemaster markups (our main independent variable). Patient characteristics related to insurance status may affect profits because different payors (public, private and uninsured) pay different rates for their care, and moreover, these characteristics may also influence chargemaster rate setting by hospitals as these prices will affect a greater share of patients within hospitals that treat a larger share of either uninsured, or privately insured, patients.² Hospital characteristics such as ownership status (across for-profit, nonprofit and public) may also matter as different hospitals may have different objective functions that they seek to maximize.^2,14,15 For example, nonprofit hospitals may be more inclined to offer charity care, and also less inclined to utilize chargemaster markups as a profit generating strategy, than for-profit hospitals. It should, however, be noted that the literature on whether different hospitals have different objective functions remains divided, with some work suggesting government and non-profit hospitals’ actions are not aligned with their favorable tax treatment.¹⁷ Market characteristics pertaining to broad (hospital referral region) market competition among hospitals may also affect both chargemaster markups, and profitability, by hospitals.¹⁸ Since all of these characteristics may influence both our outcome and independent variable of interest, they represent confounders that we need to adequately control for within our analyses.

Figure 1:

Part A (left hand side) outlines a causal directed acyclical graph diagram which summarizes our conceptual model. Part B (right hand side) provides additional details on the pathways through chargemaster price setting may affect hospital profits.

Additionally, within Figure 1A, we see that unobserved characteristics may act to influence both hospital profitability and chargemaster markups. In particular, such confounders may consist of unobserved quality or reputation characteristics of hospitals which may act to influence both chargemaster markup setting and hospital profitability. Additionally, unobserved market-level policy shocks, trends of hospital system proliferation, and transitions into new reimbursement models, might all affect both hospital chargemaster markup setting and profitability. Some of these latent characteristics may be hospital/market specific and time-invariant (e.g., hospital reputation), while others may best be thought of as time-varying factors that influence hospitals/markets in heterogenous ways (e.g., geographically targeted policies). As such, both time-invariant and time-varying factors will need to be considered as potential latent confounders within our empirical analyses.

Lastly, within Figure 1A, we see the direct effect that chargemaster markups may have on hospital profitability (given by a blue solid arrow). This is the effect that we seek to identify with our empirical analysis, and it represents the overall effect of chargemaster markups upon hospital profits.

Additionally, within Figure 1B, we delineate four potential causal pathways that may connect chargemaster markups to hospital profitability. The first of these pathways concerns the observation that chargemaster prices are commonly billed to uninsured patients and therethrough may increase profits via higher payments (or payment settlements) with uninsured patients.^1,2,4,5 Second, higher chargemasters may yield higher payments from the insured that seek care out-of-network, or who receive care at in-network facilities but are cared for by out-of-network providers.^7,8 Third, it has been noted that chargemaster prices do in many cases serve as reference price for the contractual payments between private insurers and hospitals.¹⁸ As such, higher chargemaster prices may yield increased profits by increasing payments from private payors. Fourth, higher chargemaster prices may allow hospitals to increase the cost saving value of liabilities that end up being written off as bad debt, and therethrough increased hospital profits. In summary, while each of these pathways may contribute toward higher chargemaster markups, yielding higher hospital profits, in the analyses that follow, we seek to identify the overall (i.e., combined) effect of these channels.

Methods

Study Sample

We use an unbalanced panel with an average of 3,499 U.S. acute care inpatient hospitals per year, or a total of 76,972 hospital-year observations. These hospitals are all paid by the Centers for Medicare and Medicaid Services (CMS) under the Prospective Payment System (PPS) during the years 1996 through 2017 (i.e., 22 years). This data is sourced from the CMS and reflects hospital cost report information as reported to the Healthcare Cost Report Information System by Medicare Administrative Contractors.¹⁹. Additional information regarding hospital case mix, and capacity and wage indices is sourced from the Medicare Impact Files that are available for inpatient PPS hospitals.²⁰ (Additional data details can be found within the Supplemental Appendix D, which outlines details on our analysis sample creation and shows that it is representative of the full sample of U.S. acute care inpatient hospitals).

Study Variables

Profitability Measures

Hospital profits refer to the difference between total revenues and costs. In defining hospital profitability using Medicare Cost Report data we draw on prior work and define it based on the net income from patient care services per adjusted discharge.⁹ Net income from patient care services captures profitability across both inpatient and outpatient services for all patients. To accommodate greater comparability of profitability across hospitals that differ across their volume of both inpatient stays and outpatient visits, we divide profitability by each hospitals output volume.⁹ While inpatient output can be approximated using inpatient discharge information from the Medicare Cost Reports, the Medicare Cost Report data does not include information on the volume of outpatient visits. As such, prior literature has utilized adjusted discharges, which are inpatient discharges that are scaled to approximate the overall output of care across both inpatient and outpatient care.^9,21,22 We follow the literature, and calculate adjusted discharges (which we use as our overall volume proxy) as the number of discharges for inpatient care multiplied by the ratio of total gross revenue to inpatient gross revenue.^9,21,22 (Given that there are other possible ways of defining hospital profitability we provide additional details on such definitions within the Supplemental Appendix D1, along with robustness checks across these alternative definitions within the Supplemental Appendix B3 – these are discussed further below).

Charge-to-Cost Ratio

In order to capture the degree to which charges are marked up above costs we define our hospital charge-to-cost ratio (CCR) as the ratio of total hospital gross charges (across both inpatient and outpatient care) to total costs.

Other Covariates

Drawing on our conceptual model (in Figure 1) we include control variables for hospital, patient, and market level characteristics. To adjust for hospital characteristics, we control for: (i) hospital size using the reported number of hospital beds; (ii) the level of hospital capacity using the percentage of beds that are on average occupied; (iii) hospital ownership status (across nonprofit, for-profit and government ownership); (iv) hospital teaching status (teaching, non-teaching); and (v) the urban vs. rural locality of the hospital. Second, to capture hospitals patient pool characteristics, we control for: (i) the type of payors served using the percentage of discharges that are for Medicare and Medicaid patients; and (ii) the complexity of patient cases using the hospital’s Medicare case mix index. Lastly, we control for hospital market concentration (at the hospital referral region level) using the Herfindahl-Hirschman Index (HHI) (where market-shares are based on hospital beds).

Statistical Analysis

As noted within our conceptual model (outlined within Figure 1A), identification of the overall effect that chargemaster markups have upon hospital profitability rest, firstly, upon us controlling for known (and observed) confounders. We do this by including covariates for hospital, patient pool, and market specific characteristics throughout our analyses. Second, Figure 1A indicates that unobserved hospital confounders must also be controlled for in order to obtain plausibly unbiased estimates of the chargemaster markup effect. To control for unobserved confounders, we employ both fixed effects and interactive fixed effects methods. The fixed effects identification strategy allows us to control for time-invariant unobserved hospital differences that might affect both our charge markup measure, as well as our hospital profitability. An example of such a confounder would be unobserved reputation, or quality, levels that are slow to change over time and which are likely to influence both chargemaster markups and hospital profitability. The interactive fixed effects strategy represents an important generalization of the fixed effects model (in that it nests the fixed effects model as a special case), and has the added capability of being able to control for potential cross-sectional dependence within the data.¹³ Such cross-sectional dependence appears likely within our data as it may emanate from latent policy shocks, along with trends of hospital system proliferation, and transitions into new reimbursement models – system level changes that materialize over time, and which may affect hospitals in heterogenous ways. The interactive fixed effects strategy adjusts for these latent shocks by controlling for time varying factors that may affect hospitals in heterogenous ways. This (interactive fixed effects) model represents our main (preferred) identification strategy; and the model specification is given by:

$$Y_{it}=\alpha +{\tau }_1{CCR}_{it}^{}+{\sum }_j^M{\beta }_j{x}_{it}^j+{\gamma }_i^{\mathrm{\prime }}{\mathbf{f}}_t+{ϵ}_{it}$$ (1)

In equation (1), Y_it captures our profit per discharge of hospital i in period t, CCR_it is the charge-to-cost ratio (or chargemaster markup), and ${\mathit{x}}_{\mathit{i}\mathit{t}}^{\mathit{j}}$ denotes control variable j within the set of all controls (previously defined across hospital, patient pool and market characteristics). Lastly, ${\gamma }_i^{\prime }{\mathbf{f}}_t$ represents our controlling for the combination of m unobservable time effects, f_t, whose coefficients, γ_i, are hospital specific. Details on the estimation procedure for obtaining model parameter estimates, factors and factor-loadings are provided within the Supplemental Appendix C; and results from Pesaran (2015, 2004) cross-sectional dependence tests are provided within the Supplemental Appendix A1.^23,24

Results

Sample Descriptives

Table 1 presents our analysis sample descriptives. The unit of observation is here at the hospital-year level. For this sample, we note a pooled average (inflation adjusted) profit of -$121.14 per discharge, and an average charge-to-cost ratio of 2.51. It is worth nothing that negative (mean) profits per discharge are reasonable as our main profitability measure is based on patient care services only, and as such, does not include income from other (nonoperating) hospital activities that contribute toward the overall profitability of a hospital. (Related to this, Supplemental Appendix B3 shows that mean profitability per discharge, based on overall profitability, are indeed positive).

Table 1:

Summary Statistics for Outcome Measures, Markup Measure, Hospital & Market Characteristics and Quality Measures.

	N	Mean	SD
Outcome Measure
Profit (PCS, Per Discharge)	76,972	−117.62	635.55
Markup Measure
Charge-to-cost Ratio	76,972	2.51	1.34
Hospital & Market Characteristics
Nonprofit Ownership	76,972	0.61	0.49
Government Ownership	76,972	0.19	0.39
Teaching Status	76,972	0.39	0.49
Urban Location	76,972	0.58	0.49
Number of Beds	76,972	176.62	240.61
Capacity	76,972	0.49	0.19
HHI (HRR, Beds) Hospital Level	76,972	0.14	0.12
Patient Population Characteristics
Case Mix Index	76,972	1.36	0.31
Medicare (%)	76,972	0.49	0.15
Medicaid (%)	76,972	0.11	0.10
DSH (%)	76,972	0.26	0.17

Abbreviations: PCS = Patient Care Services; DSH = Share of Patients that are Disproportionate Share; HRR = Hospital-Referral Region; HHI = Herfindahl-Hirschman Index.

Pertaining to our hospital sample – 61% are nonprofit, 19% are government owned, 39% have a teaching status designation, 58% are located in an urban setting and 9% have a referral center designation. (Additional details showcasing the representativeness of our sample of all US acute care hospital can be seen within Supplementary Appendix D3).

Time Trends

Figure 2A shows boxplots for charge-to-cost ratios for each of the years 1996 through 2017. Here, we note two trends: first, the average charge-to-cost ratio in the U.S. has increased from 1.53 in 1996 (i.e., a 53% markup) to 3.91 in 2017 (i.e., a 291% markup). This increase represents a 155% increase in the average CCR over this 22-year period. Second, we see a considerable increase within the CCR dispersion over time, indicating a trend toward greater markup heterogeneity between hospitals. Overall, the standard deviation of the CCR distribution is found to have increased by 324% between 1996 and 2017. In looking across state boundaries in 2017 (Figure 2C) we further note great state level heterogeneity in the charge-to-cost ratios. We also see a clear north to south trend with lower markups in the north and higher markups being primarily concentrated within the southern states.

Figure 2:

Figures A and B present the boxplots for the Charge-to-cost Ratio and the Profit (PCS, Per Discharge) for each of the years 1996 through 2017. Figures C and D display the state level average Charge-to-cost Ratio and Profit (PCS, Per Discharge) in 2017. The color-coding is done across six deciles for each of the two variables.

Abbreviation: PCS = Patient Care Services

Figure 2B presents similar boxplot trends for hospital profits per discharge (for patient care services). Here we note little change in the overall inflation adjusted mean profits per discharge – with an overall average profit per discharge of -$111.49 in 1996 and one of -$110.58 in 2017. However, we see a steady increase within the profit dispersion across hospitals from about 2009 onwards. Regarding profit differences across state boundaries, looking at 2017 data (Figure 2D) we again note considerable heterogeneities across states, however, we do not see the same north to south trend as before.

In Figure 3 we see that there are clear markup practice differences across hospital ownership types, with for-profit hospitals consistently having higher average markups than nonprofit and government owned hospitals. Interestingly, we note that the markup trend for for-profit hospitals has been increasing at a disproportionate rate relative to that of nonprofit and government owned hospitals. Similar trends are observed across profitability, with for-profit hospitals having consistently higher average profits per discharge than do nonprofit and government owned hospitals.

Figure 3:

Figures A and B present the average Charge-to-cost Ratio and the Profit (PCS, Per Discharge) time trends across hospital ownership status. This is done for the years 1996 through 2017. Figures C and D present the same trends across the system and non-system affiliation of hospitals. These trends are provided for the 2010 through 2017 time period.

Abbreviation: PCS = Patient Care Services.

Shifting focus to trends for system (vs. non-system) affiliated hospitals we see another set of consistent and growing trends. First, we see that markups are consistently higher for system affiliated hospitals; and we see that the trend gap between system affiliated, and independent hospitals has been increasing over time. Second, we see similar trends across profitability with system hospitals experiencing consistently higher profits than non-system hospitals, and we also see that the profitability gap across these two groups is growing across time.

Fixed Effects Analysis

Table 2 presents our main regression results across a set of specifications that utilize different sources of variation for their identification of the effect that the charge-to-cost ratio markup has on hospital profits. In column (1) we utilize both within- and across-hospital variation in the CCRs, conditional on our set of hospital, patient and market controls, as well as our time fixed effects. This specification suggests that a unit increase of the CCR is associated with a $118.77 (p<0.01; 95% CI: $106.23 - $131.30) increase in the profit per discharge. In column (2) we address concerns related to potential confounding stemming from unobserved hospital specific time-invariant factors by utilizing hospital level fixed effects. This specification removes the across-hospital variation within the data as a source of identification by only using the within-hospital variation across time for the purpose of identification (Figure S2 within the Supplemental Appendix shows graphically that this is a valid source of identifying variation as there exists considerable within-hospital variation in our data). The estimated effect of a unit increase in the CCR is here a $130.91 (p<0.01; 95% CI: $118.78 – $143.04) increase of the profit per discharge.

Table 2:

Regression Estimates Across Year Fixed Effects, Hospital Fixed Effects and Hospital Interactive Fixed Effects Specifications.

	(Model 1)	(Model 2)	(Model 3)
	Profit (PCS	Profit (PCS	Profit (PCS
	Per Discharge)	Per Discharge)	Per Discharge)
Markup Measure
Charge-to-cost Ratio	118.77***	130.91***	261.27***
	(106.23 – 131.30)	(118.78 – 143.04)	(231.54 – 291.00)
Observations	76,972	76,972	76,972
R-squared	0.11	-	-
Controls Included	X	X	X
Year FE	X	X	X
Hospital FEs	-	X	-
Hospital IFEs	-	-	X

Significance is indicated as:

^*p < 0.1

^**p < 0.05

^***p < 0.01

Note: Cluster robust 95% confidence intervals are reported within the parentheses. Clustering is done at the hospital level. HHI is defined at the Hospital Referral Region level using hospitals as the firm unit.

Abbreviations: FEs = Fixed Effects; IFEs = Interactive Fixed Effects; DSH = Share of Patients that are Disproportionate Share; HHI = Herfindahl-Hirschman Index.

Interactive Fixed Effects Analysis

While the hospital and year fixed effects specification allows us to capture time-invariant confounders, for example due to latent differences in reputation and quality, it does not allow us to capture potential confounding stemming from unobserved policies and industry trends that might affect hospitals across time in potentially heterogenous ways. Examples of such unobserved common factors within the hospital market pertain to: (i) latent time-varying quality (i.e., quality dimensions that we expect to vary across time due their importance for hospital reimbursement under value based care); (ii) market level trends of increased hospital consolidation through system affiliation; (iii) similar trends of consolidations among insurers that may put pressure on hospitals to seek revenue from non-private payor patients; and (iv) overall market shifts introduced by the passage of the Affordable Care Act (ACA). All of these factors, if not fully accounted for, can introduce bias in our CCR estimates due to the resulting cross-sectional dependence within the data. In Supplemental Appendix A1 we provide the results for the Pesaran (2015, 2004) cross-sectional dependence test, which indicate a clear presence of cross-sectional dependence within the data.^23,24 In order to account for the cross-sectional dependence due to these latent factors, we employ an interactive fixed effects specification.¹³ The results from this specification are provided within column (3) of Table 2. When accounting for a set of three factor components, we find that a unit increase in the CCR is associated with a $261.27 (p<0.01; 95% CI: $231.54 - $291.00) increase in the profit per discharge.

Heterogenous Effects Analysis

As noted within our introduction and conceptual model sections, hospital operations, objectives and in turn chargemaster markup strategies may all vary across different hospitals. Our descriptive trend analysis within Figure 2, further suggests that hospital profits may respond in heterogenous ways to a change in the CCR across different types of hospitals. To examine this hypothesis further, Supplemental Appendix A3 provides a set of results that utilize interactions between the CCR and hospital ownership, teaching status, and urban vs. rural location, in order to identify these heterogenous effects. This analysis indicates significant heterogeneities in the marginal effect of the CCR on hospital profitability across both hospital ownership and teaching status.

Limitations and Robustness Analyses

An inherent limitation of our observational study design is the potential for bias due to omitted confounding variables or model misspecification. While our interactive fixed effects approach allows us to capture both time-invariant hospital confounders and time-varying common factors it is still possible that our model can have bias due to omitted time-varying non-common hospital factors that influence both chargemaster markups and profitability. An example of such factors could be time-varying quality measures. With that said, Supplementary Appendix B2 shows that our findings are robust to the inclusion of both alternative concentration measures and rich (time-varying) hospital specific quality measures for the years 2010 through 2017.

Another limitation of this study relates to us working with cost report (accounting) data to approximate economic variables. Firstly, such a mapping, between accounting data and the economic concepts, is often imperfect. Secondly, hospital financial data (such as that within the CMS Cost Reports) may not always come from audited hospital financial statements, and as such, the data may be subject to missing information and measurement error.²¹ With this noted, we have sought to ameliorate these data limitations in two ways. First, by examining the robustness of our main findings to the use of an alternative profit measure that also incorporates profits form nonoperating activities into its definition. To this end, results within Supplemental Appendix B3 indicate that our main findings are indeed robust to alternative hospital profit definitions. Second, prior work has shown that the factor model approach pursued within this study can, importantly, ameliorate bias concerns due to systematic measurement error.²⁵

Additional robustness checks further show that our results are robust to the use of: (i) an alternative quadratic functional form model specification (see Supplemental Appendix A2), (ii) the use of an analysis sample with stricter sample balance requirements (see Supplemental Appendix A5); (iii) the use of an alternative factor model estimation strategy a la Pesaran (2006) (see Supplemental Appendix A5);²⁶ (iv) to the use of an alternative (instrumental variable) identification strategy (see Supplemental Appendix B1); and (v) to sample sub-setting that restricts our analysis to the post Affordable Care Act period of 2014–2017 (see Supplemental Appendix A7).

Discussion

We find evidence of a positive relationship between hospitals’ charge-to-cost markups and their profitability. The importance of chargemaster markups for profitability appears to have fueled a significant growth of these markups over the 22-year period of our study. Given the extent of this growth, it is important to re-iterate the significance of the chargemaster price as well as discuss recent policies that seek to target these prices.

Chargemaster Prices and the Uninsured.

Chargemaster prices matter for the uninsured because they are billed these amounts in full. As showcased within this study, these charges represent large markups over the cost of care. For example, in 2017, the median hospital charge-to-cost ratio was 3.5, while the charge-to-cost ratio at the 90^th percentile in 2017 was 6.6 – that is a 560% markup above the Medicare approved cost of care. While early reports following the passage of the ACA have indicated a decrease in the pool of uninsured patients that would be confronted with these prices, recent work indicates a reversal of these trends since 2016/2017.²⁷ These developments appear to call for policy makers to extend further focus to chargemaster markups and the growing uninsured.

Chargemaster Prices and Health Disparities.

Having to face inflated chargemaster rates may present financial challenges for individuals that are billed these rates, but they may further present societal challenges by means of increasing avoidance of medical care due to the anticipation of high costs.²⁸ What is important to highlight here is that these effects may end up disproportionately affecting minority groups. This is because racial/ethnic minorities are overrepresented within the patient population that is uninsured,²⁷ and further among patients suffering from chronic conditions such as diabetes and who are therefore in greater need of care and who have been documented to have elevated healthcare expenditures.^29–31

Chargemaster Prices and the Insured.

Recent work has further showed that these prices matter for insured patients that may receive medical care at out of network facilities, or those that are subject to surprise billings at in-network facilities.^7,8,32,33 Furthermore, a recent detailed study of hospital and insurer contracting has revealed that in about one third of cases, hospital contracts with private payors are set as a percentage off of hospitals’ chargemaster rates.¹⁸ Given the trends in Figure 1 showing that charges are evermore departing from underlying cost of care, hospital administrators (and policymakers) may wish to examine the validity of reimbursement contracts being set as a function of charges; or alternatively seek to limit these growth trends.

Chargemaster Transparency Alone Might Not Be Enough of an Answer.

Work examining state level transparency initiatives have found little to no impact of chargemaster transparency initiatives on subsequent chargemaster price setting by providers;³⁴ and while additional work is needed in examining the impact of the 2021 nationwide transparency mandate enforcements,³⁵ it seems that curbing the growth of observed chargemaster markup trends will require a full examination of the potential root causes behind these trends, such as a study of the mechanisms that cause providers to set continually higher chargemaster markups.

A number of such mechanisms appear to be suggested by our conceptual model and the preceding discussion and literature. First, if charges serve as a reference point for litigation with uninsured patients regarding final payments, then chargemaster price setting will importantly depend on the patient mix of a given hospital. Bai and Anderson (2015, 2016) present important descriptive work that appears to support such a thesis, however, additional work examining the causal basis for this link seems to be critical here.^2,3 Second, if charges matter for hospital contracting with private payors, then this too may present providers with incentives for increasing their chargemaster rates. Future studies may wish to examine this relationship by combining market level data on insurer concentration and hospital chargemaster rates. Third, it seems important to explore the way in which chargemaster price transparency information is delivered (its accessibility and ease of use), the subsequent effect that the delivery/accessibility has on chargemaster price setting by providers, and the influence that this has on the choice of provider by patients. If chargemaster reference pricing is provided effectively, then in drawing on prior studies of reference pricing (provided by insurers to insured patients) would suggest that this can present yet another important mechanism through which more downward pressure can be put on the chargemaster (markup) growth.^36–38 While the present study has focused on the identification of the overall relationship between chargemaster markups and hospital profitability, future studies may wish to examine these individual causal pathways, as well as examines their relative importance. This appears to be an important avenue for future research, especially in the advent of recent price transparency regulations within the U.S. hospital market.

Conclusion

We find evidence of a positive relationship between hospital charge-to-cost markups and profitability. These results are robust to the use of various identification strategies, to controlling for a rich set of quality measures, and to the use of alternative profit definitions.