Surprise! Out-of-Network Billing for Emergency Care in the United States

A.  The Evolution of Emergency Medicine in the United States

At present, there are more than 4,500 EDs in the United States and approximately 40,000 physicians who staff them nationwide (Hsia, Kellermann, and Shen 2011; Morganti et al. 2013). The use of EDs has risen dramatically over time. From 2001 through 2008, the use of EDs increased 1.9% each year—60% faster than concurrent population growth (Hsia Kellermann, and Shen 2011).

Over the past several decades, EDs have become one of the main pathways through which patients are admitted to the hospital. At present, a majority of a hospitals’ admitted patients entered the hospital via an ED (Morganti et al. 2013). Because EDs have become a major source of patients, hospitals now want to keep their EDs open at all hours and run them efficiently (Institute of Medicine 2006; Morganti et al. 2013) As a result, there has been a marked increase in the outsourcing of management of hospital EDs. ED outsourcing companies hire and manage physicians, manage ED operations, and take care of billing and collections. Today, roughly 65% of the physician market is outsourced (Lehrich, Kalenderian, and Nentin 2013). Among the hospitals that outsource their services, approximately one-third contract with a large national outsourcing chain, and the remainder are outsourced to smaller local firms (Dalavagas 2014).

The national market for physician outsourcing is dominated by two firms, EmCare and TeamHealth, which collectively account for approximately 30% of the outsourced physician market (Lehrich, Kalenderian, and Nentin 2013). Both firms were publicly traded until they were taken private by large private equity firms. EmCare was publicly traded until 2018, when it was bought by KKR & Co. It operates in 45 states, has 23,100 affiliated or employed physicians and health care professionals, and (according to their 2016 Form 10-K) delivers more than 18 million emergency episodes per year. More recently, EmCare has partnered with a large for-profit hospital chain and formed joint ventures where the firm and its hospital partners share in profits from physician bills (Lehrich, Kalenderian, and Nentin 2013). TeamHealth is approximately the same size. It was publicly traded until 2016, when it was purchased by the Blackstone Group.

In the aggregate, ED care is profitable for hospitals. Wilson and Cutler (2014) estimated that average ED profit margins are approximately 7.8% per patient. However, the profit margins that hospitals face for ED care vary significantly depending on how a patient’s care is funded and based on whether a patient is admitted to the hospital. Wilson and Cutler (2014) found that hospitals had profit margins of 39.6% for privately insured patients treated in EDs, whereas the profit margin for patients who were covered by Medicare, covered by Medicaid, and uninsured were −15.6%, −35.9%, and −54.4%, respectively. They also found that patients who were admitted to the hospital were significantly more profitable than those who were not. For Medicare patients, the profit margin on ED care for patients who were discharged from the ED was −53.6%, whereas the profit margin for patients who were admitted to the hospital was 18.4% (Wilson and Cutler 2014).

B.  Out-of-Network Surprise Billing

There are broadly two types of out-of-network bills. The first form of out-of-network billing results from contracting frictions between insurers and physicians. In the United States, there are approximately 55,000 ED physicians, 6,100 hospitals, and over 1,000 insurers (Kaiser Family Foundation 2019; American Hospital Association 2020). As a result, it is unlikely that every ED physician could have a contract with every insurer that covers all the patients she treats. As an example, an ED physician in a popular vacation destination could see patients from across the country. Even if she wanted to, this ED physician would struggle to enter into contracts with insurers from across the country. While an out-of-state patient’s insurer might have a contract with the hospital in the area the patient is visiting, it is possible that they might not have a contract with the patient’s ED physician. In these instances, if the physician is not engaging in a deliberate out-of-network strategy, she may accept a payment rate that is of the same magnitude as her usual in-network payments.

A second form of out-of-network billing occurs when physicians deliberately do not participate in insurers’ networks so that they can reap higher payments. As the New York State Department of Financial Services noted in their 2012 report, “A relatively small but significant number of out-of-network specialists appear to take advantage of the fact that emergency care must be delivered and [that] advanced disclosure is not typically demanded or even expected by consumers. The fees charged by these providers can, in some instances, be many times larger than what private or public payers typically allow, and are another source of consumer complaints” (New York State Department of Financial Services 2012). Indeed, a recent study found that physicians who tend not to be chosen by patients (anesthesiologists, radiologists, pathologists, and ED physicians) have the highest charges measured as a percentage of their Medicare payments (Bai and Anderson 2017).

When an insured patient sees an out-of-network physician, there are three potential outcomes. First, the insurer may pay the physician’s out-of-network bill in its entirety. This will protect the patient, but ultimately insurers will pass the cost of these higher payment rates on to all beneficiaries in the form of higher premiums. In addition, patients generally face higher coinsurance rates when they see an out-of-network provider. As a result, even if their insurer pays their physician her charge, the patient may still face substantial cost sharing. Second, the insurer may pay the out-of-network physician her usual and customary rate, which the insurer calculates on the basis of average charges or average in-network payments for the services provided. This payment is generally lower than the total billed amount. When this occurs, the physician may accept the usual and customary rate the insurer is offering and move on. Alternatively, the physician may pursue the patient to pay the difference between the charge and whatever the insurer paid. This is referred to as balance billing. Third, the insurer may not cover the costs of out-of-network care at all, leaving the patient to pay the entire physician bill herself. As we show later from our data, these physician bills can be extremely large. While there is no systematic evidence on the frequency that patients are balance billed by physicians, from 2012 to 2015, data from the Texas Department of Insurance showed that balance billing complaints in the state increased 1,000% (Gooch 2016).

There has been significant coverage of out-of-network billing in the popular press (Rosenthal 2014a, 2014b; Sanger-Katz and Abelson 2016). However, until recently, there has been no systematic evidence on the frequency that out-of-network billing occurs. Recent survey work suggests that it is fairly common for privately insured patients to be treated by out-of-network physicians.² The results of these surveys have been confirmed by recent empirical evidence. A 2014 report found that among the three largest insurers in Texas, 45%, 56%, and 21% of their in-network hospitals had zero in-network ED physicians (Pogue and Randall 2014). Likewise, in the first national study of out-of-network billing, Cooper and Scott Morton (2016) analyzed data from a large commercial insurer and found that 22% of in-network ED hospital visits included a primary physician claim from an out-of-network doctor. Using different data, Garmon and Chartock (2017) found that 20% of ED cases in which care was delivered to privately insured patients at in-network hospitals involved care from an out-of-network physician. However, as we will show below, knowing the average probability of receiving an out-of-network bill does not help diagnose the policy problem, which lies in the tail of the distribution of out-of-network billing prevalence across hospitals.

C.  EmCare and Out-of-Network Billing

There is anecdotal evidence suggesting that EmCare, the nation’s largest physician staffing company, uses out-of-network billing as a tool to raise profits. For example, on March 29, 2016, an investment advising service noted, “What EmCare actually does is take over an in-network hospital Emergency Room that is aligned with most local healthcare insurance plans and staff it with physicians who are out-of-network. … Since EmCare is out-of-network, it refuses to sign in-network agreements with local insurance providers, it 1) can charge exorbitant out-of-network reimbursement rates from the providers and 2) since it is out-of-network, it can ‘balance bill’ its patients for the difference between its prices and the amount the insurer belies is ‘usual and customary’. This is a license to print money!” (Chanos 2016)

A video of hospital administrators at Glen Rose Medical Center in Glen Rose, Texas, discussing out-of-network bills also suggests that EmCare uses surprise billing as a deliberate strategy (the transcript from the video is available in app. 1). As the hospital administrators state in the video, in order to get EmCare physicians to cease billing out of network and balance billing their patients, they would need to increase their subsidies to EmCare. To that end, one of the hospital staff says, “They [the ED physicians] bill out-of-network for most insurance… and we could expand the insurances that are covered in the ER, but it’s at a cost of about $200,000 a year to us. … [If] we require them to be in-network … then our subsidy would increase significantly.” Later, in response to discussion of the $200,000 in additional funds the hospital would have to pay EmCare, another hospital administrator replies, “We would have to pay EmCare an additional $200,000 to put those people in-network because right now billing out-of-network they’re making more money.”

Envision (the parent company of EmCare) has confirmed this strategy in their reply (posted on their website) to an earlier draft of this paper. In that reply, they state, “We dispute that ‘a hospital does not benefit directly from physicians engaging in out-of-network billing’ (page 19). Hospitals do benefit directly when higher out-of-network insurance payments rather than hospital subsidies to the emergency physicians, enable the hospital to recruit, retain, and expand high-quality board certified emergency physicians” (Envision Healthcare 2017).

A.  Insurers and Out-of-Network Billing

The physician group and the insurer bargain over the price the insurer will pay the physician for care delivered to policyholders. The insurer faces a trade-off between including more and better physicians in its network and the higher in-network payments needed to retain those physicians in the network. The decision about how broad and how highly reputable a network of providers to create (and how to handle out-of-network bills) is a function of the preferences of the purchasers of health insurance. Buyers of insurance might prefer a broad network of physicians, have a distaste for out-of-network bills, and therefore be willing to accept higher premiums. On the other hand, the buyers might prefer lower premiums and be willing to accept an insurance plan with a narrower network of providers and a higher probability of a policyholder seeing an out-of-network physician. In the extreme, the buyers of insurance could be willing to accept a plan with out-of-network ED physician billing and believe it to function as a very expensive form of patient cost sharing for accessing ED services.

When ED physicians are pursuing a deliberate strategy of billing out of network, insurers can face higher payments, higher transaction costs, and dissatisfied customers (e.g., those who received a balance bill). In general, absent specific out-of-network billing laws, the insurer’s outside option in the event of disagreement over an ED physician’s payment is litigation or the threat of litigation under the relevant federal and state statutes. The fees that physicians collect under disagreement in this setting will therefore not be competitively set by hospital demand and physician labor supply but instead be driven by the possibility of litigation over their bills as well as adverse publicity and social norms.⁴ By contrast, as we discuss, a number of states have regulations that impact ED physicians’ outside option if they bill out of network. Some states (like California, Maryland, and Connecticut) directly regulate payments to out-of-network providers. Other states (like New York and Texas) have each introduced an arbitration process between providers and insurers. Finally, there are states that do not have surprise billing protections but do have more general laws against price gouging and similar behavior. As a result, the outside option for an insurer of formally disputing an out-of-network charge will be differentially successful depending on state law. State law will impact physicians’ outside option, which will affect the rates an insurer is willing to pay ED physicians to join its network.

B.  Hospitals and Out-of-Network Billing

EDs serve as the front door to hospitals. The majority of admitted patients in a hospital at any given point in time came in via the ED. As a result, hospitals need to keep their EDs open at all hours in order to retain patients. To keep their EDs running, hospitals must recruit staff to run their EDs and arrange with physicians to provide care from inside their facilities. Alternatively, they can contract with an ED staffing company to manage the entirety of their ED, including recruiting, managing, and paying physicians. However, EDs deliver significant amounts of uncompensated care, and ED physicians regularly treat patients from whom they receive little or no compensation (Garthwaite, Gross, and Notowigdo 2018). Because of the Emergency Medical Treatment and Active Labor Act (EMTALA), EDs must provide care to patients in an emergency (Centers for Medicare and Medicaid Services 2012). As a result, hospitals must typically pay physicians a fee to work from their facilities to offset the physicians’ costs for uncompensated care and pay for the services they provide above and beyond their clinical practice (e.g., managing the ED).

Hospitals control which physicians or ED staffing firms they allow to work from inside their facilities. We assume that hospitals are aware of whether physician staffing firms engage in a deliberate out-of-network billing strategy. When ED physicians bill out of network, the ED physicians themselves (or the staffing firms for which they work) benefit from higher out-of-network payments. However, the hospitals where they work do not generally receive direct benefits from out-of-network ED physician billing. Indeed, when ED physicians remain out of network and balance bill patients, it introduces costs to hospitals, including reputational harm.⁵ Therefore, for hospitals to be willing to permit physicians to bill out of network from inside their facility, they must receive a transfer from physicians or staffing firms equal or greater to the cost of the reputational harm they incur from allowing the practice to persist.

These transfers could take a myriad of forms. For example, ED staffing firms often demand subsidies from hospitals to staff their EDs. Physician staffing firms could lower these subsidies in exchange for being allowed to bill out of network. Alternatively, ED physicians could deliver medical care in a manner that raises revenue for the hospital. This could include increasing imaging and lab testing rates (which raises hospital revenue) or increasing the rate that patients from the ED are admitted to the hospital. ED staffing firms could also raise the quality of hospital EDs, such that they attract more patients and improve the hospital’s reputation. At the extreme, physician staffing companies could enter into profit sharing agreements with hospitals, where the hospitals would benefit directly from the profits generated by physicians’ out-of-network billing.

C.  Physicians and Out-of-Network Billing

A physician or physician group faces a choice of negotiating in-network rates with insurers or going out of network, collecting higher out-of-network payments but incurring costs from engaging in the practice. In the longer term, they may see a modest reduction in the number of patients they treat if patients become aware of their out-of-network billing strategy and begin to avoid their facility. A physician or physician group must consider the incremental profit she or the group will obtain from going out of network. In a standard market with downward-sloping demand, if a physician went out of network, she would experience a significant decline in the number of patients she treats because of her higher out-of-network price that most patients would face. However, because we are examining ED physicians, we make the more realistic assumption (for this setting) that demand for ED physicians is inelastic in the short run.⁶ Therefore, in this setting, if the ED physician does not enter into an insurance network and seeks to collect her charges, she still obtains roughly the quantity of patients equal to what she would receive were she in network. As a result, we make a simplifying modeling assumption that her increase in revenue (or revenue for the ED staffing firm) is the difference between the in-network price and out-of-network effective price multiplied by the cases she performs per year. ED staffing companies may not be able to collect the entirety of their charges from all privately insured patients that they treat. For example, some insurers may not pay out-of-network physicians the entirety of their charges, and patients may have varying abilities to cover balance bills. Under this scenario, the staffing company is engaging in a form of first-degree price discrimination and seeking to collect the entirety of their charges from the patients with the ability and willingness to pay them.

Physicians likely incur costs from engaging in an out-of-network strategy. These could include fixed costs, such as physicians’ own intrinsic dislike of the practice, potential peer pressure, unpleasant meetings with stakeholders, and the cost of software necessary for billing and collection. Likewise, these could include variable costs, such as more unpleasant and time-consuming communication with patients, hospitals, and insurers; the costs of collecting on each bill; and defending against litigation.

Physicians will also have to compensate the hospital for allowing them to engage in out-of-network billing from inside their facilities. As we described, physicians can compensate the hospital from their own pockets via reducing the subsidies they require for managing a hospital’s ED services or entering into joint ventures where hospitals get a portion of physicians’ profits. A less expensive but more legally risky option for the physician is to deliver medical care in a style that benefits the hospital. However, changes in their clinical activity that benefit the hospital (such as overtesting) could open the physician or physician groups to legal risk (e.g., claims of fraudulent billing). As a result, the propensity to engage in these actions depend on the risk tolerance of these physicians.⁷

In our setting, we think ED management firms may have greater awareness of the intricacies of physician payment and better understand the benefits of setting higher charges than individual physicians operating in small group practices. In this sense, the ED management firms engage in informational arbitrage (à la Hayek). Previous work by Clemens, Gottlieb, and Molnár (2017) showed that individual physicians and physicians in small groups tend to set commercial prices that follow the Medicare payment rates. By contrast, physicians in large group practices tend to have payment rates that are less strongly correlated with Medicare payment rates. National physician management companies will likely seek a profit-maximizing price that takes advantage of ED physicians’ strong outside option. One might imagine that in equilibrium, this superior fee structure would have arrived at all hospitals. This is not the case in the United States for two reasons we can identify. First, outsourcing firms with an out-of-network strategy will not be able to enter hospitals that, because of their utility function, require compensation above what the physicians gain (e.g., some nonprofit hospitals may place a high premium on protecting patients from financial harm). Second, as out-of-network billing becomes more pervasive, the risk of regulatory backlash grows.⁸

A.  Data

Our claims data come from a large commercial insurer that covers tens of millions of lives annually. The data run from January 1, 2011, through December 31, 2015. The data are structured at the service line level and include detailed patient characteristics, a provider identifier, and the ability to link to a range of third-party data sets. We limit our analysis to episodes that occurred at hospitals registered with the American Hospital Association (AHA). Therefore, we do not include, for example, treatment that was delivered at urgent care clinics.

To construct emergency episodes, we identify emergency room visits in our data as those with a physician claim for emergency care and a facility claim with a code for emergency care that occurred on the same day. We identify ED claims for physicians as those that include a current procedural terminology (CPT) code of 99281, 99282, 99283, 99284, 99285, or 99291. We match those to facility claims by identifying claims delivered to the same patient on the same date that include a hospital service line with a revenue code of 0450, 0451, 0452, 0453, 0454, 0455, 0456, 0457, 0458, or 0459. The episode runs until the patient is discharged from the hospital. We exclude episodes with a length of stay over 30 days.

At baseline, our data include 13,444,369 ED episodes. We introduce several sample restrictions to our data to produce an analytic data set. First, we exclude episodes that were missing an AHA hospital identification number or did not come from an AHA-identified hospital. Thus, the analysis is focused on only hospital-based ED care. This restriction eliminates 1,908,710 episodes. Second, we exclude episodes for which the same physician billed as in network and out of network on separate service lines on the same claim form. This restriction eliminates 264,636 episodes. Third, we exclude episodes with duplicative insurer payments, episodes with insurer payments that were negative, and episodes for which the insurer paid $0 because the claims were denied. This restriction removes 217,267 episodes. Fourth, we exclude episodes for which the start date of the episode occurs after the end date of the episode. This restriction excludes 79 episodes. Fifth, we limit our analysis to hospitals that delivered 10 or more episodes per year and appear in all 5 years of the data. This restriction excludes 330,312 episodes. Sixth, we limit our analysis to individuals who had 6 months of continuous enrollment before their emergency episode. Having 6 months of historical data is necessary to create our Charlson comorbidity scores. This restriction excludes 1,810,245 episodes from our analysis. Finally, we winsorize the top and bottom 1% of the prices in our data.⁹ We do this to limit the influence of idiosyncratically high- and low-priced episodes.

In our data, we observe physician and hospital charges, the amount that the insurer paid, and patients’ coinsurance payments, copayments, and spending under their deductibles. We define the total amount an ED physician was paid as the sum of the insurer payment, the patient coinsurance payment, the patient copayment, and the patient deductible on physician service lines that have a CPT code for emergency services. We calculate facility payments as the sum of the insurer payment, patient coinsurance, patient copayment, and patient spending under her deductible summed across all facilities claims. All prices are put in 2015 dollars using the US Bureau of Labor Statistics Consumer Price Index.

Unfortunately, we do not observe whether patients were balance billed by physicians. Therefore, it is possible that the physician collects more in total than we can measure. To our knowledge, there are no data sets with information on the balance billing of patients. However, we construct a potential balance bill measure that is the difference between what the physician charged and what would be the median in-network payment for that case.

In addition, we construct an indicator for whether imaging occurred during an episode based on whether there are facility claims with revenue codes associated with imaging studies.¹⁰ We also identify episodes as involving an admission to the hospital if the facility claim for the episode includes a revenue code for room and board fees.¹¹

For each episode, we also observe the patient’s sex, age (measured in 10-year age bins), and race (white, black, Hispanic, and other). We also use our claims data to measure historical patient spending for 6- and 12-month periods preceding an episode. Because we do not want the emergency episodes we are analyzing to feed into the historical spending measures, we measure spending from 2 weeks before the admission date for an episode back 6 and 12 months. In addition, we used 6 and 12 months of claims data to calculate Charlson measures of comorbidity (Charlson et al. 1987).¹²

B.  Identifying Where EmCare Has Contracts

EmCare bills insurers using their contracted physicians’ National Provider Identifier (NPI) numbers. As a result, our claims data do not indicate that a particular claim is being billed by a physician employed by EmCare. Moreover, the firm does not provide a list of facilities where they have contracts. To overcome this information gap, we use data from EmCare’s own website and public documents to identify the hospitals where the firm has outsourcing contracts. We require two independent sources of information to classify a hospital as a facility that outsourced its ED services to EmCare.

Our first source of information on the hospitals where EmCare has contracts comes from the firm’s parent company, Envision. Envision posted a map on their website that included dots marking the location of hospitals where the firm had contracts (see fig. A.1). To identify hospital locations on the Envision map, we scraped the map using mapping software from ArcGIS to identify the latitude and longitude of the centroid of each point on the map.¹³ We then matched the latitudes and longitudes of these centroids to data on hospital locations from the AHA. We assumed that the AHA-registered hospital that was the shortest Euclidean distance to the centroid of each point on the Envision map was an EmCare contracted hospital.

The second source of information we use to identify hospitals that contract with EmCare is job advertisements posted by the firm. EmCare posts job advertisements on their website to recruit physicians to work at their care locations (see example in fig. A.2). The job advertisements include the name of the hospital where physicians are being recruited and the specialty of the physicians the hospital is looking to hire. We scraped the names of the hospitals and the specialty of the physicians being recruited from all EmCare’s job postings and website histories. This allowed us to create a roster of hospitals where EmCare was recruiting ED physicians between 2011 and 2015.

Ultimately, we regard a hospital as having a contract with EmCare if we are able to identify the hospital on a map from their website and found a job hiring post where an ED physician was being recruited. This strategy exploits the fact that, in general, EmCare wholly takes over an ED and participates in exclusive contracts with hospitals (Lehrich, Kalenderian, and Nentin 2013). Using this strategy, we identify 212 hospitals affiliated with EmCare. As a result, of the 3,345 hospitals in our analysis that meet our sample criteria, 6.3% outsource their ED to EmCare. On the basis of investor reports on EmCare, our sample of hospitals with contracts with EmCare represents a modest undercount of the total population of hospitals that have contracts with EmCare.

We also use the entry and exit of EmCare into and from hospitals to estimate the causal effect that entry and exit of the firm have on out-of-network billing prevalence, physician pricing, and hospital behavior. We relied on three strategies to find hospitals where EmCare entered. First, we searched the firm’s website for press releases announcing new contracts. Second, we used LexisNexis and Google to search the popular press for news stories that announced when EmCare entered or exited a hospital ED. Third, we called all hospitals where we observed that EmCare might have had a contract on the basis of our map analysis and scrapes of their job hiring pages, spoke to the staff at the ED, and inquired about when EmCare entered into a contract with the hospital ED.¹⁴ All told, as we illustrate in Table 1, we identified 36 hospitals where EmCare entered from 2011 to 2015 and three where EmCare exited a contract.

A.  Descriptive Statistics on ED Physician Payments and Out-of-Network Billing Prevalence

Our final data set is composed of 8,913,120 ED episodes that occurred between January 1, 2011, and December 31, 2015 (see Table 2).¹⁵ This represents approximately $28 billion in emergency spending. The mean in-network ED physician payment across our sample period was $320.62 (266% of what the Medicare fee-for-service program paid for the same services; Table 2). The amount that ED physicians were paid increased as a percentage of Medicare over our time period. During this period, patient out-of-pocket costs for emergency care also steadily increased, and the mean total out-of-pocket cost for an emergency episode (combining the physician and facility component) in our data was $458.69. Over 99% of ED cases in our data occurred at an in-network hospital. Table A.1 includes descriptive statistics for our analytic sample of ED episodes.

At the mean in-network hospital in our data, 25.8% of patients treated in the ED were treated by an out-of-network ED physician (Table 2). The frequency that patients at in-network hospitals were treated by out-of-network ED physicians has declined over time from 28.6% in 2011 to 21.9% in 2015. However, this average masks significant heterogeneity in out-of-network billing prevalence across hospitals and is somewhat misleading. Figure 1 shows the distribution of out-of-network billing prevalence across hospitals in our data in 2015 and summary statistics for that year. It illustrates that out-of-network billing is highly concentrated in a small group of hospitals. As we illustrate, 50% of hospitals have out-of-network billing prevalence of approximately 1%. By contrast, the out-of-network billing prevalence for hospitals in the 75th percentile of the distribution of out-of-network billing prevalence was 28%, and 15% of hospitals have out-of-network prevalence of higher than 80%. This skewed distribution is evident in 2011, 2013, and 2015 (see fig. A.3).

B.  Cross-Sectional Analysis of Hospitals’ Out-of-Network Billing Prevalence

To assess the factors associated with the variation in hospitals’ out-of-network billing prevalence, we follow the approach of Finkelstein, Gentzkow, and Williams (2016) and run a least absolute shrinkage and selection operator (Lasso) regression on a range of hospital, local area, physician market, and hospital market characteristics (a complete list and descriptions of the variables that we include in our first-stage Lasso are available in app. 3). We also include an indicator variable for whether EmCare had a contract with the hospital. The Lasso method applies a penalizing parameter to the coefficient of the explanatory variables included in the regression. We use 10-fold cross validation to choose the penalizing parameter that minimizes the mean squared error. We use this Lasso procedure to select a set of variables that we include in a second stage where we determine their conditional correlations with hospitals’ out-of-network billing prevalence.

Figure 2 presents our conditional correlations between the variables selected using the Lasso regression and the share of patients per hospital who saw out-of-network physicians between 2011 and 2015 during an emergency. The results should not be interpreted causally. In this figure, we have scaled the continuous variables so that they have a mean of 0 and a standard deviation of 1. As a result, the point estimates on our continuous variables should be interpreted as the influence of a 1 standard deviation change in the dependent variable. As figure 2 shows, the presence of EmCare at a hospital is positively correlated with the hospital’s out-of-network billing prevalence. We also observe that areas with more physicians per capita have lower prevalence of out-of-network billing. Out-of-network billing is also less common at nonprofit hospitals, teaching hospitals, and hospitals with higher amounts of technology. There is more out-of-network billing in high-population counties and regions with more economic inequality.

C.  Causal Estimates of the Effect of EmCare Entry on Hospitals’ Out-of-Network Prevalence

Our cross-sectional results suggest that out-of-network billing is significantly higher at hospitals that outsource their ED to EmCare. In this section, we estimate the causal effect that the entry of EmCare had on the likelihood that patients were treated by out-of-network physicians working from in-network hospitals. To do so, we exploit evidence we collected from press releases, news stories on the firm’s website, articles in the popular press announcing the timing of EmCare contracts, and our phone calls to hospitals to identify the dates and locations where EmCare entered into and exited hospital ED staffing contracts. We then compare outcomes before and after EmCare entered and exited hospitals. In total, we analyze the entry of EmCare into 36 hospitals between 2011 and 2015 and their exit from three hospitals during the same period. We estimate entries and exits separately. We begin by showing trends in the raw data of hospitals where EmCare entered or exited a management contract. We follow up with a regression-based analysis. Crucially, we observe no difference in the pretrends of key outcome variables before EmCare entered or exited a hospital.

In our main analysis, we estimate a hospital fixed effects model with an indicator variable, EmCare_i,t, which takes a value of 1 on and after the date that EmCare entered a hospital and returns to 0 on the dates that the firm exited hospitals if the firm lost a contract. We also run a separate estimate for the three hospitals where EmCare loses a contract. Our estimation takes the form

We compare outcomes at hospitals where EmCare entered or exited to outcomes at three sets of control hospitals: (1) all hospitals nationally that did not have EDs managed by EmCare, (2) hospitals drawn from the same states where the hospitals that experienced entry were located but did not outsource their ED services to EmCare, and (3) hospitals that were not managed by EmCare that we matched to entry hospitals using propensity scores.¹⁶ One obvious concern with our identification strategy is that treated and untreated hospitals may have differences in their trends in out-of-network billing prevalence, physician pricing, or hospital behavior prior to the entry of EmCare. However, as we illustrate, when we plot the raw data from our treated hospitals, there do not appear to be any changes in behavior prior to the entry of those firms. Moreover, that we observe significant changes in hospital behavior when EmCare ends a contract with a hospital is also suggestive that we are estimating the effects of EmCare entry and not a hospital-specific phenomenon.

EmCare enters two types of hospitals (fig. A.4). The first group (27 hospitals) has out-of-network prevalence below 90% prior to EmCare entry (the mean out-of-network prevalence in these hospitals prior to entry was 11%). The second group (nine hospitals) has out-of-network prevalence of over 90% before EmCare enters (the average out-of-network prevalence in this group is 99%).

In figure 3, we present a smoothed average using a local polynomial regression of the monthly hospital-level out-of-network ED physician billing prevalence from 1 year before EmCare entered hospitals until 1 year after their entrance (panel A) and 1 year before EmCare exited hospitals until 1 year after their exit (panel B). We measure the date of entry with 6 months of noise on either side of the entry event (gray shading). In panel A of figure 3, the raw data show a clear increase in out-of-network billing prevalence at hospitals with previously low out-of-network prevalence after EmCare entered. Looking from 6 months prior to EmCare entry to 6 months after, the out-of-network billing prevalence at these hospitals that previously had low out-of-network prevalence increases discontinuously to nearly 100%. By contrast, panel B of figure 3 shows that there is a marked decrease in out-of-network billing at the three hospitals in our sample where EmCare exited a contract almost immediately after exit occurred.¹⁷

In Table 3, we show estimates of equation (1) and identify the impact of EmCare entry and exit on hospitals’ out-of-network billing prevalence. In column 1 of Table 3, we estimate the impact of the entry of EmCare into hospitals with previously low out-of-network prevalence (those with out-of-network prevalence below 90% prior to EmCare entry). These results mirror what we observe in the raw data. We observe that the entry of EmCare into these hospitals raised out-of-network prevalence by 82.8 percentage points. In column 2, we focus on changes in out-of-network billing prevalence at hospitals that EmCare entered that previously had high out-of-network billing prevalence. After EmCare entered, there is no statistically significant change in the likelihood that a patient was treated by an out-of-network physician. This is a mechanical effect, since there was no scope for out-of-network billing to increase at these facilities. In column 3, we estimate the effect of the exit of EmCare on hospitals’ out-of-network billing prevalence. It is important to note that hospitals may have made a decision to end their contract with EmCare in order to eliminate out-of-network billing from their facility. Nevertheless, we observe that after EmCare exited a hospital, the prevalence of out-of-network billing decreased by 76.5 percentage points.¹⁸ In table A.2, we show that these results are robust to using alternative control groups.

D.  The Impact of Out-of-Network Strategies on Payment Rates

These results suggest that EmCare does not negotiate with insurers and instead utilizes its outside option and seeks to collect its charges. In panel A of figure 4, we show that after entry, EmCare raised its charges significantly. In column 1 of panel A in Table 4, we quantify these changes and show that after EmCare entered, they more than doubled physician charges, raising them by $480.13 on average. Our data contributor paid most of physicians’ out-of-network bills. As a result, after EmCare entered, we observe that the insurer payments to ED physicians increased by $391.89 (117%).

These changes also exposed patients to increased cost sharing and financial risk. Because patients typically have out-of-pocket costs that are set via coinsurance that pays a fixed percentage of the total cost of care, patient payments (e.g., cost-sharing payments) to ED physicians increased by $46.32 (92%). Collectively, we observe that the total payments to ED physicians increased by $438.20 per case after EmCare entered a hospital. This is a 114% increase in ED physician payments, and these changes occurred after EmCare entered a hospital (fig. 4). Notably, as we show in column 7, we do not observe a decrease in patient volume after EmCare enters a hospital.

While our data contributor covered most of physicians’ out-of-network charges, many insurers simply pay out-of-network providers at median in-network rates. When this occurs, physicians can bill patients for the difference between their charges and that payment (so-called balance billing). To calculate patients’ potential balance bills, we create a potential balance billing measure, which is the difference between the physician charge for the case and what would be the providers’ median in-network payment for the case (219% of Medicare rates in our data). In column 5 of Table 4, we show that the entry of EmCare raised patients’ potential balance bills (if insurers paid only median in-network rates) by $457.21 to a total of $681.54. Bills of this magnitude would be financially devastating to a large share of the population (Board of Governors of the Federal Reserve System 2016). As we illustrate in panel E of figure 4, this change also occurred immediately after EmCare entered hospitals.

The increase in physician payments was generated by price increases and substitution to higher acuity (and more generously reimbursed) CPT codes. In table A.3, we show that EmCare physicians increased their charges and total payments for all but the lowest acuity emergency CPT code. In addition to increasing their charges, EmCare physicians also increased the rate they coded ED physician services using the highest-intensity CPT code by 11.4 percentage points (47%; col. 6 in Table 4) and increased the relative value units of physician care they provided by 9% (col. 1 in table A.3). This increase in the use of high-severity coding (and the increase in physician charges) occurred after the firm entered (panel F of fig. 4).¹⁹ Moreover, as we discuss and illustrate later, these changes in coding occurred for patients with both high and low historical medical spending and high and low medical risk. We further discuss the impact of the entry of physician management companies on hospitals’ case mix in section V.F.

It is striking that at the three hospitals where we observe EmCare exit, there is a reversal in all these key outcomes after the firm’s staffing contracts ended, including a reduction in the frequency of the use of CPT code 99285. As we illustrate graphically in figure 5, immediately after EmCare exits, there is a discontinuous drop in physician charges, total payments, insurer payments, patient cost sharing, and high-severity coding. Estimates of equation (1) in panel B of Table 4 show that these changes are quite large and statistically significant. They show that relative to the quarter before exit occurred, total ED physician charges decreased by $645.76 (50%), total payments decreased by $701.68 (62%), patient cost sharing decreased by $68.68 (63%), and use of the highest severity CPT code decreased by 10.3 percentage points. We also observe a modest increase in the number of patients treated per year.

In section III.C, we argued that having the ability to go out of network without seeing a sizeable reduction in the number of patients that they treat gave ED physicians a stronger outside option in negotiations with insurers. We argued that this stronger outside option would allow them to negotiate higher in-network payments. In Table 5, we show the average in-network payments in our data made to internists for performing standard office visits and to orthopedists for performing hip replacements. We observe that, on average, internists are paid 158% of Medicare rates (col. 1) and orthopedists are paid 178% of Medicare rates (col. 2). By contrast, the average in-network ED physician in our data is paid 266% of Medicare rates (col. 3). We posited that firms that could credibly threaten to go out of network could negotiate higher payments. Indeed, we observe that in the cross section, the mean payment in our data to EmCare ED physicians (who, for the most part, do not participate in networks) is 542% of Medicare rates (col. 5).²⁰

E.  Transfers to Hospitals to Permit Out-of-Network Billing

When physicians bill out of network, it creates costs for the hospitals where they work. We hypothesized that physician management firms that use out-of-network billing as a strategy would have to offer transfers to hospitals to offset these costs. There are four categories of benefits that hospitals could receive from allowing ED physicians to bill out of network from inside their facilities. First, by allowing physicians to bill out of network, hospitals could receive a discount in the fees that they must pay a physician staffing company to manage their ED. As we described in section II.C, the transcript from administrators at Glen Rose Medical Center discussing out-of-network billing (presented in app. 1) revealed that EmCare was willing to offer a $200,000 per year concession in staffing fees to the facility in exchange for allowing them to bill out of network.

Second, hospitals, per our findings, can get additional revenue when out-of-network physicians alter their practice styles in ways that increase hospital activity (e.g., raising hospital imaging rates or admitting patients from the ED to the hospital at higher frequency). Our results presented in panel A of Table 6 are consistent with our predictions. We estimate equation (1) and find that after EmCare entered a hospital and began billing out of network for ED services, facility charges at the hospitals where they worked increased by $1,270.15 (17%), and facilities’ total payments increased by $220.11 (8%). As we illustrate in Table 6, this increase in facility payments was driven in part by a 1.1 percentage point (4%) increase in the probability that a patient received an imaging procedure (col. 5) and a 1.7 percentage point (22%) increase in the likelihood that a patient was admitted to the hospital.²¹ As we illustrate in panel F of figure A.6, this increase in admissions is visible in the raw data and occurred after EmCare entered a hospital. As we illustrate in panel B of Table 6 and figure A.7, it is striking that almost all these changes in facility activity reverse in the three cases where we observe EmCare exit from a hospital.

The modal ED treats approximately 20,000 privately insured patients per year, so the $220.11 increase in hospital payments per case estimated in Table 6 would generate an annual increase in revenue of approximately $4.4 million per hospital (20,000 × $220.21; HealthLeaders Media 2016). Wilson and Cutler (2014) estimated that privately insured patients have a profit margin for hospitals of 39.6%. Taken together, this suggests that each hospital that outsourced its ED to EmCare should make at least an additional $1.7 million per year ($4.4 million × 0.396) in profit from the changes in the way out-of-network physicians practice. Note that since this estimate does not capture profits from changes in medical care for Medicare or Medicaid beneficiaries, our estimate represents a lower bound on the returns a hospital could receive from clinical practice changes made by EmCare physicians.

Third, EmCare is increasingly entering into joint ventures with hospitals where hospitals can share in the profits of EmCare physicians (Luthi 2019). For example, according to a 2013 Deutsche Bank report, the EmCare joint venture with the Hospital Corporation of America “offers 50/50 profit sharing above a certain margin threshold, which we believe is in the 13% range” (Lehrich, Kalenderian, and Nentin 2013, 15). We found that, on average, EmCare physicians generated an additional $438.20 in physician payments per case. If EmCare were to treat 20,000 privately insured patients per year in a hospital, this would generate $8.8 million in revenue annually across all privately insured patients ($438.20 × 20,000). If we made a conservative assumption that hospitals in a joint venture with EmCare made a 1% profit on this physician revenue, this would generate an additional $87,640 for the hospital each year ($8.8 million × 0.01). Again, this estimate is a lower bound, since it does not reflect profits on Medicare beneficiaries.

Finally, EmCare could potentially bring efficiency gains to the hospitals where they gain contracts via lowering the costs of running EDs. EmCare touts that they manage the staffing, physician recruiting, and billing (https://www.emcare.com/). These could result in additional savings to hospitals.

These estimates suggest that outsourcing emergency services to EmCare conservatively brings each hospital approximately $2 million per year in additional annual profits ($200,000 + $1.7 million + $87,640). For reference, the average hospital in the United States in 2012 had a revenue of $164.3 million per year and made a profit of $12.9 million (Becker’s Hospital Review 2014). As a result, ED outsourcing would increase the average hospital’s profits by 15.5% ($2 million/$12.9 million). These gains must be offset against the costs of allowing a firm like EmCare to work from inside their facility. As we discussed in section III.B, these costs would include any reputational harm that the hospital would incur if it were discovered that they were allowing out-of-network billing at their facility and their inherent dislike of exposing their patients to financial risk.

Ultimately, a hospital would outsource their ED services to EmCare if the additional $2 million in profit that they received each year from outsourcing exceeded the sum of the reputational costs that they incurred from contracting with the firm and the cost of their inherent distaste for exposing their patients to risk. As a result, we would expect for-profit hospitals—which likely put a higher weight on profits than would government or nonprofit facilities—to be more likely to contract with EmCare. In Table 7, we present the characteristics of hospitals in our sample that did and did not contract with EmCare. We find that across all hospitals that meet our sample restrictions, 61% are nonprofit, 19% are for-profit, and 20% are government owned. Consistent with our predictions, 45% of hospitals where EmCare has a contract are for-profit facilities. Hospitals in areas with lower numbers of physicians per capita are also more likely to contract with EmCare.²²

F.  Robustness Checks

It is possible that the entry of EmCare led to subsequent changes in the case mix of patients that the hospitals treat. Indeed, EmCare advertises that a benefit of their service is that they shorten ED waiting times (Cantlupe 2013). With shorter waiting times, hospitals could potentially attract healthier patients who would have otherwise received treatment at urgent care centers. Likewise, on EmCare’s website, EmCare has also highlighted its excellence in improving the treatment of complex cases, such as stroke care (EmCare 2014). To the extent that this improves a hospital’s reputation, advertising and improvements in quality could allow that hospital to attract more complex patients. Any changes in the case mix of hospitals that EmCare entered could explain why, after the firm entered hospitals, the rates of hospital admissions, the rates of imaging tests, and the rates at which physicians coded for the most intensive services increased. Finally, EmCare could face an incentive to attract or make its patients appear riskier in order to increase the chances of the hospital where they are working becoming a trauma center. Extolling the financial benefits of being designated a trauma center, an EmCare executive wrote, “Medicare offers disproportionate funding to hospitals with trauma centers. Additionally, the Patient Protection and Affordable Care Act re-institute the trauma stabilization act, which will unlock some government funding for the development of trauma programs. On top of that, as a designated trauma center, hospitals can actually bill and collect for certain activation fees that are paid by both Medicare and private insurance companies. Those fees can be very meaningful, sometimes more than funding the trauma program itself” (Josephs 2013). As the Trauma Center Association of America (2019) notes, one criterion to becoming a trauma center is having minimum numbers of high-severity patients.

In table A.10, we analyze the impact that the entry of EmCare had on the case mix of patients that hospitals treat. We find evidence that after EmCare entered a hospital, the hospital attracted a sicker mix of patients. In columns 1 and 2, we show that after EmCare entered a hospital, the 6-month historical spending of the hospital’s patients increased by $820.39 (14%), and the 12-month historical spending increased by $1,232.60 (11%). We also find that after the entry of EmCare into a hospital, the 6-month Charlson score of patients who attend the ED increased by 7%, and the 12-month Charlson scores increased by 7.5%. In figure A.8, we show the average Charlson comorbidity score and 6-month historical spending levels of patients by month at hospitals where EmCare entered. There is no evidence of immediate changes in these outcomes after a change in management.

Crucially, however, we find the same changes in physician behavior and hospital activity at EmCare facilities appearing across patients irrespective of their health status. Thus, even holding patient severity constant, we still see an increase in quantity of care delivered after EmCare enters a hospital. In table A.11, we estimate equation (1) using several different sample restrictions and sets of controls for the health of the patients. We focus on the impact that the entry of EmCare had on the frequency that physicians coded using the CPT code for the most intensive emergency. We find that even among patients with low historical spending and no comorbidities, there was a substantial increase in the rate at which they had episodes that included physician claims coded using the highest-intensity CPT code. In column 1, we estimate equation (1) with no patient controls; in column 2, we reestimate equation (1) controlling for patients’ age, sex, and race; and in column 3, we control for patients’ age, sex, race, and their Charlson comorbidity score. Across all three estimates, the point estimate on the impact of entry on the rate of using the highest-intensity CPT code for emergency physician visits is consistent and ranges from 0.114 to 0.116. In column 4, we estimate equation (1) and limit our analysis to patients throughout our sample who have a Charlson comorbidity score of 0 (e.g., patients who have no comorbidities). In column 5, we estimate equation (1) and limit our analysis to patients throughout our data who have a nonzero Charlson score. The point estimates in columns 4 and 5 illustrate that whether or not they had comorbidities, patients were almost equally more likely to have physician visits coded using the CPT code for the most intensive emergency after EmCare entered a hospital. Likewise, in columns 6–8, we estimate equation (1) on the samples of patients in the lower third ($0–$279.60), the middle third ($279.61–$2,033.86), and the top third ($2,033.87–$115,499.30) of the distribution of historical 6-month patient health spending. Across all three subsamples, the entry of EmCare led to an increase in the rate at which patients had physician claims coded using the CPT code for the most severe emergency.

In table A.12, we repeat this analysis and examine the impact of the entry of EmCare on facility spending across different samples of the data (the sum of the allowed amounts on the physician claims). We see that there was increased facility spending across patients with and without comorbidities and with high and low historical spending. Likewise, controlling for patients’ comorbidities does little to alter the impact of the entry of EmCare on facility spending. In table A.13, we see broadly robust findings for imaging studies. After the entry of EmCare into a hospital, patients with no comorbidities are 4% more likely to receive an imaging study.

Finally, in table A.14, we analyze whether we observe higher hospital admission rates for patients with low historical spending and no comorbidities following the entry of EmCare. In column 4, we find that after EmCare entered a hospital, patients with no comorbidities were 20% more likely to be admitted to the hospital. In column 6, we find that patients with low historical spending (e.g., less than $279.60 in the previous 6 months) were 16% more likely to be admitted to the hospital after EmCare took over management of the hospital ED.

G.  Generalizability of Our Data

Our data come from a single insurer that operates across all 50 states. Our data capture nearly $28 billion in economic activity; thus, the sample is interesting to study regardless of generalizability. However, to gauge the generalizability of our results, we compare the out-of-network prevalence we observe to out-of-network prevalence presented in Garmon and Chartock (2017), the only other study that examines the out-of-network prevalence nationally.²³ Garmon and Chartock used 2007–14 data from the Truven Health MarketScan database. They focus on whether patients at in-network hospitals saw any out-of-network physicians. This measure is slightly different from our measure; we focus on the network participation of the primary physician in ED cases at in-network hospitals. Garmon and Chartock (2017) found that emergency cases that had an admission had out-of-network bills in one in five cases; outpatient emergency cases had out-of-network bills in 14% of cases. These results are similar to our results. Garmon and Chartock (2017) found out-of-network prevalence for admitted patients in Florida, Texas, and New York of 37%, 34%, and 35%, respectively. For those same states, when we focus on patients with an admission, we observe out-of-network prevalence during the 2011–15 period of 24.8%, 46.3%, and 16.1%, respectively. They also found, as we do, that out-of-network prevalence decreases over time.

A.  Policy Goals and Scaling the Effect of Policies to Address Out-of-Network Billing

A successful out-of-network policy should achieve two aims. First, a policy should protect consumers from large unexpected bills from out-of-network ED physicians whom the consumers could not reasonably avoid. Second, a successful policy should establish an environment in which the price that out-of-network ED physicians are paid for their services is either competitively determined or as close to the competitively determined price as possible. Addressing this currently missing price will protect consumers from unavoidable out-of-network fees, settle disputes between physicians and insurers over their bills, and influence in-network payments by determining ED physicians’ outside option in negotiations with private insurers.

Constraining ED physicians’ outside option could have a significant effect on total health spending via reducing their in-network payments. To produce a back-of-the-envelope estimate of the potential savings from addressing out-of-network ED billing, consider the nationwide average in-network payment rates that orthopedic surgeons negotiate with the insurer that supplied our data. Orthopedic surgeons form an interesting comparison group because, according to a recent survey, they have the highest salaries among physicians in the United States (Grisham 2017). However, whereas the average in-network ED physician payment in our data was 266% of the Medicare payment rates (and the average out-of-network payment was 637% of the Medicare payment rates), the average in-network payments to orthopedic surgeons for performing hip replacements during our sample period was 178% of the Medicare payment rates. If we assumed that our policy proposal would generate competition that lowered ED physicians’ in-network payment levels to approximate the in-network payment rate of orthopedic surgeons in our data (178% of the Medicare payment rates), this would lower total ED physician spending by 46%. If we assume that private spending is one-third of total health spending in the United States and that ED physicians account for approximately 1% of total private spending, a reasonable back-of-the-envelope calculation would suggest that addressing this issue would produce savings in the range of $5 billion annually.²⁴ While these estimates ignore general equilibrium effects, they give a scale of the savings possible via addressing out-of-network billing by ED physicians.

B.  Existing State and Federal Policies to Address Out-of-Network Billing

At present, 21 states have some laws focused on out-of-network billing, and six of those 21 have comprehensive policies that both protect consumers and include a process to determine payments from insurers to out-of-network providers (Lucia, Hoadley, and Williams 2017). New York was the first state to systematically study the issue and introduce comprehensive legislation to address out-of-network billing (New York State Department of Financial Services 2012). In a 2012 report, the state’s Department of Financial Services stated that they were receiving a growing number of complaints about out-of-network bills submitted by ED physicians and other hospital-based specialists. The report drew attention to bad actors that used out-of-network billing as a deliberate profit-seeking strategy. The New York report also highlighted key areas where policy and regulation could protect consumers and raised the need for future legislation.²⁵ In 2014, as we describe below, New York State passed a law that included patient protections and an arbitration mechanism to settle disputes between insurers and providers.

Most states’ (including New York’s) surprise billing laws include a hold harmless provision to protect patients from financial risks (Lucia, Hoadley, and Williams 2017). These hold harmless provisions stipulate that patients cannot be charged more than their usual in-network cost sharing during emergencies if they see an out-of-network provider who is working at an in-network facility. However, only nine of the 21 states with hold harmless provisions restrict providers from balance billing patients. Thus, while patients who saw an out-of-network provider would not be subject to higher cost-sharing rates, they could still be exposed to significant financial risk if physicians acted to collect the balance of their bill from them directly.

The harder policy problem for the states is choosing the missing price when there is no contract between physicians and insurers. By 2017, only California, Connecticut, Florida, Illinois, Maryland, and New York introduced state-specific methods for determining how insurers should pay out-of-network ED physicians who treat patients at in-network hospitals. In California, Connecticut, and Maryland, the missing price is determined via regulation. For example, in California, out-of-network providers are paid the greater of 125% of Medicare rates or average commercial rates in their area (California Legislative Information 2016). However, it is unlikely that a regulated price of this sort will match the true competitive price for any given transaction. As soon as the regulated price set by states differs from the market price, either the insurer or the physician will take advantage of a regulated price that favors them. If the regulated payment for providers’ out-of-network bills is greater than the expected in-network price, ED physicians will be reluctant to join networks. Likewise, if the regulated payment is below expected in-network rates, insurers will not want to form networks (see, e.g., recent experience in California described in Duffy 2019).

In Florida, Illinois, and New York, in addition to prohibiting patients from being balance billed, the states introduced an arbitration process to determine insurer payments in the event that an insurer and provider cannot reach a resolution on a payment amount in cases when an ED physician is out of network. Under the New York law, when a patient is seen out of network, the insurer makes its payment to the provider. If the out-of-network provider does not accept the payer’s offer, the provider can initiate an independent dispute resolution process. The independent dispute resolution process is judged by practicing physicians who use baseball rules arbitration. Each party submits a bid, and an arbitrator determines whether the provider will be paid either the amount requested by the provider or the amount offered by the insurer. Ultimately, this policy disadvantages providers that seek unreasonably high charges and punishes insurers that seek to make unreasonably low payments. The law also encourages physicians and payers to negotiate independently and avoid arbitration. Technically, the law applies only to fully insured insurance products, as states cannot regulate ASO plans (which account for the majority of privately insured products in the United States; Kaiser Family Foundation 2017). However, because most providers are unaware of a patient’s plan funding, their billed amount is likely chosen to reflect the possibility of arbitration.

This type of arbitration process shifts the outside option for physicians when they negotiate their in-network payments. Under the New York law, physicians cannot balance bill patients or hope to collect 100% of their charges from insurers. As a result, the 2014 New York law should both reduce the prevalence of out-of-network billing (since it is no longer as profitable a strategy) and impact in-network payments (via lowering ED physicians disagreement payoff).

Unfortunately, states’ policies (including New York’s) can apply only to the 40% of commercially insured individuals in the United States who are enrolled in fully insured employer-sponsored health plans (Kaiser Family Foundation 2017). Sixty percent of individuals in the United States with commercial insurance are enrolled in plans offered by firms that self-insure. Because of the Employee Retirement Income Security Act of 1974, state-based protections for out-of-network policies do not apply to these enrollees. At the federal level, protections for consumers are limited. The Affordable Care Act amended Section 2719A of the Public Health Service Act and required health plans to cover emergency services without prior authorization and irrespective of network status.²⁶ Unfortunately, the provision still allows providers to balance bill patients for the difference between their charges and the insurer payment.

C.  Analyzing the Impact of New York State’s Law

We use our claims data to test the impact of New York State’s efforts to reduce surprise billing on ED physicians’ out-of-network billing prevalence, ED physicians’ in-network payment levels, physician charges, and facility payments. As table A.15 shows, our data include 323,936 ED episodes delivered at New York hospitals between 2011 and 2015, which captures approximately $1 billion in emergency health care spending. Our data capture changes before and after their report on March 7, 2012; the passage of their law on March 31, 2014; and implementation of the law on the March 31, 2015.²⁷

To identify the impact of New York State’s efforts to address out-of-network billing, we compare key outcomes in New York before and after their policy push with outcomes in 32 control states that do not regulate the missing price when there is no contract between physicians and insurers and that have over 5,000 ED visits per year in our data.²⁸ We show how key outcomes in New York State changed during the period from March 7, 2012, when the Department of Financial Services report was published, and after March 31, 2015, when the surprise billing protections took force in the state.

We begin by producing event study–style difference-in-difference estimates that take the form

In addition, we estimate a more traditional difference-in-difference estimator that takes the form

We take two approaches to calculating the precision of our estimates of both equations (2) and (3). First, we present standard errors that are clustered around hospitals. However, because we are comparing treatment effects in one state (New York) to outcomes in 32 control states, there are potential concerns that traditional clustering methods would be inappropriate for estimating precision with so few degrees of freedom and only a single treated group (Donald and Lang 2007). As a result, in our second approach, we follow Buchmueller, DiNardo, and Valletta (2011) and implement a permutation test in the spirit of Fisher (1935). To do so, we compare the treatment effects we observe in New York State to 32 placebo treatment effects we observe when we estimate equation (2) independently, and we sequentially use each of those 32 control states as the placebo treated state instead of New York. We then present how our treatment effects for New York State compare with the distribution of placebo treatment effects that we observe when the control states are used as the treatment group.

D.  The Impact of New York State’s Out-of-Network Billing Laws

Figure 6 presents estimates of equation (2) and shows quarter by quarter changes in the prevalence of out-of-network physicians, in-network ED payments, ED physician charges, and facility payments in New York and in the 32 control states from 2011–15. As we illustrate in panel A of figure 6, the prevalence of out-of-network ED physicians in New York and control states followed similar trends before the 2014 passage of New York’s out-of-network protection law (although there was a modest reduction in out-of-network prevalence in New York following the release of the Department of Financial Services report on March 7, 2012). However, almost immediately after the law was passed (and before the law’s implementation date), there was a marked reduction in the prevalence of out-of-network ED physicians in New York. We show the point estimates underlying figure 6 in table A.16. They imply that the prevalence of out-of-network ED physicians in New York in the fourth quarter of 2015 was 11.7 percentage points lower than it was in the first quarter of 2011.

Figure 7 shows the distribution of out-of-network prevalence across hospitals in 2011 and 2015. The out-of-network prevalence in New York in 2011 was 25.0%. Four years later, the rate was 6.4%. As the figure illustrates, the reduction in out-of-network providers was driven by reductions in out-of-network prevalence across nearly all hospitals, including those that previously had high prevalence of out-of-network billing.

Column 1 in Table 8 presents our long-difference estimates of equation (3). The point estimates imply that the introduction of the package of reforms in New York State reduced the out-of-network prevalence from 2011 to 2015 by 12.8 percentage points off a base out-of-network prevalence in New York of 14.5% (implying a 88% reduction in out-of-network prevalence). These estimates are precisely estimated using both of our approaches to infer statistical significance. We present estimates from alternative functional forms of our difference-in-difference estimator in table A.17. Our main result is robust across alternative specifications.²⁹

As we illustrate in panel B of figure 6, in-network ED physician payments in New York and control states were similar until the second quarter of 2013, when there was a large and statistically significant reduction in payments in New York. This reduction occurred two quarters before the state’s arbitration law was passed and a year after the Department of Financial Services’ report. It is likely that the 2012 report, in highlighting bad actors and foreshadowing future policy action, gave insurers stronger bargaining leverage with providers. By the fourth quarter of 2015, per estimates presented in table A.16, in-network ED physician payments in New York were $52.93 lower than they were in the first quarter of 2011, when they were 260.22 (a 20% reduction).

Our long-difference estimates of equation (3) on ED physicians’ in-network payments are presented in column 2 of Table 8. They suggest that the New York reforms lowered in-network physician payments by $44.97 (15%). These estimates are precisely estimated using both of our approaches for identifying standard errors. This reduction in payments is consistent with predictions that the law would lower ED physicians’ disagreement payoff in negotiations with insurers over in-network payments, which should lower the prices reached in the negotiations. It is also notable that this effect was observable in New York, where—in addition to engaging in arbitration—insurers have to identify how their proposed out-of-network payments are scaled relative to usual and customary payments (the 80th percentile of physician charges).³⁰

We also analyze the impact of the New York reforms on ED physicians’ charges and facility payments in our sample. Our results are presented in panels C and D of figure 6 and columns 3 and 4 of Table 8. We do not observe any statistically significant increases in these outcomes throughout our sample period.

Evidence from New York State suggests that introducing a hold harmless provision and arbitration over insurers’ payments to out-of-network physicians can lower the frequency of out-of-network billing and the level of physicians’ in-network payments. This result shows how changing physicians’ outside option in negotiations alters their ultimate negotiated payment. Nevertheless, the New York State law is administratively complex, costly to administer, and has the potential to be gamed. If patients receive a surprise out-of-network bill and are charged out-of-network rates, they must be aware that the protections exist and fill out the form included in appendix 4. Likewise, the state has to fund and administer the arbitration process in perpetuity. Moreover, because states cannot regulate ASO products, the New York protections offer formal protection only to individuals covered by fully insured insurance products. And perhaps the biggest challenge with this policy is that, because parties must show how their bids relate to usual and customary payments (e.g., charges, which physicians set themselves), there is scope in the long-run that physicians could game the system and increase their charges over time, which would improve their outcome in arbitration. As a result, the long-run impact of arbitration may differ from what we found in the short run.³¹