Context: Hospital cost shifting—charging private payers more in response to shortfalls in public payments—has long been part of the debate over health care policy. Despite the abundance of theoretical and empirical literature on the subject, it has not been critically reviewed and interpreted since Morrisey did so nearly fifteen years ago. Much has changed since then, in both empirical technique and the health care landscape. This article examines the theoretical and empirical literature on cost shifting since 1996, synthesizes the predominant findings, suggests their implications for the future of health care costs, and puts them in the current policy context.
Methods: The relevant literature was identified by database search. Papers describing policies were considered first, since policy shapes the health care market in which cost shifting may or may not occur. Theoretical works were examined second, as theory provides hypotheses and structure for empirical work. The empirical literature was analyzed last in the context of the policy environment and in light of theoretical implications for appropriate econometric specification.
Findings: Most of the analyses and commentary based on descriptive, industrywide hospital payment-to-cost margins by payer provide a false impression that cost shifting is a large and pervasive phenomenon. More careful theoretical and empirical examinations suggest that cost shifting can and has occurred, but usually at a relatively low rate. Margin changes also are strongly influenced by the evolution of hospital and health plan market structures and changes in underlying costs.
Conclusions: Policymakers should view with a degree of skepticism most hospital and insurance industry claims of inevitable, large-scale cost shifting. Although some cost shifting may result from changes in public payment policy, it is just one of many possible effects. Moreover, changes in the balance of market power between hospitals and health care plans also significantly affect private prices. Since they may increase hospitals’ market power, provisions of the new health reform law that may encourage greater provider integration and consolidation should be implemented with caution.
Keywords: Cost shifting, Medicare, hospital charges, health insurance, health policy
The definition, existence, and extent of hospital “cost shifting” are points of debate among participants and stakeholders in discussions of health care policy and reform. The academic literature defines the term precisely and characterizes the range of its effect. Even though that literature has grown considerably in recent years, not since the mid-1990s has it been systematically reviewed and summarized (Morrisey 1993. 1994. 1996 ; see also Coulam and Gaumer 1991 ). In this article, I update those older reviews, summarize the relevant features of and changes to health care policy, and place the results in today's policy context.
That hospitals charge different payers (health plans and government programs) different amounts for the same service even at the same time is a phenomenon well known to economists as price discrimination (Reinhardt 2006 ). That hospitals charge one payer more because it received less (relative to costs or trend) from another also is widely believed. This is a dynamic, causal process that I call cost shifting. following Morrisey (1993. 1994. 1996) and Ginsburg (2003). among others. Price discrimination and cost shifting are related but different notions. The first depends on differences in market power, the ability to profitably charge one payer more than another but with no causal connection between the two prices charged. The second has a direct connection between prices charged. In cost shifting, if one payer (Medicare, say) pays less relative to costs, another (a private insurer, say) will necessarily pay more. Whereas cost shifting implies price discrimination, price discrimination does not imply that cost shifting has occurred or, if it has, at what rate (i.e. how much one payer's price changed relative to that of another).
That hospitals shift their costs among payers is intuitively appealing. Public payments—from Medicare or Medicaid—go down (again, relative to cost or trend, qualifiers that I will omit from now on), and as a consequence, private payments go up, taking health insurance premiums along with them. Karen Ignagni, the president and CEO of America's Health Insurance Plans (AHIP), described cost shifting as follows: “If you clamp down on one side of a balloon, the other side just gets bigger” (Sasseen and Arnst 2009. http://www.businessweek.com/magazine/content/09_41/b4150026723556.htm ). According to Dobson, DaVanzo, and Sen, it is a simple hydraulic effect: “As some pay less, others must pay more” (2006, 23).
Is this intuition correct? Are costs immutable and simply shifted from one payer (that pays less) to another (that necessarily pays more)? If providers shift costs, by how much do they do so? When casually expressed or generously interpreted, the idea of cost shifting conjures up a dollar-for-dollar trade-off; that is, one dollar less paid by Medicare or Medicaid results in one dollar more charged to private payers. At least one recent health insurance industry–funded report (PWC 2009 ) assumed this level of cost shifting.
Some health care policy stakeholders have an interest in convincing policymakers that cost shifting is both inevitable and large. Continuing the cost-shifting assumption (and ignoring a countervailing assumption of profit maximization, to which I will return later), if public payments are relatively less generous, then hospitals will raise private prices more than they would otherwise. In turn, insurers’ premiums for policies and self-insured firms’ health care costs would rise more quickly, making the private purchase and sponsorship of health care coverage relatively more difficult for consumers and firms, respectively. Thus, convincing policymakers to be concerned about cost shifting also is in the interest of the privately insured, employers, and the insurance and hospital industries, all of whom benefit from (or, at least, are not harmed by) higher public payments. Individuals and firms would rather not spend more for care, which cost shifting implies; insurance companies do not want to charge higher premiums; and hospitals would prefer higher public payments for their services.
Of course, if costs could be shifted significantly, public payment policy would have little leverage on total health care costs. In that case, cost shifting would amount to price adjustments that would force private payers to subsidize most of the public programs’ payment shortfalls. Thus, the extent of cost shifting is important. Is it dollar-for-dollar, or is it less? If less, by how much? In other words, how much leverage does public payment policy have on total health care costs? How much does it influence private prices and premiums? For how much cross-subsidization does it account? The literature, as I will show, answers these questions.
The literature offers several broad conclusions that modify Morrisey's (1993. 1994. 1996) main finding that cost shifting was small to nonexistent. First, based on theoretical considerations alone, the conditions necessary for cost shifting are possible but circumscribed. Furthermore, if there is cost shifting, it cannot always and forever be large and persistent. Second, the empirical literature finds that to the extent it has occurred at all, cost shifting usually is at a low rate. Instead, the vast majority of public payers’ shortfalls are accommodated by cost cutting. not cost shifting. Third, private payment-to-cost ratios are influenced by many factors other than public payment rates. Thus, changes in the former cannot, and should not, always or fully be explained by changes in the latter. Fourth, the rate of cost shifting largely depends on the intensity of price competition in the private market for hospital services, that is, the relative market power of hospitals and health care plans. Indeed, one cannot assume that estimates of cost shifting from one era or one market will apply at another time or in another place. Fifth and finally, private and public prices and margins can influence each other. In other words, the direction of causality between private and public payment levels goes both ways: they are jointly determined.
Besides reviewing the literature in this article, I provide a framework informed by theory for empirical specifications of hospital cost-shifting analysis. In that framework I identify the control factors and estimation techniques required to obtain unbiased cost-shifting estimates. (Additional technical detail pertaining to this framework is found in Frakt 2010a .) I also examine each empirical study in light of this framework. Although no study (on any subject) is perfect, some are stronger than others. The stronger studies that I identify provide the most credible estimates of hospital cost shifting and indicate how the phenomenon varies with market structure. Ultimately, by considering the full body of work—imperfect as each individual effort may be—I have been able to draw some robust conclusions.
In addition to public payer shortfalls, providing care to the uninsured may lead to hospital cost shifting and affect private premiums, although the estimates vary. Families USA (2005) estimated that private insurance premiums were about 10 percent higher in 2005 due to the use of health services by the uninsured, whereas both Kessler (2007) and Hadley and colleagues (2008) found less than a 2 percent effect. The remainder of my article focuses on hospital cost shifting from Medicare and Medicaid to private payers and does not cover what may be caused by the uninsured.
For decades, concerns about cost shifting have played a role in the consideration of hospital payment policy. According to Starr (1982, 388), in the 1970s, “commercial insurance companies worried that if the government tried to solve its fiscal problems simply by tightening up cost-based reimbursement, the hospitals might simply shift the costs to patients who pay charges.” A 1992 report by the Medicare Prospective Payment Assessment Commission (ProPAC) asserted that hospitals could recoup from private payers underpayments by Medicare (ProPAC 1992 ). If that were so, hospitals would not need to fear inadequate government payments. Yet somewhat paradoxically, around the same time, hospitals used the cost-shifting argument to call for higher public payment rates (AHA 1989 ). More recently, during the debate preceding passage of the new health reform law—the Patient Protection and Affordable Care Act (PPACA)—two insurance and hospital industry–funded studies (Fox and Pickering 2008 ; PWC 2009 ) and one peer-reviewed publication (Dobson et al. 2009 ) reasserted that half to all public payment shortfalls were shifted to private payers.
The issue of cost shifting is certain to arise again in the near future. Even though cost shifting was debated during consideration of the PPACA, public payment policy is not settled, nor will it ever be. The new health reform law includes many provisions designed to reduce the rate of growth of public-sector health care spending. For instance, among the law's provisions, annual updates in payments for Medicare hospital services will be reduced; payments for them will be based partly on quality measures; and payments for preventable hospital readmissions and hospital-acquired infections will be lowered (Davis et al. 2010 ; Kaiser Family Foundation 2010 ). In aggregate and over the ten years between 2010 and 2019, the Congressional Budget Office (CBO) estimated that the savings from lower Medicare hospital payments would be $113 billion (CBO 2010a ).
In addition, Medicaid eligibility will expand in 2014 to all individuals with incomes below 133 percent of the federal poverty level. The CBO has estimated that by 2019, Medicaid enrollment will grow by 16 million individuals (CBO 2010b ). Because some of these new Medicaid beneficiaries would otherwise have been covered by private plans (a crowd-out effect; see Pizer, Frakt, and Iezzoni 2011 ), the lower Medicaid payments relative to private rates may increase incentives to shift costs. Conversely, to the extent that the expansion of Medicaid—as well as the equally large (CBO 2010b ) expansion of private coverage encouraged by the PPACA's individual mandate and insurance market reforms—reduces the costs of uninsurance and uncompensated care, the law may also reduce hospitals’ need to shift costs. Nevertheless, if past experience is any guide, when some of the PPACA's provisions are implemented, they are likely to be challenged by the hospital and insurance industries using cost-shifting arguments.
Much of the commentary in the literature pertaining to public and private payments to hospitals and their relationship refers to time series such as those depicted in Figure 1 (see, e.g. Dobson, DaVanzo, and Sen 2006 ; Lee al. 2003 ; Mayes 2004 ; Mayes and Hurley 2006 ; Zwanziger and Bamezei 2006 ). The figure shows the aggregate payment-to-cost ratios for all hospital-based services financed by private payers, Medicare, and Medicaid from 1980
through 2008. Except, perhaps, between 1980 and 1985, the private payment-to-cost ratio is negatively correlated with that of public programs. This is indicative of cost shifting, although other hypotheses are consistent with the evidence; that is, it may be coincidental or driven by other factors. As I suggest later, much of this may be explained by changes in hospital costs and changes in hospitals’ or plans’ price-setting power due to market size, reputation, and other factors relating to “market clout.”
Aggregate Hospital Payment-to-Cost Ratios for Private Payers, Medicare, and Medicaid, 1980–2008.
Figure 1 breaks the years 1980 to 2008 into five spans of time by four lines, marked (A) through (D). These five eras correspond to periods over which the characteristics and structure of the health care market (including hospitals’ and plans’ market power) and policy landscape differed due to identifiable legislative or market events. In the following discussion, I focus on changes in Medicare policy and payments. Medicaid payments tend to track Medicare payments, as Figure 1 shows.
The Golden Stream (before 1983)
Policymakers have struggled with Medicare financing since the program's early years. The original design of hospital payments reimbursed hospitals retrospectively for all services at their reported costs plus 2 percent for for-profits and plus 1.5 percent for nonprofits (Weiner 1977 ). These so-called return on capital payments were eliminated in 1969 (U.S. Senate 1970 ), and the cost reimbursement system that replaced them included a so-called nursing differential that paid hospitals an additional 8.5 percent above inpatient nursing costs (Kinkead 1984 ). The 8.5 percent nursing differential was reduced to 5 percent in 1981 (SSA 1983 ) and was eliminated altogether by 1984 (Inzinga 1984 ). Thus, from the inception of the program into the 1980s, hospitals could earn greater Medicare revenue and profit simply by increasing their reported costs or a portion of them (inpatient nursing costs in the case of the nursing differential) (Mayes 2004 ). 1 With no incentives for hospitals to contain costs, the system was described as “a license to spend, … a golden stream, more than doubling between 1970 and 1975, and doubling again by 1980” (Stevens 1989. 284).
Meanwhile, indemnity plans were the norm in the private sector. Without the leverage of network-based contracting (in which some providers could be excluded) and with payments rendered retrospectively on a fee-for-service basis, the private sector also had no success in controlling costs. In 1982, network-based managed care plans 2 emerged when California passed a law allowing health insurance plans to selectively contract with hospitals. This statute was widely emulated elsewhere, thereby sowing the seeds for managed care's role in controlling costs in the 1990s (Bamezai et al. 1999 ).
Thus before 1983, attempts by public and private payers to control hospital costs were largely unsuccessful. In general, both rose over time, consistent with the positive correlation between the two that persisted until about 1985, which is evident in Figure 1. In the relationship between hospitals and their payers, hospitals had the lion's share of power. Price competition did not exist, and hospitals attracted physicians and patients with expensive, nonprice amenities and services (Bamezai et al. 1999 ).
Incentive Reversal (1983–1987)
With a goal of reducing domestic spending, the Reagan Administration targeted Medicare's hospital payments. Then Secretary of Health and Human Services Richard Schweiker became enamored of New Jersey's hospital prospective payment model, based on diagnosis-related groups (DRGs), and accordingly used it for Medicare's system (Mayes 2004 ). Under Medicare's prospective payment system (PPS), each hospital admission was assigned to one of almost five hundred DRGs, each of which was associated with a weight based on the average costs of treating patients in that DRG in prior years. The payment to a hospital for an admission was the product of the DRG weight and a conversion factor. Medicare could (and did) control the amount of payments to hospitals by adjusting the growth rate of the conversion factor and/or adjusting the relative DRG weights (Cutler 1998 ).
The critical element of the PPS was that prices were set in advance of admissions (i.e. prospectively), thereby putting hospitals—not Medicare—at financial risk for the cost of an admission. Rather than paying hospitals more if they did more, as the earlier system had done, the PPS encouraged them to do less and to pocket any surpluses of prices over costs. The reversal of incentives was designed to control costs, and the conversion factor and DRG weights were the policy levers for doing just that.
The PPS was phased in over four years. Hospitals quickly learned how to reduce lengths of stay and, thereby, costs. Since PPS payments were based on historical costs, the early years saw a spike in aggregate payment-to-cost ratios, as shown in Figure 1 (Coulam and Gaumer 1991 ).
Medicare, Congress's Cash Cow (1987–1992)
By 1987, the PPS was fully phased in, and Congress began using its policy levers to extract huge savings from Medicare and apply them to deficit reduction. The legislative mechanism was the annual budget reconciliation process. Robert Reischauer, director of the Congressional Budget Office (CBO) from 1989 to 1995, explained how the PPS was viewed and used by Congress:
Medicare was the cash cow! … Congress could get credited for deficit reduction without directly imposing a sacrifice on the public…. And to the extent that the reduction actually led to a true reduction in Medicare services, it would be difficult to trace back to the Medicare program or to political decision-makers. (Mayes 2004. 157–58)
Aggregate Medicare hospital payment-to-cost ratios fell every year from 1987 to 1992 because hospitals did not restrain costs as quickly as payments were adjusted (Guterman, Ashby, and Greene 1996 ). During this period, as Medicare margins fell, private pay margins grew. The effects of managed care had not yet been fully felt in the commercial market, leaving private purchasers vulnerable to hospitals’ market power. If there ever was a time when market conditions were ripe for cost shifting, this was it.
The Ascendance of Managed Care (1992–1997)
The role of market power in setting prices is clear when considering the experience of the 1990s. The business community, desperate to end the annual double-digit percentage increases in premiums, changed course by no longer offering traditional indemnity plans and instead encouraging the growth of managed care. Beginning in 1993, the majority of enrollees in private plans (51%) were covered by managed care, a number that grew rapidly thereafter; by 1995, 70 percent of enrollees were in managed care plans (Mayes 2004 ). As Robert Winters, head of the Business Roundtable's Health Care Task Force from 1988 to 1994, remembered, “What happened in the late 1980s and in the early 1990s, was that health care costs became such a significant part of corporate budgets that they attracted the very significant scrutiny of CEOs…. More and more CEOs [were] saying, ‘Goddammit, this has to stop!’” (Mayes 2004. 162–63).
What stopped it was network-based contracting. The willingness of plans and their employer sponsors to exclude certain hospitals from their networks strengthened plans’ negotiating position. That is, to be accepted into plans’ networks, hospitals had to negotiate with plans on price. In this way, the balance of hospitals’ and plans’ market power shifted, resulting in the downward private payment-to-cost ratio trend between 1992 and 1997 illustrated in Figure 1 .
By contrast, public payers’ payment-to-cost ratios rose in the early 1990s. But this is not a (reverse) cost-shifting story because there is no evidence that public payments increased in response to decreasing private payments. Instead, the dynamics are better explained by changes in cost. Guterman, Ashby, and Greene (1996) found that the growth of hospital costs declined dramatically in the early 1990s, from above 8 percent in 1990 to below 2 percent by mid-decade, perhaps because of the pressures of managed care, a point echoed and empirically substantiated by Cutler (1998). The rise of hospital costs continued at low rates through the 1990s, averaging just 1.6 percent per year between 1994 and 1997. By contrast, Medicare payments per beneficiary to hospitals, which had been partially delinked from costs under the PPS, increased by 4.7 percent per year (Mayes and Hurley 2006 ). Thus, the movements in Figure 1 's time series confound the effects of price and cost, which—along with obscuring market power effects—gives a false impression of large, pervasive cost shifting. Put simply, there are many ways for public and private payment-to-cost ratios to change, and the causal connection between prices (cost shifting) is just one of them.
The Managed Care Backlash and the BBA (1997–2008)
With so much room for costs to fall, managed care plans profited relatively easily for several years, negotiating with hospitals to accept lower increases in payments and reducing subscribers’ hospital use (Reinhardt 1999 ). But plans’ revenues fell throughout the 1990s as price competition squeezed inefficiencies and surplus from the system. In an attempt to maintain their profitability, plans imposed greater restrictions on enrollees, subjecting them to more stringent utilization reviews, tighter networks, elimination of coverage for certain services, and higher cost sharing (Mayes and Hurley 2006 ; Rice 1999 ).
These cost-saving measures became increasingly unpopular, and a managed care backlash ensued. In response, the states and the federal government enacted managed care reform and consumer protection laws (Sorian and Feder 1999 ). By 1997, the era of managed care's strong restraints on cost had ended and, with it, the low premium increases they had delivered earlier in the decade. America entered the current age of health care plans, with less restrictive network contracting embodied in the preferred provider organization (PPO), which became the norm. Although plans and hospitals still negotiate on price, because consumers dislike restrictions on their choice of providers, leverage shifted away from plans and toward hospitals. Reinforcing this shift, the number of hospital mergers rose in the late 1990s (Vogt 2009 ), and by the turn of the century, private payment-to-cost margins began to climb.
Coincidental with the popular rejection of managed care, Congress turned its attention to the budget deficit and again sought savings from Medicare. In 1997, it passed the Balanced Budget Act (BBA), which promised $115 billion in Medicare savings between 1998 and 2002 by eliminating retrospective cost-reimbursement for postacute care, long-term hospital services, and hospitals’ outpatient departments (Wu 2009 ). Thus, hospitals’ Medicare payment-to-cost ratios declined as those of private payers rose, but perhaps only coincidentally. The latter was facilitated by a shift in market power; the former by policy change. The extent to which they are causally related cannot be determined from Figure 1 alone.
In summary, Figure 1 reveals a negative correlation between public and private payment-to-cost ratios since 1985. Although this suggests cost shifting, it does not prove it. Other hypotheses are consistent with the evidence. Historical changes in hospital costs and the balance of hospitals’ and plans’ market power may explain all or some of the data. Only careful empirical analysis can reveal the causal effect of public prices on private prices, free of the confounding effects of changes in market power and hospital practices that impact costs.
To identify those studies providing empirical analyses of cost shifting or theoretical predictions of the phenomenon, I used Google Scholar to search the academic literature from 1996 to the present with the search string: health (payment OR rate) (Medicare OR Medicaid) “cost shift.” More than six hundred documents satisfied these search criteria, from which I selected only those that were not included in Morrisey's 1996 review, appeared in peer-reviewed journals, and either offered theoretical treatment of or attempted to estimate the size of a cost-shifting effect. The list was augmented by relevant papers published in 1996 or later cited by or citing those found in the Google Scholar search and meeting the aforementioned criteria. Although I did include papers pertaining exclusively to nursing homes’ and physicians’ cost shifting, they are not part of this review. They are, however, described in Frakt (2010a) .
The final papers pertaining to hospitals’ cost shifting are listed in Tables 1 and and2. 2. Table 1 cites those that provide theoretical developments, and Table 2 cites those with empirical findings. Those papers that include both are listed in both tables.