Background of Changes in Hospital Financing and Organization

Rising health expenditure and tight public budgets have led policymakers in many countries to seek ways to improve the performance of healthcare organizations. In most countries, hospitals in particular are thought to have huge potential to improve their performance and are increasingly being held accountable for such improvements. Indeed, since the 1980s policymakers have identified a wide range of policy interventions to this effect.

Some of these interventions have been initiated at the government level, including: payment systems, such as those based on diagnosis-related groups (DRGs); reforms in hospital planning, such as the centralized commissioning and decommissioning of hospital services for different regions; and the introduction of quality assurance systems. In health systems that are more market-oriented, such as in the United States, Germany, and Switzerland, measures to improve the performance of hospitals may take the form of individual decisions made by the hospitals themselves, albeit within regulatory and payment systems that may give hospital owners incentives to choose certain options over others. Examples include a hospital’s entry into a health system, privatization, cooperation with other hospitals or healthcare organizations, and specialization in certain services. While for instance in Sweden the regional government decides whether a specific hospital should specialize in a certain area, in Germany hospitals make this kind of decision themselves (at least within the framework of capacity planning undertaken at the level of each of Germany’s 16 states). The interventions presented in this article were chosen based on their policy relevance and the availability of empirical evidence, and are as follows: the introduction of DRG-based hospital payments; the entry of hospitals into multi-institutional arrangements, privatization that can occur in these situations, and the entry of hospitals into looser cooperation arrangements; and, last, the specialization of hospitals.

The purpose of this article is to give an overview of studies investigating the impact on hospital performance of key reforms in the sector. In doing so, the article aims to group these studies into research streams, to assess their validity, and to synthesize their results. The article will identifies the drivers of the effects associated with, or resulting from, certain reforms. The policy implications of our findings and avenues for further research are subsequently presented and discussed.

Definition of Hospital Performance and Methodological Approaches

Definitions of hospital performance vary considerably depending on context. Those used by medical researchers often differ from those used by economists. In the medical literature, hospital performance is frequently understood to refer to quality of care, whether in terms of processes or outcomes. A measure of the former is the initiation of treatment within a specified time after arrival for certain emergency conditions. Some measures of the latter are in-hospital or post-hospitalization mortality rates, and the rate of readmission to hospital. In contrast, the economic literature frequently defines hospital performance in terms of financial performance and efficiency, although recently more and more economic papers have emerged that measure hospital performance as quality of care (Gaynor, Laudicella, & Propper, 2012). Efficiency can be defined as a measure of how well an organization produces output (i.e., treatments or cases) from a given amount of input, especially human and technological resources. It can be broken down into technical, cost, and profit efficiency. This article focuses primarily on studies that use parameters of financial performance, parameters of efficiency, or both.

Studies that analyze financial performance are usually undertaken in competitive hospital markets characterized by a large proportion of for-profit/nonprofit hospitals that have access to capital markets, as is the case for example in the United States and Germany (Büchner, Hinz, & Schreyögg, 2016; Ramamonjiarivelo, Weech-Maldonado, Hearld, Pradhan, & Davlyatov, 2018; Shen, 2003). By ownership, hospitals can generally be divided into private hospitals, which are invested in and owned by private entities (i.e., individual or several private owners), and public hospitals, which are invested in and owned mainly by public entities such as governments (Tiemann & Schreyögg, 2009). Private hospitals can be further classified as for-profit or nonprofit. Typical measures of financial performance across different ownership types are operative cash flow, equity ratio, profit before tax, and debt ratio. The assumption underlying studies of financial performance is that hospitals behave as profit-maximizing entities in competitive hospital markets even if they are forced to reinvest their surpluses, as is the case with nonprofit and many public hospitals. It can also be argued that in markets in which the state does not provide sufficient funding for capital costs, public and nonprofit hospitals are forced to generate surpluses to cover these costs and to ensure survival of the organization (Shen, Eggleston, & Lau, 2007).

The assumption that hospitals of all ownership types behave as profit-maximizing entities is, however, highly controversial in the literature because different ownership types may be associated with different internal incentive structures. Indeed, there is a large body of literature on theoretical assumptions about differences in behavior between organizations with different ownership types. For instance, property-rights theory assumes that ownership of a firm involves two essential rights: (1) the right to control the firm and (2) the right to appropriate its profits (i.e., the financial surplus). Accordingly, the defining difference between hospitals that are for-profit and those that are public or nonprofit is that the latter two are precluded from distributing their surplus to those in control of the organization. This is known as the non-distribution constraint (Hansmann, 1988).

This being said, agency theory assumes that managers (or agents) seek to maximize their own utility rather than that of the organization or its owners (or principals). As a result, in hospitals of all three ownership types, hospital owners face a principal–agent problem. However, it is assumed that for-profit hospitals are able to address this more effectively, as their owners can use profits as their measure of the manager’s success, limiting any deviation from their interest by making the manager’s compensation a positive function (Newhouse, 1970; Fama & Jensen, 1983).

Another important point to consider is that private entities have access to capital markets and thus have an incentive to generate high profits to attract investors. In contrast, public and nonprofit hospitals are assumed to be generally averse to profit maximization. This is especially the case for public hospitals, which can have their profits taken away by public owners and spent on other public goods (Pestieau, 2009). It is therefore reasonable to assume that—depending on a country’s regulatory environment—using financial performance criteria to measure hospital performance may lead to bias if ownership types are not taken into account.

In some situations, the most meaningful measure of performance might be a combination of financial performance and efficiency. When an organization is privatized, for example, it is expected to increase its profitability. One of the measures that privatized organizations often take to do so is to reduce their number of staff, which for instance was found by Heimeshoff, Tiemann, and Schreyögg (2014) to be the case in Germany. Organizational changes such as this tend to be directly visible in efficiency measures (e.g., staff reduction as decreasing inputs), whereas the effect of organizational changes on financial performance often takes some time (e.g., Tiemann & Schreyögg, 2012).

With regard to their methods, studies assessing how changes in hospital financing and organization affect hospital performance can be divided roughly into two streams. The first is linked to program evaluation literature (Abadie & Cattaneo, 2018) and focuses on establishing a causal link between interventions and effects while considering selected performance indicators, such as costs, revenues, length of stay, and measures of profitability. These studies use difference-in-differencess regression models and occasionally also instrumental variable approaches or some kind of matching, for example with propensity scores (Clemens & Gottlieb, 2014; Gaynor et al., 2012; Shen, 2003). The other stream of literature measures changes in hospital efficiency while attempting to capture the production process of a hospital as comprehensively as possible through its inputs and outputs in contrast to selected performance indicators (Büchner, Hinz, & Schreyögg, 2015; Rosko, Proenca, Zinn, & Bazzoli, 2007). Often, studies in this category do not establish a causal link between interventions and effects. These studies measure technical, cost, and profit efficiency and use data envelopment analysis, stochastic frontier analysis, and, more recently, partial frontier analysis. While data envelopment analysis remains by far the most common approach to measuring the efficiency of hospitals (Kohl, Schoenfelder, Fügener, & Brunner, 2018), stochastic frontier and partial frontier analysis are gaining favor as they can be used to overcome certain weaknesses of data envelopment analysis, such as its treatment of outliers (Hollingsworth, 2008; Varabyova, Blankart, Torbica, & Schreyögg, 2017).

Effects of the Introduction of DRG-Based Hospital Payments on Hospital Performance

There is no doubt that case-based payment systems, such as those that use DRGs, have had a major impact on hospital markets. Previous systems were based predominantly on fee-for-service payments, per diem payments, global budgets, or some combination of these. The shift to DRG-based payments was radical and its impact can mainly be attributed to three factors (Street, O’Reilly, Ward, & Mason, 2011):

The third reason in particular leads to a different incentive structure for hospitals. The aim of yardstick competition is to provide a benchmark for providers, usually based on average market costs, giving them an incentive to reduce costs (Shleifer, 1985). For details on how DRGs are calculated in different health systems, refer to Schreyögg, Stargardt, Tiemann, and Busse (2006). If providers outperform others, they generate a surplus, whereas they generate a deficit if their costs per case are higher than the benchmark. This creates a continual incentive to reduce costs. To ensure an iterative process of cost reduction, all DRG systems have some kind of yearly or biyearly updating mechanism (Schreyögg et al., 2006; Street et al., 2011).

Given this radical change in incentive structures, one would expect hospitals to have improved their performance after DRG-based payments were introduced. Although several studies on this topic exist, only a few of them examine how the introduction of DRG-based hospital payments may have affected indicators of performance within the framework of a natural experiment. One reason for this may be the methodological challenge of establishing a causal link between such payment systems and hospital performance. Indeed, countries that have introduced DRG-based payment systems have usually done so nationwide and after a short phasing-in period. This means that there is no comparable control group of hospitals, unaffected by the intervention, that is available for analysis. Although psychiatric care hospitals might, at first glance, appear to be a suitable control group because they were usually excluded from such reforms, they have a completely different case-mix (i.e., patients are often discharged without being cured, in the narrow somatic sense of that word) and the optimal length of stay remains unclear for many psychiatric conditions. Another possible control group might be the hospitals that remained under the old payment system and did not opt into the new DRG-based payment schemes during the period in which the new payment system was gradually phased in. However, even if selection effects are considered, the phasing-in periods have to be regarded as transitions during which different rules apply and the incentives of the new system have not yet been fully implemented. Last, the considerable length of the phasing-in periods in most countries presents its own methodological challenges. Because these periods usually lasted a couple of years, other hospital reforms were often initiated in parallel, making it difficult, if not impossible, to distinguish between the effects of each.

The few studies that do examine the impact of DRG-based payments on hospital performance have yielded mixed results. In Portugal (Dismuke & Sena, 1999, 2001), Sweden (Gerdtham et al., 1999a, 1999b), and Norway (Biørn et al., 2003; Hagen et al., 2006), the introduction of the new payment system appears to have led to increases in technical efficiency. In Switzerland, where the new payment system was introduced in a staggered fashion across different cantons, Widmer (2015) found increases in cost efficiency, albeit in a small sample of hospitals. In each of these studies, efficiency gains were realized through reduced inputs (fewer staff, shorter length of stay), increased outputs (more admissions), or both.

A number of other researchers, however, did not observe a positive impact on performance. Studies on hospitals in the United States (Borden, 1988; Chern & Wan, 2000) and Austria (Sommersguter-Reichmann, 2000) did not find any increase in technical efficiency. Herwartz and Strumann (2014) found a significant decline in technical efficiency in a sample of German hospitals. For Norway, Biørn et al. (2003) and Hagen et al. (2006) found mixed evidence on the impact of the new payment system on cost-efficiency.

While the heterogeneity of these results may be due in part to the different methods used in each study to estimate efficiency, it is probably also a function of very different health systems and contexts in the different countries. Each country has different reimbursement mechanisms; payment mechanisms, such as additional reimbursement components, that differ in their relevance; and, of course, a host of different and often simultaneous healthcare reforms. It should also be mentioned that none of the studies examined a possible trade-off between quality and efficiency after the introduction of DRG-based payments, even though some researchers have found such trade-offs in hospital care independent of DRGs (e.g., McKay & Deily, 2008).

Even if there is no convincing evidence that DRG-based payments have improved hospital performance, it is likely that yardstick competition has encouraged hospitals to adopt some of the organizational patterns seen in other markets. This may have facilitated other changes at the hospital level, such as cooperation, mergers (privatization), and specialization, all of which are examined in later sections of this article.

Another, newer stream of research looks at how hospitals have responded to changes, especially changes in price, within existing DRG-based payment systems. Depending on the changes in question and context in which the sample of hospitals was located, studies in this stream have found causal links to the upcoding of patients into more profitable diagnoses and increases in the number of admissions, as well as to changes in treatment intensity, clinical pathways, and the range of services (Acemoglu & Finkelstein, 2008; Clemens & Gottlieb, 2014; Cutler, 1995; Dafny, 2005; Duggan, 2000). One of the main challenges in this stream of research is to find the best approach with which to distinguish between hospital responses at the intensive margin (e.g., upcoding) versus the extensive margin (e.g., increase in admissions). Although this stream of research is related only indirectly to hospital performance, it underscores how important it is to monitor how hospitals respond to changes within existing DRG-based payments systems and whether, and to what extent, such systems have unintended effects.

In summary, several studies have investigated how the introduction of DRG-based hospital payments may have affected efficiency in hospitals. Their results, however, are mixed, which may be explained by the different methods employed by the studies and, perhaps more importantly, the different national-policy contexts in which their samples of hospitals were situated. In cases where gains in efficiency were observed, they were realized through reduced inputs (fewer staff, shorter length of stay), increased outputs (more admissions), or both. This research stream suffers from two structural problems: (1) Usually there is no control group available for comparison because DRG-based systems are generally rolled out nationwide; and (2) the period during which most DRG-based payment systems are phased in is long, which means that other reform initiatives have generally been introduced in parallel, making it difficult if not impossible to distinguish between the effects of each. Another stream of research examines the causal link between price changes for given treatments and hospitals’ responses to them. The results of such studies suggest that price changes have had unintended effects and that careful monitoring of them is important. Future research should build upon these studies and disentangle effects at the intensive (e.g., upcoding) versus the extensive margin (e.g., increase in admissions).

Effects of Multi-Institutional Arrangements on Performance

Effects of Specialization on Performance

Specialization is one of the most discussed topics in health policy, and many decision makers expect specialization to improve performance and increase quality of care at the same time. It is therefore not surprising that a number of different incentives have been devised to encourage the specialization of hospitals. For instance, the introduction of DRG-based payments has been a major driver of specialization. These payments encourage hospitals to specialize in DRGs, particularly involving elective procedures, that entail low production costs. Doing so may ultimately reduce aggregate hospitalization costs if the specialized hospitals are efficient (Dranove, 1987).

The general idea that specialization can improve performance dates back to Adam Smith and his work on the division of labor (Smith, 1776). Taylor also proposed enhancing organizational performance by deploying workers based on their individual skills (Taylor, 1911). While Smith and Taylor focused on individual specialization, March and Simon proposed distinguishing between individual and higher-level (i.e., organizational) specialization (March & Simon, 1958). For the hospital market, Clark and Huckman (2012) pointed out the potential of linking both approaches in order to measure and operationalize specialization at the level of the operating unit, such as departments, units, or other suitable structures in hospitals.

Investigating the impact of specialization on hospital performance is challenging, however, because it is not a directly observable variable and must therefore be operationalized using administrative data. There are several approaches to doing so, all of which construct a measure based on information about hospital patients grouped into different categories according to their diagnoses (Lindlbauer & Schreyögg, 2014). The most common of these measures in the hospital context is the Information Theory Index (Farley, 1989; Farley & Hogan, 1990), which can also be understood as a measure of diversification (Kim et al., 2015). The Information Theory Index compares the proportions of patients in different diagnosis categories in a hospital with the average proportions for these categories in all hospitals in a country or region. Using the Information Theory Index, Zwanziger, Melnick, and Simonson (1996) developed a similar measure based on the distance between a hospital’s patient case-mix and a predetermined baseline. Others have used the Herfindahl–Hirschman Index to operationalize specialization as the concentration of diagnoses within hospitals or operational units, such as departments (Baumgardner & Marder, 1991). Another concept was put forward by Daidone and D’Amico (2009), who used an index similar to the Gini coefficient to measure how the proportion of patients in each diagnosis category in a hospital deviated from a uniform distribution.

Lindlbauer and Schreyögg (2014) made two observations about specialization measures like these, which were based on proportions. First, increasing the number of patients in only one category increases its proportion but also decreases the proportions of other diagnosis categories, although there may have been no change in treatment behavior in the latter. Second, relying solely on a proportion-based measure can make it appear that only small hospitals are highly specialized, whereas large hospitals, such as academic medical centers, may be highly specialized as well. To address these limitations, they developed an alternative measure of specialization known as “category medical specialization,” which considers information on the volume of patients in each diagnosis category in addition to concentration. Using this measure, they were able to identify both small and large hospitals that offered specialized hospital services.

Several studies have investigated the relationship between hospital specialization and hospital performance. These use primarily cross-sectional data and do not treat specialization as an event (e.g., one that takes place after new regulation). The majority of these studies focus on efficiency. Among the studies that investigated the relationship between hospital specialization and efficiency, data envelopment analysis and stochastic frontier analysis have been the approaches of choice. As a measure of specialization, these studies use the Information Theory Index (Herwartz & Strumann, 2012; Lee, Chun, & Lee, 2008; Linna & Häkkinen, 1999), the Gini coefficient (Atella et al., 2012; Daidone & D’Amico, 2009) or category medical specialization (Lindlbauer & Schreyögg, 2014). Carey, Burgess, and Young (2008) examined individual specialty hospitals to reduce complexity. These studies used data from hospitals in the United States, Finland, Italy, Korea, and Germany. Lee et al. (2008) and Daidone and D’Amico (2009) found specialization to be positively associated with technical efficiency, whereas Atella et al. (2012) found the relationship to be negative, and Herwartz and Strumann (2012) found that the relationship changed over time. Linna and Häkkinen (1999) found a positive relationship between specialization and cost-efficiency, while Carey et al. (2008) found that individual specialty hospitals were not associated with superior cost-efficiency compared to full-service hospitals. Lindlbauer and Schreyögg (2014) found different results depending on the definition of specialization. When measured using the Information Theory Index and the Gini coefficient, specialization was found to be negatively associated with technical efficiency, whereas it was found to be positively associated with technical efficiency when measured using category medical specialization. The standardization of processes is suggested in the papers to be the primary driver of improved efficiency.

A few papers also investigated the impact of specialization on financial performance. Eastaugh (2011) found that increased specialization among hospitals in the United States was associated with a 10% decline in unit cost per admission. Kim et al. (2015) found that, when identified using the Information Theory Index, specialization had a positive effect on income per bed, whereas it had a significant positive effect on income per specialist, but a negative effect on income per bed when identified using the Herfindahl–Hirschman Index. Finally, Schneider et al. (2007) investigated whether the presence of more specialized hospitals in regional markets had an effect on the costs and profitability of general hospitals. They found that the presence of such hospitals reduced the costs and increased the operating margin of general hospitals and concluded that specialization in the hospital market seems to have important competition effects. Gonçalves and Barros (2013) investigated economies of scope in the provision of auxiliary clinical services that are typically carried out within in-patient facilities in Portuguese hospitals, by estimating a flexible cost function. They found economies of scope for some of the examined services, but not for all.

In summary, while a number of studies have looked at the impact of specialization on performance, the effects remain largely unclear. The main reason for this may be the complexity of measuring specialization. As the Information Theory Index, the Herfindahl–Hirschman Index, Gini, and category medical specialization all measure different forms of specialization, more research is needed, such as simulation studies, to compare the different concepts and to disentangle the differences between them. These various concepts of specialization and the heterogeneity of samples used in different countries may explain why the studies published to date have yielded such mixed results. Finally, future studies should attempt to investigate specialization as an event using program evaluation techniques, even though self-selection is a limitation in this context.

Discussion of Studies and Promising Avenues for Future Investigation

This article presents an overview of studies that have investigated both the major interventions that have taken place since the 1980s in the financing and organization of hospital services and their impact on hospital performance. The validity of the studies and their results are highly heterogeneous. Table 1 provides an overview of our key findings for each of the research streams, covering the main effects, drivers, and validity of methods. The table also highlights the challenges in each research stream and the scope for further research.

Table 1 shows that, overall, surprisingly few studies have examined the impact of policy reforms on hospital performance. None of the research streams has a rich body of literature comparable to that seen in other fields of health economics. Chief among the numerous reasons for this is a lack of data—for example, the absence of control groups of non-DRG hospitals—and the complexity of the research designs that would be required to investigate, say, the different forms of multi-institutional arrangements that are prevalent in the hospital sector. Of all the evidence, that on the effects of privatization is the strongest. Studies in this research stream use longitudinal designs with control groups and have found robust increases in efficiency and financial performance. Evidence on the entry of hospitals into health systems and the effects of this on efficiency is similarly strong. Although the other three streams of research also contain well-conducted studies with valuable findings, they are predominantly cross-sectional in design and therefore cannot establish causation. Whereas the effects of introducing DRG-based hospital payments and of specialization remain largely unclear, we can conclude that vertical and horizontal cooperation are likely to have a positive effect on efficiency and financial performance. The drivers of improved efficiency or financial performance are very different depending on the reform or intervention being investigated, but reductions in the number of staff and improved bargaining power in purchasing stand out as being of particular importance.

We identified several promising avenues for future investigation. One of these is situated within a new area of research examining the link between changes in the prices of treatments and hospitals’ responses to them. As there is evidence of unintended effects of DRG payments, future studies should attempt to distinguish between changes in hospitals’ responses at the intensive margin (e.g., upcoding) versus the extensive margin (e.g., increase in admissions). When looking at the effects of entering into a health system and of privatizations, there is still considerable need for research. With privatizations, in particular, the underlying processes are not yet fully understood, and the potential trade-offs between increases in performance and changes in the quality of care have not been sufficiently examined. For instance, studies such as that by Bloom et al. (2015), who found a positive impact of competition on management quality, are needed for privatizations to understand the underlying processes. The main challenge for research on privatization is small sample sizes. It is therefore hardly surprising that the few studies that do exist in this area focus on the very large hospital markets in the United States and Germany. It would be useful, however, to widen the scope of this research and to pursue the same research questions in studies examining the effects of entering into health systems in general rather than focusing exclusively on privatizations. Last, there is substantial need for further papers in the areas of multi-institutional arrangements and cooperation, as well as specialization. In both research streams, natural experiments investigated using program evaluation design are lacking. One of the main challenges here, however, is that cooperation and specialization cannot be directly observed but rather must be constructed based on survey or administrative data. Moreover, there are multiple definitions of specialization, and each of the measures a different form of specialization. More research is clearly needed to disentangle them.

This review has some important limitations. First, the interventions presented in this article were chosen purposively based on a subjective assessment of their policy relevance, real-world impact, and the availability of empirical data. There are, of course, many other interventions that have been undertaken by governments and hospitals in recent years that have also had an impact on hospital performance. Examples include changes in hospital capacity planning (e.g., Karmann & Rösel, 2016), the introduction of selective contracting and other competitive models, for instance in California (e.g., Mobley, 1998), and the introduction of quality assurance and quality certification systems (e.g., Lindlbauer et al., 2016). Second, this review is a narrative one and is not based on a systematic review of the literature. Although we searched for literature in a structured manner in the most common databases, further studies may be available for the different research streams.

Policy Implications

The findings of this review have important implications for health policy. It is often assumed that the major reforms in the hospital sector since the 1980s have led to improved hospital performance. This review suggests, however, that the evidence in this regard is mixed, and is strong only for the effects of a hospital entering into a health system and of privatization. This may come as a surprise to policymakers. DRG-based hospital payments are, however, an important case in point. Almost all high-income and many other countries followed the example of the United States and introduced DRG systems, naturally with certain country-specific adaptions in calculation and payment mechanisms. Yet it has rarely been questioned whether DRG-based hospital payments actually lead to superior performance. Studies showing the unintended effects of these systems have emerged only recently, long after most DRG systems were fully implemented. While there is no evidence that these unintended effects are severe, it has become clear that setting the right incentives and carefully monitoring each step are much more crucial than had been thought. Policymakers would therefore be well advised to tread more carefully with future policy reforms, and when changing incentives in the hospital sector. Policy interventions would benefit greatly from the running of pilot trials that use control groups and defined evaluation periods before such interventions are implemented across the hospital sector as a whole.