« Back to Results

Pay for Performance in Health Care: International Evidence

Paper Session

Sunday, Jan. 7, 2018 8:00 AM - 10:00 AM

Marriott Philadelphia Downtown, Meeting Room 403
Hosted By: International Health Economics Association
  • Chair: Adam Wagstaff, World Bank

An Incentive Program with Almost No Incentive: An Overlooked Benefit of Pay-for-Performance

Shiko Maruyama
,
University of Technology Sydney
Chunzhou Mu
,
University of Technology Sydney
Kees van Gool
,
University of Technology Sydney

Abstract

The idea of “pay for performance” (P4P) has been advocated and applied for more than two decades. The large literature has concerned whether it actually improves health outcomes and found mixed evidence, and researchers tend to conclude that how best to pay for performance remains a compelling question, notwithstanding numerous forms of incentive programs experimented around the world. A problem in this literature is that although the idea of P4P is to reduce inefficient use of healthcare resources (typically overuse in traditional pay-for-volume), researchers have only studied its impact on certain outcomes rather than the efficiency gain from P4P. This paper offers a new view that P4P can lead to efficient allocation of healthcare resources even without improvement in health outcomes, by studying a unique physician incentive program introduced in Australia in 2001 to promote effective management of chronic diseases, the diabetes Service Incentive Payment (SIP), which attracts an incentive of A$40 per patient per year after completing a lengthy cycle of diabetes care - almost negligible for general practitioners (GPs). We evaluate the effect of the diabetes SIP on various healthcare utilization and cost outcomes of diabetics, using a large survey linked to multiple years of detailed administrative medical records. More specifically we identify the causal effect of the SIP availability by exploiting geographical variation in the penetration of SIP completion, measured by the fraction of diabetics-GP pairs that claimed SIP reward in each postal area, based on the idea that SIP penetration measures the level of GPs’ knowledge about and familiarity to the SIP in each area, which is presumably exogenous from each GP’s point of view. To account for potential confounding differences across areas, we use pairs of diabetics and GPs who are ineligible for the SIP as a control group, and employ a difference-in-differences method.

Pay-for-Performance and Selective Referral in Long-Term Care

Toshiaki Iizuka
,
University of Tokyo
Haruko Noguchi
,
Waseda University
Shinya Sugahara
,
University of Tokyo

Abstract

We examine how pay-for-performance (P4P) affects long-term care (LTC), focusing on adult day-care services. LTC expenditure is expected to increase dramatically, and P4P for LTC could potentially improve welfare by providing better outcomes at lower costs. However, little is known about the effects of P4P on LTC. We exploit a natural experiment in Japan in which a local government (Shiga prefecture) introduced an outcome-based bonus payment in addition to nationally uniform fee-for-service payments. We construct unique matched user-care manager-provider data using the universe of LTC claims data in which care managers’ referral decisions are directly observed. At the population level, using all other prefectures as a control, we find weak evidence that P4P affects LTC outcomes; however, the impact appears very small if it exists at all. Because P4P does not affect LTC expenditures, we conclude that the overall effect of current P4P is negligible. In contrast, at a disaggregated level, within-treatment comparisons find stronger evidence of user selection. After P4P, LTC outcomes improve more when care managers refer users to affiliated providers as opposed to non-affiliated providers. Moreover, care managers refer more users whose care levels are relatively more easily improved to affiliated providers. A lack of risk adjustment and vertical integration of care management and day-care service appear to explain the selection. A successful P4P would need to address these issues.

Physician Performance Pay: Experimental Evidence

Daniel Wiesen
,
University of Cologne
Jeannette Brosig-Koch
,
University of Duisburg-Essen
Heike Hennig-Schmidt
,
University of Bonn
Nadja Kairies-Schwarz
,
University of Duisburg-Essen
Johanna Kokot
,
University of Duisburg-Essen

Abstract

We present causal evidence on the effect of performance pay on medical service provision from an artefactual field experiment with a representative sample of German resident primary care physicians. In the experiment, we introduce performance pay, which is adjusted according to patients’ severities of illnesses, to complement capitation. Performance pay is granted if a health care quality threshold is met. In line with standard theory, we find that performance pay significantly reduces underprovision of medical services, and, on average, it increases the patients’ health benefit. The magnitude of these effects depends, however, on patients’ characteristics. Findings are robust towards variations in levels of performance pay. Beyond standard theory, we find evidence for a crowding-out of altruistic behavior when physicians receive performance pay. Physicians’ characteristics such as gender and practice location significantly affect crowding-out of altruistic behavior.

Showmanship: The Relationship Between Health Care Provider Generosity and Provider Effort

Nivedhitha Subramanian
,
Duke University
Manoj Mohanan
,
Duke University

Abstract

Healthcare in many parts of the developing world is characterized by high access to care, but very low quality of care. This paper uses novel data from rural Bihar, India, to explore the relationship between healthcare provider generosity and quality of care delivered by health care providers. The data include a lab-in-field generosity game and measures of healthcare provider performance when providers do and do not know that they are being observed. We find that providers exhibit high levels of generosity to a health-related NGO in the area and high levels of effort when they know that they are being observed. However, when measured using standardized patient methodology – i.e. when providers do not know that they are being observed – measures of provider generosity are not correlated with high levels of provider effort.
Discussant(s)
Karen Eggleston
,
Stanford University
Aditi Sen
,
Johns Hopkins University
James Rebitzer
,
Boston University
Adam Wagstaff
,
World Bank
JEL Classifications
  • I0 - General