ADVERSE EFFECTS OF PERFORMANCE BASED FINANCING

The Massachusetts health insurance law (2006), the Oregon health insurance experiment (2008), and Medicare’s value-based physician payment system (2013) reflect the U.S. attempts to raise cost efficiency in the health sector. Yet, the analyses of the preliminary results of the Medicare’s value-based financing reveal various adverse effects, including deterioration of quality for the best-performing U.S. hospitals. Arguably, the undesirable effects may be attributed to a linear degressive reimbursement schedule in the performance-based financing. Overall, the theory suggests that degressive performance-based rates might be exploited as a tool to raise quality and stimulate efficiency. However, the way the rates are diminished is a key issue to avoiding the unwanted effects of the remuneration mechanism.

The paper addresses the adverse effects of performance-based financing by exploiting the unique nationwide administrative data (2007-2012) for recent Japanese experience of introducing step-down (length-of-stay dependent) degressive rates in its per diem prospective payment system. The novelty of the paper is the use of a variety of diagnoses and a large sample of hospitals for empirical estimates of dynamics of hospital efficiency (proxied by length of stay) and hospital quality (proxied by planned early readmission rate) in the reformed hospitals. The analysis employs a recent administrative database from Japan’s Ministry of Health, Labor, and Welfare (August 21, 2012) on annual hospital/major diagnostic category level aggregated information for all patients, discharged in July-December 2006-2010 and July 2011-March 2012. Merging the annual files by hospital name (checking for any change of name due to restructuring, mergers, and closures), we construct an unbalanced panel of 697 hospitals, which have submitted data since 2007.

Owing to historical links between hospital departments and a certain medical university, and hierarchical relations within each department, our analysis assumes that a hospital strongly adheres to its practice patterns. Consequently, for each major diagnostic category (a group of diagnoses based on ICD-10 with minor modifications in Japan) we employ dynamic panel data models and focus at the reform effects at different percentiles of hospitals. As robustness check we employ cross-section analogues of the panel data models, which enables incorporating a number of time-invariant hospital variables, differenced out in dynamic panel data estimations. The analysis controls for major hospital characteristics (the number of beds as hospital size and proxy for capital; the number of hospital departments as proxy for diversity; the time-invariant dichotomous variables for rural, emergency, university hospitals, for the presence of MRI or CT scanners), ownership, the status of designated local public hospital and the presence of the independent hospital accreditation.

The results indicate persuasive evidence supporting the adverse effect of degressive rates in performance-based schedule. The average length of stay significantly increases for hospitals in percentiles 0–25 of the pre-reform nationwide length of stay. While average length of stay declines at hospitals in percentiles 51-100, it is accompanied by a significant increase of planned readmission rate within 42 days after discharge.

As a remedy to the problem built-in degressive reimbursement schedule, the paper discusses the applicability of the “best practice” rate-setting.