Introduction Directory UMM :Journals:Journal of Health Economics:Vol19.Issue5.Sep2000:

Ž . Journal of Health Economics 19 2000 731–753 www.elsevier.nlrlocatereconbase Consumer satisfaction and supplier induced demand Fredrik Carlsen a, , Jostein Grytten b a Department of Economics, Norwegian UniÕersity of Science and Technology, NTNU, 7491 Trondheim, Norway b Dental Faculty, UniÕersity of Oslo, Oslo, Norway Received 4 March 1998; received in revised form 10 February 2000; accepted 2 March 2000 Abstract This study examines the relationship between supply of primary physicians and con- sumer satisfaction with access to, and quality of, primary physician services in Norway. The purpose is to throw light on a long-standing controversy in the literature on supplier Ž . inducement SID : the interpretation of the positive association between physician density and per capita utilization of health services. We find that an increase in the number of physicians leads to improved consumer satisfaction, and that the relationship between satisfaction and physician density exhibits diminishing returns to scale. Our results suggest that policy-makers can compute the socially optimal density of physicians without knowl- Ž . edge about whether SID exists, if one accepts the controversial assumption that consumer satisfaction is a valid proxy for patient utility. q 2000 Elsevier Science B.V. All rights reserved. JEL classification: I18 Keywords: Satisfaction; Supplier inducement; Financing; Physicians

1. Introduction

Despite several decades of intensive theoretical and empirical research, the Ž . extent of supplier inducement SID and policy implications remain controversial Corresponding author. Tel.: q47-73-59-1931; fax: q47-73-59-6954. Ž . E-mail address: fredrik.carlsensvt.ntnu.no F. Carlsen . 0167-6296r00r - see front matter q 2000 Elsevier Science B.V. All rights reserved. Ž . PII: S 0 1 6 7 - 6 2 9 6 0 0 0 0 0 4 4 - 8 ŽReinhardt, 1985; Phelps, 1986; Feldman and Sloan, 1988; Rice and Labelle, . 1989 . One reason for the controversy is probably the heavy focus on physicians’ motiÕes, and whether they act as perfect agents for patients. Since most patients do not have and never will gain medical expertise, we may never be able to discern whether patients would have demanded the amount of care they actually receive if physicians and patients were equally informed. Ž . Recently, Labelle et al. 1994 have argued that more attention should be paid to the consequences of SID. If additional health services result in improved health status or better access to health care, then SID may be beneficial to society irrespective of physicians’ motives for generating more services. Along the same Ž . lines, Ryan 1992 argues that patient satisfaction may be improved by additional supply of health services, even if the additional services are induced. Adequate data on the effect of health services on health status are often not available, particularly for primary physician services. Alternative sources of information about the benefits of health care are surveys about consumer satisfac- tion with their health care system. There are at least three reasons why surveys Ž may provide valuable information to policy-makers Pascoe, 1983; Ware and . Davies, 1983; Cleary and McNeil, 1988; Hall and Dornan, 1988 . Firstly, along with improved health status, satisfaction is an ultimate outcome of health care. Secondly, there are several dimensions of health services which patients can observe and evaluate, such as travelling distance, waiting time before an appoint- ment, the physical environment and the interpersonal skills of the staff. Thirdly, satisfaction surveys provide information about patient behaviour. Studies show that satisfied patients are more inclined to comply with recommended treatment and keep appointments and are less inclined to shop around for a doctor than Ž dissatisfied patients Kincey et al., 1975; Berkanovic and Marcus, 1976; Linn et . al., 1982; Marquis et al., 1983; Stewart, 1995 . This paper employs survey data to study the relationship between supply of primary physician services and consumer satisfaction in Norway. Our study is directly related to the long-standing controversy in the SID literature, about the proper interpretation of the association between physician density and per capita utilization of health services. While most studies find a positive relationship between supply of services and utilization, there is no agreement as to whether this reflects SID by physicians to protect their income, a supply response to unob- served variation in population health or an effect of the price andror volume of Ž services on access to and quality of health care Fuchs, 1978; Tussing, 1983; Rossiter and Wilensky, 1984; Stano, 1985; Cromwell and Mitchell, 1986; Feldman . and Sloan, 1988; Rice and Labelle, 1989; Dranove and Wehner, 1994 . Being no exception, Norwegian studies of SID have also produced conflicting results. While studies based on aggregate data indicate that SID is present, micro data analyses using patients or physicians as the unit of observation have produced Ž . the opposite conclusion Grytten et al., 1995; Carlsen and Grytten, 1998 . How- ever, if higher physician density is associated with higher satisfaction with primary physician services, policy-makers may conclude that SID, if it exists, does not constitute a practical problem in Norway, independent of what motives physicians may have. The paper is organized as follows. In Section 2, a simple theoretical model is developed to explain the meaning and consequences of SID when primary physician services are funded and regulated by the state, as is the case in Norway. The section also contains a discussion about how policy-makers can use survey data to calculate the socially optimal density of physicians without having detailed knowledge about the market for primary physician services. In Section 3, the survey data and our econometric specification are presented. The empirical results are presented in Section 4, and Section 5 provides concluding remarks.

2. Theoretical framework