Reconsidering theories and evidence of supplier induced demand

This paper seeks to make three contributions to the literature on supplier induced demand (SID). They concern, firstly, the criterion for accepting or rejecting SID. Secondly, its theoretical basis and thirdly the results of two Australian studies designed to test the existence of SID in the Australian medical market. In the first part of this paper (Section 2) it is argued that SID may be described as a ‘default theory’ which gains its plausibility from its ability to explain and predict service use. From this perspective SID remains an acceptable, instrumental, theory until it is shown to be unnecessary and that the alternative, more orthodox analytical framework has superior explanatory power. Despite this justification the persuasiveness of the theory is increased if there is a plausible explanation of patient and doctor behaviour which is consistent with it. It is argued that present agency theories are not fully satisfactory. In the second part of the paper two suggestions are made with respect to behaviour based upon the importance of uncertainty facing both a doctor and a patient. It is shown that medical uncertainty (as distinct from asymmetrical information) can explain both patient and doctor behaviour. In contrast with many theorists, SID is shown to be consistent with the ethical behaviour of a professionally and financially motivated doctor optimising demand shift in a competitive and dynamic market. The first empirical study summarised here used Australian cross sectional data. It is used as a vehicle for commenting on the problem of identification raised by Ramsey and Wassow (1986). The second study draws upon the unique patient and doctor incentives in Australian hospitals which allow a discriminatory test of the relative importance of patients’ and doctors’ incentives. The paper draws heavily upon a recently published article by the authors (Richardson and Peacock 2006). It includes arguments which were developed after this publication. These chiefly relate to the default justification of SID in the first section.