Referrals in Healthcare Markets: An Experiment

Inefficiencies are pervasive in healthcare markets due to asymmetric information between patients and physicians. One source of inefficiency may be the general practitioners’ (GP) referral behavior of patients to specialists (e.g., Chandra and Skinner, 2012). Due to their remuneration, GPs may have an incentive to refer too few patients leading to undertreatment for some patient types. On the other hand, if GPs receive a substantial payment from the specialist for referring patients, GPs may refer all patients leading to inefficiently high treatment costs. Referral payments from specialists to GPs have been debated controversially as a means to enhance efficiency in physician-to-physician referrals. In all major Western healthcare markets, referral payments are prohibited as they are considered to be unethical. In the theoretical economics literature, however, referral fees are shown to increase market efficiency (see, e.g., Pauly, 1979, Garciano and Santos, 2004). Clean empirical evidence on how referral payments affect physicians’ behavior and market efficiency is still lacking. In order to provide causal evidence, we address the following research questions by means of a laboratory experiment: How does the introduction of a referral fee affect GPs’ diagnosis effort? How does a referral fee affect GPs’ referral decision? How is the market efficiency affected by a referral price?

We introduce a novel experimental design and an illustrative theoretical model to derive  behavioral predictions. In our framed laboratory experiment, we consider two types of patients:  suffering from a minor or a severe illness. GPs decide about whether to exert effort to diagnose a patient and whether to treat or to refer a patient. GPs can only heal minor illness, specialists heal both illnesses but at higher costs. Specialists always have to provide the utility maximizing treatment for a patient. Both physicians are paid per capitation and bear their treatment costs. In the main experimental treatments, we exogenously vary the referral fee: no payment, reflecting the current policy (baseline treatment), intermediate level (treatment 1) and high level (treatment 2).Results in fact provide evidence for our theoretic predictions and suggest that intermediate levels of referral payments are most efficient. In the baseline treatment, GPs treat both patient types. Hence, inefficiencies arise from undertreatment of patients with severe illness. Under high referral payments, GPs mostly refer both patient types. This results in unnecessary specialist treatments of patients with minor illnesses. Our behavioral results suggest that the referral payments between specialists and GPs can be an effective means to enhance efficiency in medical service provision.