Do financial incentives shape physicians' prescription behaviors

Commissioned by: The Institute of Economics, Zagreb 
Project duration: February 17, 2017 – September 30, 2017 
Project manager: Ivan Žilić
Collaborator: Alexander Ahammer

Brief outline: 
Ideally, physicians are perfect agents. They diagnose and provide treatments in the way patients would if they had perfect information. In reality, however, we observe profound variations in the provision of medical care which cannot be explained by demand-side heterogeneity. Understanding sources of these variations is crucial for policy making, since welfare losses may occur when patients are over or undertreated. In this work we focused on one specific potential source of variation—financial incentives. In Austria, physicians are allowed to dispense drugs themselves in the form of on-site pharmacies, which makes them entrepreneurs and agents at the same time. On-site pharmacies are permitted primarily for the purpose of ensuring unhindered access to medical drugs in areas where no regular pharmacies exist. However, maintaining an on-site pharmacy allows physicians to earn a mark-up on every drug they prescribe. This is sometimes referred to as the “gray area of medicine“—physicians may exploit their informational advantage to prescribe medication the patient’s health status does not actually require, for the purpose of maximizing own income.   

We used administrative data from the Upper Austrian Sickness Fund to check whether having an on-site pharmacy altered the prescription behavior of Upper Austrian general practitioners (GPs). The Upper Austrian Sickness Fund covered the universe of medical consultations in Upper Austria containing 23,820,854 medical consultations. To identify a causal parameter of interest, we used a novel identification strategy that gave us a sample of randomly allocated patient–physician matches. In particular, we restricted our sample to drugs prescribed on weekends and public holidays where GPs in Austria rotate to provide out-of-hours service in order to ensure the provision of basic health care. Using this novel approach which accounted for patient-GP sorting, as well as fixed-effects model, we were able to identify the causal effect of maintaining an on-site pharmacy on drug and medicine dispense behavior. The results of our research contributed to the scarce literature on the effect of physician self-dispensing on prescription behavior, as well as to the public debate on how to improve efficiency of healthcare systems while providing universal and quality care.