C. VAN DE VOORDE ET AL fund to estimate the price elasticity of three types lization effects of co-payments. It is not impossi of users for three types of doctor visits using ble that any effect of supplier-inducement lags panel data methods. We estimated highly signifi- several years behind the initial consumer/patient cant out-of-pocket price elasticities for the general demand reaction, and partly or fully offsets initial population in the range from -0.39 to -0 28 demand reductions. Second, in order not to jeop for GP home visits, -0.16 to -0.12 for GP ardize equitable access to health care for those office visits and -0.10 for specialist visits. These with high needs of care, the co-payment increase estimates were generally lower-and in the differ- was introduced simultaneously with an income- ences model even non-significant--for the groups related maximum annual amount of cost sharing of elderly and disabled(WOPI). By lack of price per insured. This obviously mitigates the financial variation, no robust and significant price elastic- consequences of cost sharing for intensive users come elderly and disabled(WOPI, which is over time, also dilute the initial shock effect exempt from much of the co-payments In summary, this paper does provide some ex- We can draw the following conclusions. First, ternal validation of the Rand HIE results in a despite the obvious negative effects of demand context with favourable conditions for supplier- reductions, on average, physician incomes in a induced demand. It is clear that even in such a e setting, we find highl Significant context, the utilization of physician services is not negative short-term price elasticities for the ma- completely price inelastic, and that a price in jority of the population. This implies that, at least crease, therefore, does not only invoke cost shif in the short run, not all price-induced demand ing to patients, but also some demand reduction reduction can be countered by physician-induced An exception is the use of those elderly and demand for additional services. Since we could disabled groups not on low incomes. which ap- hat the longer-term consequences are, but the served reduction in utilization is mainly, in the available aggregate data suggest that pre-reform care is impossible to tell from this study. The time trends continue after the co-payment rise, high rates of doctor visits in Belgium-especially albeit at a lower level. Second, the general active those at the patients home--suggests that the population appears to be more price-sensitive han the WOPI group, for which we only found appropriateness of the utilization level and mix certainly deserves more detailed consideration small and non-significant price elasticities. This is but demand-side cost sharing is only one ap- consistent with the Hie finding that those in proach to the issue. Supply side cost sharing with the general population) are, ceteris paribus, ation than the current fee-for-service---warrants for the general population the utilization of GP at least as much consideration by Belgian health home visits is significantly more price responsive han the demand for (GP and specialist)office visits. This finding is consistent with those of Chiappori et al. 3, 14], who examined a similar ACKNOWLEDGEMENTS French natural experiment, where a 10% co-pay ment introduction did induce a negative demand The authors would like to thank Joris Diels, Jos Kesenne. Guy reaction for gp home visits. but not for office Tormans and Remi Van De Reyd of the Federation of Chri visits. They attributed this higher price respon- tian Mutualities in Brussels for providing the data used in thi siveness to the larger share of the monetary price paper, and Inge Duchesne and cindy vandoren for excellent in the total cost of a home visit, as opposed to an research assistance in an early stage of office visit van Vliet, Edward Norton and other participants in the Eu- Policy implications of our results have to be ropean Workshop on Econometrics and Health Econom drawn with some caution for several reasons. One Tinbergen Institute, Amsterdam, for comments on an earlie that owing to the limited number of post- the EUs BIOMED II Programme(contra act BMH4-cT98 reform observations, it is difficult and dangerous 3352) for the project 'Economic determinants of the distribu- o make any longer-term predictions on the uti- tion of health and care in Europe Copyright a 2001 John Wiley Sons, Ltd Health Econ.10:457-471(2001)470 C. VAN DE VOORDE ET AL. fund to estimate the price elasticity of three types of users for three types of doctor visits using panel data methods. We estimated highly significant out-of-pocket price elasticities for the general population in the range from −0.39 to −0.28 for GP home visits, −0.16 to −0.12 for GP office visits and −0.10 for specialist visits. These estimates were generally lower—and in the differences model even non-significant—for the groups of elderly and disabled (WOPI). By lack of price variation, no robust and significant price elasticities could be obtained for the group of low-income elderly and disabled (WOPI), which is exempt from much of the co-payments. We can draw the following conclusions. First, despite the obvious negative effects of demand reductions, on average, physician incomes in a fee-for-service setting, we find highly significant negative short-term price elasticities for the majority of the population. This implies that, at least in the short run, not all price-induced demand reduction can be countered by physician-induced demand for additional services. Since we could only observe the short-run effects, we cannot say what the longer-term consequences are, but the available aggregate data suggest that pre-reform time trends continue after the co-payment rise, albeit at a lower level. Second, the general active population appears to be more price-sensitive than the WOPI group, for which we only found small and non-significant price elasticities. This is consistent with the HIE finding that those in lower health (as the WOPI certainly are compared with the general population) are, ceteris paribus, less responsive than those in good health. Third, for the general population the utilization of GP home visits is significantly more price responsive than the demand for (GP and specialist) office visits. This finding is consistent with those of Chiappori et al. [3,14], who examined a similar French natural experiment, where a 10% co-payment introduction did induce a negative demand reaction for GP home visits, but not for office visits. They attributed this higher price responsiveness to the larger share of the monetary price in the total cost of a home visit, as opposed to an office visit. Policy implications of our results have to be drawn with some caution for several reasons. One is that owing to the limited number of postreform observations, it is difficult and dangerous to make any longer-term predictions on the utilization effects of co-payments. It is not impossible that any effect of supplier-inducement lags several years behind the initial consumer/patient demand reaction, and partly or fully offsets initial demand reductions. Second, in order not to jeopardize equitable access to health care for those with high needs of care, the co-payment increase was introduced simultaneously with an incomerelated maximum annual amount of cost sharing per insured. This obviously mitigates the financial consequences of cost sharing for intensive users (like, for example, the chronically ill), but may, over time, also dilute the initial shock effect. In summary, this paper does provide some external validation of the Rand HIE results in a context with favourable conditions for supplierinduced demand. It is clear that even in such a context, the utilization of physician services is not completely price inelastic, and that a price increase, therefore, does not only invoke cost shifting to patients, but also some demand reduction. An exception is the use of those elderly and disabled groups not on low incomes, which appears rather price insensitive. Whether the observed reduction in utilization is mainly in the domain of ‘frivolous’ and/or ‘unnecessary’ use of care is impossible to tell from this study. The very high rates of doctor visits in Belgium—especially those at the patient’s home—suggests that the appropriateness of the utilization level and mix certainly deserves more detailed consideration, but demand-side cost sharing is only one approach to the issue. Supply side cost sharing— including alternative systems of provider remuneration than the current fee-for-service—warrants at least as much consideration by Belgian health policy makers. ACKNOWLEDGEMENTS The authors would like to thank Joris Diels, Jos Kesenne, Guy Tormans and Remi Van De Reyd of the Federation of Christian Mutualities in Brussels for providing the data used in this paper, and Inge Duchesne and Cindy Vandoren for excellent research assistance in an early stage of this study. They are grateful to two anonymous referees and to Erik Schut, Rene´ van Vliet, Edward Norton and other participants in the European Workshop on Econometrics and Health Economics, Tinbergen Institute, Amsterdam, for comments on an earlier version of the paper. EvD acknowledges financial support of the EU’s BIOMED II Programme (contract BMH4-CT98- 3352) for the project ‘Economic determinants of the distribution of health and health care in Europe’. Copyright © 2001 John Wiley & Sons, Ltd. Health Econ. 10: 457–471 (2001)