طرف عرضه و طرف تقاضا هزینه به اشتراک گذاری در تجدید ساختار بیمه سلامت اجتماعی : کدام موثرتر است؟

Microeconomic theory predicts that if patients are fully insured and providers are paid fee-for-service, utilization of medical services exceeds the efficient level (‘moral hazard effect’). In Switzerland, both demand-side and supply-side cost sharing have been introduced to mitigate this problem. Analyzing a panel dataset of about 160,000 adults, we find both types of cost sharing to be effective in curtailing the use of medical services. However, when moral hazard mitigation is traded off against risk selection, the minimum-deductible, supply-side cost sharing option ranks first, followed by the medium-deductible demand-side alternative, making the supply-side option somewhat more effective.

One of the main goals of health care financing systems is to promote efficient levels and types of care (Ellis and McGuire, 1993). If patients are fully insured and providers are paid fee-for-service, they desire larger than optimal quantities of health care services, connoted ‘moral hazard’. Zeckhauser (1970) and Zweifel and Manning (2000) have analyzed how demand-side cost sharing (in the guise of deductibles or co-payments) can be used as a corrective. However, demand-side cost sharing exposes consumers to financial risk, contradicting the very objective of insurance. Unless limited by a stop-loss, it also makes beneficial procedures unaffordable to some patients (Nyman, 1999). In addition, it might be considered unfair towards the chronically ill. These considerations have created interest in the alternative of supply-side cost sharing (in the guise of capitation or prospective payment). Because of their information advantage, providers of medical care can influence the demand for their services to a greater extent than other professionals (Arrow, 1963). Moreover, providers are less vulnerable to risk than patients because they can pool treatment cases. However, supply-side cost sharing might also promote a reduction in quality or the denial of beneficial but costly services, a phenomenon commonly termed stinting (Newhouse, 2002). Both demand-side and supply-side cost sharing have been empirically examined in terms of their effectiveness. The novelty of this paper is that it directly compares the expenditure effects of demand-side and supply-side cost sharing (and combinations thereof), using contract variants offered by the same health insurer. This has the advantage that many side conditions (underwriting policy, billing procedure) are kept constant. Moreover, the paper complements Lehmann and Zweifel (2004), who construct a proxy for unobserved health status from prior health care expenditure (HCE), by the two-stage residual inclusion estimator (2SRI, Terza et al., 2008). In this way, risk-selection effects are more fully controlled for. Finally, it extends the set of instruments influencing choice of plan but not HCE by including the premium for the baseline contract, the potential premium reduction for a restricted plan, the individual's credit record, and years of membership with the same fund. For the capitated plan, an additional instrument is a dummy indicating whether or not an Independent Practice Association (IPA) was operative in the individual's county of residence. The data come from Switzerland, a country where consumers have annual free choice of plan with no employer involvement. The chronically ill are not precluded from switching due to open enrollment. Low-income individuals (about 30 percent of the population) are eligible for premium subsidies. Receiving the subsidy, they are less likely to choose high-deductible plans, because a reduction in the income transfer to the sick state is particularly disadvantageous if the income effect on medical consumption is strong (Nyman, 1999). On the other hand, managed-care type plans that are too restrictive compared to premium charged will not be chosen. The remainder of this article is structured as follows. Section 2 contains an overview of the empirical literature. The policy setting is described in Section 3, while Section 4 is devoted to a description of the data base. In Section 5, we explain the econometric methods used to separate moral hazard from risk-selection effects and to deal with the very skewed distribution of the HCE data. The estimation results are presented in Section 6. Section 7 discusses policy implications in view of related literature, while the final Section 8 contains a summary and conclusions.

Managed competition in social health insurance aims at creating incentives for insurers to increase efficiency and to respond to consumer preferences while preserving solidarity between high- and low-risk types (Van de Ven et al., 2007). Therefore, it is important to know whether contractual innovations such as deductibles or capitated IPA plans achieve ‘true’ cost savings rather than merely serving as a means for risk selection. This research measures and compares the impacts of demand-side cost sharing (through voluntary deductibles) and supply-side cost sharing (through prepaid IPA plans) on individual health care expenditure (HCE), controlling for risk-selection effects. The data comes from a large panel of Swiss adults covering the years 2003–2006. Since unobserved health status influences both contract choice and HCE, a proxy is constructed from HCE during the first three years of the observation period, complemented by the residuals from the contract choice equation (the two-stage residual inclusion method proposed by Terza et al. (2008)). Higher annual deductibles and IPA plans are both found to achieve marked reductions of moral hazard. An increase in the annual deductible by CHF 200 (some EUR 133, from minimum to medium) is estimated to decrease the probability of positive HCE by almost 1 percentage point, while the IPA alternative might even be associated with an increase. In return, it achieves a reduction of positive HCE by some 12 percent, compared to only 7 percent of the medium deductible. Increasing the deductible by CHF 700 (some EUR 466) reduces the probability of reporting HCE by about 3.7 percent and the amount of positive HCE by about 18 percent. However, this effectiveness of demand-side cost sharing comes at the price of substantial risk-selection effects. Because voluntary cost sharing plans are especially attractive to low risks, such plans might lead to market segmentation and hence higher premiums for high risks. The most favorable ratio of moral hazard attenuation over risk selection is achieved by the IPA with the minimum deductible, amounting to 1:2.76 (CHF 355/980; HCE amounts to CHF 4610 for the baseline contract). The next-best alternative is the medium-deductible FFS contract with a ratio of 1:3.56 (CHF 250/891). According to this criterion, supply-side cost sharing is somewhat more effective than the demand-side alternative. Still, this research is subject to several limitations. First, since the data set only comprises individuals who were with one and the same insurer from 2003 to 2006, it fails to measure risk-selection effects associated with changes between competing insurers. Second, even ‘within’ risk-selection effects may not be controlled for perfectly. There is no guarantee that the HCE equation is correctly specified for the three preceding years, a necessary condition for obtaining residuals that serve as good proxies for unobserved health. The same caveat applies to the residuals of the contract choice equations. Thus, estimates of expected HCE reductions achieved by higher deductibles and IPA plans could still be biased. Third, deductible options have price and income effects. As shown by Nyman (1999), only the former should be counted as inefficient consumption. Finally, results relating to IPA plans have limited generality as long as they cannot be linked in detail to the incentives faced by participating health care providers. Nevertheless, the findings of this study permit one to draw the conclusion that allowing insurers to offer plans with both demand-side and supply-side cost sharing does generate ‘true’ savings in Swiss social health insurance. After controlling for risk-selection effects, both variants are estimated to achieve marked reductions in moral hazard that can be passed on to consumers in the guise of premium reductions without jeopardizing insurers’ solvency.