آیا برنامه های سلامتی خطر-انتخاب ؟مطالعه ممیزی در بیمه سلامت اجتماعی آلمان
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|25545||2012||10 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Public Economics, Volume 96, Issues 9–10, October 2012, Pages 750–759
This paper evaluates whether health plans in Germany's Social Health Insurance select on an easily observable predictor of risk: geography. To identify plan behavior separately from concurrent demand-side adverse selection, I implement a double-blind audit study in which plans are contacted by fictitious applicants from different locations. I find that plans are less likely to respond and follow-up with applicants from higher-cost regions, such as West Germany. The results suggest that supply-side selection may emerge even in heavily regulated insurance markets. The prospect of risk selection by firms has implications for studies of demand-side selection and regulatory policy in these settings.
In competitive markets with regulated premiums, health plans have economic incentives to exploit predictable, unpriced heterogeneity in risk by selecting individuals who are low-cost within a premium group. This “cream-skimming” is inequitable and inefficient, and limits payers' ability to leverage high-powered payment systems to encourage efficiency in production (Newhouse, 1996 and van de Ven and Ellis, 2000). In addition to the policy concern, the possibility of supply-side selection also has implications for research on consumer behavior in competitive insurance markets, as unobserved concurrent activities by insurers can confound empirical tests of demand-side adverse selection (e.g., Fang et al., 2008 and Finkelstein and Poterba, 2006). This paper uses an audit approach to examine whether health plans select on geography, an easily observable predictor of unpriced risk. Regulatory approaches to contain risk selection aim at limiting plans' access to information about risk types, restricting mechanisms for selection and reducing the potential gains from selection by adjusting payments to more closely reflect individuals' expected costs. However, selection may yet emerge in markets with strict supervision and risk adjustment. In practice, even sophisticated adjustment methods are unable to eliminate all variations in risk, leaving substantial residual incentives for selection (Shen and Ellis, 2002b) which may even increase in the comprehensiveness of the adjustment formula (Brown et al., 2011). Moreover, while risk adjustment may mitigate gains from selection it can decrease incentives for efficiency by moving the payment system closer to cost-based reimbursement (van de Ven and Ellis, 2000). An optimal payment structure may maintain some uncompensated heterogeneity to balance this trade-off. Managing the selection-efficiency trade-off from risk adjustment is particularly challenging in the case of geography. The location of enrollees is readily observable by plans and correlated with expenditure risk, two conditions that facilitate cream-skimming. However, managing geography is not straightforward. On the one hand, geography has practical appeal as a simple composite index of costs, and accounting for spatial variations can contain potentially large selection incentives. On the other hand, geography is merely correlated with a multitude of cost drivers that regulators may prefer to address separately. In particular, risk adjustment should compensate only for legitimate differences in health care needs or resource costs, e.g. costs due to morbidity or input prices. Plans should be at risk for factors that they can potentially manage, such as practice styles or moral hazard.1 In actuality, geographic variations are due to both legitimate and objectionable factors (e.g., Fisher et al., 2003). The resulting policy trade-off can lead to residual incentives for selection on geography. Geographic variations are pervasive in many settings, and regulating this specific selection-efficiency trade-off is a recurring concern. In the US, geographic variations in health care spending have been well established and recognized by regulators (Dartmouth Atlas Working Group, 2011). In the Medicare Advantage (MA) program, private health plans receive a risk-adjusted capitation payment to assume the costs associated with providing benefits covered by the traditional fee-for-service (FFS) Medicare. The capitation payment varies according to the enrollee's risk (based on a range of disease conditions) as well as the county-specific base rate, which is a function of historical spending in the FFS program. The differences in the county rates are substantial. As illustration, in 2011 the unweighted MA rates per “aged” beneficiary ranged from $729 in Des Moines county, IA, to $1505 in Saint-Bernard, LA (KFF, 2011). Geography also features in insurance programs for the non-elderly population. The Patient Protection and Affordable Care Act (PPACA) of 2010 requires the use of risk adjustment for the individual and small-group markets both inside and outside the state health insurance exchanges. The law does not stipulate whether a geographic adjuster should be included but explicitly suggests Medicare Advantage's methodology as a model for adjustment in the exchanges. Similar to the Massachusetts health insurance exchange, PPACA also requires adjusted community-rating, allowing limited premium variation based on a number of factors, including rating areas. The risk adjustment and rating areas therefore require definitions and policy decisions on geography. In the context of MA and the exchanges, geography is mostly discussed as policy instrument to encourage plan entry (Mcguire et al., 2011). However, plans may also exploit any within-area differences and mismatches between rating and actual market areas. The threat of geographic risk selection is particularly serious since additional rules, including the remaining allowed rating factors, could artificially generate heterogeneous premium groups within rating areas and induce selection even in presence of complementary regulation (Pauly, 1984). Establishing geographic cream-skimming is ultimately an empirical question. The German Social Health Insurance (SHI) provides a useful context to identify selection on geography in a heavily supervised environment. Health plans in the SHI are not allowed to collect medical histories as part of the enrollment process, and they cannot refuse any applicant or vary premiums, benefits or provider networks. As in MA and the insurance exchanges, payments to plans are adjusted for the morbidity of enrollees. However, geographic variations remain a source of heterogeneity and a motive for cream-skimming. The risk adjustment system accounted for East/West differences until 2007, but a 2009 reform explicitly excluded geography from the payment formula. In an opinion on the reform, the German Constitutional Court recognized the existence of spatial variations in costs and their financial implications for plans (BVerfG, 2005). However, it argued that legitimate variations due to morbidity are sufficiently compensated by the new formula and that plans should face incentives to actively manage variations due to regional inefficiencies or patient preferences. As consequence of this policy, health plans have incentives to exploit geography to improve their risk structure and financial standing. The aim of this paper is to assess empirically whether plans act on the prevailing financial incentives to select on geography by focusing their recruitment efforts on applicants from low-cost areas such as East Germany. To separately identify cream-skimming from potentially concurrent demand-side adverse selection, I implement a double-blind audit study in which health plans are presented with fictitious applicants who have different addresses but are otherwise identical. I measure response rates for letters, emails, and phone calls, as well as the weight and stamp value of letters as proxies for insurers' resource expenses. The findings indicate that plans are more likely to respond to applicants from East Germany, a result consistent with cream-skimming even in this tightly regulated setting. The paper also highlights the value of the audit approach for examining firm behavior and is, to my knowledge, the first audit study of selection behavior by health insurers.
نتیجه گیری انگلیسی
The German Social Health Insurance is characterized by uniform benefits and community-rated premiums, and a limited set of tools for sickness funds to manage care and costs. In this setting funds face incentives to cream-skim low risks to improve their financial performance. Importantly, funds can easily observe the geographic location of applicants which is correlated with variations in costs but remains unaccounted for in the risk adjustment system. This paper considers whether sickness funds in the SHI select on geography. I implement an audit study in which funds receive requests for contract materials from applicants across Germany, and find that, for some response modes, funds are less likely to respond and follow-up on requests from applicants located in areas that are financially less attractive, such as West Germany. This differential response is unlikely to be due to marketing that is tailored to consumer differences, and could operate through varying recruitment efforts across office locations. From a policy perspective, the current SHI approach to geography is not on the frontier of the selection-efficiency trade-off: since the funds are restricted in their ability to manage regional inefficiencies, there is no scope for gains from imposing geographic risk. In this particular context, the welfare loss from selection is likely to be relatively small. Since funds are less likely to respond at all to high-cost applicants, the results imply that consumers choose among different sets of plans. Since the benefit package and provider networks are mostly identical across plans, this limited choice generates only small losses to consumers. In the worst-case scenario of full risk segmentation, high-cost applicants are stuck in plans that provide no supplemental benefits and charge the maximum supplemental fee of 1% of income, i.e., up to 36.75 Euro per month at the income threshold of 3675 Euro. Importantly, in Germany even shunned applicants maintain insurance coverage from their previous sickness fund. In the US, Medicare beneficiaries have the traditional FFS program as fallback if they cannot gain access to a MA plan. This is not the case in the US non-elderly market where supply-side risk selection can lead to significantly higher welfare costs as individuals may be unable to obtain adequate risk protection. US plans may also be more aggressive than their German counterparts which are subject to limits on total profits. The persistence of cream-skimming in a market as heavily regulated as the Social Health Insurance raises the question of additional strategies to contain selection by reducing the potential gains and available mechanism to select. First, the regulator could increase monitoring and penalties for this behavior, in effect raising the costs of skimming. Second, the risk adjustment could be enhanced by including geography, at the (efficiency) cost of also accounting for sources of variation that should be not compensated. In general, potential losses may be tolerable, particularly if individual access is given a large weight relative to welfare costs from entrenching inefficient provider and patient behavior (see also Glazer and McGuire, 2009). On a practical level, geographic adjustments involve a choice of the level of geography and the time horizon of adjustment. As Newhouse (1996) notes, the skewness of medical expenditures makes small-area adjustments unstable unless spending is averaged over longer time periods, as in the US Medicare program, trading off signal and noise. Third, the regulator could allow plans to use other tools to reduce costs, including the ability to negotiate with providers and to manage care. If plans can effectively reduce regional variations with these instruments, they may find geographic cream-skimming a relatively less appealing strategy. Finally, in the German context, some sources of regional variations could be eliminated by equalizing prices across regions and compensating funds for the income-related caps on the supplemental fee and copays. An alternative strategy has been proposed as part of “managed competition” (Enthoven, 1993) where a “sponsor” serves as intermediary between patients and plans, and in this role also manages enrollment. Enthoven explicitly envisions that this setup prevents screening and selection of applicants by plans, and acts as clearinghouse for information and transactions. This approach could also be used by the US health insurance exchanges, which can provide applicants with marketing material and withhold applicants' details until they are signed by a plan. This leaves room for alternative selection strategies such as dumping but these may be more easily detectable. The possibility of supply-side selection is also relevant to empirical research on demand-side selection in markets for health, long-term care and car insurance, and annuities (e.g., Chiappori and Salanie, 2000, Fang et al., 2008 and Finkelstein and Poterba, 2006).17 Observational studies of selection by consumers may need to account for concurrent actions by the insurer that are outside the pricing decision. For instance, Finkelstein and Poterba (2006) test for selection by considering whether individual characteristics that are not used in the pricing of annuities are nonetheless correlated with both the demand for insurance and risk occurrence. As in the German SHI, these “unused observables” can result from prohibitively high transactions costs and regulatory restrictions or, as in the case of the UK annuity market, from voluntary restraint by insurers. The possibility of concurrent supply-side selection on non-price margins could have important implications for these tests and provide a rationale for why firms forego valuable information in their pricing decisions.