Within the discussion about the increasing expenditures in health insurance, the overutilization of medical care is often attributed to the existence of a moral hazard problem. Since moral hazard has a great impact on health insurance policies, there is a growing interest in the economic literature to identify and to measure its effects. Although the problem of overconsumption of medical care does not mean moral haz- ard per se, the determination of the latter may reduce its scope and help to mitigate the problem of overutilization. The main objective of this paper is an empirical evi- dence of the moral hazard phenomenon. By analysing the economic literature on moral hazard in health insurance this paper seeks for examples of its empirical evi- dence, whereby the emphasis lies on distinguishing between the demand-oriented (especially ex-post) and the supply-oriented (external) moral hazard.
„Insurance, by lowering the price of medical care, induces people to get something good for themselves without having to pay its full cost. Knowing they are insured, they use medical care they wouldn’t use if they had to pay the full cost.” (Stone, 2011, p. 888) According to the concept of the homo oeconomicus each individual is a utility maximizer, acting rationally and optimize all available opportunities. However, the increase of utilization in health insurance is usually not viewed as rational but car- ries a strong implication of immorality. When people use more health care services, they act selfish and fail to take social responsibility (Stone, 2011, p. 888). The eco- nomic literature describes this possible increase in usage of medical care by lowering the marginal cost to the individual as moral hazard (Pauly, 1968, p. 535).
In the meantime, moral hazard has become a well-known political term and the most powerful mean to influence health policies (Germany is not an exception). In this con- text there was a lot of discussion in recent years about measures which could reduce the increasing demand of health care and the associated “cost explosion” in the German statutory health insurance (Bertelsmann Stiftung et al., 2006/Hoh, Hone- kamp, 2010). And although the increase in expenditures for health care does not mean moral hazard per se, the identification of the latter may help to mitigate the problem of overutilization. However, the empirical determination of moral hazard is due to its complexity rather problematic. The traditional (conservative) economic lit- erature on moral hazard in health insurance targets the consumers of health care and does not cover the implications on the supply side. But as health care demand is supply-induced, it is not less important to analyze moral hazard of the health care providers. Thus, it is important to distinguish between the demand-oriented and the supply-oriented moral hazard to quantify its effects. The objective of this paper is to review former empirical data, which measure moral hazard in health insurance from these two different perspectives.
This seminar paper is organized as follows. The first section begins with the origin of the term and the definition of moral hazard in health insurance. The second section is going to perform some examples of empirical evidence. It will review the data distinguishing between the demand-side and the supply-side moral hazard. The third section will give a brief conclusion to the discussed topic.
2. The phenomenon of moral hazard
Moral hazard is the foundational idea of the insurance branch, which has been stud- ied by many economists and has influenced insurance policies. Here is an example: The person who buys an insurance wants to be protected against the monetary damages. As the insured thinks that the insurance will cover his potential losses, he may act more risky than he has to carry possible costs by himself. The observed phenomenon is no less important in the health insurance branch. Moral hazard be- came an important question in the modelling and build-up of the German health in- surance system created in the 19th century (Guinnane, Streb, 2010, p. 4). The main point of moral hazard in health insurance could be described as follows: when people are insured, they use more medical care services than they don’t have health insur- ance (Stone, 2011, p. 887).
According to Dembre and Boden (2000) the term “moral hazard” was first used in the 17th century in the discussions of insurance to explain “the possibility that insurance would encourage the insured party to take on additional risk in a way which could not effectively be monitored” (Dow, 2010, p. 3). The term had negative connotation meaning immoral actions of the involved parties. From the 1960s the expression “moral hazard” appeared in the mainstream economic literature. In health insurance this term was spread through the papers of Kenneth Arrow (1963) and Mark Pauly (1968), who used the expression to refer to the subjective attitude to risk. In his arti- cle “Uncertainty and the Welfare Economics of Medical Care” Arrow focused on the incomplete market of health insurance and the significance of the uncertainty as un- quantifiable risk. As Arrow explained, the incidence of illness is a mean event and the efficiency of treatment is uncertain. Furthermore, the consumer and the supplier have different information on the latter (information asymmetry). In the view of the above he offered possible solutions limiting moral hazard under uncertainty, as licensing standards for health care professionals or concepts of trust and delegation (Arrow, 1963, pp. 964-966).
In “The Economics of Moral Hazard” Pauly shifts the focus away from uncertainty and ethics towards more conventional analysis of the incentives: “The problem of moral hazard in insurance has, in fact, little to do with morality, but can be analyzed with orthodox economic tools” (Pauly, 1968, p. 531). He emphasized that the increasing usage of health care services through insured is not immoral but rational: “The re- sponse of seeking more medical care with insurance than in its absence is a result not of moral perfidy, but of rational economic behaviour. Since the cost of the individual's excess usage is spread over all other purchasers of that insurance, the individual is not prompted to restrain his usage of care” (Pauly, 1968, p. 535). Pauly considered that it is not possible to eliminate moral hazard completely, but such cost sharing models as deductibles and coinsurance could limit its scope.
All in all, the articles of Arrow (1963) and Pauly (1968) show important findings about the moral hazard theory:
1. The market of health insurance is incomplete. No single insurance policy is the best or the most efficient.
2. The pricing theory of medical care industry differs from the competitive norm. From the consumer’s point of view the price of medical services is zero.
3. As the price mechanisms differ from the norm, the result of this is the overcon- sumption on the demand side. So, the consumer doesn’t care about the price of medical care paid by the insurance or spread over all other insured.
4. And as the consumer doesn’t care about the price, there is an additional prob- lem: The supply-induced demand of medical care leads to the overconsump- tion on the supply side (Arrow, 1963 and Pauly, 1968).
Thus, the welfare implications of health insurance are ambiguous. One the one hand, the income transfer from healthy to the ill increases welfare, on the other hand the rational response of the individuals - to maximize their utilization - works welfare decreasing (Koc, 2005, p. 98).
In the literature on insurance theory there are several types of moral hazard. First of all, it can be divided into legal and illegal. The illegal moral hazard falls within the definition of insurance fraud and will not be discussed in this paper. The legal form includes the internal moral hazard, which takes place between two parties of the in- surance contact and the external moral hazard arising in connection with behavioural changes of health care providers (e. g. health care professionals or pharmacists). Furthermore, the internal form can be divided into - ex-ante and ex-post moral haz- ard. The ex-ante moral hazard refers to behavioural patterns of an insured, which increases the extent or the likelihood of sickness before the insured gets sick, e. g. unhealthy lifestyle, the lack of willingness to preventive measures or carelessness. The problem of ex-post moral hazard is based on the possibility of influencing the extent of health care costs by the insured after the sickness episode, e. g. costs for different treatment methods (Hoh/Honekamp, 2010, pp. 12-13). The following sec- tions are mainly dealing with the internal ex-post and the external moral hazard.
3. The empirical identification of moral hazard
Although moral hazard is one of the important findings in health insurance, it is rather difficult to verify its existence empirically, especially in case of ex-ante moral hazard. For instance, it is difficult to prove that a smoker increases his cigarette consumption when he has health insurance or a mountaineer acts more risky when he is insured. The situation is a bit different in case of ex-post moral hazard. The decrease of de- mand for physiotherapy or rehabilitation services due to greater coverage shows that consumers use these services only if they don’t have to pay for them. Indeed, the existence of moral hazard can only be identified in an indirect way, e. g. measuring, how growing coinsurance affects health care demand (Hoh/Honekamp, 2010, p. 13).
The price elasticity of health care demand conditional on health status is a quote common practice to quantify moral hazard effects in health insurance (Boes/Gerfin, 2013). The price sensitivity of demand measures the changes in utilization and ex- penditures and the results have been used to estimate changes in welfare. The ef- fects of overconsumption on welfare are very important in debates on the designing of cost coverage models in health insurance (Frick/Chernew, 2008, p. 3). However, since the demand for health care is usually measured by total costs for health care, not units of health care services, the major challenge in measuring the price sensitiv- ity is the definition of the price. Keeler et al. (1977) demonstrated that consumers use marginal, not the effective price making their decisions on consumption. Using the shadow price leads to less price responsiveness of the consumer. However, due to the complexity of the model the effective price has not been used in applied work un- til recently (Boes/Gerfin, 2013). The determination of the correct price is not the ob- jective of this paper. Using the existent data, this paper should provide a review of the former empirical results.
3.1 Full insurance and health care demand
One of the most powerful research papers, which studies how the demand for medi- cal services responds to the amount of out-of-pocket expenses is the RAND Health Insurance Experiment (HIE). Although the experiment results were performed in the middle of 1970s, many assumptions and recommendations on coinsurance are still based on the RAND results. The main goal of the experiment was to study the changes in demand patterns of consumers with different insurance contacts. The par- ticipants of the RAND experiment were randomly chosen family members (n=5809) of six US sites. Participating families were assigned to one of the insurance plans with coinsurance rates of 0, 25, 50 and 95 percent. Each insurance contract had an upper limit of annual out-of-pocket expenses 5, 10 or 15 percent of the family income, up to 1000 dollar in the maximum (Manning/Newhouse et al., 1987, p. 4).
The results of the HIE show that higher coinsurance rates decrease the demand for medical care. Table 1 provides the sample means for several measures of use of health care services - physician visits, outpatient, inpatient and total expenses.
Table 1: Sample means for annual use of medical services per capita
(Cf. Manning/Newhouse et al., 1987, p. 19)
Abbildung in dieser Leseprobe nicht enthalten
According to HIE results the per capita expenses in the insurance plan without out-of- pocket expenses are 45 percent higher than those with 95 percent coinsurance rate. The insurance plans with the coinsurance rates of 45 and 50 percent lie between these two extremes. A similar trend was observed in the outpatient expenses. The outpatient expenses of the free insurance group were 67 percent higher than of the 95 percent group, whereby outpatient visit rates to physicians were 66 percent higher. The largest statistically significant decrease in outpatient expenses between the other groups occurs between the contacts with 0 and 25 percent coinsurance rate. The changes in the usage of inpatient services could be observed more weakly:
The inpatient expenses on the free insurance plan were 30 percent higher than those on the 95 percent plan (Manning/Newhouse et al., 1987, p. 18).
The RAND experiment is surely one of the most important studies in health economics. It is remarkable to note that the results of it are largely consistent with nonexperimental literature, which indicated that price elasticity of demand for health care vary from -0.1 to -2.1. The results of the RAND experiment show that price elasticities are in the rage of -0.1 to -0.2 (Manning/Newhouse et al., 1987).
The indicated connection of the demand for medical services and the amount of out- of-pocket expenses can be represented with the theoretical model of J.-M. Graf von der Schulenburg (2000), later used by Christoph Vauth and Henning Osmers (2004) in their research paper. They consider an estimation model of treatment costs (K) of an individual, in which K(0) stands for treatment costs by full insurance and K(α) for costs by coinsurance with a different coinsurance rate α. The variable β describes by what percentage the costs would reduce in case if the insured has out-of-pocket ex- penses. The coinsurance model, where β equals zero indicates that K(0) and K(α) differs only by the amount α paid by the insured, however, the total treatment costs remain the same. The model of treatment costs (K) is given by:
Abbildung in dieser Leseprobe nicht enthalten
According to it the percentage of cost changes β is: αሻ ሻ
Abbildung in dieser Leseprobe nicht enthalten
Under assumption that the treatment costs are identical with the cost conditions in the insurance contract following substitution is possible: β = 100-φ, whereby:
Abbildung in dieser Leseprobe nicht enthalten
The new variable T(α) means the given insurance contract (tariff) with a coinsurance rate of α.