68 Assuming ? rst mover advantage, we play a Stackelberg game with the brand-name producer as price leader and the generic competitor as price follower. We will ? nd that both RP and MSAs are adequate frameworks to improve competition compared to simple proportional co-payment arrangements, but prices as well as manufacturer surpluses will be persistently lower under MSAs. Combined with serious classi? cation problems which come up with RP, our results clearly suggest a system of MSAs as the superior control instrument. The chapter is structured as follows: section 4.2 gives a short introduction into generic competition, which is the basis for RP and MSAs. In sections 4.3 and 4.4, functionality and empirical experiences of the two control-instruments are presented. Subsequently, in section 4.5 we start with the economic analysis. Extending a simple vertically differentiated duopoly model, we deduce equilibrium conditions and compare the respective impact of RP and MSAs on manufacturers’ price-setting behaviour, competition and surpluses. Section 4.6 compares the model results. Based on these results, in section 4.7 a reform proposal is given. Finally, section 4.8 draws the conclusions. 4.2. Generic Competition Once a pharmaceutical patent expires, generic replacement drugs may enter the market. Generics are exact copies of original branded drugs and contain the same active chemical ingredients. The existence of those perfect substitutes and the negligibly small economies of scale in the production of pharmaceuticals (Caves et al. 1991) lead to the expectation that the generic entry would take away the possibility of monopoly rents of brand-name incumbents. Prices should decrease to marginal costs. However, empiric studies of the US-market show a different picture:53 Contrary to expectations, Grabowski and Vernon (1992) revealed that prices for the 18 high sales-volume branded drugs they investigated increased by an average of 7 % one year after and 11 % two years after generic entry.54 At the same time, prices for generics signi? cantly decreased and were on average 35 % lower than their initial entry price after two years. Grabowski and Vernon’s descriptive results were proved by Frank and Salkever (1997), who used an econometric model to test a sample of 32 drugs which lost patent protection during the early to mid-1980s. They found evidence that brand-name prices increase, whereas generic prices strongly decrease after the 53 We refer to the US-market because it is widely unregulated and thus generic competition is not distorted due to speci? c public control instruments. In the US, the pharmaceutical sector is mainly characterised by market-driven controls through private health care insurers using managed care drug bene? t programs (see Danzon 1997). 54 The phenomenon that branded drug producers increase prices as a response to generic entry is called “Generic Competition Paradox”. The thinking is that demand is divided into two segments: price insensitive (“loyal”) and price sensitive (“non-loyal”) consumers. Since the generic drugs capture the non-loyal patients, manufacturers of brand-name drugs raise their prices in the price insensitive segment (Frank and Salkever 1992). 69 market entry of the latter. In a recent study, Marco (2005) also came to the conclusion that, on average, the prices of branded drugs increase in the face of generic entry. Not all studies have concurred. Caves et al. (1991) analysed 30 drugs which lost patent protection between 1976 and 1987 and they observed decreasing prices, but the amount of decline is quite small and depends on the number of generic alternatives. They suggested a decline of brand name price by 4.5 % on average per three generic entrants. Additionally, the generic price is about 50 % lower than the brand-name drug price. Once again a convergence of prices is minimal. The reasons for this seeming contradiction between theoretical predictions and empirical impacts of generic competition is caused by the demand- and supply-sided particularities of the pharmaceutical market:55 Although branded drugs and their generic equivalents are perfect substitutes, from the perspective of consumers and physicians they are not equal (Merino-Castello 2003, p. 7). In fact, uncertainty about the generics’ quality leads to an arti? cial vertical differentiation that segments consumers’ demand. Patients show loyalty towards drugs they had been using before. Thus, from their point of view, the switching costs associated in changing medical treatment results in a higher willingness to pay for a product of known quality (Klemperer 1995). The importance of the switching costs depends on the length of the period patients have already consumed the branded drug. Coscelli (2000) showed that patients who change medical treatment are mostly those who have only recently begun to take the drug. The impact of the arti? cial product differentiation is even intensi? ed since consumers usually do not face the full price of the prescribed drug, but only a proportional co-payment. The reduced price sensitivity results in a moral hazard and hence medical overtreatment. In addition, the supply side of the pharmaceutical market is characterised by agency and information imperfections. It is not patients, but physicians who determine the consumption choice. However, in general doctors neither bene? t ? nancially nor must they fear monetary losses from their prescription decisions. Consequently, they get no direct return on the investment for searching for information about the availability and ef? cacy of alternative pharmaceuticals (Hellerstein 1998, p. 111). They have no incentive to make the effort to convince their patients about the equivalency of expensive branded and cheaper generic drugs. Indeed, empirical studies deliver evidence that physicians have little knowledge of actual drug prices (Kolassa 1995). In sum, they do not behave as perfect agents for the patients and show habitual persistence in their prescribing behaviour (Coscelli 2000, p. 367), because they know about the price insensitivity of consumers, since mainly a third-party, the health insurer, bears the higher costs (Dranove 1989). This behaviour is commonly known as the supplierinduced-demand effect. Agency and information imperfections as well as the arti? cial vertical product differentiation give producers of branded drugs the continuing ability to in? uence prices even when generic products have entered the market. The period of exclusivity due to 55 A detailed overview is given in López-Casasnovas and Puig-Junoy 1999, pp. 14 - 17. 70 patent protection allows the innovator to build up a high-quality reputation and to capture high market shares. Thus, the ? rst mover pricing advantage results in a strong price-setting power in the pharmaceutical market even after patents have expired (Scherer and Ross 1990, p. 585). 4.3. Reference Pricing as an Insuf? cient Surrogate for Market Competition The previous section shows that the effects of generic competition on the price level of branded drugs are low or at least controversial. Besides the remaining price-setting power of the incumbents, after the patent protection expires insuf? cient co-payment arrangements are the critical factor for the failure of generic competition to contain pharmaceutical spending. If people do not have to pay the entire drug price out-ofpocket, but only a small proportional deductible, the impact of the price mechanism is limited. This failure to drive prices lower lead to concerns about the share of pharmaceutical expenditure on public funds and provoked Germany to introduce RP in 1989. Since then, many jurisdictions have followed or implemented similar therapeutic substitution programs for reimbursable drugs into their social health insurance systems. The objectives are always the same: promoting price competition and strengthening cost consciousness by means of enforcing ? nancial pressure. Although reimbursement limits for drugs and instruments for encouraging the consumption of cheaper medicines are nothing new,56 RP is different. It is based on the assumption that drugs can be classi? ed into medication groups which are therapeutically equivalent and clinically interchangeable and that a uniform reimbursement level for every group can be established (Schneeweiss 2007, p. 18). Although there are different kinds of possible realizations, in general RP is characterised by ? ve features (Lopéz-Casasnovas and Puig-Junoy 1999, p. 7): Groups of drugs are de? ned in terms of their interchangeability. They may or may not include patented products. A ceiling for the amount reimbursable for every group of drugs is determined by the third-party payer (public or private insurer). The respective reimbursement ceilings are calculated from the domestic prices of the drugs within the same speci? c group. The co-payment ceiling depends on the price of the selected drug and may be avoided if the drug does not exceed the reference price. The concept of interchangeability, the selection criteria for group composition and the reimbursement ceilings are frequently reviewed and changed when necessary. 56 The number of cost-containment instruments is legion. For a comprehensive overview see Mossialos (1998), p. 88. An international comparison is given in Ministry of Health (2006b). For Germany see Schreyögg et al. (2004).