51 time-series dataset has been gathered from the SHI Pharmaceutical Index. It is comprised of the years from 1983 to 2004, so 22 data points are being used for the regression. The data have been de? ated with the Pharmaceutical Price Index (year 2000 = 100 %).39 Endogenous variable: Real SHI-Drug Expenditure per Insurant (Expt). 3.5.2. The Exogenous Variables It is usually assumed that there exist two exogenous determinants for drug expenditure: age structure of the population and technological progress. Additionally, the basic model takes co-payments of the insurants into consideration. For the next step we try to include political interests by using different dummy variables. Thus the following expenditure determinants come into consideration: a) age structure, b) technological progress, c) co-payments and d) political interests. a) Age structure: Pharmaceutical consumption is heavily depending on age.40 Despite comparably high infantile drug usage due to children’s diseases and a temporarily boost between an age of 15 to 19 (mainly caused by sex hormones for girls), drug consumption stays at a very low level of below 200 De? ned Daily Dosages per year on average. But after the lowest average, between the age of 20 to 25, drug consumption begins to increase slowly but consistently. Beyond the age of 40 there is a dramatic, exponential boost to the acceleration of drug use. The maximum is reached between the age of 85 and 90, which averages more than fourteen times of the drug consumption of a 20 years old person. In 2003, more than 50 % of the total drug expenditure were prescribed to people beyond 60 years old, although they represented only 26 % of the insured people (? gure 3.2). 39 Source: SHI Drug Index of the Scienti? c Institute of the Health Care Fund, published in Schwabe and Paffrath (volumes 1985 - 2006). 40 Although state of health depends on age, it is absolutely controversial if the ageing of the population demographic leads directly to increasing health care expenditure; therefore long-ranged prognoses of drug expenditure development are highly precarious (see chapter 2). But for the analysis in this chapter the dynamic interrelation between age and drug expenditure has only minor relevance because we are only interested in temporary ? uctuations of expenditure development. 52 Figure 3.2: Drug Consumption in De? ned Daily Dosages per Head 2003. 0 200 400 600 800 1000 1200 1400 1600 0- 4 5- 9 10 -1 4 15 -1 9 20 -2 4 25 -2 9 30 -3 4 35 -3 9 40 -4 4 45 -4 9 50 -5 4 55 -5 9 60 -6 4 65 -6 9 70 -7 4 75 -7 9 80 -8 4 85 -8 9 ? 9 0 D ef in ed D ai ly Do sa ge s Age Source: Nink/Schröder (2004a), p. 1105. Figure 3.3: Age Dependency Ratio 1983 - 2005. 0,20 0,21 0,22 0,23 0,24 0,25 0,26 0,27 198 3 198 5 198 7 198 9 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Age Ratio 60 + Year Source: Based on data from the Federal Statistical Of? ce. From 1991 including the new German Laender. The relationship between age and drug consumption can be added into the regression by using the ratio of insured people who are older than 60 years to the total number of insurants (? gure 3.3). This age limit is of particular importance because it equals the average retirement age in Germany and thus prejudices the amount of gross income and the contribution assessment basis respectively. Exogenous variable 1: Age Dependency Ratio 60 (Age??t). The coef? cient is expected to have a positive sign. 53 b) Technological progress: Including technological progress is dif? cult because the introduction of new control instruments encourages the pharmaceutical companies to change their product-, research-, and price-policy. Hence technological progress becomes indirectly an endogenous variable of the political decision-making process (Blankart and Wolf 2005). Thus strategic market behaviour inhibits using, for example, the number of newly introduced patent-protected drugs as a measure for medical progress. Figure 3.4 illustrates this problem. Figure 3.4: Turnover Share of Patent-Protected Preparations on Total Drug Consumption, 1993-2003. 0 5 10 15 20 25 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Turnover Shar e Year Innovative Preparations Analoga Source: Nink and Schröder (2004b), p. 161. Patent protection itself does not prove the quality of a newly introduced drug. Real innovations, i.e. formulas which have pharmacologically new active ingredients, must be distinguished from formulas with merely marginal modi? cations, so-called “analoga” or “me-too drugs”. The companies’ product-mix, this means the decision to produce either highly innovative drugs or analoga, is mostly dependent on the regulatory environment. This can be seen by taking a look into the SHI’s history of reform acts: the exclusion of patent-protected drugs from reference pricing in 1996 led almost immediately to an increase of turnover shares for analoga and to a decrease of real innovative drugs. This was just a result of strategic market behaviour of pharmaceutical manufacturers, since research costs for analoga are comparatively low and the absence of any price control guaranteed high margins (see Schreyögg et al. 2004). In return, the re-inclusion of analoga resulted in a decreased market shares for them. This development already started in 2002 because it was very likely at that time that the government would implement the re-integration into the reference price scheme. Consequently, innovations in the pharmaceutical sector hinge on politics and cannot be directly implemented as an exogenous variable into our regression. To compensate for this, like Breyer and Ulrich (2000) instead of the number of patent-protected drugs, we use the basic wage total, which determines the amount of 54 contributions, of the SHI members as a surrogate for productivity advancements and technological progress.41 Since unemployed dependants and children were exempted from contributions, we do not concentrate on the number of insured persons but on the number of SHI members .42 Exogenous variable 2: Real Basic Wage Total per SHI member (BWaget). The coef? cient is expected to have a positive sign. c) Co-payments: Co-payments immediately raise the direct burden for patients. Accordingly, under the premise that on average pharmaceuticals are normal goods, it can be assumed that higher out-of-pocket payments decrease drug consumption. Since the sum of co-payments depends on the extent of total drug expenditure, we use the share of co-payments on SHI drug expenditure per insured person (? gure 3.5). Figure 3.5: Share of Co-Payments on Total Drug Expenditure per Insured Person 1983 - 2005. 0 2 4 6 8 10 12 14 16 198 3 198 6 198 9 19 92 19 95 19 98 20 01 20 04 Sh ar e o n To ta l Ex pe nd itu re s i n % Year Source: Based on Schwabe/Paffrath (several volumes). 2004 and 2005 estimated. From 1993 including the new German Laender. Exogenous variable 3: Share of Co-payments on Total Drug Expenditure per Insurant (CPayt). The coef? cient is expected to have a negative sign. 41 Using basic wage total as a proxy for increasing productivity may underestimate the cost pressure due to technological progress. Especially in the health care sector, technological progress is dominated by cost driving product innovations. An estimated linear increasing variable may be adequate to compensate the underestimation (Breyer and Ulrich, 2000). Therefore, we include a time trend into the cointegration relationship of the ECM in section 3.5.5. 42 In the SHI, only their members are liable for contributions. Children and unemployed spouses are exempted from charges; thus, the amount of insurants exceeds the amount of members. 55 d) Political interests: The impact of politicians’ interests is considered in our model by including a political dummy variable. Crucial to the speci? cation of the dummy variable is the choice of the assumed political behaviour. Scienti? c literature particularly distinguishes between partisan and opportunistic motives. Partisan theory traces back to Hibbs and is based on the assumption that political parties serve the interests of their core constituencies.43 Accordingly, in respect to traditionally strong relationships between right-wing governments and the medical industry as well as the medical fraternity, increasing expenditure in periods with a Christian-Liberal coalition in power can be supposed, whereas disburdening the patients at the expense of the suppliers should appear in the years of left-wing governments. Thus the dummy variable takes the value of plus one in years of a Christian-Liberal coalition (1983 - 1998) and minus one in years of a Red-Green coalition (1999 - 2004) (? gure 3.6a). Opportunistic theory can be dated back to Nordhaus (1975) who assumes that selfinterested politicians use macroeconomic policies to gain votes. The ambition of politicians is to cut health care expenditure by aiming at decreasing or at least stabilizing the non-wage labour costs, reducing unemployment as well as acquiring votes by disburdening the insured persons. In contrast to burdening physicians, who act as opinion multipliers due to their intensive contact to patients, it should be very attractive for politicians to cut costs at the expense of the pharmaceutical industry. On the one hand, pharmaceutical expenditure are one of the largest cost-pools in the SHI and on the other hand, political decisions against the powerful pharmaceutical industry are very popular among the public. One can suppose that savings at the expense of the pharmaceutical industry are more likely to gain votes than bene? t cuts or increasing contribution rates. The dummy variable which re? ects opportunistic behaviour takes the value of minus one in years of federal elections and zero in years without elections (? gure 3.6b). Unlike the usual scheduling of federal elections in the second half of the year in 1983 and 1987, the election took place in the year’s ? rst quarter. For this situation, opportunistic politics is already assumed for the pre-election year, i.e. 1986 (no value for 1982 because of data limitations). Next we test a combination of partisan and opportunistic motives like Frey and Schneider described in 1978. They showed that political parties serve the interests of their electorate in years without elections, but behave opportunistically in election years. Thus the dummy variable takes the value of plus one in years when expansionary policies are anticipated, i.e. in years without federal elections and with a rightwing government in power, and minus one when a drug expenditure contraction is expected, i.e. in the years of elections or in years where a left-wing government is in power (? gure 3.6c). In the last step we try to include the interests of the pharmaceutical industry itself. The industry is aware of the more supply-side friendly orientation of right-wing governments. The Liberals are an especially important political mediator for the industry. 43 An overview of partisan theory is given in Hibbs (1992). 56 b. Opportunistic Behaviour. c. Partisan and Opportunistic Behaviour. d. Partisan, Opportunistic and Corporatistic Behaviour. -1 0 1 198 3 198 4 198 5 198 6 198 7 198 8 198 9 199 0 199 1 199 2 199 3 199 4 199 5 199 6 199 7 199 8 199 9 200 0 200 1 200 2 200 3 200 4 -1 0 1 198 3 198 4 198 5 198 6 198 7 198 8 198 9 199 0 199 1 199 2 199 3 199 4 199 5 199 6 199 7 199 8 199 9 200 0 200 1 200 2 200 3 200 4 -1 0 1 198 3 198 4 198 5 198 6 198 7 198 8 198 9 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 -1 0 1 198 3 198 4 198 5 198 6 198 7 198 8 198 9 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 Figure 3.6: Political Dummy Variables. a. Partisan Behaviour. 57 From the manufacturers’ perspective, the probability for unpleasant reform acts would increase with a left-wing government in power. Consequently, we suppose that industry has incentives to support the election chances of a Christian-Liberal coalition. Of course, it cannot be assumed that the pharmaceutical industry is able to counteract politics permanently.44 Nevertheless it uses various instruments in order to in? uence, at least temporarily, drug expenditure: since advertising is substantially restricted for prescription drugs, manufacturers have concentrated on the information of experts, i.e. physicians, pharmacies and hospitals to have in? uence on prescription purchasing behaviour. The physician as the agent of the patient has a twofold advantage: neither does the patient have the necessary information about the drug market, nor does he have the medical knowledge to assess the pharmacologic effectiveness of the drug concerning his disease. Thus the demand is determined by the supplier. However, even medical experts lack suf? cient information because there are 40,000 known pharmaceutical products in Germany and a great number of new drugs which ? ood the market each year (Schwabe 2004a). The pharmaceutical industry tries to exploit this de? ciency “to position the drugs in such a way that they will be prescribed”.45 In Germany, there are 16,000 sales people engaged in the pharmaceutical industry and they visit physicians and pharmacies about 25 million times per year (Glaeske and Janhsen 2007, p. 8 ff.). Together with public relations, information events and advertising for over-the-counter drugs, the pharmaceutical industry spends almost 30 % of its overall turnover for promotion.46 The massive attempts to convince physicians to prescribe their products are only one part of the pharmaceutical industry’s market strategic behaviour. The industry also tries to evade regulatory instruments by means of coordinated actions in their pricesetting and product policy (Perschke-Hartmann 1994). Since a reciprocity of interests between government and industry exists during a Christian-Liberal coalition, “corporatistic” behaviour must be implemented only in the years of a left-wing government. We assume that the possibilities for in? uence by industry are narrowed during a Red-Green era, thus industry is hardly able to perpetually counteract cost-cutting ambitions of a Red-Green government. It is feasible, though, that at least in election years industry can consciously manipulate drug expenditure e.g. by more marketing activities. To compensate for this, we lift the dummy variable from minus one to plus one in 2002. Thus we test corporatistic behaviour in combination with partisan as well as opportunistic politics (? gure 3.6d). Exogenous variable 4: Political dummy variable (PDumt) which considers partisan, opportunistic and corporatistic behaviour. In all cases the coef? cient of the dummy variable is expected to be positive. 44 If the industry were able to increase drug expenditure permanently, then of course it would try to do it. 45 Citation from a medical representative job description at P? zer (http://www.p? zer.de, Dec. 2006). 46 In comparison to this, the automobile industry pays approximately 6 % of the turnover for advertising expenditure (Breyer et al. 2003, p. 422). 58 3.5.3. Model Speci? cation I – First Differences We use OLS regressions to test the existence of a statistically signi? cant relationship between drug expenditure and political interests. We hypothesise the following relationship: . (3.1) Thus we assume that the amount of drug expenditure in year t per insured person (Expt) is dependent on the age structure of the insurants (Age??t), the basic wage total per SHI member (BWaget), the share of co-payments on total drug expenditure (CPayt) and politicians’ interests (PDumt). Since the estimations are based on annual values (from 1983 to 2004) we presume static relationships. All monetary variables represent real values de? ated with the pharmaceutical price index (2000 = 100 %). Starting in 1993, the data for the new German Laender are included. Before we can start the regression analysis, it is ? rst necessary to check the time series for stationarity, i.e. if their means and variances are constant over time. If the time series is non-stationary, the problem of spurious regressions arises and OLS results will be inef? cient. Therefore we use the Augmented Dickey-Fuller (Dickey and Fuller 1979) and the Philips-Perron (Philips and Perron 1988) tests to verify the existence of unit roots. It can be supposed that the time series of drug expenditure, age ratios and total wages are probably trend dependent. Accordingly, a trend variable has been included into the stationary tests. Only co-payments reveal no consistent time trend and hence only the constant term has been considered. Table 3.1: Stationarity Tests. Augmented Dickey-Fuller Philips-Perron Variable Regression t-Statistic t-Statistic log (Exp) Level 1st diff erence c, t c -3,2 -4,612*** -2,887 -5,835*** log (Age??) Level 1st diff erence c, t c -0,781 -2,851* -1,047 -2,811* log (BWage) Level 1st diff erence c, t c -0,715 -3,247* -0,715 -3,215* log (CPay) Level 1st diff erence c c -1,67 -3,944*** -1,67 -4,312*** Note: *** Non-stationarity rejected at a 1% critical level; * at a 5% or 10% critical level. c = constant, t = trend; t-statistics based on MacKinnon critical t-values. As table 3.1 shows, it is not possible to reject the existence of a unit root for any of the time series. However, non-stationarity can be rejected for all variables in ? rst differences. So we can assume that the variables are I(1), i.e. they are integrated of order one, and follow a random walk.