MAX-PLANCK-INSTITUT FÜR GESELLSCHAFTSFORSCHUNG MAX PLANCK INSTITUTE FOR THE STUDY OF SOCIETIES

Transcription

1 MAX-PLANCK-INSTITUT FÜR GESELLSCHAFTSFORSCHUNG MAX PLANCK INSTITUTE FOR THE STUDY OF SOCIETIES MPIfG Discussion Paper 05/9 Ending Cooperation: A Forma Mode of Organizationa Change in German Pharmaceutica Interest Representation Andreas Broscheid MPIfG Discussion Paper

2 Andreas Broscheid Ending Cooperation: A Forma Mode of Organizationa Change in German Pharmaceutica Interest Representation MPIfG Discussion Paper 05/9 Max-Panck-Institut für Geseschaftsforschung Kön Max Panck Institute for the Study of Societies Coogne November by the author(s) MPIfG Discussion Paper ISSN MPIfG Discussion Papers are refereed schoary papers of the kind that are pubishabe in a peer-reviewed discipinary journa. Their objective is to contribute to the cumuative improvement of theoretica knowedge. The papers can be ordered from the institute for a sma fee (hard copies) or downoaded free of charge (PDF). Downoads Go to Pubications / Discussion Papers Max-Panck-Institut für Geseschaftsforschung Max Panck Institute for the Study of Societies Paustr Coogne Germany Te Fax info@mpifg.de

3 Abstract This paper uses a game theoretic mode to investigate the spit-up of the main German pharmaceutica association into an association for mutinationa, researchbased companies and an association representing sma and medium-sized firms. In order to expain the breakdown of cooperation among group members, the paper empoys a mode that combines coective action and bargaining. The mode suggests that changing issues can pay an important roe in organizationa change. In the case of the German pharmaceutica industry, the key issue changed from pharmaceutica registration to cost contro. With respect to the former issue, the different factions in the pharmaceutica industry had compementary interests and were abe to find a compromise position. The atter issue ed to a confict over the redistribution of scarce resources between the factions; formay, this meant that a issue dimensions were equay important to a factions. As a resut, it became impossibe to reaize poicy gains through a poitica compromise. Zusammenfassung Die fogende Studie untersucht den Austritt forschender Arzneimittehersteer aus dem Bundesverband der Pharmazeutischen Industrie (BPI) zum Jahreswechse 1993/1994 und die darauf fogende Gründung des Verbands forschender Arzneimittehersteer (VFA). Unter Zuhifenahme eines spietheoretischen Modes, das Verhandungstheorie mit der Theorie koektiven Handens verbindet, wird hier argumentiert, dass Änderungen im Issuekontext organisatorischen Wande verursachen können. Im Fae der deutschen Pharmainteressen wurde die Zuassung von Arzneimitten as dominantes Issue von der Frage der Kostenkontroe abgeöst. Im Fae der Arzneimittezuassung konnten die verschiedenen Hersteergruppen Kompromisspositionen erreichen, während im Fae der Kostenkontroe zu einem schwer ösbaren Verteiungskonfikt zwischen innovativen, OTC- und Nachahmerprodukt-Hersteern führte, der auf die Gründung getrennter Verbände hinausief.

4 4 MPIfG Discussion Paper 05/9 Contents Introduction 5 The BPI and pharmaceuticas poicy before A theory of coaition buiding in interest groups 10 Basic assumptions 14 Constant association poicy: The free-rider outcome 17 Association poicy as a function of member contributions 21 Summary of resuts 22 Back to German pharmaceutica interests 23 Changing issues: Safety versus price contro 23 Seective incentives: The changing Red List 26 Concusion 27 Appendix: Forma presentation of mode, proofs 30 A.1 Basic assumptions 30 A.2 Constant c 31 A.3 c as a function of bˆ 33 References 34

5 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 5 Introduction In 1993, more than 100 German and internationa pharmaceutica companies eft the German Federa Association of the Pharmaceutica Industry (Bundesverband der Pharmazeutischen Industrie, BPI), the association that had represented their interests since The resut was the foundation of two new pharmaceutica associations, one for the generics industry, and another for the innovative, mutinationa companies. The atter group, the Association for Research-Based Pharmaceutica Companies (Verband Forschender Arzneimittehersteer, VFA), quicky repaced the BPI as the most infuentia German pharmaceutica association. Its members comprise 60% of the prescription drug market, and it accordingy pays an important roe in nationa and internationa fora, such as EFPIA (the European Federation of Pharmaceutica Industries and Associations) and recenty in the German Roundtabe in the Heath Care Sector (Frankfurter Agemeine Zeitung 1993b,d,f). Here is a puzze: Why did an association that had been successfu and powerfu for more than 40 years suddeny break apart? The iterature on interest group formation and maintenance does not provide concusive answers. The puraist tradition (McFarand 1991; Truman 1951; Voge 1989) emphasizes the roe of socia confict and the strength of the enemies of business: If industry interests are threatened, they wi organize. This is not the case here; the crisis occurred in a period when the pharmaceutica industry was on the defensive. Despite the externa threat, the industry did not ray around its major association but spit up into different factions. Rationa Choice theorists in the Osonian tradition argue that interest group formation and maintenance is the resut of a set of seective incentives (Moe 1980; Oson 1965; Wison 1973). This woud suggest that groups fai because members are no onger interested in their services. This does not fit the BPI s case. On the one hand, the spit-up in the pharmaceutica industry was accompanied by furious intra-group conficts members seemed to be very interested indeed. On the other hand, the BPI retained a arge number of members, which suggests that at east some seective incentives might sti have worked. Buiding on the Rationa Choice approach to interest group formation and maintenance, I propose a forma mode that wi hep us to understand cases ike the one encountered in the German pharmaceutica industry. Simiar to Oson, I assume that interest groups have to overcome coective action probems, which they can do with the provision of seective incentives. However, in contrast to other coective action modes, I assume that group members do not agree on which coective good they shoud pursue. The coective good consists in the pursuit of a poitica position, and group members differ I thank Pieter Bouwen, Simone Burkhart, David Coen, Jürgen Feick, Fabio Franchino, Lothar Krempe, Andrew McFarand, Wofgang Streeck, and the participants in seminars at the Max Panck Institute for the Study of Societies and the Schoo of Pubic Poicy at University Coege London for their hep and feedback. Aso, I thank severa members of German pharmaceutica associations for the opportunity to interview them for this project. A shortcomings are the author s faut.

6 6 MPIfG Discussion Paper 05/9 in their evauation of different poicies. Therefore, the group has to negotiate a common position, in addition to deaing with coective action probems. I wi show that under certain conditions, group organizers can design institutiona mechanisms that use poicy confict to overcome coective action probems. If conditions change, groups that reied on such institutiona mechanisms to ensure cooperation might fai. Why shoud we care? First, if we want to know who infuences poicy, we need to know who is we organized, since we-organized interests can communicate their message more ceary and infuence poitica choices more easiy. Second, it has been observed that the infuence of particuar socia actors changes over time (McFarand 1991; Voge 1989). In order to understand the reasons for these fuctuations in infuence, we have to find the factors that might change. In this paper, I propose that the poitica agenda is one such factor that infuences the abiity of groups to provide incentives for different interests to cooperate. Third, the case under study is substantivey important. The BPI represented one of the important German industries of the twentieth century. Its crisis has aarmed other groups, which are afraid of simiar things happening to them. 1 If we want to understand the process of German pharmaceuticas poicy in the second haf of the 1990s, we have to investigate the factionaization of pharmaceutica interest representation. The section foowing this introduction provides some substantive information on the functioning of the BPI before the 1994 spit-up, on the main issues that have been important for the association since the 1960s, and on the attempts made in 1993 to reorganize and save the association. This wi serve as a background for the subsequent theoretica discussion and the formuation of a game theoretic mode. The concusions drawn from the mode wi then be used to anayze the empirica case; conversey, the empirica case provides us with a first informa assessment of the usefuness of the mode. The forma discussion of the mode, incuding a proofs, can be found in the appendix. The BPI and pharmaceuticas poicy before 1994 The confict between research-oriented mutinationa firms and pharmaceutica SMEs that ed to the spit-up of the BPI focused on organizationa issues. Therefore, it wi be usefu to survey the BPI s forma structure. 2 Founded in 1951, the BPI is the odest post-war pharmaceutica association in Germany. 3 In 1984, it represented 506 member companies, which accounted for 95% of phar- 1 Persona interview with a German interest association representative, Apri 17, If not otherwise noted, the foowing is based on Groser (1985). 3 The more speciaized Bundesfachverband der Heimitteindustrie (federa sector association of the proprietary medicines industry, BHI), which originay represented ony the over-the-counter sector, was founded a few years ater. In the eary 1980s, it was renamed Bundesfachverband (now Bundesverband) der Arzneimitte-Hersteer, BAH.

7 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 7 maceutica production in Germany. Since its membership fees constitute a proportion of income (0.12% of turnover, of which 0.02 percentage points are used as membership fees for the German chemica association), the BPI had comparativey arge resources and a comparativey arge staff. Athough membership fees are weighted by firm size, the BPI s decision making processes are rather egaitarian. The association s main governing body, the genera assemby, consists of 80 to 120 deegates seected by the BPI s regiona associations. The number of deegates seected within a region depends on its reative membership size, which means that the principe one firm one vote is the basis of the interna decision making process. The genera assemby seects the executive committee and some of the board members. The hierarchica structure of the BPI is based on region, not on sub-sectora characteristics. Before the German re-unification, the association was divided into 10 regiona associations, whose boundaries corresponded to the regiona divisions of the German chemica industry association (VCI), and roughy to the German Länder. (After re-unification, the regiona structure was consoidated into 7 regiona associations.) The regiona associations seect some of the members of the association board. In addition to the regiona associations, the BPI contains severa poicy-oriented and sub-sectora committees and sections that dea with specific issues such as over-the-counter drugs, phytopharmaceuticas, or veterinary medicine. These fora offer opportunities for direct firm participation. Simiar to other arge German business associations, the BPI has not merey been a obbying group but has aso engaged in sef-reguatory activities, for exampe in the areas of advertising 4 and information directed at physicians. In 1981, the BPI estabished a codex for its members that contained rues and guideines for advertisement. 5 Vioation of these rues was punishabe by excusion from the association. This constituted a rather powerfu seective incentive, since association membership was the precondition for incusion in the so-caed Red List of medicines. This ist of a medicines produced by BPI members, aong with their indications, was sent to practicing physicians every year. The fact that the Red List constituted a powerfu incentive to join the BPI is documented by a court case reported by Groser. This case invoved a former BPI member who was excuded from the association due to the vioation of some of its advertising reguations. 6 The company sued to have its products incuded in the Red List, but ost. Subsequenty, it decided to adhere to the advertising rues and rejoin the BPI (Groser 1985: 49). 4 In Germany, the advertisement of pharmaceutica products to the genera pubic is stricty imited; pubic advertisement of prescription medicines, for exampe, is prohibited. 5 The codex aso reguated other areas, such as producer participation in procedures to monitor the post-marketing performance of medicines. 6 This case took pace before the formuation of the codex, and referred to restrictions regarding the size of medicine sampes that coud be given to physicians.

8 8 MPIfG Discussion Paper 05/9 Taken at face vaue, the confict of 1993 was about demands by a group of companies that were organized in the so-caed Medica-Pharmaceutica Research Society (Medizinisch Pharmazeutische Studiengeseschaft, MPS) to change the BPI s organizationa structure. 7 The main point of criticism was the power of the regiona associations, which, it was argued, ed to a cumbersome decision making process. In addition, arge research-oriented firms compained that their infuence in the BPI was smaer than their financia contribution (Frankfurter Agemeine Zeitung 1993c). Even after the VFA had been founded by the members of the MPS (and other arge firms), the BPI secretariat tried to transform its association into an umbrea group that woud be abe to integrate the VFA as a corporate member. This attempt faied to obtain the required two-thirds majority of the December 1993 BPI assemby (Frankfurter Agemeine Zeitung 1993a,b). The difficuty with such a purey organizationa expanation of the 1993 confict and its consequences is that the BPI was abe to work very effectivey for more than four decades before the break-up. Aso, for a ong time, the industry had been divided into arge companies invoved in research-oriented production for an internationa market and SMEs focusing mosty on generics and other ess research intensive medicines for a domestic market (incuding phytopharmaceuticas and homeopathic medicines, which are reativey popuar in Germany). It is not obvious why these reativey stabe factors ed to the radica change in the eary 1990s. In order to understand the change, we need to ook at the changing issues that dominated pharmaceuticas poicy. In the ate 1960s and eary 1970s, the main issue was the registration of pharmaceuticas, and the criteria used for this process. Beginning in 1961, pharmaceuticas had to be registered with the Federa Bureau of Heath (Bundesgesundheitsamt), and the sae of heath-threatening medicines was prohibited. However, it was not necessary for producers to prove the safety and effectiveness of their medicines. In 1964, foowing the thaidomide affair, documentation of safety and effectiveness were required, but the reguations were sti not as strict as in other countries, such as the United States (Murswieck 1983: ). As internationa trade in pharmaceuticas increased, there was aso increasing pressure to adjust registration requirements according to WHO guideines. This pressure came, in part, from arge, export-oriented companies, whereas smaer companies were reuctant to agree to stricter reguatory requirements (Westpha 1982: 213 ff.). In 1976, a new pharmaceuticas aw introduced stricter registration requirements that at east party foowed the American mode 8 (Murswieck 1983: ). 7 The main activities of the MPS focused on pubic reations and support of basic research rather than on poitica obbying (Groser 1985). 8 Exceptions are in the areas of phytopharmaceuticas, homeopathic, and other traditiona medicines.

9 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 9 Since the ate 1970s, the so-caed heath care cost exposion has dominated heath care poicy, particuary pharmaceuticas poicy. The 1977 Heath Insurance Cost Containment Act reduced the reimbursement of medicines for certain indications (common cod etc.), increased prescription fees for patients, and introduced imits on the overa pharmaceutica expenses of sickness funds (Webber 1988). In 1989, the first of a series of heath reform aws passed by the center-right government introduced a negative ist of medicines that were excuded from reimbursement by statutory sickness funds. In 1991, this ist was extended to incude not ony medicines for indications that were routine and not dangerous, but aso medicines with ow economic vaue or ow therapeutic effectiveness. Furthermore, the 1989 aw introduced a system of so-caed fixed prices for medicines using the same therapeutic substances (this system was ater to be extended to medicines with simiar therapeutic substances). Sickness funds woud reimburse ony the fixed prices, and patients had to pay the difference between the fixed and the actua price (Schneider et a. 1993). 9 The price contro measures of 1977 and 1989 primariy affected producers of non-innovative medicines. These medicines were either subject to price contros, or they had to be paid for by patients, which reduced the eagerness of doctors to prescribe them. In contrast, the 1992 Heath Care Structura Reform Act affected the arge, innovative companies. It introduced a budget for the pharmaceutica expenses of the statutory sickness funds, increased patient co-payments for medicines, and imposed a five-percent price reduction on medicines not subject to fixed prices (Banke/Perschke-Hartmann 1994; Kirkman-Liff 1999; Perschke-Hartmann 1993). Since physician compensation woud be reduced if budgets were exceeded, physicians were carefu when prescribing particuary expensive medicines a practice that particuary affected the innovative, patented medicines. Furthermore, innovative medicines tended not to be subject to fixed prices, and were therefore subject to the mandatory price reduction. It is interesting to see that the three major issues in pharmaceuticas reguation since the 1960s provoked a division of interests between arge, innovative, mutinationa companies and sma and medium-sized companies targeting the German domestic market. The question is: Why did it take unti the 1990s for these conficts to ead to the formation of separate associations for the two factions of pharmaceutica producers? The foowing theoretica considerations attempt to provide an answer by investigating the interaction between the structura characteristics of the BPI and the nature of the saient issues. 9 The fixed prices woud be determined by the associations of physicians and sickness funds (Webber 1989).

10 10 MPIfG Discussion Paper 05/9 A theory of coaition buiding in interest groups Beginning with Oson (1965), the existing iterature has focused on interest group creation and maintenance rather than group faiure and change. 10 The reason for this is we known: The puraists had assumed that interest group formation was basicay automatic, and Oson argued that there was a probem that is, a coective action probem. Faiure was considered the defaut; group creation had to be expained. The question of group faiure and change was addressed, if at a, ony as a minor topic, or impicity. Wison (1973: 31), for exampe, mentions a number of factors that constitute threats to organizationa maintenance. However, the issue has not received systematic attention. The ack of attention to interest group faiure is not such a terribe probem, since expanations of group success aso impy expanations of faiure. I wi buid here on these impicit expanations of faiure, and for that purpose I wi summarize the main arguments that we can find in the iterature. However, I wi maintain that these arguments either do not hep us understand the case being investigated in this paper, or are not sufficienty formaized to be appied to specific cases. This paper intends to remedy that situation. Oson s forma mode suggests ony one condition for the provision of coective goods: The existence of a privieged group that is, the presence of one group member whose benefits from the provision of the coective good are so high that she prefers to provide the good on her own. Since it is cear, empiricay, that there are circumstances under which non-privieged groups (Oson cas them atent groups) organize, Oson provides additiona arguments to expain group formation. The most important of those arguments reate to the size of the groups and the provision of seective incentives. The theory of seective incentives is a straightforward extension of Oson s mode: Contributions to the provision of a coective good are associated with the receipt of a private good. Equivaenty, the refusa to contribute to the coective good subjects the defector to punishment. As a resut, the coective good is the by-product of the provision of private goods. In contrast, the argument that sma groups are more ikey to overcome coective action probems is not ceary inked to Oson s forma mode. His argument smaer groups have ower organizationa and monitoring costs, and they can use socia incentives does not directy foow from his main argument. Later studies show that the sma-group argument is not necessariy compatibe with Oson s basic mode (Sander 1992). 10 In fact, the question of interna interest group poitics has been somewhat out of fashion. Research has focused more on the poitica infuence of competing obbyists (Andersen/1995; Austen-Smith 1993; Austen-Smith/Wright 1992; Wright 1996) or on anayses of systems of interest representation (Fakner 2001; Gray/Lowery 1988; McFarand 1991; Streeck/Schmitter 1991) and not on organizationa foundation or maintenance. This means that the question of group faiure has been even further negected.

11 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 11 Authors such as Wison (1973) and Moe (1980) extend the concept of seective incentives, identifying various types of them. Wison distinguishes between materia, specific and coective soidary, and purposive incentives (32 47). Moe emphasizes non-economic incentives: In addition to purposive and soidary incentives, he mentions norms, peer pressure, and the impact of authority. The emphasis on seective incentives, however, prompted the question of whether the argument did not repace one coective action probem with another: If seective incentives were provided to induce cooperation, there were positive externaities invoved, and the seective incentives therefore aso had characteristics of coective goods. The soution was found in organization theory, particuary in the concept of the entrepreneur. Poitica entrepreneurs were identified as those actors who organize contributions to coective goods and suppy seective incentives (Frohich et a. 1971; Moe 1980; Wison 1973). The behavior of entrepreneurs, in turn, was expained by the existence of opportunities to profit from the organization of groups. With respect to interest groups, entrepreneurs coud make a iving as association secretaries. What are the impications that we can draw from the Oson-Wison-Moe ine of argument for an expanation of interest group faiure? First, Oson s size argument woud suggest that group maintenance becomes more difficut as the number of potentia group members increases. This argument has to be interpreted subty, however, since Oson s concept of group size is somewhat unusua: The size of a group is the minimum number of members who can gain from cooperation. 11 Therefore, one hypothesis might be (roughy): As those members of the group who vaue the coective good most ose interest in the good, it is more ikey that the group wi dissove. With respect to interest groups, this means that groups decine if the poitica process and its outcomes diminish in importance for their members. Second, the seective-incentives argument impies that interest groups fai if the seective incentives they provide ose vaue or cannot be provided anymore. This argument is more compex and difficut than it seems to be at first gance. We shoud expect entrepreneurs to suppy other seective incentives if one shoud fai. However, it is possibe that the suppy of new seective incentives might be too costy for the entrepreneur to pursue. In addition, it might not even be possibe to produce new seective incentives that can repace the ones that have ost their vaue. For the researcher, it is very difficut to determine which possibe seective incentives are avaiabe in a particuar case. In fact, the definition of the set of avaiabe seective incentives is part of the game of interest group formation; the entrepreneur s imagination and inventiveness are important factors in this regard. Therefore, it is very difficut to investigate empiricay whether it was the ack or the cost of seective incentives that ed to group faiure. 11 For a smaer number of members, it woud not be worth cooperating because their costs of cooperation exceed the aggregate gain from the coective group. The smaest group is a privieged group; this impies that there is one member who prefers to provide the coective good on her own.

12 12 MPIfG Discussion Paper 05/9 Third, the roe of the entrepreneur suggests potentia expanations for interest group faiure. If the suppy of abe secretaries in a particuar sector decines, some groups might ack effective organizers as staff members retire or join other organizations. The reason for such a decine in competent entrepreneurs coud be the reative decine of the profitabiity of group organization as compared to other entrepreneuria projects. Again, this might be reated to ack of interest in the group, for exampe as poitics becomes ess important for an industria sector. Empiricay, however, this argument is not particuary credibe: In particuar, the existence of arge consutancies that offer commercia organizationa support indicates that interest groups are commerciay worthwhie, from an entrepreneur s perspective. Furthermore, their existence suggests that entrepreneuria resources are avaiabe. There are two further strands of argument about the formation of interest groups that shoud be mentioned here. First, empiricay oriented studies of interest groups have noted the importance of the interaction between a group and its environment (Moe 1980; Schmitter/Streeck 1999). With respect to poitica interest groups, the roe of the government can be very important. In the European Union context, for exampe, the European Commission has been important as a sponsor of some groups (for exampe, BEUC). The roe of business associations in reguatory poicy making can ead to rather powerfu seective incentives (for exampe, because physician reimbursement by statutory sickness funds is channeed in Germany through regiona doctors associations, ony physicians who are association members can receive reimbursement). The decine of some groups, therefore, can be the resut of government poicy; an exampe is the decine of British abor unions as a resut of neo-ibera poicies. Dereguation can aso ead to a decine in the sef-reguatory activities that groups perform, and hence to a oss of seective incentives. In addition, the change in saient poicy issues might reduce the reevance of poitics to the group members, and thereby reduce the vaue of coective behavior. Lasty, theories of conditiona cooperation argue that cooperation can be an equiibrium outcome of games in which the decision of one actor to cooperate can be made conditiona on the cooperation of other actors. For exampe, Axerod shows that in repeated Prisoners Diemma games, cooperation can be a Nash equiibrium strategy (1984). 12 In order for this to happen, however, the actors have to have a arge time horizon. A possibe expanation for group faiure, therefore, coud be that it has become more ikey that the interaction between groups wi end in the near future. In other words, group faiure coud be the resut of a sef-fufiing prophecy. The probem with this argument is that it is difficut to determine what ending interaction woud mean in an interest group context. In forma terms, it means that the game ends (or is ikey to 12 Other conditions for conditiona cooperation have been noted. Hardin (1982), for exampe, discusses overapping group activities that is, the possibiity that a set of actors wi interact in different contexts. Defection in one context can then be punished by defection in another context.

13 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 13 end) in the near future. Does this mean that the reevance of the poitica arena is ikey to vanish for a particuar group? Does it mean that some group members wi cease to exist for exampe, due to economic probems? Some expanations of group faiure that I deduced from the existing iterature are not appicabe to the case that I am investigating here. The argument that group members ose interest due to the decining reevance of poitics for their industry is not usefu because the breakup of the BPI happened at a time when the government was intervening heaviy in the pharmaceutica market: Poitics was highy reevant; the breakup was accompanied by strong poitica conficts within the group. It does not seem pausibe that it was a ack of poitica entrepreneurship that caused the crisis. In fact, the BPI s executive director during the crisis, Hans Rüdiger Voge, was activey trying to change the association in a way that coud accommodate the dissenters demands (Frankfurter Agemeine Zeitung 1993e). In addition, the fact that the BPI s spit-up ed to the formation of two new associations indicates that there was not a ack of entrepreneurship. With respect to conditiona cooperation, the interaction between different actors in the pharmaceuticas sector was not ikey to end any time soon and it hasn t ended yet (athough some actors now beong to different associations). Which eements of the existing theories are usefu for the case under study? First of a, seective incentives are important. I wi show that one of the probems turned out to be the fact that the most powerfu seective incentive was not a purey private good. However, the faiure of seective incentives cannot provide the compete picture. If it is ony the oss of seective incentives that eads to group faiure, we shoud observe a reativey peacefu dissoution members simpy ose interest. As I have aready mentioned, this was not the fact in the present case. Second, the nature of poitica issues is important. We wi see that changing poitica issues payed a major roe in destabiizing the association. A third eement that wi be important in the present anaysis is the heterogeneity of group members. However, I approach heterogeneity in a manner that is different from Oson s. For Oson (and other coective action theorists), heterogeneity refers to the question of how highy different actors vaue the benefits of coective action. These vauations are basicay fixed. I argue, in contrast, that the heterogeneity of group members does not stem from the fact that some demand more or ess of a good, but rather from the fact that different actors demand substantivey different goods. In order to mode the substantive heterogeneity of interests in a group, I make use of concepts empoyed in theories of bargaining. In particuar, I argue that we can use socaed spatia modes to anayze the different poitica options that different actors prefer, and we can determine which compromise positions ead to different eves of utiity for the various actors (Eneow/Hinich 1984). As a resut, I combine an investigation of coective action and bargaining.

14 14 MPIfG Discussion Paper 05/9 The combination of bargaining and coective action aows me to introduce a type of mechanism for conditiona cooperation that is different from the one proposed by Axerod. In my mode, the actors can make their contribution to the coective good conditiona on their infuence in the bargaining process. This means that under certain conditions, coective action can be individuay rationa. This is where the nature of poitica issues comes in: The mechanism of conditiona cooperation that I propose is strong (that is, a actors stricty prefer to cooperate) ony if the poitica issues and, accordingy, the actors utiity functions have certain characteristics that I wi define in the foowing mode. Basic assumptions Let us assume that there are three factions 13 attempting to infuence pharmaceuticas poicy. We ca one of them SME and another MNC. Athough the names are purey coincidenta and do not infuence the resuts of this investigation, et us say that SME is a faction of sma and medium-sized pharmaceutica companies, and MNC consists of mutinationa, research-oriented companies. The third faction that tries to infuence poicy, SHI, can be thought of as representing statutory heath insurances. At the start of the game, SME and MNC form a coaition, which we ca BPI. We are mainy interested in the question of whether SME and MNC wi continue their coaition or whether they wi dissove it. We are not interested in SHI s strategy; we incude it in the mode to increase reaism, athough, as it turns out, we coud just as we eave it out of the mode. The different factions make poicy proposas on pharmaceuticas poicy; the factions that are part of a coaition make a common poicy proposa. I formaize this with a spatia mode, using weighted Eucidean distances as a basis for utiity functions. I assume that there is a two-dimensiona poicy space. For exampe, et us say that the issue is cost contros. Then we can say that one dimension represents the overa reduction in medicine prices, and a second dimension represents whether the price reduction affects mosty innovative medicines or generics. As a further exampe, consider the issue of pharmaceuticas registration. Again, we can identify a two-dimensiona poicy space. One dimension is the strictness of registration contros, and the other dimension is the degree of governmenta or industria contro of reguatory decisions. For each faction, we can identify the highest poicy preferences using so-caed idea points that is, points in the poicy space that identify each faction. I use the symbos i SME, i MNC, and i SHI, to identify the idea points of the three factions. 13 I use the term faction here in the Madisonian sense factions in society. I prefer this term to other possibe choices groups, interests, etc. because it is genera, indicates difference of interest, and we do not easiy confuse factions with organized associations.

15 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 15 The poicy benefit each faction receives from a poicy outcome is based on the distance between its idea point and the outcome. We conceptuaize the poicy outcome as a point, o, in the (two-dimensiona) poicy space, so that o = (o x,o y ). We then say that the poicy benefit for a faction I use SME as an exampe is the negative of the distance between its idea point and the outcome: i SME o SME = (i SME,x o x ) 2 + (i SME,y o y ) 2. (1) Now, this conceptuaization of SME s benefit assumes that the faction evauates both poicy dimensions of the issue equay. This might make sense for an issue such as cost reduction: SME wants itte overa cost reduction, and arger cost-reduction for innovative medicines (which tend to be produced by the arger companies). However, this might not be true for other issues. Take, for exampe, pharmaceuticas registration: It has been argued that MNCs preferred stricter registration requirements that woud increase their products internationa competitiveness. In addition, they preferred a reativey high eve of government contro to assure confidence in internationa markets (in contrast, SMEs preferred a ower eve of government contro than the CMEs; cp. Westpha [1982: 217 ff.]). However, strictness of registration requirements was the more important issue from the MNCs perspective, whereas the impementation of the requirements was secondary. Let us say that strictness is represented by the x-axis of the poicy space (outcome o x ), and government contro by the y-axis (outcome o y ). Then, MNC s poicy benefit woud be i MNC o MNC = α MNC,x (i MNC,x o x ) 2 + α MNC,y (i MNC,y o y ) 2 (2) with α MNC,x > α MNC,y. In technica terms, the poicy benefit is conceptuaized as a negative weighted Eucidean distance (Eneow/Hinich 1984). The different factions have obbying budgets, which I denote by b MNC, b SME, and b SHI, respectivey. I assume that the factions spend a of their obbying budgets. The rationae for this assumption is the fact that, once obbying budgets are aocated, their costs are sunk and it is rationa to spend them. 14 The main question that the mode poses is whether SME and MNC maintain their coaition, in the form of an association. I assume that there is a poitica entrepreneur who organizes the association, and who prefers for the association to exist. 15 The entrepreneur has to submit an association poicy proposa and offer seective incentives. There are severa actors who can be such poitica entrepreneurs. Usuay, it is an association s secretariat that pays this roe, by proposing and negotiating association positions and arranging for the services that an association provides (such as information about important poicy issues, studies on industry deveopments, etc.). 14 Stricty speaking, then, this is not an assumption but a proposition that can be proven; I state it as an assumption to save space. 15 We can formaize this by saying: If the association is successfuy estabished, the entrepreneur receives a utiity of 1; otherwise, she receives a utiity of 0.

16 16 MPIfG Discussion Paper 05/9 I assume that the factions, or the coaition/association, are agenda setters in the sense that one of their proposas becomes poicy. This assumption is based on the consideration that more resources are required from poicy makers mosty in terms of expertise to deveop independent poicy proposas. Such a resource-based argument is particuary credibe in highy technica poicy areas such as heath care poicy. The agenda setting roe of factions can be impicit in the sense that they may not make a proposa at the beginning of the poicy process. They may just as we defend the status quo, which impies that they do not propose a new poicy soution. They are agenda setters, however, in the sense that one of their (impicit) proposas wi become the government s poicy choice. The probabiity that a faction s poicy proposa wi become aw depends upon the resources that the faction puts into obbying for the proposa. This probabiity is based on Tuock s ratio probabiity mode (1980). As an exampe, assume that MNC spends its entire budget obbying for its idea point, i MNC, and that nobody ese obbies for this poicy. Then, the probabiity that i MNC wi become aw is p b MNC ( imnc ) =. (3) bmnc + bsme + bshi Now, assume that the entrepreneur has proposed association poicy c, and SME decided to contribute a share of its budget ca it bˆ SME to support c. Simiary, MNC decides to support c with bˆ MNC. Then, the probabiity that c becomes aw is bˆ MNC + bˆ SME p( c) =. b + b + b MNC SME SHI The utiities that the factions reaize at the end of the game consist of the possibe poicy benefits (that is, the negative Eucidean distances), weighted by the probabiities of their impementation, pus the seective incentives, s, provided by the entrepreneur. Taking MNC as an exampe, u MNC = p(c) i MNC c MNC p(i SME ) i MNC i SME MNC p(i SHI ) i MNC i SHI MNC +s (4) ( i MNC i MNC = 0, therefore this term drops out of the equation). The game sequence can be summarized as foows: A poitica entrepreneur proposes a coaition (association) between SME and MNC. This incudes proposing a common poicy position and offering seective incentives for joining the association. The two factions consider which share of their obbying resources they shoud contribute to the coaition/association, and which share they shoud spend on obbying for their own most preferred poicy.

17 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation A actors obby, and the government chooses one of the proposed poicies and impements it. Utiities are reaized. The forma soution of the game is presented in the appendix; in the foowing section, I provide a ess technica discussion of the resuts. First, I discuss the resuts that occur when the entrepreneur presents a constant association poicy that is, a poicy that cannot be changed by the factions that join the association. This produces the cassica free-rider resut. After that, I present the resuts that occur under a mechanism that aows for association members to infuence the association poicy proposa. In particuar, I examine institutiona mechanisms under which the ocation of the proposa is contingent on the contributions of the association members. Constant association poicy: The free-rider outcome Let us assume that the entrepreneur proposes a specific coaition poicy, c, that is a constant point in the poicy space. This means that this point wi be association poicy, independent of which faction is a member of the association, and which share of its obbying resources it contributes. We can say: Proposition 1 Assume that c is a constant and s = 0. Then bˆ SME = bˆ MNC = 0. In other words, neither SME nor MNC wi contribute to the association if they cannot infuence the common poicy and if there are no seective incentives. This is simiar to the cassic free-rider outcome that individuay rationa actors tend not to contribute to common goods. The proof for this proposition is simpe. Take SME as an exampe: Since MNC s contribution to c is a constant that is, not infuenced by SME s contribution SME maximizes its expected poicy benefit by obbying for its most preferred poicy. Since the equivaent hods for MNC, there wi be no cooperation between the two actors. The free-rider resut stated in Proposition 1 differs from Oson s coective-action probem. First, Oson does not necessariy concude that there wi be no cooperation. Instead, he argues that contributions to the coective good woud be suboptima, athough under certain circumstances they coud be positive. Here, the resut is that there wi be no contributions to the association poicy. Second, Oson assumed that the common good woud be preferred by both actors. Here, this is not aways the case, as I show beow. The question, then, is: When does an association benefit both actors? In other words, when do the two factions become victims of a Prisoners Diemma situation in which cooperation woud be beneficia but not individuay rationa?

18 18 MPIfG Discussion Paper 05/9 In order to answer this question, we have to compare the factions expected utiities under independent obbying with their utiities under cooperation. For argument s sake, et us assume that each faction either spends its entire budget on its own poicy proposa, or contributes everything to the association. Then, in the case of SME, the benefits of mutua cooperation are weaky arger than its utiity from non-cooperation if b SME+ b MNC i SME c SME b SME + b MNC + b SHI b SHI i SME i SHI SME b SME + b MNC + b SHI b MNC i SME i MNC SME b SME + b MNC + b SHI b SHI i SME i SHI SME, b SME + b MNC + b SHI which can be simpified to b MNC i SME i MNC SME i SME c SME. (5) b SME + b MNC Simiary, from MNC s benefit: b SME i SME i MNC MNC i MNC c MNC. (6) b SME + b MNC It is easy to show that equations 5 and 6 hod with equaity if ĉ = ( b SME b i SME,x + MNC imnc,x, b SME + b MNC b SME + b MNC b SME b isme,y + MNC imnc,y b SME + b MNC b SME + b MNC ). (7) That is, c ~ is a point on the ine connecting i SME and i MNC. Now, if we find a point in the poicy space that both MNC and SME prefer to ĉ, then a coaition between the two actors produces poicy benefits, and Proposition 1 refers to a situation resembing a Prisoners Diemma. I give the geometric intuition of the argument here; the forma proof is in the appendix. Consider the poicy benefit measure proposed in equation 2. If the α parameters in this equation are equa, the actor vaues both issue dimensions equay. We can pot equation 2 on a system of coordinates that represents the two issue dimensions; the resut is a circuar indifference curve, consisting of a points that suppy the actor with the same poicy benefit.

19 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 19 Figure 1 Two actors with equa evauation of poicy dimensions MNC ĉ SHI SME If both SME and MNC have circuar indifference curves 16 that is, if they vaue the issue dimensions equay their indifference curves that contain ĉ wi intersect exacty at ĉ, but at no other point (see Figure 1). Since each actor stricty prefers a points (and ony those points) that are within the indifference curve to those that are on or outside the indifference curve, there are no poicy choices that both SME and MNC prefer to ĉ. As a resut, the ack of cooperation under proposition 1 does not ead to any expected poicy osses. Now, consider the situation in which not a α parameters are equa. In particuar, assume that α SME,x α MNC,x. α SME,y α MNC,y In this case, the indifference curves of SME and MNC that go through ĉ intersect such that there are points inside both curves. As a resut, there are poicy positions that both SME and MNC prefer to ĉ. This means that cooperation can ead to positive poicy gains, and the ack of cooperation in proposition 1 describes a Prisoners Diemma. The foowing emma describes the conditions for this to happen: α SME,x α Lemma 1 If MNC,x α SME,y α MNC,y, then mutua cooperation can ead to positive poicy gains, compared to independent obbying by SME and MNC. 16 An aternative condition is that α SME,x = α MNC,x α SME,y α MNC,y.

20 20 MPIfG Discussion Paper 05/9 Figure 2 Two actors with unequa evauation of poicy dimensions c MNC SHI SME Substantivey, this means that poicy gains are possibe if both factions evauate the issue dimensions differenty, so that they can trade poicy concessions on one issue dimension against poicy gains on another dimension. Compare figure 2: In the case in which the indifference curves are ova-shaped, both MNC and SME prefer c to ĉ. Independent of the shape of the factions indifference curves, if association poicies are constant, the entrepreneur has to provide seective incentives. There are different ways to conceptuaize seective incentives. For exampe, the entrepreneur coud pay a seective incentive to faction (I use and k as widcards for the different factions) that is a continuous function of its contribution in support of the association poicy: s(bι). In other words, if an actor contributes a arger share of his budget to the coaition, then he wi receive a arger seective incentive; the function s(bι) determines to what extent contributions to the coaition transate into seective incentives. We can then show (see the appendix): Lemma 2 If s = f ( b ˆ ), then b ˆ = b if s(ˆ b ) > bˆ b SME 1 + b MNC i ~ c,0 bˆ b where c ~ is the common poicy proposa. (8) This means that the entrepreneur can induce a faction to contribute its entire resources to the coaition if the seective incentives react strongy enough to changes in contributions. Aso, Lemma 2 shows that (margina) seective incentives have to increase as the distance between a faction s idea point and the proposed coaition poicy increases. This is very intuitive: An actor makes arger poicy concessions when he joins the group ony if his concession is compensated by a arger seective incentive.

21 Broscheid: Organizationa Change in German Pharmaceutica Interest Representation 21 Association poicy as a function of member contributions We have confirmed that coaition buiding between poitica actors eads to free-rider probems, even if the two actors have different interests. The two main soutions to the free-rider probem are seective incentives (which are part of the mode) or conditiona contributions. I propose an impementation of the conditiona contribution mechanism that is based on the conditiona definition of the common good that is, the poicy position. This is a practice commony encountered in interest groups: The idea is that those actors who contribute more to the interest group have a stronger infuence on determining its poicy goas. At the very east, actors who do not contribute do not have the right to infuence the group poicy. I incorporate the idea of conditiona association poicies by assuming that the poitica entrepreneur does not propose a particuar poicy, but rather estabishes a decision making process that determines c. Under this process, c moves coser to i SME if SME contributes more of its budget to the association, and it moves coser to i MNC if MNC contributes more. Here, I propose a very simpe version of this argument. Suppose we use the foowing mechanism to determine c : c ={ i SME i MNC ~ c k if bˆ SME = b SME and bˆ MNC = 0 if bˆ MNC = b MNC and bˆ SME = 0 if bˆ MNC = b MNC and bˆsme = b SME if bˆ MNC = 0 and b SME = 0 In effect, the entrepreneur gives each faction the choice of contributing its entire budget to the association, or of contributing nothing. If both factions contribute their budgets, the association wi pursue a common poicy; if ony one faction contributes its budget, the entrepreneur wi pursue the contributing faction s idea poicy. Of course, there are other mechanisms that we coud think of. In particuar, a more genera mechanism woud be one in which each faction spits up its budget between a private poicy and the association poicy in any way it ikes. The soution to this mechanism, however, depends on the exact shape of the reationship between poicy and contributions. In some cases, this makes the derivation of soutions very difficut. In other cases, it can be shown that the factions either prefer to contribute their entire budgets or nothing at a. As a resut, under a number of conditions, our simpifying mechanism is endogenous to the game. In addition, the simpe mechanism that I propose here is fairy reaistic. Association fees tend to be a fixed amount determined by the associations. Members either join and pay