Houda HAFFOUDHI, LAEP

Transcription

1 Te logic of twolevel games wit endogenous lobbying : te case of international environmental agreements Houda HAFFOUDHI, LAEP alss , version 1 11 Dec Maison des Sciences Économiques, boulevard de L'Hôpital, Paris Cedex 13 ttp://mse.univparis1.fr/publicat.tm ISSN :

2 THE LOGIC OF TWOLEVEL GAMES WITH ENDOGENOUS LOBBYING: THE CASE OF INTERNATIONAL ENVIRONMENTAL AGREEMENTS Houda HAFFOUDHI Abstract alss , version 1 11 Dec 2007 International environmental agreements (IEAs) are increasingly important in a globalized economy. Te aim of our paper is study te effect of political pressure groupslobbies on te size and stability of IEAs. To tis purpose we use te framework of twolevel games to explain ow national political situation influences te decisions of governments at te international negotiations arena. We present an endogenous lobbying model in wic we assume tat lobbies try to influence te policy coice of governments by offering political contribution in return for policy compromise. Indeed, we use te "interest based explanation" of international environmental policy to describe te incentives of countries to join an agreement. Tis approac classifies countries in four categories: pusers, bystanders, intermediate and draggers. We found tat, wen government gives te same weigt to contribution and to social welfare, te contributions from te industrial lobby give incentives to government (Pusers, intermediate) to participate in te grand coalition making it stable. Our results suggest tat in order to sustain te grand coalition, weak global environmental agreements i.e. tose involving small abatement targets sould be negotiated. Te result is similar if governments are more interested by political contribution. However, if governments care less about political contribution tan about social welfare, industrial contribution is not enoug to limit te free riding incentives of eac type of government. In tis situation, pusers are te more expected to sustain a small stable coalition. Keywords : noncooperative game, interest group, coalition teory, environmental Policy. JEL Classification : C720, D720, D780, Q280 Résumé Cet article a pour objet d étudier l influence exercée par les groupes de pression lors des négociations internationales et son impact sur la taille et stabilité des accords environnementaux internationaux (AEIs). Pour ce faire, nous présentons un modèle de lobbying endogène dans lequel nous supposons que les groupes de pression recourent à la contribution politique pour influencer les coix politiques de leurs gouvernements. Notre modèle montre que, lorsque le gouvernement donne le même poids à la contribution et au bienêtre social, la contribution politique du groupe de pression industriel donne assez de LAEP, Université Paris I PantéonSorbonne. ouda.affoudi@univparis1.fr. Te autor is grateful to all participants in te 13 t Annual Conference of EAERE (Budapest, June 2004) and 1 st conference of AERNA (Vigo, June 2004) for comments on an earlier version presented. Tis paper also benefited from discussion wit te participants of te EAEREFEEM Summer Worksop Political Economy of te Environment (Venice, September 2003). Any remaining errors in te paper are our responsibility

3 motivations aux gouvernements pour participer à la grande coalition qui devient alors stable. Nos résultats suggèrent que pour soutenir la grande coalition, des accords environnementaux globaux faibles ceux qui impliquent des cibles d abattement faibles devrait être négociés. Le résultat est semblable si les gouvernements sont plus intéressés par la contribution politique. Cependant, si les gouvernements se soucient plutôt du bienêtre social, la contribution industrielle n'est pas assez importante pour limiter les motivations de free riding des gouvernements. Motsclés: Jeux noncoopératifs, groupes de pression, téorie de la coalition, politique environnementale. Classification JEL : C72, D72, D78, Q28 1 Introduction alss , version 1 11 Dec 2007 Some of te most important environmental problems urgently calling for solution are problems related to global pollution. Environmental problems suc as ozone depletion, climate cange and marine pollution ave been te focus of intense negotiations at te international level over te past two decades. International environmental agreements (IEAs) are examples of collective action to tackle suc global problems. Beyond teir specific interest, tese agreements are also important in te context of coalition formation teory. Tis teory focuses on coalition formation mecanisms, i.e. on te incentives tat lead to selfenforcing international environmental agreements and define te number of te signatory countries. Te presence of a strong freerider incentive prevents most IEAs of being stable and/or effective. For studying tese problems, noncooperative game teory as proved to be a very fruitful approac. Furtermore, tis approac sows tat wen countries reac a stable agreement, tis involves a small number of participants. Basically, te models of Barrett (1994), Bauer (1992), Carraro and Siniscalco (1992, 1993) and Hoel (1992) employ a stability concept borrowed from te oligopoly literature (D'Aspermont, et al., 1983) were a coalition is said to be stable if no country wants to accede to te coalition (external stability) and no country wants to leave te coalition (internal stability). All works studying te problem of IEAs stability and abatement level ave te inconvenient of assuming tat governments maximize welfare function. However, recent events in te international policy arena ave illustrated te extent to wic organized groups condition environmental policy, bot at national and multilateral level. Industry and environmentalist lobbies ave been extremely influential. For example, in te USA, tey ave eld different positions on some issues, suc as multilateral emission cuts 1. On oters, suc as te compliance of foreign legislation wit American environmental standards, teir objectives ave often coincided 2. Tus, we argue tat tere are political constraints tat bind te ands of national governments tat participate in te international negotiation process. Te aim of our paper is to study te effect of political pressure groupslobbies on te size and stability of international environmental agreements. 1 Wile green lobbies ave exercised "considerable influence on te negotiations" at Kyoto conference in favor of multilateral reductions in greenouse emissions (Financial times, December 11, 1997), a broad coalition of corporations, unions and economic lobbies as organized " one of te most intensive campaigns ever mounted on a single political issue, seeking to convince tat American curbs on greenouse gas are unfair and damaging to te economy " (Financial Times, September 10, 1997) 2 For example, bot ave demanded compliance of foreign legislation wit American environmental standards on incidental catcing of dolpins set out in te Marine mammal protection Act. 2

4 alss , version 1 11 Dec 2007 To analyse government beavior at te international policy, Sprinz and Vaatoranta (1994, 2002) presented te approac of "interest based explanation" Tey explain tat government position can be deduced from information about te country's ecological vulnerability and abatement cost. Tis analysis suggests tat countries will act as «pusers» for substantial emission reduction wen teir ecological vulnerability to environmental pollution impact is ig and teir abatement costs for pollutant emissions are low. In contrast, countries caracterized by ig abatement costs and low ecological vulnerability can be expected to act as «draggers» in suc negotiation due to te low benefitcost ratio of pursuing emission reduction. Countries wit bot ig ecological vulnerability and ig abatement cost are caugt in between te former two groups as tey face an "intermediate" benefitcost ratio, wile countries tat are neiter affected by te environmental problem nor face ig abatement cost will act as bystanders in international environmental negotiation. Tey expect tat puser countries take more stringent environmental positions tan intermediate countries do, wile te latter group is expected to favor environmental protection more often tan draggers. Te likeliood of bystanders' supporting environmental protection sould fall between tose for pusers and draggers; owever, no direct comparison wit te intermediate group seems to be appropriate on teoretical grounds. Tis study give an interesting explanation to government decisions at te international level but present te same inconvenient tan te previous works, it doesn't consider national political actors and teir effects on government policy. Since we are mainly concerned wit a positive analysis of bot coalition formation and te size of a stable IEA, we assume ten tat autorities, rater tan seeking social welfare objectives, pursue teir own selfinterest motives maximizing teir political support. Our paper sares te interest of an increasing political economy literature, wic examines te influence of interest groups on policymaking 3. Most studies ave focused on te role of producer groups in te determination of trade policy. In tis area, te political contributions approac of Grossman/Helpman (1994, 1995 and 1996) is a sort of a standard model. A more recent body of literature, wic includes Frederiksson (1997), Aidt (1998) and Conconi (2003) studied te political economy of environmental policy. Tese studies adopt te political contribution approac to study te impact of environmentalists and producer interest on environmental policy. But none examine te effect of lobbing by environmentalists and producer groups on te formation and stability of IEA. In te majority of recent literature, lobbying is modeled as a "menu auction" were exogenously given lobby groups offer policy makers contribution scedules, representing binding promises of payment, depending on te cosen policy (Berneim and Winston 1986, Besley and Coate 2001, Dixit, Grossman and Helpman 1997, Grossman and Helpman 1994 and 1996). In tis paper, we propose an alternative model of endogenous lobbying were given te set of existing lobbies; te government cooses te lobbies wit wic it will bargain over policy in excange for contribution. (Felli and Merlo 2002, 2003) Te aim of te current paper is to develop an international framework in wic te Government's decision about IEAs' participation and abatement level (international level) are influenced by pressures of interest groups wo organize a collective action troug electoral contributions (national level). It is, terefore, assumed tat lobbies try to influence te policy coice of te governments by offering tem political contribution in return for policy compromise. To tis end, we use te twolevel games framework, wic acknowledges te interplay between domestic politics and international relations (Putnam, 1988). Tis paper is, based on te notion tat lobby groups pressure constitutes a political constraint tat te governments face at ome, and sapes te outcome of international agreements. Ten, in te 3 See Person and Tabellini (2000), for extensive review of tis literature. 3

5 national level we ave a lobbying game. We ave te bargaining game between government and lobby groups presented earlier. Eac group propose its political contribution and government try to coose wic national political coalition it will coose to maximize its objective function at te international level. Ten, government considers lobby support wen it decides about bot its participation to te international environmental agreement and its abatement level in two following stages. Hence, in te international level, we ave te coalition game in wic governments decide noncooperatively weter or not to sign te agreement. Two groups of countries emerge at te end of tis stage: te signatories and te non signatories. In te tird and last stage, governments play te noncooperative Nas emission game, were governments, wic sign te agreement, play as a single player and divide te resulting payoff according to a given burdensaring rule. We organized te remainder of te paper as follows. In section 2, we present te analytical framework. Section 3 sets out te basic model. In section 4, we determine te abatement level equilibrium. In te section 5, we present te results from our stability analysis. In section 6, we report on results of various sensitivity analysis. Section 7 summarizes te main findings, draws policy conclusions and concludes wit some remarks about future researc issues. alss , version 1 11 Dec An analytical framework Te economic literature on international environmental agreements tends to treat te participant in international negotiations as monolitic and benevolent governments tat sincerely represent te common interest of teir country (see, eg., Barrett, 1997). Wile tis approac as yielded many important insigts, it appears somewat incomplete and inappropriate, for analyzing te international environmental agreements' size and stability. In particular, it leaves out te idea tat governments often ave interests opposed or aligned wit tose of teir domestic constituents, and tat it is te incentive embodied in elections and oter political control systems tat ultimately determine wat tese governments can and will do at te negotiation table. Tese ideas ave long been recognized by political scientists and public coice scolars, and ave been formalized in te teory of twolevel games. In most international bargaining situations, negotiators attempt to find an agreement acceptable not only to te foreign countries wit wic tey are bargaining, but also to te majority of teir domestic interests. Negotiators often find temselves simultaneously engaging in domestic and international bargaining. Ten te goal of tis article is to present a formal model wic reveals ow domestic and international factors interact to sape international environmental cooperation between nations. Tis model examines te interaction between international environmental negotiations and a simple domestic political situation; it is a two level game. Suc games ave been discussed before, and a few studies ave attempts to formalize te notion (Scelling, 1960; Walton and McKersie, 1965; Evans, Jacobson and Putman, 1993; Putnam, Formal models include Iida, 1991 and 1993; Mo 1991; Morrow 1991; Lomann 1993) but rare are tose wic ave studied international environmental negotiations. As te name of te teory suggests, te game is played at two levels: te international level, were te executives of countries involved in cooperation meet to negotiate te terms of an international environmental agreement, and te national level, were a political market constrains te set of politically acceptable actions available to te national representative during te negotiation at international level. Te general structure of tis game is illustrated in te following Figure: 4

6 Figure 1 : Twolevel Games (source: Putman, 1988) Négociation Internationale Marcé politique Pays 1 AEI Marcé politique Pays 2 alss , version 1 11 Dec 2007 Te national political markets impose constraints on te representatives in many ways. We ave te electoral incentives of te executive. Because te electorate is not a direct participant, it cannot directly influence te international bargaining, but it can influence it indirectly troug election. Ten government, in te international bargaining game as to make a proposal tat is acceptable to is domestic constituents (Morrow, 1991). Wile voters are concerned about many different policy issues, tey take a basic interest in te issue of environmental problems, altoug priority placed on it differs between countries (CEU, 1999). Successfully reacing an IEA at te international level is, terefore, important for national politicians wo want to demonstrate progress in te implementation of climate policy, for example, and, ultimately, for winning election. Te adoption of emissions targets involves, as te same time, te benefits and cost perceived by different voters. Voters may well reward politicians for reacing international agreements, even toug te "real" benefits of action against climate cange are igly uncertain and would, in any case, not materialize until far into te future. On te oter and, greenouse gas abatement is costly, and te costs are borne by firms and ouseolds immediately. Ten wen governments seek to implement climate policy, tey risk losing votes from voters armed by abatement decision. Interest groups affect also government decisions (Olson 1965). Special interest groups: green lobby and industrial lobby in particular Business associations and environmental NGOs are able to affect te beavior of politicians by providing information, by financing election campaigns, or by bringing climate cange problems to te forefront of te minds of te voter (Grossman and Helpman, 2001). All tese political factors are taken into account wen te executives of te countries meet at te international level to decide weter or not tey will accept to participate in te IEA, and define te abatement level tat individual representatives would consider politically acceptable. Te model presented is an example of nested game. We can tink of international negotiation as consisting of domestic and international games tat are played simultaneously, tat is, players take a single action applicable to bot games. 5

7 3 Model Consider a world of i=1,...,n countries, eac of tem emits a pollutant tat damages a sared environmental resource. Eac Government i's objective function is presented as follow (1) Ui( qi, qg) = B( Q) ACi( qi) + σ i Ci ( qi, qg) Were B(Q) denotes abatement benefits, AC i (q i ) is country i's abatement cost, q i is i's abatement and Q represents te global abatement suc tat Q= qi, ( Q a). C i >0 i represents te monetary contribution given by lobby to te government. l i represents te national coalition of interest group wit wic government prefers to negotiate. σ i >0 measures te intensity of eac government's preferences over contribution wit respect to environmental policy 4. li alss , version 1 11 Dec 2007 Te country i's current abatement benefits are assumed to depend on current total abatement as follows 2 Q BQ ( ) = b aq suc tat Q a. 2 Were a and b are positive parameter suc tat b represents te slope of eac country's marginal abatement cost curve. Eac country's abatement costs are assumed to depend on its own abatement level. For country i, te abatement cost function is assumed to be given by c AC ( q ) = q 2 2 i i i Te parameter c>0 represents te slope of eac country's marginal abatement cost curve. Tis cost isn't sufficiently ig to make producer profit negative. In tis paper, we propose an alternative model of lobbying were te elected policymaker cooses te lobbies tat participate in te policymaking process. Tis is te sense in wic lobbying is endogenous in our model. We assume tat given te set of existing lobbies, te government cooses te lobbies wit wic it will bargain over policy in return for contribution (Felli and Merlo, 2002). We model environmental policy making as te outcome of a political process tat involves not only elected government but also nonelected political agents know as lobbies. We assume tat tere are 2 lobbies: Environmentalist and industrial respectively E and I wic differ wit respect to teir policy preferences. Eac lobby as a most preferred policy outcome q I and q E respectively. 4 if σ i =0 governments are purely policymotivated and lobbying is irrelevant, tat is te case usually presented in environmental policy literature, were B(Q) AC i (q i ) represents country i global welfare. In tis case government abate q g is most preferred policy. 6

8 Eac lobby is assumed to be able to sign binding contracts on environmental policy coice wit government in excange for contribution transfers. Notice tat te government as te option of not signing any contract and of implementing is most preferred policy q g. We suppose tat only environmentalists ave environmental concerns and tat teir ideal a abatement level is q E =.Te environmentalist's current benefit from abatement decision N takes te following form: (2) V E ( q, q ) = ν E ( q, q ) C E ( q, q ) i i E i E i i E b Were ( q, q ) B( Q) D( q, q ). D( q, q ) ( q q ) 2 generated by te abated emission. E 2 ν i E = i i E i i E = E i, represent te damage alss , version 1 11 Dec 2007 Ten, environmentalists' preferences depend on te global benefit generated by te total abatement realized by all countries and on te damage caused by te nonabated emissions consisting in te difference between teir ideal point (q E ) and teir country current abatement level (q i ). Wen environmentalists gain from te increases in total (global) abatement effort, tis means tat an environmentalist group in a country j will support its own government even toug abatement effort as been made in some country i. Tis means tat government will be rewarded by its environmentalist group for aving incited oter government to participate to te collective abatement effort. But, global benefit isn't sufficient to explain wy an environmentalist group is supporting its own government. We assume tat it will only do so, wen its own government undertakes additional abatement effort. To introduce tis condition, we suppose tat environmentalists are armed by te damage caused by te nonabated emissions consisting in te difference between teir ideal point (q E ) and teir country current abatement level. Ten environmentalist group reduces it government support bot wen government doesn't take any abatement decisions and wen its abatement decision is lower tan environmentalist ideal abatement level (q E ). Ten te more te government abatement level is closed to te environmentalist group ideal point, te more it will be supported. Industrialist groups are always armed by teir government abatement decisions and teir abatement cost is assumed to depend on its own abatement level and nor one else's and it takes te following form: (3) V I (, ) I (, ) I i qi qi = ν qi qi Ci ( qi, qi) I c 2 Were ν ( qi, qi) = ACi( qi, qi). ACi( qi, qi) = ( qi qi) et qi = 0. Ten industrialist group 2 abatement cost is no else tan its country's abatement costs. We denote { I, E} Λ= te set of lobbies. Let {{ },{ I},{ E},{ I, E} } = 7

9 be te collection of all possible coalition of lobbies wit wom government may coose to participate to te IEA and to bargain over abatement policy and contribution. Eac possible coalition l i is associated wit a willingness to pay, ω l ( q, ) i i qg, for any policy q i te government may coose to implement instead of is most preferred policy q g ; (4) ωl ( q, ) (, ) i i qg = Ci qi qg suc tat ω{ } ( q, q ) 0 = i g li Given te preferences of a lobby specified is equation (2) above, te willingness to pay of lobby l i for any abatement policy q i implemented by government is (5) C (, ) (, ) i qi qg = ν qi q ν ( qg, q) alss , version 1 11 Dec 2007 C i ( qi, q g) is te monetary value of utility gains (or loss) wit respect to te status quo tat lobby obtains if government cooses abatement level q i instead of its most preferred policy q g applied in te status quo. Te status quo is ere defined to be government decision in absence of any lobbying, q g. Given lobby groups preferences specified in equation (2) et (3), eac lobby political contribution takes te following forms: b C q q = B Q B q + 2 q q q q q I c 2 2 Ci ( qi, qg) = q q 2 q ( ) g i I qg q i 2 E 2 2 i ( i, g) ( ) ( g) 2 ( ) g i E g i Te willingness to pay ωl ( q, ) i i qg of national political coalition l i, for any policy q i tat government may coose instead of its ideal policy q g, takes te following form: (6) li Ci ( qi, qg) = α B( Q) B( qg) + b ( q q ) S1 ( q ) g i g qi S2 2 b b were S 1 =, S 2 = q suc tat b = b si = E et b = c si = I b b. Ten, for all li li c λ + λ wit λ = et b 0,1,0,1 respectively. Te parameter α indicates te existence of te possible coalition of = {{ },{ I},{ E},{ I, E} }, S 1 takes as values { 0,1,,1 } S 2 takes as values { } environmentalist lobby group in te national political coalition l i. α=1, if government consider te political contribution proposed by te environmentalist lobby group and α=0 if not. Ten, for all government political coice tere is a willingness to pay. Our model does not explain te process of lobby formation. We simply assume tat only two groups of citizens overcome te freeriding problem described by Olson (1965) and get politically organized: a proportion of te population, te "environmentalists", wo form a 8

10 national green lobby and te industrialists, wo form producer lobby. Tis model is first concerned wit establising te effects of political support motives on te determination of emission abatement and ten te number of countries signing a stable IEA. 4 Te equilibrium Following te approac of te noncooperative game teory of coalition formation, we assume tat countries decide simultaneously in te last two stages of our model. In te first stagete coalition game tey decide noncooperatively weter or not to sign te agreement. Te equilibrium number of countries participating in an IEA is ten derived by applying te notions of internal and external stability of a coalition originally developed by D'Aspremont et al. (1983). In te second stage, tey play te noncooperative Nas emission game, were te countries, wic sign te agreement, play as a single player and divide te resulting payoff according to a given burdensaring rule. alss , version 1 11 Dec 2007 We use a feedback resolution to resolve our two stage non cooperative game. We begin by te determination of signatories and non signatory countries abatement. Ten, we determine te size of stable coalition wic depends on national lobby support. Finally, we deduce wic national political coalition will maximize eac government payoff at te equilibrium. We suppose tat tere are two groups of countries. We assume tat s identical governments s s sign an agreement and Ns do not. Let Q denote te abatement level of te coalition, and q i s s denotes te abatement of any individual signatory, suc tatq = sq i. In a similar manner, eac nonsignatory government's abatement is q ns i yielding a total abatement of all non ns ns signatories Q = ( N s) q i.global abatement level is Q= ( N s) q ns s i + sqi We assume tat countries decide simultaneously in bot stages. Te nonsignatories beave noncooperatively wen signatories coose teir abatement level by maximizing teir collective payoff function. Tat is, nonsignatories coose q ns i by solving te following maximization problem. (7) Uns ( qns, qg ) = B( Q) ACns ( qns ) + σ ns Cns ( qns, qg ) lns Explicitly, eac nonsignatory government objective function takes te following form: 2 Q c 2 Uns ( qns, qg ) = b aq qns + σns ν ( qns, q ) ν ( qg, q) 2 2 lns Te nonsignatory abatement level verifies te following equation: (7 ) MB( Q) + σ MC ( q, q ) = MAC ( q ) ns ns ns g ns ns lns were MB : marginal global benefit, MC : marginal Contribution and MAC : marginal abatement cost. 9

11 Signatory s governments coose Qs by solving te following maximization problem. (8) Us( qs, qg) = B( Q) ACs( qns) + σ s Cs ( qs, qg) s N s N s N ls Explicitly, signatory s objective function takes te following form: Q c 2 2 s 2 2 Us( qs, qg) = b aq qs + σs ν ( qs, q) ν ( qg, q) s N s N s N l were Q= sq. s + ( N s) qns. Government member of te coalition considers not only its national contribution but also contributions received te oter member to remain in te coalition. Signatory s governments abatment level verifies te following equation : alss , version 1 11 Dec 2007 (8 ) MB( Q) + σ s MCs ( qs, qg) = MACs( qs) s N ls Using équations [7 ] et [8 ], we can resolve te second stage of te game. Te result is presented by te following proposition: Proposition 1 : Eac government abatement level decision depends not only on its national political process but also on te political process in te foreign country. Te individual abatement level of signatory and non signatory government take te following form : s s ns s + s σss2χns N s σnsαns σss2 sσsα s s ns s σnss 2 qs = a+ ϕ ϕ ns 2 ns s ( 1+ σ ) σnss2 χs s ( 1 sσsαs ) σnss2 ( 1 σnsαns ) σ nsαns χ + s ss qns = a+ ϕ ϕ were 2 s ns c ϕ = s ( 1+ sσsαs ) χns + ( N s)( 1 + σnsαns ) χs + χsχns, χs = s. σss1 + λ, χns = σnss1 + λ, λ =. b ( 1 sσα ) χ + ( ) ( 1+ ) ( 1+ ) Hence, for coalition structure s, MB( Q) + σ s MCs ( qs, qg) = MACs( qs) olds for a s N ls member i c in equilibrium and for a singleton j MB( Q) + σ MC ( q, q ) = MAC ( q ). Tus, joining coalition s as te advantage tat ns ns ns g ns ns lns own abatement efforts are matced by oter members and ence iger benefits, but also 10

12 means iger abatement costs and iger/lower lobby contributions. Bot effects determine weter a coalition is stable, wic is cecked in te following section. nc c We ave always q > 0 and q > 0 wic guarantees tat our solutions are interior. Te remaining problem is to determine te number of signatories to te selfenforcing IEA. 5 Te size of stable IEAs We now proceed wit te determination of te size of te stable IEA, denoted by s*, using te internal and external stability conditions. Recall tat te internal stability condition ensures tat if a government were to defect unilaterally, its gains from free riding would be outweiged by te adjustment (due to its defection) of abatement levels of te remaining members of te IEA. Te external stability condition ensures tat no oter non signatory government finds it beneficial to unilaterally join te IEA; formally, te internal and external stability conditions satisfy tis definition: alss , version 1 11 Dec 2007 Definition: An IEA consisting of s signatories is selfenforcing if * * * * M s ( s ) > M ns ( s 1) et M s ( s + 1) < M ns ( s ) A full caracterization of solution cannot be obtained analytically for tis functional specification. However, simulations reveal a very simple and compelling relationsip between c σ s, σ ns and λ =. b We test all coalition structures for stability using our political economy model. To simplify our presentation, we suppose tat b takes tree values (0.01, 1,100), c takes as values (0.01, 1, 100). In our bencmark scenario, wen governments are immune to te pressure of lobby groups (i.e. σ=0), it turns out tat tere are only one stable coalition of small size (s*=2) suc tat λ>1 (b=0.01 and c=100). Tere were only internally stable coalitions, but none of tese is externally stable. Tis stresses te presence of strong freerider incentives in our framework. Tis situation, owever, is different once te government considers political pressure factors lobby contributions into its payoff function. In te following we report on te results for tis exercise, considering te case of σ s =σ ns =1 for bot signatory and nonsignatory governments. Figure 2: Stable coalitions under different lobby contributions Nonsignatories Signatories Lobby Environmentalist Industrial 2 ( ) Grande coalition ( ) Grande coalition Environmentalist and Industrial Grande coalition ( ) Environmentalist 2 (b>c) 2 ( ) Grande coalition 7 ou 2 ( ) Industrial 2 ( ) Singleton 2 ou 5 (b<c) Singleton Environmentalist and Industrial Singleton 2 ( ) Grande coalition 5 ( ) 11

13 From figure 2, we can observe tat once governments include in teir decisions te level of political pressure tat te lobby groups exert in te form of monetary contributions stability and size of coalition increase. Tere are no nontrivial stable coalitions in tirteen of our scenarios. Among tese, te grand coalition is stable in five of te cases. We also can observe tat te extent of tis improvement is determined mainly by te decisions of signatories and teir abatement and cost caracteristics. Indeed, stable coalition, if it exists, appens globally wen b =0.01 and c=100. More specifically, wen signatories continue to be immune to lobby influence, small coalition is stable independently of nonsignatories' decision about wit wic lobby tey bargain. Te situation does not improve muc wen nonsignatories bargain wit environmental lobby group or bot lobbies; in tis case only one possibility gives a large stable coalition, tat is wen nosignatories are immune to lobby contribution. If signatories decide to accept only te contributions of te industrial lobby, te stable coalition emerge independently of signatories' decision about wit wic lobby tey bargain. We ave coalition of size two and five if signatories ave te same national political coice and te grand coalition (i.e. an agreement of all countries) if not. Differently to te oters cases, te grand coalition is stable for all values of b and c. alss , version 1 11 Dec 2007 Proposition2: Once governments include in teir decisions te level of political pressure tat te lobby groups exert, stability and size of coalition increase. Basically, industrial lobby contribution reduces te incentive of free riding and can incite signatory governments to sustain te grand coalition wen te nonsignatory government as different national political preferences. To explain government decisions wic emerge from our model, we ave to determine te solution of our to stage game. Figure 2 gives all stable coalitions under different lobby contribution but didn't specify wic one maximizes government payoff function. Te solution depends on abatement cost and benefit parameters, and is given by table 1. Table 1. Lobby coalition wic maximize government payoff under Stable coalitions ecological vulnerability (b) abatement cost (c) 0.01 Grande coalition Grande coalition Grande coalition 1 7 Grande coalition Grande coalition Grande coalition (a) 12

14 abatement cost (c) 0.01 {{I,E}, {E}} 1 {{I,E}, {E}} ecological vulnerability (b) {{I}, } ; {{I}, {I,E}} {{I}, } ; {{I}, {I,E}} 100 {{I,E}, {E}} {{I,E}, {E}} (b) {{I}, }; {{I},{I,E}}; {{I},{E}} {{I}, }; {{I},{I,E}}; {{I},{E}} {{I}, }; {{I},{I,E}}; {{I},{E}} alss , version 1 11 Dec 2007 Table 1 presents te solution of our game. It gives government decision about its participation to IEA and wit wic lobby it bargain to maximize its payoff function. Te grand coalition, wen signatories receive support only from te industrial lobby, represents te solution of our model if b is ig and b>c. Neverteless, tere are oter stable coalitions, namely a coalition of size seven and te singleton coalition, wic emerge at te equilibrium, wen b<c and signatories are supported by environmental lobby or bot lobby. Government position can first be deduced from information about te country's ecological vulnerability (b) and abatement cost(c) (Sprinz and Vaatoranta, 1994). By combining tese two indicators, governments can be classified into four categories, as follow: Figure 3: Prediction of interestbased explanation from government negotiating position abatement cost Sources : Sprinz and Vaatoranta 1994, 8 ecological vulnerability Low Hig Low Bystanders Pusers Hig Draggers Intermediates Te interestbased explanation proposed by Sprinz and Vaatoranta (1994, 2002) suggests tat countries will act as «pusers» for substantial emission reduction wen teir ecological vulnerability to environmental pollution impact is ig and teir abatement costs for pollutant emissions are low. In contrast, countries caracterized by ig abatement costs and low ecological vulnerability can be expected to act as «draggers» in suc negotiation due to te low benefitcost ratio of pursuing emission reduction. Countries wit bot ig ecological vulnerability and ig abatement cost are caugt in between te former two groups as tey face an "intermediate" benefitcost ratio, wile countries tat are neiter affected by te environmental problem nor face ig abatement cost will act as bystanders in international environmental negotiation. Te interest based explanation allows an initial understanding of possible positions taking by countries in international environmental negotiations. However, it doesn't explain wy we ave stable coalition wen government considers its national political coalition support, wereas, in te base case, (witout lobbies' influences) te singleton is te most frequent situation. 13

15 alss , version 1 11 Dec 2007 Following tis approac, we can deduce (from table 1, (a)) tat contribution elps to reduce te free rider incentives present in te coalition wen countries do not receive any support from lobbies. Countries caracterized as pusers do not ave any incentive to form a stable coalition in te "base case". However, wen countries consider industrial lobby support tey are willing to participate to te grand coalition (table 1, (b)). Indeed, eac government receives a positive contribution from industrial lobby, because tey perform an abatement level lower tan tat resulting from te scenario (O,O)encefort also called as "base case". Given teir ig level of marginal benefit, pusers are interested in iger levels of global abatement. Te level of abatement in te "base case" corresponds to te government's most preferred level of abatement (q g ). Te industrial lobby benefits from te decision of countries to ave a lower abatement level tan in te "base case", ence is willing to make a positive contribution to compensate for te associated costs (table 2). Intermediates present te same incentive tan pusers in te "base case". Considering industrial lobby support tey are also incited to participate to a large agreement but given teir ig level of marginal abatement cost, teir abatement level is lower tan tat of pusers. Tis ensures tat eac government receives a level of contribution from te industrial lobby sufficient to compensate it for te forgone benefits due to te lower level of global abatement compared to "base case" situation. Countries tat are expected to act as bystanders do not ave any incentive to form a stable coalition in te "base case". Altoug tey may participate to a selfenforcing IEA wen tey consider industrial lobby contribution (grand coalition, Figure 1), tey maximize teir gain wen tey avoid a coalition. Eac government prefers to beave as singleton coalition witout considering lobby contribution. Wen tey consider bot lobbies support, tey receive positive contribution from environmental lobby as tey abate more tan in te "base case" and tey lose small support from industrial lobby as, given teir small marginal abatement benefit, tis increase in small. Globally tey gain more in tis situation compare to te grand situation wen tey receive contribution from industrial lobby. However, given teir low marginal benefit, bystanders prefer to be immune to lobby influence and not to increase teir abatement level. Tis government maximizes teir gain wen tey remain in te status quo. On te contrary, draggers may ave incentives to participate to an agreement in te base case. In our bencmark scenario, wen governments are immune to te pressure of lobby groups, it turns out tat te coalition of size two is stable (Figure 2). For instance, given teir ig marginal abatement cost, eac government is interested in low level of abatement and forms a small coalition wen costs are very ig (wit c=0.01 and b=0.01). Wen governments consider bot lobbies contribution, tey receive enoug compensation to form a coalition of size seven (table 1, a, b). Given its costbenefit structure, eac signatory's government will abate a medium level of abatement but iger level tan qg. Tis ensures tat, on te one and, tis government receives a ig level of contribution from te environmental lobby sufficient to compensate it for te forgone due te iger level of global abatement compared to te "base case" situation; and, on te oter and, ey loose a small support from industrial lobby. Globally, draggers receive a positive contribution from bot lobbies enoug to sustain an agreement. 14

16 Proposition 3 Te size of stable coalition depends not only on national political coalition but also on benefit cost structure of eac government. Considering political contribution, pusers are te most expected to sustain a large coalition (industrial contribution) Draggers are te less expected to form a stable coalition (environmental or two lobbies contribution). Intermediate are caugt in between te former two groups (industrial lobby contribution). Neverteless, eac government decision cange wen it canges its national political coalition. alss , version 1 11 Dec 2007 However, wen costs are very ig, government expected to beave as draggers maximize teir gain wen tey avoid te agreement. Te incentives of free riding are very important in tis situation. Altoug in te base case, we ave small stable coalition wen costs are ig, considering environmental lobby support, te contribution isn't enoug to reduce free riding incentives. Given te ig marginal abatement cost, eac dragger acting as a singleton as incentive to ave low level of abatement. Eac government receives positive contribution from industrial lobby, due to te lower level of abatement compares to te situation of "base case" sufficient to break te coalition. Table 2: Results for stables coalitions for eac type of government (wit and witout lobby contributions) Countries Abatement (no contributions) Pusers (a)* 9, ,990 Intermediates 9,901 Draggers(1) 990, ,198 Draggers (2) (b) 0,099 10,097 Bystanders 9, ,100 Abatement (wit contributions) 9, ,900 9,9 990,000 5,884 6, ,819 0,131 0,197 19,575 9, ,000 net Gain (no contributions) ,50 99,97 50,0051 net Gain (no contributions) Contributions C I s =0, , C I s =0,9704 5, C s E = 691,66 5, C I =4,81 C ns E = 691,69 287,92 C s E = 95,32 287,22 C I =0,37 49,9974 C ns E = 95,33 C s E = 0,3583 C s I =0,0048 * : Similar result for te oter cases. (a) : caes b=100 and c=1, (b) : case b=0,01 and c=100, Draggers (1):te two oter cases, : don t exists Table 2 gives te abatement level and payoff for signatories (grand coalition) and/or non signatories (small coalition or singleton coalition respectively) and te total abatement level in te case of small coalition. From table 2, it is evident tat witout te contribution of te industrial lobbies te grand coalition is not stable. Pusers gain more tan intermediates by participating to a large agreement. Draggers are te most gainer by sustaining a selfenforcing IEA wen tey consider bot lobbies contribution. In addition, we find tat puser countries take more stringent environmental positions tan intermediate countries do, wile te latter 15

17 group is expected to favor environmental protection tan draggers. Te likeliood of bystanders' supporting environmental protection fall between tose for pusers and draggers; owever, no direct comparison wit te intermediate group seems to be appropriate on teoretical grounds. Proposition 4: A large stable coalition corresponds to a weak agreement an agreement tat not involve a ig level of abatement. Compare to te status quo. A stringent environmental agreement, if tey emerge, will be sustained only by a small number of countries. alss , version 1 11 Dec 2007 All regions, wit te exception of part of draggers, would gain by leaving te coalition and becoming free riders. If countries do not respond to political pressures, and given teir benefit and cost structure, te increase in abatement efforts wen joining te coalition will impose tem a severe burden tat is not compensated by te augmented global benefits. Bystanders are always interested in acting as a singleton wit and witout lobby influence wereas draggers ave te same incentives wen teir abatement costs are ig. Te situation canges once countries accept te contributions from te industrial lobby (pusers and intermediate) or bot lobbies (draggers). No country as an incentive to leave te coalition. Tis result migt appear strange given tat te industrial lobby is interested in aving te lowest abatement possible. But tere are two factors tat elp to rationalize tis outcome: te cange in te abatement efforts and te increase in payoffs due te contributions. As we can observe abatement levels are lower in te case wen countries only bargain and receive contributions from te industrial lobby. Wen governments are expected to beave as draggers, tey receive globally a positive contribution from bot lobbies enoug to compensate te increase to te cost due to te increase in te abatement. 6 Sensitivity Analysis A typical feature of simulation metod is tat results depend on parameter values, wic are subject to some uncertainty. Given te large number of parameters tat enter our model, some selection is necessary for a sensitivity analysis. We believe tat tere is ig uncertainty concerning te weigt given by eac government to political contribution compared to social welfare from abatement level and to tat given by lobby group. Hence, we conduct a sensitivity analysis were we cange te value of δ. For instance, to capture te idea tat government cares relatively more about contribution tan about social welfare, we assume tat δ>µ. As µ=1, ig δ means tat government are more interested by political contribution tan lobbies. Ten, we consider tree cases to analyze te influence of government preferences on teir decision at te international environmental negotiation. Tat is, we consider: δ s =0.2 & δ ns =0.5; δ s =1.5 & δ ns =2 and δ s =1.5& δ ns =0.5; to study te sensitivity of our stable coalitions. Tese tree cases represent respectively te situation in wic bot signatories and non signatories care relatively less about contribution tan about environmental policy compared to lobbies, te situation in wic tey care more about contribution tan about environmental policy and te situation in wic signatories are more interested by contribution were non signatories care rater about its social welfare from abatement level. 16

18 Proposition 5: Government decision at te international environmental negotiation depends not only of its benefit cost structure but also on te weigt tat it gives to political contribution compared to social welfare. Industrial lobby contribution reduces government incentives to beave as free rider. Neverteless, te size of stable coalition is small if governments are more interested by social welfare and te grand coalition is stable if governments are rater interested by political contribution. In addition, if signatory and non signatory present different political weigt, pusers are te more expected to sustain a large stable coalition; wereas draggers form a stable partial coalition if tey receive a enoug contribution from environmental lobby group to compensate teir ig cost. Table 3. Lobby coalition wic maximize government payoff under stable coalitions (scenario δ s =0.2 & δ ns =0.5) alss , version 1 11 Dec 2007 abatement cost (c) ecological vulnerability (b) Singleton Singleton Singleton Singleton Singleton First, not only for our base case but also for te scenarios (δ s =0.2 & δ ns =0.5) no large coalition is stable regardless of te definition of stability. Only pusers sustain a stable coalition of size equal 2 5 (table 3). Te abatement policy applied in tis situation is lower tan tat of status quo. Altoug pusers receive positive contribution from industrial lobby, tis compensation is not enoug to avoid te free riding incentives of eac government. Differently to te situation were governments give te same weigt to social welfare and contribution, event if governments ave te same lobby support, te fact tat tey are more interested by environmental policy can not give incentive to government, especially pusers, to sustain a large stable coalition. Table 4. Lobby coalition wic maximize government payoff under Stable coalitions (scenario δ s =1.5 & δ ns =2) abatement cost (c) ecological vulnerability (b) Singleton Grande coalition Grande coalition 1 Singleton Singleton Grande coalition Singleton Singleton 5 Te same result wen signatories continue to give low weigt to contribution compared to non signatories (0.5, 1.5) or (0.5, 1). Except for puser, we ave partial stable coalition (s = 48 or 28 respectively) wen signatory receive contribution from industrial lobby and non signatories from environmental lobby 17

19 Second, wen governments give more weigt to contribution over environmental policy (δ s =1.5 & δ ns =2), te grand coalition is often stable 6 (table 4). Basically, intermediates and bystander ave te same incentives to form a grand coalition and tey maximize teir payoff if tey receive contribution from industrial lobby. Concerning draggers, wen tey ave ig cost tey maximize teir gain wen tey accept te contribution from industrial lobby and in tis case te grand coalition is stable. Given teir cost benefit structure and te weigt tat tey give to contribution, draggers receive enoug compensation from industrial lobby group to remain in te coalition. Lobbies transfer reduces in tis situation te incentives of free riding very important in te case of draggers. Contrary to te case (δ s =1 & δ ns =1) partial cooperation is not stable and environmental lobby contribution maximize te gain of draggers but is not enoug to give tem incentive to remain in te coalition. Pusers will realize te same gain bot in te situation of small stable coalition (s=2), wen signatory are immune to lobby influence and non signatory receive environmental lobby support (O,E) and in te situation of grand coalition s=gc were signatory receive support from industrial lobby group independently of non signatory coice. Pusers are indifferent between te two situations, because tey don't cange teir beavior of abatement. alss , version 1 11 Dec 2007 Table 5. Lobby coalition wic maximize government payoff under Stable coalitions (scenario (δ s =1.5 & δ ns =0.5)) abatement cost (c) ecological vulnerability (b) Singleton Grande Grande coalition coalition 1 Singleton Singleton Grande coalition Singleton Singleton Tird, te result is closely related to te second point. A conclusion tat as been derived from te situation in wic signatory are more interested by contribution tan non signatory, is tat te grand coalition is stable wen government receive contribution from industrial lobby (table 5). Pusers, given teir ig marginal benefit, sustain a large coalition. Concerning intermediates and bystanders tey may form a large partial coalition but tese coalitions are only internally stable. Finally draggers, given teir ig marginal cost, tey may sustain a stable coalition (s=14) if tey receive positive contribution from environmental policy enoug to compensate te cost related to te increase in teir abatement level. Globally, in te different case presented earlier government as incentives to consider lobbies contributions and ten to remain in te coalition. Pusers adopt less constringent environment policy ten in te situation witout contribution (Table 6). Ten, tey globally form a large coalition. Draggers ave to reduce iger level of pollution tan in te situation witout national support. Tey receive ig level of contribution from teir environmentalist lobby groups enoug to compensate teir abatement costs and to increase teir gains. 6 Te result is te same for all situation in wic signatory give more weigt to contribution tan non signatory (1.5, 1), (1.5, 1.5), (2, 1.5). 18