Domestic politics and climate change: international public goods in two-level games

Abstract

We use the theory of two-level games to explore how domestic constraints affect the outcome of bargaining games over national contributions to an international public good such as global climate change, and to discuss the implications for the Schelling conjecture. We model the international negotiations on two dimensions—domestic and foreign contributions to the international public good—and extend the basic two-level model by examining a nonzero sum, two-dimensional conflict model on level one that includes characteristics of both conflict and cooperation. Our main results suggest that if the domestic game is a ratification game (as in the presidential system of the US), then contributions do not exceed those in a benchmark game without domestic constraints. But if the domestic game is an election game (which is more important in the parliamentary system of most continental-European countries), contributions can actually be higher than the benchmark.

Notes

¹ We thank Charles F Mason, Tom Crocker, Tommy Stamland, Owen Phillips, Eric Nye, Kevin Siqueira and seminar participants at Clarkson University, and the Public Choice meetings in Charleston for helpful comments on earlier versions. Shogren thanks the Norwegian University of Life Sciences and GREQAM for the financial support. All errors remain ours.

² Other global public goods include global justice, trade agreements, international financial stability, cultural heritage, health, knowledge, telecommunications, cyberspace, peace and foreign aid (see for example Kaul et al 1999). See Gilligan (2004) for a discussion of international public good negotiations with more than two countries.

³ Carraro and Siniscalco (1998, 570) and Folmer et al (1998, 21) were among the first to recognize a need for the literature on international environmental agreements to ask about the motivation of acting governments and include the relationship between international agreements and national voters.

⁴ Currarini and Tulkens (1997), Siqueira (2003) and Eckert (2003) model transboundary pollution games with domestic audiences. Using a core-theoretic framework, Currarini and Tulkens focus on the stability of an international agreement when the objective function of the negotiator depends on the aggregate pay-off of a winning coalition in power. Siqueira shows a domestic consumption tax on emissions could be inferior if implemented by a government that does not coordinate its policies with other governments. See our comments on Eckert in the text.

⁵ Even though this definition is consistent with Milner and Rosendorff (1997), it is a loose interpretation of Schelling's point (1960, 28). A government tries to change the domestic constraints by changing domestic preferences or institutions to improve the bargaining position (Humphreys 2007 is a recent theoretic study of this idea). In Milner and Rosendorff and our model, preferences and institutions are exogenous. For a discussion on when the Schelling conjecture holds, see Tarar (2005) and Hammond and Prins (2006).

⁶ Two-level games have been used to analyse international trade (Grossman and Helpman 1995; Milner and Rosendorff 1996, 1997), arms control (Morrow 1991) and macroeconomic policies (Lohmann 1993).

⁷ Reminiscent of the ‘monopolistic agenda-setter’ literature, the main difference is that the agenda-setter cannot set monopolistic agendas, even though she is a political monopolist in her own country. See for example Romer and Rosenthal (1979), Banks (1990a, 1993) and Morton (1988).

⁸ Such a retrospective-prospective voting rule has been used in the political science literature; see for example Bueno de Mesquita et al (1999).

⁹ Consider briefly the main difference between our model and those within the literature on strategic delegation bargaining games (see for example Fershtman et al 1991; Cai 2000). The bargaining agent does not get any utility directly from the bargaining outcome but only indirectly via a compensation scheme designed by the principal; in contrast, our agent (the government) has its own intrinsic preferences that might differ from those of the principal (the voter). The reason why the principal hires an agent in a bargaining or any other noncooperative game is nevertheless somehow related to Schelling's ideas: the principal can commit himself and show this commitment to the counterpart. See also Segendorff (1998), who uses a level-one game similar to ours.

¹⁰ Iida (1993, 1996), Morrow (1991), Milner and Rosendorff (1996) and Mo (1994, 1995) examine conflict models with one political issue. Putnam (1988) also focuses on these models, noting that things change with the inclusion of a second dimension.

¹¹ See also Grossman and Helpman (1995), Milner and Rosendorff (1997), Segendorff (1998), Mansfield et al (2000) and Dai (2006).

¹² Popular ratifications, of course, play a role in decision-making in European countries. Recent international agreements on the constitution of the European Union and the common currency have been exposed to national ratifications in several countries, such as the recent Ireland vote. For academic evaluations of this process, see Hug (2004), Pahre (2004) and Hug and Schulz (2005).

¹³ See for example Lizzeri and Persico (2001), Milesi-Ferretti et al (2002), Persson (2004), Persson and Tabellini (1999, 2003, 2004a, 2004b) and Persson et al (2000).

¹⁴ The assumption that the aggregation of the international public good follows a summation technology can matter. The affect of domestic constraints on the supply of public goods when the aggregation is based on a different technology (for an example of a best-shot or a weakest-link scenario see Sandler 1997, chapter 2) remains an open question.

¹⁵ The ‘ideal point’ of each country is given by its own Nash equilibrium contribution and by an infinite contribution of the other country.

¹⁶ In the following we will use the feminine pronoun for the Home government H, the male pronoun for the median voter and the neuter pronoun for the foreign government F. See Tarar (2001) for a discussion of the case if F faces a domestic constraint as well.

¹⁷ This is a key assumption. See for example Mansfield et al (2000) who assume the median legislator plays the noncooperative game with the foreign government.

¹⁸ This case is similar to the tariff-setting game in Milner and Rosendorff (1997), with the difference that in their game F and V play the noncooperative game instead of F and H.

¹⁹ This becomes clear when one looks at their Figure 1, in which they compare a point x on the contract curve with a point b that is the constrained optimum: in their Figure 1 H would prefer b over x but x cannot be the result from an unconstrained optimization á la (2) to begin with.

²⁰ Mo (1994) finds that domestic constraints are a bargaining advantage if the domestic constituents are neither too weak nor too strong.

²¹ Mansfield et al (2000) avoid this by stating in their footnote 9 that ‘we require that the international negotiators choose a ratifiable offer that lies on [PO] when such an agreement exists’. See the discussion in Dai (2002) and Mansfield et al (2002).

²² For a discussion on how far the ‘divided government’ discussion also concerns parliamentary democracies see chapter 10 of Alesina and Rosenthal (1995), Milner (1997, 37–42) and Pahre (1997). See also the recent work on the effects of parliamentary versus presidential government forms on domestic public good contributions cited in footnote 13.

²³ Two other papers model two-level games with an explicit electoral stage: Morrow (1991) also assumes that the voter has a certain cut-off utility for the case a treaty (on arms control in his paper) between the two countries has been signed and a different cut-off utility for the case such a treaty has not been signed. These cut-off utilities, however, are independent from any electoral considerations on the voter's part. Governments and voters in Smith and Hayes (1997) have preferred points and indifference circles around them: if one government gains, the other loses, which is not so in our public good environment.

²⁴ Stage 0 could be a signalling game in an electoral campaign. See for example Banks (1990b) or Harrington (1992).

²⁵ The assumption that voters use such a retrospective-prospective voting rule can be traced back to Downs (1957). See also Williams (1994) and Bueno de Mesquita et al (1999). Also note that there is an implicit assumption in this model that a voter does not ‘punish’ a government if he or she fails to come up with an agreement. This assumption seems reasonable in cases where the nonagreement would not get a lot of media coverage.

²⁶ Including the additional constraint that V has to be better off compared to the Nash equilibrium does not change the main results. Note also that the second constraint in (11) means the political topic the countries are bargaining over is not important enough for H to risk losing elections. One could imagine that H is only partially willing to give up his convictions. This could be modelled by adding a positive constant to the left-hand side of the second inequality, and the results on the following pages, while fundamentally still intact, would be weaker with a larger constant.

²⁷ In all these three cases, ‘closer’ does not refer to a Euclidean distance but to a difference in terms of utility.

²⁸ The empirical study by Jensen and Spoon (2007, 1) finds policy preferences of governing parties, in particular green parties when in the government, ‘are significant predictors of [domestic] progress towards GHG [greenhouse gas] emission targets’.

Additional information

Notes on contributors

Jason F Shogren

 1  1 We thank Charles F Mason, Tom Crocker, Tommy Stamland, Owen Phillips, Eric Nye, Kevin Siqueira and seminar participants at Clarkson University, and the Public Choice meetings in Charleston for helpful comments on earlier versions. Shogren thanks the Norwegian University of Life Sciences and GREQAM for the financial support. All errors remain ours.