Electoral Rules and Government Spending in Parliamentary ...

1 Quarterly Journal of Political Science, 2007, xx: 1 34 Electoral Rules and Government Spendingin Parliamentary Democracies Torsten Persson1, Gerard Roland2and Guido Tabellini31 IIES, Stockholm University; CEPR; NBER; and CIAR;E-mail: of Economics and Political Science, UC Berkeley; CEPR;E-mail: Bocconi University; CEPR; CES-Ifo; and CIAR;E-mail: present a theoretical model of a Parliamentary democracy where Electoral com-petition inside coalition governments induces higher Spending than under singleparty governments. Policy preferences of parties are endogenous and derived fromopportunistic re-election motives. The Electoral rule affects Government Spending ,but only indirectly: proportional elections induce a more fragmented party systemand a larger incidence of coalition governments than do majoritarian evidence from post-war Parliamentary democracies strongly supportsthese : Electoral Rules ; party systems; coalition governments; fiscal policy; Electoral Code: JEL Classifications: H00, D72, D78.

2 A longer version of this paper has circulated under the title: How do Electoral Rules shape partystructures, Government coalitions and economic policies? This research is supported by CIAR, theSwedish Research Council, Bocconi University, MIUR and the Communaut Fran aise de Belgique,as well as a grant from the Center for Business and Policy Studies financing data collection. Weare grateful to Nolan McCarty, two anonymous referees, Antonio Merlo, Matthias Messner, RogerMyerson, Fausto Panunzi, Ken Shepsle and Dan Trefler, K re Vernby, and seminar participants atBerkeley, Bocconi, CERGE-EI, CIAR, IIES, KUL, MIT, Pennsylvania, LSE, UC Irvine, UIUCand Yale for their helpful comments, and to Krister Lundell, Andrea Mascotto and Jose MauricioPrado Jr.

3 For their skillful assistance with xxxxxxxxx; DOI xxxxxxxxxxxxx 2007 T. Persson, G. Roland and G. Tabellini2 Persson, Roland and TabelliniHow do Electoral Rules influence Government Spending ? We address the question buildingon three distinct lines of long research tradition in political science has studied the influence of electoralrules on party structures. Various models of Duverger s law predict that majoritarianelectoral rule leads to a two-party system and many empirical studies have been devotedto testing this prediction. This literature implies that, via the party system, majoritar-ian elections produce single-party governments more often than proportional elections,which instead produce fragmentation of political parties and coalition, or minority, , recent empirical research in economics has shown that Electoral Rules exerta strong effect on fiscal policy: majoritarian elections are associated with smaller gov-ernment Spending , smaller budget deficits and smaller welfare states, than proportionalelections.

4 An important question left open in this research concerns the mechanismbehind these reduced form results. Is the association between Electoral Rules and policyoutcomes due to different party structures and types of governments under alternativeelectoral Rules ? Or is it due to the direct effects of Electoral Rules on the incentives ofpoliticians, for given party structures?2No clear and rigorous answer to these questionshas been formulated so third line of research on which we build concerns the so-called common poolproblem in fiscal policy. It is commonly argued that coalition governments spend more(or run larger budget deficits) than single-party governments, because each party in acoalition does not fully internalize the fiscal costs of Spending .

5 Even though this claimfinds some support in the data,3it has not really been derived from first , a single-party Government is assumed to behave as a unitary decision maker,while a coalition Government faces a collective choice problem. But why should a singleparty representing several groups in society behave any different than a coalition ofparties representing the same groups? Moreover, the economic literature on the commonpool problem takes the motivation of parties, as well as the number of parties, as given,with no link to the Electoral paper is an attempt to provide the missing links between these existing strandsof research. We present a theoretical model where Electoral Rules shape the equilibriumnumber of parties and thus spill over on Government formation and fiscal policychoices.

6 When testing the empirical predictions of the model, we show that the indi-rect effects of Electoral Rules on number of parties and type of Government drive1 See, for instance, Cox (1990, 1997), Laver and Schofield (1990), Lijphart (1984a, 1984b, 1994, 1999),Palfrey (1989), Powell (1982, 2000), and Taagepera and Shugart (1989).2 The incentive channel is explicit in recent contributions by economists studying how electoralrules influence the composition (Lizzeri and Persico 2001, Persson and Tabellini 1999) or size(Milesi-Ferrettiet ) of Government Spending . Persson and Tabellini (2003, 2004a) presentreduced-form empirical results see also the commentary by Acemoglu (2005) and the survey byPersson and Tabellini (2004b).

7 3 See, , the contributions by Poterba and von Hagen (1999), in particular the chapter by Kon-topoulos and Rules and Government Spending in Parliamentary Democracies3the result that majoritarian elections lead to less public Spending than main theoretical result shows that the distinction between single-party and coali-tion governments is indeed central for the size of public Spending . In our model, politi-cians are opportunistic and the policy preferences of parties endogenous. The centralmechanism is that voters can discriminate between the parties of a coalition governmentat the polls, while they cannot discriminate between different factions of a single-partygovernment.

8 This creates Electoral conflict anelectoral common pool problem withina coalition Government , but not within a single-party Government . A recent and veryinteresting paper by Bawn and Rosenbluth (forthcoming), to which we owe considerableinspiration, discusses a similar idea in a less formal this result, we endogenize party formation and the type of Government . Specif-ically, we provide a simple model example where primitive groups of politicians decidewhether to form large or small parties. Under proportional elections these choices leadto a more fractionalized party system than under majoritarian elections. The mechanismin our model generating this Duvergerian result is thus the strategic choices by activepoliticians, rather than strategic voting by the electorate the common mechanism inthe literature.

9 Here, we follow the call of Riker, who in his survey of Duverger s lawstated (Riker 1985, p. 764)The direction one must go, I believe, is to turn attention away from the expected utilitycalculus of the individual voter and to the expected utility calculus of the politician andother more substantial participants in the model example is not a general and comprehensive theory covering the entirepolitical process from party formation to economic policy choices, but it has sharp andtestable predictions: PR induces higher Spending than majoritarian elections, but onlythrough more party fragmentation and higher incidence of coalition Government . Inother words, if we hold the type of Government constant, the Electoral rule has no directeffect on public believe that the specificity of our predictions illustrate the usefulness of formalmodeling as a basis for empirical work.

10 The theoretical predictions are confronted withpolitical and economic data from up to 50 Parliamentary democracies in the post-warperiod. We present empirical estimates relying on the cross-country variation in the data,as well as the within-country variation around reforms of Electoral systems. Both typesof estimates support our theoretical result that PR induces more Government Spending ,but only indirectly, via party formation and the incidence of coalition governments. Theoverall effect is similar in magnitude to earlier empirical results on reduced form: a fullscale shift from majoritarian to proportional elections raises overall public Spending byroughly 5% of general issues studied in this paper have been the focus of many contributionsby political to the best of our knowledge, no other paper (except Bawnand Rosenbluth, forthcoming), has tried to integrate in a single model endogenous4 Persson, Roland and Tabelliniparty formation, Government formation and fiscal policy choices.