Presentation on theme: "EITM 2007 Institutions Week John Aldrich Arthur Lupia."— Presentation transcript:
EITM 2007 Institutions Week John Aldrich Arthur Lupia
The modal form of democracy is parliamentary The legislature chooses the government. The government = the prime minister and the cabinet. To survive, a government must retain the support of a legislative majority.
Coalition government is the modal form in parliamentary democracy The number of parties earning seats varies. In many countries, it is rare that a single party earns a majority of seats. Post-election bargaining among political parties determines the distribution of cabinet posts. Policy making requires cooperation among parties that will later compete electorally.
Coalition Governance Birth How do coalitions form? Who is in the coalition? Life How do coalitions delegate? How do they manage joint problems of governance and rivalry? Death When and under what circumstances do coalitions terminate?
Research on Coalition Governance Until very recently, not a primary focus of modeling. Activity is increasing, particularly amongst younger scholars. Modeling allows efficient comparative work. Studies that integrate theoretical models with large scale empirical work are very rare.
Coalition Duration M. Most parliamentary governments can end at any moment. When and how they end has a broad societal impact. NH. CDs causes and consequences are independent of structural attributes, critical events, country-specific factors, negotiation dynamics. P. Vary by paper. C. Increasingly integrated theoretical and empirical models yield better explanations of coalition duration.
Browne, et al (1986) M. Why do governments dissolve before their time? NH. Structural attributes largely determine a governments duration. P. Governance contains stochastic elements. C. Stochastically occurring critical events explain more variation.
Browne Premises The probability, p, of a critical event occurring in a given time interval is both low and invariant across time intervals ( a Poisson process). Partition the CIEP into N intervals (Np=1). P(X=r|N, p) = [e-Np(Np)r]/r! P(X(t) dissolved|X(0) undissolved) = 1-e-Npt. The baseline expectation is of a constant flow of events. The stated null hypothesis is that observed distributions will not follow this model.
Browne Data 238 cabinets from 12 European countries 1945-1980. Caretaker governments excluded.
Browne implications Cabinets with multiple members, minority governments possess less ability to deflect critical events. Structural attributes approaches not rejected, but stochastic explanations improve explanations. Unidentified, however, are the descriptive attributes of such events…
King, Alt, Laver and Burns (1990) M. Structural attributes versus critical events. NH. A unified explanation is no better. No cabinets are more durable than others. The real story is country-specific. P. Unifies the two approaches. Includes censoring, institutional and country-specific variables. C. The approaches are reconcilable. Unified explanation superior.
KABL Premises Critical Events: Y i = e - yi Y i – a random variable describing cabinet duration length, the rate of event occurrence 1/ expected duration Structural Attributes: Y i = x i + i Could duration be generated by a normal distribution?
KABL Premises Is cabinet durability constant for the entire history of a country? The termination hazard has systematic and stochastic components. Durations are independent. Governments lasting longer than 12 months before the CIEP ended partly because of its shadow.
King, Alt, et. al. Conclusions Model 1.1. Browne et al. Baseline Model 1.2. Censoring improves the fit. Model 1.3. Include country and structural attributes. Even better fit. Majority status increases duration. Number of formation attempts reduce duration.
KABL Premises Model 2.1. Best Country attributes only model. Models 2.2 -2.3 Structural attributes added. Improved fit. Comparing best models, country-specific effects disappear. Number of formation attempts corresponds to less durability. Table 3 shows the fit. Q: Censoring and strategy?
Warwick (1992) M. Are termination hazard rates constant? NH. Yes. P. Event history model applies. What is the termination rate given survival at time t? (t)=exp( x(t)) 0(t) (t) – hazard rate; 0(t) baseline rate after x considered. Censor only involuntary terminations. T1. Without other factors, (t)/ t >0 for most countries. T3. W/ other factors, baseline rate increases w/ term length.
Bargaining is Essential If no party can impose its preferences on others, bargaining is required. Parties bargain in the shadow of citizen opinions and political rivals. The quality of coalition governance depends on what politicians know and how they react to uncertainty about each other, the voters, and the future.
Coalition Termination and the Strategic Timing of Parliamentary Elections M: What determines the timing and nature of coalition terminations? What factors affect government membership and portfolio allocation? Null Hypotheses: 1. Parties terminate cabinets when they expect electoral gains. 2. Structural attributes or critical events are sufficient to explain the timing of governmental transitions. A strategic approach adds nothing. 3. Size is power in parliamentary bargaining.
Lupia-Strom Premises There are three parties. Two are in the incumbent coalition. A legislative session is in progress Parties have common knowledge and expectations Parties have policy preferences and receive benefits from being in government Electioneering and bargaining are costly
Result 1 OLD. "Parties terminate cabinets when they expect electoral gains". Grofman and van Roozendaal (1994: 158) NEW. A party with favorable electoral prospects can gain from non-electoral means. If other partners want to avoid elections, offer a cabinet reshuffle. Good electoral prospects for one party are not sufficient to cause a parliament's dissolution.
Result 2 OLD: The governing coalition will comprise the largest and smallest legislative parties (Austen-Smith and Banks 1988 and Baron 1991). NEW: The offer depends on walk away values. All else constant, parties prefer a coalition that generates greater gains. a larger share of such gains. A formateur will choose partners with bad electoral prospects, (2) high election-related transaction costs, (3) high discount rates, (4) bad coalition alternatives, and, (5) assets that would be lost after termination or dissolution.
Example Premises Party A has 49 seats. Party B has 48 seats. Party C has 4 seats. Any coalition including C produces value. Any coalition without C produces no value. Results J C has the fewest seats, but the largest walk- away value. J The only sustainable outcome involves a contract giving party C almost all of the power.
Walk Away Values A walk away value is what a person, such as a party leader, can gain without an agreement with other players. The electoral connection has its impact here. Who gets what depends on the walk away values of potential coalition partners. Walk away values are why size need not equal power in coalition bargaining.
The Effect of Other Factors If an institutions, preferences, resources or country- specific factors are to affect coalition governance, they must affect a pivotal actors walk away value. Example: Institutions formateur rules size/composition rules internal party rules independence of the judiciary and civil service
Implication OLD. The time it takes a coalition to negotiate tells us something about the quality of governance. NEW. When time can be used strategically, a lengthy process implies... Lack of information about walk-away values. Low discount rates/high walk-away values. Nothing about effectiveness of the agreement.
Tested in Daniel Diermeier and Randy T. Stevenson. 1999. Cabinet Survival and Competing Risks AJPS 43: 1051-1068. Sanford C. Gordon. 2002. Stochastic Dependence in Competing Risks AJPS 46: 200-217. Bernard Grofman and Peter van Roozendaal. 1995. Exit models, hazard rates and government duration. With empirical application to government duration in the Benelux countries (1945-1994). ISCORE paper.
Diermeier and Stevenson (2000) M. Can institutional and bargaining considerations improve empirical work on cabinet termination? NH. Critical events, structural attributes, non-strategic approaches or static models are sufficient to answer the question. P. Stochastic version of Lupia and Strom. C. Dissolution hazards increase. Replacement hazards do not.
Lupia and Strom (1995) Conclusions Favorable electoral prospects are neither necessary nor sufficient for termination or dissolution. Bargaining advantages often attributed to party size can be the result of other factors. Given a particular crisis, government replacement depends on the electoral cycle.