Tsebelis G. Veto Players: How Political Institutions Work

Подождите немного. Документ загружается.

301

several times in the last 15 years. These peculiarities have led students of the EU to characterize

it as a “sui generis” system (Westlake (1994: 29), Nugent (1994: 206)). Finally, EU institutions

include very complicated provisions. Decisions in the Council of ministers are made by a triple

majority, while two more institutions the European Parliament and the Commission participate in

the decisionmaking process quite frequently as veto players. For these reasons, I consider a

successful analysis of this fluid and unusual system as a demanding test for the theory in this

book. Should veto players analysis provide us with interesting and accurate insights about the

EU, the theory will have gone through a quite demanding test. The reader will see that the

structure of the legislative process has changed several times shifting power among the main

institutional actors. In addition, these changes have affected the role of other actors, like

bureaucracies and the judiciary, exactly the same way as in other countries as the first half of this

book has led us to expect, and exactly as the previous chapters of the second half have

demonstrated.

302

CHAPTER 9: GOVERNMENT STABILITY

As we saw in Chapter 4 government stability is an important variable for the study of

parliamentary systems. For example, Lijphart (1999: 129) considers government duration as a

proxy for “executive dominance” and differentiates his approach from what he calls the

“prevalent” point of view according to which “cabinet durability is an indicator not just of the

cabinet’s strength compared with that of the legislature but also of regime stability.” Huber and

Lupia (2000) argue that government stability increases ministerial efficiency because a minister

expected to stay in place will be respected by the bureaucracy.

Actually the formation of government coalitions and the duration of the corresponding

governments has probably been one of the most prolific branches in the literature of politics in

advanced industrialized democracies. Starting with the work of Riker (1962) coalition theorists

discovered the significance of “minimum winning coalitions”

124

and then proceeded to define a

series of other concepts useful for the study of coalition formation: “minimum size,” “minimal

range,” “minimum connected winning,” “policy viable.”

125

Empirical work on the durability of

different governments flourished (Dodd (1976), Sanders and Herman (1977), Robertson (1983),

Schofield (1987), Laver and Schofield (1990), Strom (1988), King et. al. (1990), Warwick

(1994)). Some of this work was based on the “numerical” composition of different governments

(number of seats they controlled, majority or minority status) other parts included the policy

positions of parties (either all of them, or only the ones composing the government), most of it

included additional information relevant to government formation (whether the government had

to receive an investiture vote from parliament, how many attempts at government formation were

made before a successful government).

124

Coalitions stop controlling a majority of seats in parliament if they lose a party member.

125

See Lijphart (1999: 91-96) for definition and discussion of all these concepts.

303

I will review this literature, point out the latest findings, and confront it with the main

expectation generated by the veto players theory: that policy stability leads to government

instability. In order to be able to generate specific predictions from the theory I will have an

extensive discussion of a fundamental (in game theoretic models) but elusive (in empirical work)

concept: the “status quo.” I will develop the difference between policies (like the ones we studied

in Chapter 7) and ouctomces (like the ones I did in Chapter 8) further, and explain why what

used to be a satisfactory situation in a country in the past may now require significant changes.

From this starting point I will study the implications for government survival. Governments are

the actors who are responsible for such adjustments, and since their composition affects their

capacity to act, it ultimately affects their probability of survival. As a result of this analysis,

government duration in parliamentary regimes will be linked to the configuration of veto players.

Once this link is established, I will revisit Lijphart’s analysis of the connection between

government duration and executive dominance introduced in Chapter 4, and explain why

government duration empirically correlates with executive dominance despite the fact that there

is no logical connection as we saw in Chapter 4.

This chapter is organized in three parts. The first part reviews the literature on

government duration and examines whether it depends on characteristics of the parliament or the

government of a country. The second uses the theory presented in this book to explain the

empirical findings. The third uses the findings in previous chapters of this book to explain why

government duration correlates with executive dominance although they are not logically related.

I. The Literature on Government Stability

Most of the literature on government duration correlates it with parliament characteristics

like the number of parties and their ideological distances from each other. More recent analyses

304

focus on government characteristics. In this section I will compare and contrast the two

approaches, but I will start by explaining what the literature measures when it discusses

“government stability.”

1. Government stability. While many authors have written about government stability,

they have not applied the same defining conditions for what a government replacement is. For

example, different authors do not agree if the situation where a government is replaced by

another with the same party composition should be counted as a single government or two

different ones.

More precisely, there are four different criteria used in the literature: whether the party

composition of a government changes, whether there is a formal government resignation,

whether there is a change in prime minister, and whether there is an election. While all authors

accept the first criterion, variations exist with respect to all the others. Dodd (1976) and Lijphart

(1984) accept only the first criterion as a necessary and sufficient condition for the change in

government. The second most frequently used criterion for government termination is an

election. Laver and Schofield (1990: 147) justify this criterion because an election changes party

weights in parliament, and consequently modifies the bargaining environment where coalition

formation takes place. The other two criteria have serious drawbacks for comparative analysis:

formal resignation is required in some countries and not in others, and resignation of a prime

minister may or may not be for political reasons, so there is wider disagreement with respect to

these two criteria.

Which one of these criteria is the most appropriate? I think that every author selects a

criterion that makes the most sense on the basis of his view of how the process works. For

example, if the view is that government composition depends on the relative power of different

305

parties, inclusion of elections as a criterion is a very reasonable choice. If it is perceived that

most prime minister (PM) resignations are political events even if actors involved are claiming

that they resign “for personal reasons” inclusion of PM resignation makes sense too. In Chapter 7

I adopted the Lijphart (1984) and Dodd (1976) criterion of party composition of governments by

creating a list of “merged” governments but replicated the calculations with the more traditional

definitions with the same qualitative results. Lijphart (1999) averaged the duration results

generated by his own criterion with the criteria applied by most of the literature and used

“average duration” generated this way to calculate his “executive dominance” index. We will

return to that point in the last section of this chapter.

2. Palriamentary features affect government stability. No matter what the criterion of

duration, it is usually correlated with parliamentary characteristics. For example, even in the

initial impetus for the development of coalition theories (Riker (1962) based on cooperative

game theory) a series of “policy blind” models were assuming that the coalitions formed would

be “minimum winning” in parliament so that ministerial portfolios would not be allocated to

parties that were not needed for a majority in parliament.

Subsequently, policy position criteria were introduced, and the underlying model

revolved around improving one’s position in the cabinet. Most of the time this meant increasing

a party’s portfolio share, although some analysts like De Swaan (1973: 88) maintained that “an

actor strives to bring about a winning coalition in which he is included and that he expects to

adopt a policy that is as close as possible… to his own most preferred policy.”

The implicit or explicit argument in all the approaches was that different parties will

force a government to resign when they have good chances to be included in the next

government and obtain better position. As a result, the characteristics of the parliament that

306

produce the governments enter into play. If a party was centrally located in the parliament, if it

were large, if other parties were dispersed or clustered were good predictors of the probability

that a party will be included in the next government. Here are the variables affecting government

duration on the basis of this literature: the number of parties in the political system (Duverger

(1954)), the “effective number of parties” in a system, the presence of anti-system or other

“extreme” parties, the degree of ideological polarization or “cleavage conflict” (all of these

conditions make it more difficult to form and maintain governments). Finally, a formal

investiture requirement eliminates some governments that might have survived for a while

otherwise (Laver and Schofield (1990: 147-48)). With respect to characteristics of the

government itself, the results of empirical analyses indicated that “minimum winning” status

increased government longevity, while Sanders and Herman (1977) and Schofield (1987) did not

find evidence that ideological compactness of the government affected its longevity.

All these approaches leave according to Laver and Schofield (1990: 155): “Two

important loose ends… The first is that there is considerable unexplained variation between

systems in the average duration of cabinets. The second is that the duration of cabinets seems to

be unrelated to policy matters, despite the fact that party policy greatly enhances our ability to

explain the formation of governments in the first place.” In order to address these two problems

Laver and Schofield introduce “bargaining environment” as an independent variable. They show

that in single dimensional policy space (left-right) one can divide different countries in unipolar

centrist, unipolar off center, bipolar and multipolar. They then demonstrate that countries with

centrist unipolar bargaining system (like Luxemburg and Ireland), or with bipolar system (like

Austria and Germany), have governments that last significantly longer than countries with

multipolar bargaining environments (like the Netherlands, Finland, Italy); countries with

307

unipolar (off center) systems (like Norway, Sweden, Iceland) have intermediate levels of

government longevity. Laver and Schofield (1990) were one of the last “deterministic models”

126

of government duration, and they pushed the method as far as it has reached.

The deterministic approach was criticized by the “events approach” models that are based

on the idea that actual dissolutions of governments are caused by random events that could not

have been anticipated by the actors (Browne et.al. (1984)). The initial attempt of the events

approach was not to focus on the causes of government duration, but to explicitly model its

randomness. Events models focused on the conditional probability that a government will be

terminated given that it had survived for a certain period of time. This conditional rate of

termination (hazard rate) assumed to be constant across countries became the dependent variable

in most of the analyses, but the empirical results were poor: hazard rates were different across

countries (the only exceptions were Belgium, Finland, Italy, and Israel).

King et al (1990) were the first to present a model that unified the two approaches. It

included the causal arguments of deterministic approaches along with the superior methodology

of the events approach: the model made hazard rates a function of the characteristics studied by

the deterministic models. In other words the new model was assuming that governments fall as a

result of random events, but the capacity of different governments to survive was a function of

different characteristics prevailing in the party system of the country. The results of this unified

model indicate that fragmentation of the party system and the polarization of the opposition are

the regime attributes the most strongly associated with cabinet duration.

As a result of the King et al. findings, Laver and Schofield (1990: 161) conclude: “…we

note that the fragmentation and polarization of the party system appear as the important variables

in an analysis that controls for a wide range of matters and even takes account of the impact of

126

Models that assume government duration to be a function of the independent variables included in model.

308

random shocks. These, of course, are the variables we identified as being important parameters

of the stability of the bargaining system and therefore liable to have an impact on cabinet

stability.”

In conclusion, the introduction of more advanced methodology did not alter the

conclusions of the coalition literature by 1990: the two main characteristics that affect

government survival in parliamentary systems are features of the party system of a country:

fragmentation and polarization. Both of these variables represent characteristics determined by

the parliament that selects the different governments, and both of them have to do with the

bargaining environment prevailing in this parliament.

3. Government features affect government stability. Several years later Paul Warwick

(1994) presented a serious criticism of the above results, and an alternative specification of the

underlying model. He moved the explanation of government survival from the parliament to the

government. He criticized both the parliament polarization variable and the parliament

fragmentation variable, and replaced them with similar variables describing governments.

With respect to polarization, Warwick argued that it did not necessarily affect the

complexity of the bargaining environment. In fact, it had the opposite effect: “Laver and

Schofield believe that the polarization variable reflects the overall complexity of party positions

in the party system; the more complex this array, the more vulnerable the distribution of

bargaining power to slight perturbations. But what the variable actually measures is the

proportion of parliamentary seats held by extremist parties, and since extremist parties are

normally considered unsuitable coalition partners by prosystem parties – “non-coalitionable” in

Laver and Schofield’s (1990: 200-1) terms – their presence should narrow the range of coalition

alternatives, other things being equal” (Warwick (1994: 46)). Warwick’s alternative explanation

309

of polarization: “Given the noncoalitionable status of extremist parties, governments formed in

such systems usually must either encompass an ideologically diverse array of prosystem parties

and/or settle for being minoritarian; either way, they are vulnerable to early collapse or

termination. Apart from accounting for the sign of the polarization coefficient, this explanation

has the advantage of locating the proximate cause of government survival in a particular

government attribute, rather than associating it with a general feature of the larger parliamentary

environment” (Warwick (1994: 47)).

With respect to fractionalization, Warwick argued that the measure should reflect the

situation inside the government: “King et. al.’s interest in size or fractionalization extended only

to the size of the party system; they never tested the size of government. Once government size is

taken into consideration, however, it eliminates the significant role played by effective party

system size, indicating that large party systems tend to experience greater instability because the

governments they produce are themselves large. This refinement on the King et. al. model

implies that if there is validity to the bargaining-environment idea, it is the bargaining

environment within the government that matters, not the larger parliamentary bargaining

environment” (Warwick (1994: 47)).

Warwick introduces the variable “ideological diversity” that is similar to what I called

“range” in Chapter 7 and “ideological distances” in Chapter 8. It measures the ideological

diversity of the government coalition on the basis of a series of indicators including left-right,

clerical-secular, and regime support. The introduction of this variable turns polarization (that is,

the size of extremist parties) into an insignificant independent variable for majority governments.

For minority governments the opposite is true: while the ideological diversity of the government

is not significant, polarization is. Warwick (1994: 66)) explains the difference as follows:

310

“Although polarization shows a highly significant impact on government survival in both

majority and minority situations before ideological diversity is introduced (Models 1 and 3), a

comparison of Models 2 and 4 shows that it survives the introduction of ideological diversity

index only in minority government situations. Correspondingly, the significant effect conveyed

by the ideological diversity index is confined to majority governments.”

To summarize the arguments: Most of the literature up to 1990 was explaining

government duration in parliamentary democracies on the basis of characteristics of the party

system mainly ideological diversity and fractionalization. The reason that they were focusing in

characteristics prevailing in the parliament that was producing each government was that the

underlying argument was that parties would determine their behavior on the basis of the

probabilities they had to be introduced in the new government, and these probabilities are

determined by characteristics of the party system. Warwick (1994) performed crucial

experiments and introduced both characteristics of the parliament and of the government. The

result was that the number of parties in parliament was replaced by the number of parties in

government, that was in itself replaced by the ideological distances between parties in

government for majority governments. For minority governments (that are usually single party

governments) the ideological diversity of parliament remains a strong explanatory variable.

Warwick interprets his findings as indicating that what determines survival is bargaining within

governments. I will go over these findings on the basis of veto players theory.

II. Veto players and Government Stability

So far in this book I have been presenting results on the basis of spatial models without

being specifically concerned about the position of the status quo. At the theoretical level

(Chapters 1 and 2) I generated propositions that held for every position of the status quo, while at