Tsebelis and Money: Bicameralism

Tsebelis and Money. 1997. Bicameralism. Cambridge: Cambridge University Press.

In Brief

"This book argues that the interaction between the two chambers in bicameral legislation is central to comprehending behavior within each chamber, a point thus far neglected in the study of bicameral legislatures. The authors examine how the bicameral legislatures of some fifty countries produce legislation. They use both cooperative and noncooperative game-theoretic models to understand the interactions between the chambers observed in these fifty countries. Cooperative models are used to establish that bicameral legislatures, when compared with unicameral ones, increase the stability of the status quo and reduce intercameral differences to one privileged dimension of conflict. Noncooperative game-theoretic models are used to investigate the significance of a series of institutional devices governing intercameral relations: where a bill is introduced, which chamber has the final word, how many times a bill can shuttle between chambers, and whether conference committees are called. These models are corroborated with data from the French Republic, supplemented with case studies from Germany, Japan, Switzerland, the United States, and the European Union." [from the back cover]

Chapter-by-Chapter Notes

Chapter 1: Efficiency vs Redistribution

There are two dimensions along which a bicameral legislature can change policy.

1. The efficient dimension: a decision that makes both houses better off (thus, common interests of the two houses). This may lead to both higher quality and more stable legislation (pgs 37-40).
2. The political dimension (i.e. redistribution): a decision that is closer to the preferences of one house than the other. This may help prevent a tyranny of the majority/agenda-setter/minority by giving a voice to more groups (pgs 35-6).

In the present literature, many scholars have advanced arguments in favor of bicameralism, especially arguing that it promotes stability and better legislation. Pages 35-40. Of course, stability is the opposite of responsiveness, so there is a tradeoff.

The book's key questions, based on a historical and intellectual review of bicameralism in chapter 1:

1. Does bicameralism actually have the predicted effects? Are ideas in the current literature correct and mutually consistent?
2. Do specific institutional features affect legislative outcomes? Which ones?

Bicameral institutions will be examined along the two dimensions above, efficiency and redistribution.

Chapter 2: Measuring Bicameralism

This chapter presents an empirical classification of bicameral institutions into a dataset. To classify the strength of each country's bicameralism, we should examine two main variables:

1. Upper house membership. Lower houses are almost always elected by universal suffrage for a term, unlike upper houses, which are elected/chosen through many different mechanisms. The houses cannot be assumed to have identical preferences even if they are controlled by the same party: differences in rules in the two houses may make a difference too. (e.g. in US Senate, it takes a 60% vote to end a filibuster, so the House and Senate have different median voters even if they have identical composition).
2. Methods of resolving inter-house disputes (i.e. which house is superior, or are they equal?). There are three basic ways of resolving disputes btw the houses. (a) The navette, or shuttle system: the bill goes back and forth until both houses agree on the same thing. There may be a limit on how many times a bill can shuttle back and forth. (b) Conference committees, which run like a unicameral legislature with delegates from both houses; these committees then return the bill to their respective houses for a closed (no amendment) debate (usually). (c) Unicameral mechanisms, like a joint session or a decision by one house. These mechanisms can be understood with models of unicameralism.

These classifications give rise to two empirical questions to be examined in the book.

1. How do these diverse institutional differences among bicameral systems affect legislative outcomes?
2. When the upper house is weak relative to the lower house, what is its role? Does it have any impact on legislation?

Introduction to Part II (chaptesr 3-5): Overview of the Model

See pg 74 for a brief model to understand the roles of efficient and political conflict. An 'efficient' move is one that shifts the status quo towards the line connecting the medians of the two houses. A 'political' move is one that shifts the status quo along this line towards one of the houses' ideal points. This book will use both cooperative and noncooperative game theory.

• Cooperative game theory assumes agreements are enforceable, thus focuses only on the decision itself.
• Noncooperative game theory assumes agreements are not enforceable, thus a focus on institutions, order of play, etc.

Chapter 3: The Dimensionality of Intracameral Conflict

The chapter examines "three different solution concepts of cooperative game theory -- the core, the uncovered set, and the tournament equilbirium -- and [demonstrates] that all three solutions point to one predominant dimension of [intercameral] conflict" (78). In English: Even if members of each house have preferences along more than one dimension, the conflict between the houses will center (primarily) around one dimension. This sets the stage for chapter 4's unidimensional noncooperative analysis.

Chapter 4: The Model of Intracameral Bargaining

The model: The authors examine a divide-the-dollar game between two players (houses), first with complete information, then with incomplete information. (Complete information: players know each other's payoffs; incomplete information: they don't.)

• A divide the dollar game with complete information: the less patient player (the one that discounts the future more) gets less (because he is willing to concede more now than try to get more later--it's worth the same to him). Thus, if the lower house has the final word, the upper house gains power if the allowable number of negotiating rounds is higher. Similarly, if the default solution is a conference committee, the more powerful house loses power as the number of negotiating rounds increases. Also, if there is a limited number of times a bill can shuttle back and forth, the house that introduces a bill first has an advantage.
• A divide the dollar game with incomplete information becomes a bargaining game, so no further analysis is necessary.

In English, this is the main logic of the model: The two houses negotiate primarily along one dimension; for example, imagine they need to pick a number between 0 and 100. This is analogous to "dividing a dollar" and deciding how much each player gets (between 0 and 100 cents). There is a key difference, though: you don't know whether your opponent wants the whole dollar, part of it, or none of it (because the "dollar" might be the size of the deficit or something). One player might want to pick "80," the other might want "20," but you don't know the other player's preferences--so he might be bluffing. Because each player discounts the future, each player is a bit impatient to make a deal. The less patient player will get less of what he wants (see page 100).

This model leads to a few predictions:

• Prediction 1: "The relative power of each house in bicameral legislatures is a function of institutional constraints (number of possible iterations, stopping rules, who initiates the process) and the impatience of each legislature to reach a deal." (If the lower house ultimately decides, the upper house's power increases as the number of shuttle rounds increases. The house making the initial proposal has an advantage.)
• Prediction 2: "The number of actual negotiating rounds in bicameral legislatures increases with the uncertainty of one house about the other house's willingness to compromise and with the time discount factor of each house." (104)

Thus, this model allows the authors to make point predictions about the outcomes of bicameral bargaining (unlike the model in chapter 3). It can also offer comparative statics predicitons for small changes in procedure. This model is unidimensional and assumes that each legislature has a discount factor.

In sum: The strength of each chamber's bargaining position is a function of institutional constraints, each player's patience, and the level of uncertainty regarding this impatience (uncertainty leads to protracted bargaining). To test this model's predictions, we need to gather data on the following variables:

• Dependent variables: How long will bargaining last? who will get closer to their ideal point?
• Independent variables: How many rounds of bargaining are allowed? How patient is each house? How much does each house know about the other player's preferences and patience? Which house proposes the bill? How are disagreements resolved?

Chapter 6: A Test of the Model

To test the model in chapter 4, this chapter presents a case study of the French Fifth Republic. France is used because the composition of the Senate remained fairly constant, but the composition of the National Assembly varied wildly. They begin by operationalizing their variables:

• Operationalize discount factor as the size/cohesiveness of the ruling coalition in each house (since a smaller coalition means you fear losing power, so you want to get your agenda enacted asap). Also, impatience grows over the course of the legislative term.
• Operationalize uncertainty: uncertainty is highest when coalition controls around 20 percent of seats, and during the middle of the legislative period.
• The Senate has fairly constant composition, so its discount factor is assumed to be known. However, uncertainty about the Assembly's discount factor is high if the ruling coalition is either very high or very low.

They then examine several sets of predictions:

• Model 1: length of navette process: longer if Assembly's discount factor is less clear
• Model 2: similar, but applied to opposition party
• Model 3: combines 1 and 2
• Model 4, 5: add learning and electoral cycles to the equation
• Model 6: tests whether the strength of the coalition/opposition affects the length of the navette process in a linear way.
• Results: they all work pretty well.

To summarize this chapter: The length of the shuttle/negotiation process depends on each house's impatience to end the bargaining, and on uncertainty about this impatience.

Chapter 7: The Process of Intercameral Bargaining

Chapter 6 tested the model's (given in chapter 4) predictions; now we test its underlying assumptions. In essence, chapter 6 was intended to demonstrate correlation among the model's independent and dependent variables, and chapter 7 is used to strengthen our belief in the model's internal logic.

Although the authors recognize that there are too few cases to make a causal inference (p 174), they argue that the results are nonetheless consistent with their model's hypotheses, suggesting that their model is at least plausible. Thus, they claim to have evidence that "legislative houses are driven by impatience to pass legislation and that variation in levels of impatience, along with institutional rules governing the navette process, affect the power of the Senate to influence outcomes." (174)

The authors claim to have refuted Lijphart's earlier claim that weak upper houses are useless: an upper house can exert influence simply through "patience" even if it is weak. They also claim to have completely refuted the theory that Senatorial (upper-house) wisdom and expertise matter.

The "Presidential Attributes" theory does accurately predict eventual outcomes, but it doesn't explain why the Senate still has influence when preferences are dissimilar. But impatience does predict in all cases.

Chapter 9: Implications for Future Research

Future researchers should consider these questions:

• Is preservation of the status quo (i.e. stability) good or not? What is its connection to government stability?
• Applied to U.S. Congress: we need studies of the relationships between the houses, committes, and conference committees.
• Also, several methodological questions.