APPLICATION: PIVOTAL POLITICS 1
A. Goals Pivotal Politics 1. Want to apply game theory to the legislative process to determine: 1. which outcomes are in SPE, and 2. which status quos would not change in SPE (i.e., the gridlock interval). 2. We will build up, with models for 1. One chamber, without 2/3rds override finished last week. 2. One chamber, with 2/3rds override start slide 59. 3. One chamber, with 2/3rds override, and filibuster pivot. 3. Assume all actors are rational and complete information. 2
One Chamber, Veto override Assume: median of chamber proposes, president accepts or rejects, veto override. Game Sequence: 1. Median of chamber (M) proposes bill b. 2. President (P) may veto or sign. 3. Congress can override veto with 2/3 majority (v L left pivot, V R right pivot of single chamber). Now go to slide 59 from last week 3
Pivotal politics (Krehbiel 1998) Median Legislator 1. Median legislator proposes a bill propose bill b Filibuster Pivot 2. Filibuster pivot filibusters or invokes cloture (passes bill) filibuster cloture (pass) q veto President sign 3. If bill passes, President signs or vetoes Veto Pivot b 4. If bill vetoed, Veto pivot overrides or sustains sustain override q b 4
Analysis of vetoes and proposals q < p q p v L f L m f R v R First, graph overrides President could sign or veto, because she cannot affect the outcome (if m proposes rationally, it will pass). Second determine whether president signs or vetoes. -Same as before. 5
Analysis of vetoes and proposals q < p q p v L f L m v R Favor: W f L (Q) = W fl (Q) W fr (Q); Furthermore, W V L (Q) = W VL (Q) W fl (Q). f R Third, graph what filibuster pivots favor. Note: in these cases, the preferences of the override pivot defines the outcome. 6
Analysis of vetoes and proposals q < p q p v L f L m m will propose m because m is in W V (Q) which L will pass. Hence, m is the outcome. f R v R Fourth, consider what m would propose. -Same as before. 7
Analysis of vetoes and proposals p < q < v L p q v L m v R f L f R First, graph overrides President is indifferent between signing and vetoing because W V (Q) will be the outcome in L either case. Second determine whether president signs or vetoes. -Same as before. 8
Analysis of vetoes and proposals p < q < v L p q v L m v R f L Favor: W f L (Q) = W fl (Q) W fr (Q); Furthermore, W V L (Q) = W VL (Q) W fl (Q). f R Third, graph what filibuster pivots favor. Note: in this case, the preferences of the override pivot defines the outcome. 9
Analysis of vetoes and proposals p < q < v L q v L m v R x p f L f R Fourth, m will propose x because x is the element closest to m that is in W V L (Q). Hence, x is the outcome. consider what m would propose. -Same as before. 10
Analysis of vetoes and proposals v L < q < f L < m p v L q f L m f R v R First, graph overrides. W VL (Q) W VR (Q) = No overrides. 11
Analysis of vetoes and proposals v L < q < f L < m p v L q f L m f R v R President vetoes because m wants to move the bill to the right. Second, determine whether president signs or vetoes. 12
Analysis of vetoes and proposals v L < q < f L < m q p v L f L m f R v R Third, graph Favor: W f L (Q) = W fl (Q) W fr (Q). what filibuster pivots favor. 13
Analysis of vetoes and proposals v L < q < f L < m p v L q f L m f R v R m cannot propose anything that passes, so m proposes a throw away (i.e. any x: x > q) same as before. Fourth, consider what m would propose. 14
Analysis of vetoes and proposals m < q < f R p v L f L m q f R v R First, graph overrides. W VL (Q) W VR (Q) = No overrides. 15
Analysis of vetoes and proposals m < q < f R p f L m q f R v L President signs anything in W m (Q) because he prefers that to q. v R Second, determine whether president signs or vetoes. 16
Analysis of vetoes and proposals m < q < f R p f L m q f R v L q cannot be get past filibuster pivots, because f L < q < f R. Generally: any q: f L < q < f R cannot be defeated. v R Third, graph what filibuster pivots favor. 17
Analysis of vetoes and proposals m < q < f R p f L m q f R v L v R m cannot propose anything that gets past filibuster pivots, so m proposes any x this outcome is new. Fourth, consider what m would propose. 18
Analysis of vetoes and proposals m < f R < q < 2f R -m p v L f L m q New case f R v R First, graph overrides W VL (Q) W VR (Q) = No overrides. President signs anything in W m (Q) because he prefers that to q. Second, determine whether president signs or vetoes. 19
Analysis of vetoes and proposals m < f R < q < 2f R -m f R p v L f L m q v R Favor: W f R (Q) = W fl (Q) W fr (Q). Third, graph what filibusters favor. 20
Analysis of vetoes and proposals m < f R < q < 2f R -m x p v L f L m f R q m will propose x because x is the element closest to m that is in W f (Q). Hence, x is the outcome (this is new). R v R Revisiting, what will the president do? Note, or any q in [f R, 2f R -m], the outcome is the point furthest left in W f R (Q). Only the filibuster pivot on the far side comes into play in the model. Fourth, consider what m would propose. 21
SPNE policy outcome Pivotal Politics Summary 2f R -m f R m v L p 2v L -m p v L m f R 2f R -m Status quo (q) The status quos that cannot be defeated are between v L and f R -- an wider range than without the filibuster pivot. All outcomes will be between v L and f R. Extreme status quos are still dictated by m. 22
SPNE policy outcome Pivotal Politics Summary 2f R -m f R m v L p 2v L -m p v L m f R 2f R -m Status quo (q) Gridlock interval This range is called the gridlock interval because status quos in this interval do not change. 23
Comparison of EIG across models Median voter F M V P Note: P is on right. We had him/her on left. Veto with override F M V P Pivotal politics F M V P Full gridlock Partial convergence to M Full convergence to M 24
Empirical implications Gridlock interval: the set of points in equilibrium under the rules of the game. If the gridlock interval becomes bigger than previous Congress, less legislation should pass. If the gridlock interval becomes smaller than previous Congress, more legislation should pass. Krehbiel tests this by assuming the distribution of q is uniform and looking at the volume of major legislation. 25
Bush Sr. Clinton Clinton Clinton Clinton W. Bush W. Bush W. Bush W. Bush Obama 26
Krehbiel (1998) Pivotal Politics 27
Gatekeeping Model Gatekeeping model (Denzau and Mackay, AJPS 1983) Committee has monopoly power regarding the introduction of a bill in its jurisdiction. If a bill is sent to the floor, it is considered under an open rule (i.e., unlimited amendments). Cox and McCubbins (2005) apply the gatekeeping model to argue that the majority party has negative agenda power even if it cannot influence votes on the floor. 28
Gatekeeping model L is the legislative median 29
Analysis of committee s decision Suppose L < C Propose bill if and only if b* = L is closer to C than the status quo q: q C L C L C 30
Policy outcomes of gatekeeping model SPNE policy outcome 2C-L C L L C 2C-L Status quo (q) Note: block everything in this area (i.e. it s the EGI). 31
pivotal politics vs party cartel 32
Summary Spatial models of legislative politics rely heavily on extensive form games (including complete and perfect information) Agenda setting model proposal power may be constrained by the veto player, explains gridlock Veto override Congress constrained by preferences of veto player closest to president Pivotal politics adds filibuster pivot, greater gridlock, dampens convergence to median voter from both directions Gatekeeping reverses sequence, blocking power prevents some policy movement towards median voter 33
Chiou and Rothenberg (2003) Four Models 1. Preference (or Basic): same as Krehbiel, except a. Two filibuster pivots, b. Two chambers, c. Senate median proposes. 2. Party Agenda Setter: same as (1), except a. Senate majority party median proposes, b. Both floor medians have to accept or reject, 1) Previously, if one chamber proposed, it wouldn t accept or reject. 3. Party Unity: same as (2), except a. Everyone in a party votes identical to their party s median in the chamber. 1) The size of the two parties is now relevant for filibuster pivots and override pivots. 34
Chiou and Rothenberg (2003) Four Models 1. Presidential Leadership: same as (3), except a. The president s party will never oppose the President in a vote. 35
Chiou and Rothenberg (2003) 84 th Congress (1955-57) MP S - the number of majorityparty members in the House and Senate. C - the number of legislators required to end filibuster. EGI corresponds to the SPE on the 45 degree line. Not always the case that the party unity model produces the widest EGI. Presidential leadership model not shown. 36
Chiou and Rothenberg (2003) 37
Test Dependent Variable: Chiou and Rothenberg (2003) Legislative Gridlock: total number of failed proposals divided by number of Congressional proposals (Binder). Two others, in separate regressions. Independent Variables: Gridlock interval: SGI measured using DW-NOMINATE. Budget: budget deficit as a percentage of federal outlays. Pubic Mood: some public liberalism index. Divided Government: one or more of the chambers are controlled by a party different from the President. 38
Chiou and Rothenberg (2003) Preference Model 39
Chiou and Rothenberg (2003) Party Agenda Setter 40
Chiou and Rothenberg (2003) Party Unity 41
Chiou and Rothenberg (2003) Presidential Leadership What do you think? 42