The veto as eectora stunt EITM and a test with comparative data Eric ITAM, Mexico City Apr. 2, 203 eitm@mpsa
Motivation A Form: see EITM in action forma mode 2 comparative statics 3 fasifiabe impications 4 test B Substance: Why vetoes? Caused by position taking (cf. Grosecose&McCarty 200), not uncertainty (cf. Cameron 2000) 2 Veto overrides enter the anaysis, branch symmetry 3 Comparative research design
Minor tweak to a standard mode Agenda setter mode appied to exec. eg. reations (Romer&Rosentha 978, Kiewiet&McCubbins 988) Add propensity 0 π to take positions accept x Nature: π propose x e veto v sustain x 0 not x 0 override x π = 0
Minor tweak to a standard mode Agenda setter mode appied to exec. eg. reations (Romer&Rosentha 978, Kiewiet&McCubbins 988) Add propensity 0 π to take positions accept x Nature: π propose x e veto v sustain x 0 not x 0 override x π = 0
Minor tweak to a standard mode Agenda setter mode appied to exec. eg. reations (Romer&Rosentha 978, Kiewiet&McCubbins 988) Add propensity 0 π to take positions Nature: π a = x a = x 0 e x 0 a e = x a e = x 0 x v a v = x 0 a v = x x 0 x π =
Dua motivation standard campaign (exicographic) trade-off posturing-ony 0 π τ u i (ω a) = ( π) PoicyGain i (ω) + π Position i (a) = ( π) (Poicy i (ω) Poicy i (x 0)) + π Position i (a) if π = 0 : u i (ω) = PoicyGain i (ω) Sma-π constraint removes confict between PoicyGain and Position
Dua motivation standard campaign (exicographic) trade-off posturing-ony 0 π τ u i (ω a) = ( π) PoicyGain i (ω) + π Position i (a) = ( π) (Poicy i (ω) Poicy i (x 0)) + π Position i (a) if π = 0 : u i (ω) = PoicyGain i (ω) Sma-π constraint removes confict between PoicyGain and Position
Equiibrium of the stunts game x = y (x) = if x 0 < or min(e 0, v 0 ) or { < x 0 < min(e 0, v 0 ) & 0 < π < τ} or τ < π min(e 0, v 0 ) + ɛ if < min(e 0, v 0 ) x 0 & 0 π < τ x 0 if < x 0 < min(e 0, v 0 ) & π = 0 accept if x v & π = 0 or x e & π 0 veto otherwise z (x) = τ (x) = { override if x v sustain otherwise 0 if e v & { < x 0 v or 2v x 0 2e } or x 0 e v 2 e x 0 ɛ if e v & e x x 0 0 2e 2 v x 0 ɛ if < v < e & v x x 0 0 2v if v e or x 0 or 2e x 0.
When to expect stunts π = 0 (standard) sma-π (campaign) x = ω x a hopeess proposa ω 0 x 0 e v 0 x 0 e v gains from trade x = ω 0 e v x 0 x = ω 0 e v x 0 A tiny drop of position-taking eaves setter mode s outcomes unchanged, but accounts for vetoes and overrides
Letting x 0 vary x ω x 0 e v x = ω = τ = x 0 0 e 0 + ɛ e 0 + ɛ 2 e x 0 ɛ x 0 z 9 z 0 z z 2 x 0 0 e v (2e )
Letting x 0 vary x = ω e 0 e v x 0 x = ω = τ = x 0 0 e 0 + ɛ e 0 + ɛ 2 e x 0 ɛ x 0 z 9 z 0 z z 2 x 0 0 e v (2e )
Comparative statics I: v e x = ω = τ = z z2 z3 sustained veto x0 (assemby stunt) 0 v e (2e ) x = ω = τ = x0 0 v0 + ɛ v0 + ɛ 2 v x0 ɛ x0 II: < v < e z4 z5 z6 z7 z8 override x0 0 v e (2v ) (2e ) (executive stunt) x = ω = τ = x0 0 e0 + ɛ e0 + ɛ 2 e x0 ɛ x0 III: e v z9 z0 z z2 x0 0 e v (2e ) (executive no veto stunt) If, e = f (div.gov) and, v = f (maj.above.override) testabe hypotheses foow
Comparative statics (sma-π and e) Veto probabiity/incidence: δpr.veto δ Override probabiity/incidence: δpr.ovr δ < 0; δpr.veto δe 0; δpr.ovr δe (Inequaities reverse when > e). > 0; and δpr.veto δv 0; and δpr.ovr δv 0. 0.
Veto institutions in state governments Pocket veto No pocket veto Item veto CO GA FL ID PA IA SD TX MI MNNM NY ND MD UT NE e WI WV a OH KY WY TN AK b CA NJ AL WV b CT DE OR HI IL b SC KS AK a VA LA MA MS MOMT WA IL a OK a AZ OK b AR no veto NC ce IN No item veto VT NH ME RI Q = 2 Q = 3 5 Q = 2 3 NC d NV Q = 3 4
Comparative data needed to test Share required to override a veto A Status / 2 3 / 5 2 / 3 sessions Div. Govt. above override 32 6 5 8 Div. Govt. beow override 7 34 28 Div. Assemby 5 38 9 9 Unif. Govt. beow override 5 23 9 Unif. Govt. above override 63 34 9 26 Tota 00 00 00 00 N 70 28,067,365 Institution party system interaction among observed sessions
Veto incidence in states egisative sessions 979 99 456 Frequency 0 0 20 30 40 50 60 70 80 90 0 50 00 50 200 250 300 350 400 450 Veto count mean = 6, sd = 40 vetoes
Four modes of veto counts in state egisative sessions MODEL MODEL 2 MODEL 3 MODEL 4 parameter parameter parameter parameter estimate a p- estimate a p- estimate a p- estimate a p- Variabe (robust std. error vaue c (robust std. error vaue c (robust std. error vaue c (robust std. error vaue c in parentheses) b in parentheses) b in parentheses) b in parentheses) b Constant 2.79 (.28) <.00 3.7 (.49) <.00 3.38 (.46) <.00 3.53 (.22) <.00 PainDG.6 (.3) <.00.54 (.2) <.00.53 (.) <.00.75 (.08) <.00 SuperDG.3 (.3).02.36 (.3).0.53 (.3) <.00.95 (.2) <.00 DividedAssemby.08 (.).5.05 (.).69.02 (.).89.9 (.0).05 n(eect.proximity).08 (.03).02.07 (.03).03.03 (.03).29.03 (.02). n(bispassed).00 (.03) <.00.97 (.06) <.00.97 (.05) <.00.92 (.03) <.00 ItemVeto.72 (.5) <.00.44 (.5).003 PocketVeto.23 (.09).02.5 (.08).08 SpeciaSession.2 (.25).62.06 (.23).8 ProfessionaAssemby.58 (.09) <.00 EconomyGrew. (.).30 State fixed effects (not reported) Wad test of ni paramet.:,08 <.00,084 <.00,24 <.00 4,797 <.00 LR test that α = 0: 2.0 0 4 <.00.9 0 4 <.00.5 0 4 <.00 5.5 0 3 <.00 Log ikeihood = 3,723 3,699 3,670 3,347 N =,365,365,365,365 Notes: (a) Negative binomia method of estimation. (b) Huber 967; White 980. (c) Two-taied tests.
MCMC repica of mode 2 Coefficient for: super.dg pain.dg div.assemby super.ug n(eection.proximity) Bayesian credibe intervas.5 0.5
Mode, hypotheses, and test summary n(veto.count j ) = β 0 + β super.dg j + β 2 pain.dg j + β 3 div.assemby j +β 4 super.ug j + β 5 n(eection.proximity j ) +... + error j Test Hypothesis Coef. Prediction resut eve a Divided government surge β 2 + + <.00 β + +.00 Supermajority thrust β 4 + +.067 b β & β 4 + +.003 Size and status b β β 4 + +.02 Eectora puse β 5 + +.00 (a) One-taied test. (b) Wad test.
Mode, hypotheses, and test summary n(veto.count j ) = β 0 + β super.dg j + β 2 pain.dg j + β 3 div.assemby j +β 4 super.ug j + β 5 n(eection.proximity j ) +... + error j Test Uncertainty Hypothesis Coef. Prediction resut eve a prediction Divided government surge β 2 + + <.00 + β + +.00 Supermajority thrust β 4 + +.067 b β & β 4 + +.003 Size and status b β β 4 + +.02? Eectora puse β 5 + +.00 (a) One-taied test. (b) Wad test.
Govt status Expected vetoes per 00 bis passed 4 5 6 7 8 9 0 2 3 Pain unified govt. Div. Assemby Pain divided govt. Super divided govt. Super unified govt. Years to next eection Expected vetoes per 00 bis 4 3 2 0 0 2 3 4 5 6 7 8 9 0 2 Pain divided govt. Pain unified govt.
Summing up Next steps: TSCS: Beck&Katz (995) extends to negbin regression? 2 Go hierarchica, mode overdispersion at state-eve 3 Try Richman s (20) status quo direct test Food for thought (forum?): 4 Pubication: EITM too much for one paper? 5 Fence-sitting? Comments/critiques very wecome Thank you!
Summing up Next steps: TSCS: Beck&Katz (995) extends to negbin regression? 2 Go hierarchica, mode overdispersion at state-eve 3 Try Richman s (20) status quo direct test Food for thought (forum?): 4 Pubication: EITM too much for one paper? 5 Fence-sitting? Comments/critiques very wecome Thank you!