Who Blinks First? Legislative Patience and Bargaining with ...jhp2121/publications/WhoBlinksFirst.pdfLegislative Patience and Bargaining with Governors When legislators and governors

55Who Blinks First

LEGISLATIVE STUDIES QUARTERLY XXXIV 1 February 2009 55

THAD KOUSSERUniversity of California San Diego

JUSTIN H PHILLIPSColumbia University

Who Blinks FirstLegislative Patience andBargaining with Governors

When legislators and governors clash over the size of American state govern-ment what strategic factors determine who wins Efforts to address this questionhave traditionally relied upon setter models borrowed from the congressional literatureand have predicted legislative dominance We offer an alternative simplification ofstate budget negotiations that follows the ldquostaring matchrdquo logic captured by divide-the-dollar games Our model predicts that governors will often be powerful but thatprofessional legislatures can stand up to the executives when long legislative sessionsgive them the patience to endure a protracted battle over the size of the budget Inthis article we present our analysis of an original dataset comprising gubernatorialbudget proposals and legislative enactments in the states from 1989 through 2004The results indicate strong empirical support for our predictions

Who is more influentialmdashlegislators or governorsmdashwhen theybargain over the size of American state budgets What institutionalfeatures and strategic contexts help to determine each grouprsquos level ofsuccess In any system of separated powers understanding thebargaining process between the legislative and executive branches iscrucial to predicting policy outcomes and uncovering the determinantsof political power In this article we explore legislative-executiveconflict across the American states using a bargaining model that differsfrom the models utilized in much of the existing literature A new datasource on gubernatorial budget proposals and legislative enactmentsprovides the testing ground for our modelrsquos empirical implications

Efforts to assess the budgetary influence of legislators and chiefexecutives have traditionally relied upon setter or spatial models ofpolicymaking In these models the outcome of interbranch bargainingis a function of the various playersrsquo preferences the order of interac-tions and the location of status quo policy (Romer and Rosenthal 1978)

56 Thad Kousser and Justin H Phillips

Typically the legislature is treated as a monopoly proposer submittingldquotake it or leave itrdquo offers to an executive who possesses an absoluteveto The executive is then forced to choose between the appropriationsfigures contained in the bill and the reversionary or status quo pointThis reversion is almost always assumed to be the previous yearrsquosspending plan maintained in the absence of legislative-executive agree-ment on a new budget through a continuing resolution

In spatial models the legislaturersquos proposal power combined withits ability to credibly threaten to keep expenditures at the status quolevel gives the legislature substantially greater influence overbudgetary outcomes than the executive holds For instance using aspatial model of presidential-congressional bargaining Kiewiet andMcCubbins (1988) have shown that when the president prefers smallerexpenditures than Congress proposesmdashthe circumstance most favor-able to the presidentmdashthe president exerts only a limited influenceover budgetary outcomes When the president prefers a higher level ofexpenditures the president has no influence at all Kiewiet andMcCubbinsrsquos insights have received additional support from a subse-quent investigation by McCarty and Poole (1995)

In the study of American states applications of setter modelsalso predict legislative dominance In their influential analyses of statebudgeting under divided government Alt and Lowry (1994 2000)amended the spatial model developed by Kiewiet and McCubbins toaccount for the balanced-budget requirements that exist in most statesIn Alt and Lowryrsquos model the legislature and governor must reachagreement on fiscal balance (whether there is a surplus deficit orbalanced budget) in addition to fiscal scale Alt and Lowry also addedan assumption backed by Lowry Alt and Ferreersquos (1998) empiricalwork that fiscal imbalance results in significant electoral losses forthe governorrsquos copartisans in the legislature

Alt and Lowryrsquos model like the Kiewiet and McCubbins modelsuggests executive weakness When there is interbranch disagreementover the size of the budget the legislature can use its monopoly proposalpower to threaten the governor with fiscal imbalance by passing acontinuing resolution rather than a new budget Since in this modeldeficits or surpluses put the governorrsquos copartisans in the legislatureat risk the governor will be forced to make significant concessions tothe legislature on fiscal scale in return for a balanced budget Afterreviewing the empirical predictions of their model under different fiscalcontexts and configurations of party control Alt and Lowry concludedthat state legislatures are even stronger than Kiewiet and McCubbins(1988) predicted Congress to be Although governors can achieve some

57Who Blinks First

of their fiscal goals when members of their party control one or bothhouses of the state legislature chief executives must make severe con-cessions when they bargain with a legislature fully controlled by theopposition party When each party controls one branch according toAlt and Lowry (2000 1043) ldquoIn no case does the governor achieve asignificant shift in the budget target in the direction of her ideal pointrdquo

While spatial models and their progeny have unquestionablyprovided important insights into legislative-executive bargaining webelieve that these models are not the most appropriate simplificationof budgeting negotiations in most American states Their portrayal ofgubernatorial weakness contradicts much of the existing scholarshipin the state politics literature Case studies (Bernick and Wiggins 1991Gross 1991) surveys of political insiders (Abney and Lauth 1987Carey et al 2003 Francis 1989) and other qualitative works (Beyle2004 Rosenthal 1990 1998 2004) all point to the extraordinary powerof governors many even refer to the governor as the ldquochief legislatorrdquoAccording to these analyses governors can and often do dominatethe legislature with respect to the eternal question of how much to taxand spend

Additionally the conclusion that the legislature can force thegovernor to accept an unfavorable deal largely depends on the assump-tion that the reversion point in the absence of a budget agreement isthe status quo preserved through a continuing resolution Continuingresolutions although frequent in federal budgeting (Fenno 1966Meyers 1997 Patashnik 1999) are not common or important consid-erations in state budget negotiations Only nine states permit someform of continuing resolution (Grooters and Eckl 1998) and even thesemeasures are labeled ldquominibudgetsrdquo (Connecticut) ldquointerim budgetsrdquo(New York) or ldquostopgap fundingrdquo (Pennsylvania) None can becomepermanent and the players in budget negotiations do not hope or fearthat they will avoid crafting a new budget

We would argue that a late budget with all of the political andprivate costs that it entails is the relevant reversion that drivesinterbranch negotiations In most states a delayed budget triggers anautomatic shutdown of the government (Grooters and Eckl 1998) Inall states it generates unfavorable press and usually a special legislativesession Public polls conducted in California1 New York2 and NewMexico3 have all demonstrated that a late budget cuts deeply into theapproval ratings of both branches The possibility of voter disapprovalevens the field on which the budget bargaining game is played Neitherbranch likes a delayed budget agreement or a government shutdownso both sides face incentives to deal The legislaturersquos proposal power


erodes when legislators cannot fall back on an acceptable status quoThis legislative limitation should make governors more powerful inthe budgetary process than spatial models predict a dynamic suggestingthat an alternative model should be sought for describing state budgetmaking

In this article we offer an alternative simplification and discussour tests of its main implications Our theory is based on formal modelsdevised by Rubinstein (1982 1985) and Osborne and Rubinstein (1990)and applied to state budget bargaining by Kousser (2005) Our modeltreats the outcome of interbranch bargaining as a function of the insti-tutional capacities and constraints of the legislature We view budgetbargaining as a ldquostaring matchrdquo in which the political and personalcosts of a delayed budget swamp the influences of proposal power andstatus quo policies Because the governor and legislature face sharedcosts of delay they both have incentive to reach an agreement quicklyNegotiations are carried out informally behind closed doors ratherthan in a sequence of bills sent to the governorrsquos desk In the staring-match dynamic that this negotiation creates the identity of the ldquowinnerrdquodepends on relative levels of patience or endurance Governors canprevail in this game if they are willing to endure longer budget nego-tiations than their legislative opponents can stand

In our application of the divide-the-dollar game governors arepatient bargainers but legislative patience is treated as a function ofprofessionalization The governorship in all states is a full-time andwell-paid job governors can afford to engage in long and protractednegotiations over the budget State legislatures on the other hand varywidely in session lengths Legislators receive sizable salaries to meetnearly year-round in states such as California New York Illinois Ohioand Massachusetts but they meet as briefly as two or three months ayear in New Mexico Georgia Utah and Kentucky and earn only smallsalaries or per diems Legislators in these less-professionalizedchambers usually hold second jobs to which they must return soonafter the legislative session These individuals pay high opportunitycosts if their governor vetoes their budget and calls them in to a specialsession These costs make ldquocitizenrdquo legislators less patient relative tothe governor and their counterparts in more-professionalized legisla-tures and give the governor a bargaining advantage Our staring-matchmodel predicts that the governor will be more successful whenbargaining with citizen as opposed to highly professionalized legis-latures And since relatively few state legislatures are highlyprofessionalized governors should generally be quite powerful in thebudgetary arena

59Who Blinks First

Clearly we are not the first to argue that full-time legislaturesexert greater influence over budgetary matters than their part-timecounterparts but our treatment of professionalization differs signifi-cantly from the models in much of the existing literature Tradition-ally professionalized legislaturesmdashhouses with longer sessions highersalaries and plentiful staff support (King 2000 Squire 1992)mdashareconsidered more powerful because they possess an increased intelli-gence capacity (Rosenthal 1990) These legislatures usually have alarge staff dedicated exclusively to fiscal policy a revenue-estimatingcapability independent of the executive branch and a sizeablecontingent of experienced legislators These features are believed toreduce the governorrsquos traditional informational advantages and enhancelegislative independence and assertiveness (National Conference ofState Legislatures 2005) While professionalization may indeed havethese effects we argue that its real advantage is that long sessionsmake legislators willing to endure extended interbranch negotiationsover the size of the budget

To test the predictions generated by our abstraction of thebudgeting process we estimated an econometric model of the out-comes of interbranch bargaining over the size of the state budget Weused an original dataset of annual gubernatorial budget proposals andthe corresponding legislative enactments culled from various issuesof the National Association of State Budget Officersrsquo Fiscal Survey ofStates We examined data for all states over a 16-year period fiscalyears 1989 through 2004

Our analysis revealed striking evidence of gubernatorial strengthin budgetary negotiations Across all types of states and legislaturesour econometric estimations show that the chief executiversquos proposedbudget has a positive and statistically significant effect on the budgetthat is ultimately passed and signed into law Most important we foundgubernatorial influence to indeed be inversely related to legislativeprofessionalization Among states with citizen houses there is nearlya one-to-one relationship between the size of the gubernatorial proposaland the size of the enacted budget In states with professional legislativebodies the magnitude of gubernatorial influence falls by approximatelyhalf These results are consistent with the expectations of our staring-match model and provide systematic empirical evidence that thissimplification of budget bargaining may be more appropriate for thestate context than the more traditionally utilized setter or spatial models

In the next section we present our staring-match model of statebudget bargaining in greater detail We discuss the logic of the gameits assumptions and its predictions Next we present our estimation


of an econometric model of the outcomes of legislative-executivebargaining and interpret the results We break down the componentsof professionalism to show that professional legislatures are generallymore powerful than citizen houses and it is longer sessions ratherthan higher salaries or more staff support that grants them this powerWe then consider the potential endogeneity of gubernatorial budgetrequests We conclude by exploring the implications of our analysisfor the study of state politics

Legislative and Gubernatorial Influence on State Budgeting

A Staring-Match Model of the Appropriations Process

To analyze the outcomes of legislative-executive bargaining overthe size of the budget we applied the framework of the divide-the-dollar games developed by Rubinstein (1982 1985) and Osborne andRubinstein (1990) Our application of their games treats bargainingbetween a governor and state legislature as a staring match ldquoblinkingrdquomeans signing or passing a proposal that closely reflects the demandsof onersquos opponent Hereafter we refer to this treatment as ldquothe staring-match modelrdquo The winner is determined largely by the relative patiencelevels of the players which we argue are functions of their institu-tional characteristics

The game we describe here like its spatial counterparts is highlystylized and abstract lacking the detailed discussion of the appropria-tions process contained in many descriptive analyses of state budgeting(cf Garand and Baudoin 2004 National Association of State BudgetOfficers 2002 and Rosenthal 2004) Yet this abstraction is useful forconveying the logic of our argument in a simple direct mannerFurthermore like the gamersquos basic intuition many of the assumptionsmade in the game conform nicely to budget bargaining at the statelevel and are consistent with observations made by qualitative studiesand in interviews with legislative staff A more-detailed discussion ofthe assumptions necessary to apply the staring-match model to statebudget negotiations along with proofs of the propositions we presentcan be found in Kousserrsquos work (2005 ch 6)

Since this model has been used less frequently than spatial modelswe will review its assumptions and notation There are two players agovernor (PG) and a legislature (PL) each behaving as if it were aunitary actor Although this assumption ignores important intrabranchdivisions each branch has rules for aggregating internal preferencesinto a final position justifying the common use of this assumption in

61Who Blinks First

models of interbranch bargaining (Alt and Lowry 2000 Cameron 2000Kiewiet and McCubbins 1988) When both the legislature and thegovernorrsquos office are controlled by one political party their disagree-ments may be fewer than under divided government But because thereis still likely to be interbranch conflict whether government is unifiedor divided4 we assumed that the branches must agree on how to ldquodividethe dollarrdquo of the budget The branch winning the biggest share of thedollar (in the game) exerts the most control over the size of the statebudget (in our application of the game) The division of the dollar isrepresented as an offer of (XL XG) and XL can fall anywhere in theinterval [0 1] Rounds of play are numbered as T = 0 1 2

In the most natural application the game begins with the legisla-ture proposing how to divide the budgetrsquos figurative dollar (Thegovernor could also begin these informal negotiations and as we laterdemonstrate the logic of the game would be the same and the divisionof the dollar would remain largely unchanged) Faced with this offerof a budget with a given size the governor either accepts and signs itor sends the game into its next stage The governor begins the secondstage with a counteroffer5 but even if the legislature immediatelyaccepts it the agreement has been delayed one round and both sidesreceive a payoff that is discounted according to their patience levelsThe discount factor is conventionally denoted by δ

Rounds of alternating offers continue until one player acceptsthe otherrsquos proposal For every round that a bargain is delayed theutility a player receives from her or his portion of the dollar is equal tothat portion multiplied by δ Assuming that this discount factor remainsconstant from round to round we would designate the value of anagreement in round t to PL at the beginning of the game as XLδt Whena playerrsquos patience is set at δ = 09 the player will be indifferent betweenreceiving 45 cents in one round and getting 50 cents in the next because50 cents deflated by 09 gives the player 45 cents of utility

We employed this deflation because continuing resolutions arerare in states and a delayed budget deal is costly to both branches6

When the players fail to adopt a state budget on time the governor andlegislaturersquos public images are harmed Even when the delay does notrun afoul of constitutional requirements the failure to pass a new budgetis politically infeasible because it denies the legislature and governorthe chance to create new programs and alter the composition ofspending In either case the status quo is a nonstarter and the rever-sion point that dictates the playersrsquo incentives is a delayed budgetEach player is thus willing to give up some of the dollar to reach anagreement early Failure to reach any agreement is of course the worst


possible outcome for both players giving each zero utility Thesefeatures characterize Rubinsteinrsquos basic bargaining game

Proposition 1 In a game satisfying all of these assumptions and whereboth players face the same discount factor δ there existsa unique subgame perfect equilibrium7 PL will alwayspropose the division (XL XG) and accept an offer onlyif it is better than or equal to YL Whenever it is PGrsquosturn to make an offer PG will propose (YL YG) andalways accept an offer that matches or beats XG Inequilibrium PL proposes (XL XG) in round t = 0 andPG accepts

1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +

The proof of this proposition is outlined by Osborne andRubinstein (1990 45) and traced out for the state politics applicationby Kousser (2005 233ndash37) Proposition 1rsquos implication for state politicsis that governors will not face a severe bargaining disadvantage becausethey lack formal proposal power In contrast with setter models inwhich gubernatorial power depends on a governorrsquos spatial prefer-ences relative to the legislaturersquos and the level of fiscal imbalance in astate8 in our version of Rubinsteinrsquos model governors can receivesome of what they want no matter which direction they wish to movethe size of government and no matter who begins the bargainingRegardless of which branch makes the first offer power over the budgetwill be divided quite equitably Both branches bargain in the shadowof a late budget and the political penalties it can bring Both are eagerto avoid delay and whichever branch can move first makes a fair offerthat it knows the other branch can afford to accept In the most straight-forward application of the staring-match model this offer comes whenthe legislature passes a budget bill But even if the governorrsquos publicbudget proposal which often receives much media attention and setsthe agenda for later negotiations is thought of as the first offer thetheoretical prediction for the division of the dollar does not changeradically As the payoffs demonstrate the ldquofirst moverrdquo advantage thataccrues to the branch making the initial offer is small when both playersare relatively patient and not tremendously large even when they arein a hurry to pass a budget When both players discount payoffs thatare delayed one round by a factor of 09 the first mover receives 526

63Who Blinks First

cents of the dollar and the other branch gets 474 cents Even when thediscount factor equals 07 the division of the dollar is still a somewhat-equitable 588 cents to 412 cents Regardless of which branch isthought of as making the first offer Proposition 1 leads to the followingempirical implication

Hypothesis 1 Governors will exert a powerful influence over the sizeof state budgets

Varying Legislative Patience

The basic model assumes that governors and legislators possessthe same patience level but this assumption may not always hold trueIn particular the members of a citizen legislature should be signifi-cantly less willing to engage in protracted budgetary disputes with thegovernor than their more-professionalized counterparts The rationalehere is that in addition to political costs that both branches pay whenthere is budgetary gridlock lawmakers serving in a less-professionalized legislature face private costs of delay These costswill decrease the legislaturersquos patience and advantage the governor

There are of course several relatively professionalized statelegislatures These chambers resemble the US House of Representa-tives they meet in lengthy sessions their members are well paid andthe legislature employs numerous nonelected staff In states such asCalifornia New York and Michigan there are few if any restrictionson the number of days the legislature may meet as a result lawmakersare in session much of the year Furthermore legislators serving inthese chambers receive annual salaries in excess of $75000 as wellas generous per diems (Council of State Governments 2005) Theselawmakers can therefore treat legislative service as a career and do notneed second jobs even as the session length makes holding a secondjob close to impossible

Most state legislatures however are notably less professionalizedIn these chambers the number of days that legislators are allowed to meetis often constitutionally restricted On average regular sessions are limitedto approximately 90 calendar days per year in extreme cases sessions areconstrained to no more than 60 or 90 days biennially Compensation forservice in most chambers is also low or nonexistent To support them-selves and their families legislators in citizen chambers usually holdsecond jobs to which they must return soon after the legislative session

As a result members of a part-time body face high opportunitycosts when they fail to reach agreement on a budget with the governor


In the absence of such an agreement legislators are usually forcedinto what may be a time-consuming special session and are preventedfrom pursuing their private careers or personal lives The prospect ofleaving their day jobs to resolve budget conflict should make membersimpatient On the other hand governors pay much lower private costswhen they veto a bill at the end of a session They may force a specialsession stalling whatever private travel or governing plans they mighthave9 but because all governors are paid well to do their job full-time10 they can endure round after round of negotiations Participantsin gubernatorial negotiations with the less-professional legislaturespoint out the paramount importance of this dynamic A senior advisorto Oregon Governor John Kitzhaber explained that ldquoAs session goeson the wait is in our favorrdquo11 In New Mexico a special session calledby Governor Gary Johnson to resolve the 2000 budget standoff ledlegislators to grouse take political heat and ultimately accede to manyof the governorrsquos demands12 We therefore expected professionalchambers to be able to match the governorrsquos endurance whereas part-time bodies would be vulnerable to threats of a veto and extendednegotiations One piece of descriptive evidence consistent with ourexpectation is that budget standoffs although rare occur primarily inprofessional full-time legislatures13

This potential asymmetry in the patience levels that the branchespossess can be formalized in an extension of the basic Rubinstein modelin which the two players have different discount rates When a citizenlegislature negotiates δL will be lower than δG If the governorrsquosadvantage in patience is large then it will swamp the advantage thatthe legislature holds from moving first Proposition 2 is simply a less-general form of Proposition 1 in which discount rates are allowed to vary

Proposition 2 In a game similar to the basic game but where playersface individual discount factors δL and δG there existsa unique subgame perfect equilibrium PL will alwayspropose the division (XL XG) and accept an offer onlyif it is better than or equal to YL Whenever it is PGrsquosturn to make an offer PG will propose (YL YG) andalways accept an offer that matches or beats XG Inequilibrium PL proposes (XL XG) in round t = 0 andPG accepts

1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ

δ δ δ δminus minus

=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First

The exact payoffs that the legislature receives at different levelsof legislative patience are displayed graphically in Figure 1 One cansee how steeply the legislaturersquos share of the dollar drops as its membersbecome less and less patient assuming that the governor has a discountrate of δ = 09 Although we do not investigate variations in thegovernorrsquos patience level here it likely changes with such factors asapproval ratings the timing of the next election and the governorrsquospolitical ambitions Investigations of these variables may providefruitful further tests of the implications of the model

The solid line maps the payoffs when the legislature begins thebargaining and the dotted line shows the results if the governor movesfirst The gap between these linesmdashthe first-mover advantagemdashisrelatively narrow What really determines who will control the budgetis the legislaturersquos patience A professional legislature that can crediblythreaten to wait the governor out in a special session will win gaining525 cents if it is equally as patient as the governor If the privatedemands on members of a citizen legislature reduce their discountrate to 05 then they will get a mere 182 cents even when they movefirst Hypothesis 2 states the specific testable implication of thistheoretical finding

0

01

02

03

04

05

06

010203040506070809

Legis la ture Makes Initia l P ro po s a l

Go verno r Makes Initia l P ro po s al

FIGURE 1Payoffs for Legislatures with Different Levels of Patience

Legislaturersquos Share of the D

ollar

Legislaturersquos Patience Level

Legislature makes initial proposal

Governor makes initial proposal


Hypothesis 2 The influence that governors exert over the size of theirstate budgets will grow as the level of legislativeprofessionalism in their states declines

Before discussing the tests our hypotheses it is worth notingthat the centrality of patience or discount rates in our model is one ofthe features that most clearly distinguishes it from existing analysesSpatial approaches to legislative-executive bargaining at both thenational and state levels rarely consider the potential effects that shiftsin discount rates may have on outcomes (Alt and Lowry 1994 2000Kiewiet and McCubbins 1988 McCarty and Poole 1995 but see Banksand Duggan 2006) Even when the patience levels of the players areallowed to vary spatial models predict no effect Primo (2002) forinstance examined how some of these dynamics might affect Romerand Rosenthalrsquos (1978) model He found that even when spatial modelsare extended to multiple stages of bargaining discount rates do notfactor into the equilibrium Primorsquos results suggest that impatient citizenlegislatures should not face a bargaining disadvantage becauseldquoimpatience and time preferences may not be key features of politicalbargainingrdquo (421)

Testing Predictions of the Staring-Match Model

We tested our hypotheses by systematically examining therelationship between the size of the governorrsquos proposed increase intotal per capita state expenditures (measured as a percentage of theprevious yearrsquos budget) and the size of enacted spending changemdashthat is the change contained in the budget ultimately adopted by thelegislature and signed into law (again measured as a percentage of theprior yearrsquos budget)14 Unlike most of the existing literature our studygauges gubernatorial power by directly measuring the policy prefer-ences of governors rather than assuming that their party affiliationstell us exactly what they want

Prior studies of variation in state policy outputs that reliedexclusively on measures of party control as a proxy for gubernatorialand legislative preferences gained their causal traction from theassumption that Democrats always and everywhere want governmentto expand or that the magnitude of policy disagreements between thetwo major parties is constant across states (Alt and Lowry 2000 Dye1966 1984 Garand 1988 Hofferbert 1966 Kousser 2002 Smith 1997Winters 1976 but see McAtee Yackee and Lowery 2003 for a relax-ation of the latter assumption) None of these studies has found that

67Who Blinks First

governors can move policy in their preferred direction in a statisti-cally significant manner

Instead of using party affiliations as a proxy we chose to measureexecutive preferences directly by recording the percentage change inper capita expenditures that the governor proposes at the beginning ofthe year Our choice is similar to Clarkrsquos (1998) strategy of gatheringgubernatorial recommendations for agency budgets from 20 states andCanes-Wronersquos (2001) use of presidential budget proposals to analyzefederal bargaining A dataset that systematically gathered legislativebudget proposals would be ideal but no such dataset exists Still usinggubernatorial proposals for spending changes as an independentvariable and legislative enactments as our dependent variable we cansee how far different types of legislatures shift policy from what thegovernor wants This is the same empirical strategy that Kiewiet andMcCubbins (1988) employed in their influential study of presidential-congressional bargaining

As mentioned previously we collected our dataset of gubernato-rial budget proposals and enacted state budgets from various issues ofThe Fiscal Survey of States a publication of the National Associationof State Budget Officers (NASBO) Each year NASBO conducts twosurveys of state budget officials to identify trends and changes in statefiscal policy The spring survey gathers information concerning thegovernorrsquos proposed general-fund budget and the autumn survey iden-tifies details of the enacted budget (usually Table A-3 in both reports)Our analysis includes data for all states over a 16-year period fiscalyears 1989 through 2004 Data prior to fiscal year 1988 are unavail-able Since NASBO consistently reports data in current dollars weconverted the values for each year into 2000 dollars using the ConsumerPrice Index for all urban consumers (CPI-U)

Evaluating the predictive power of the staring-match model alsorequired us to identify an appropriate measure of the professionalizationof state legislatures A venerable literature in state politics has demon-strated the importance of variation in legislative salaries sessionlengths staff support and other resources (Berry Berkman andSchneiderman 2000 Fiorina 1994 Hamm and Moncrief 2004 Karnigand Sigelman 1975 Roeder 1979 Squire and Hamm 2005 Thompson1986) Many researchers follow Squire (1992 72) or King (2000 329)combining the components of legislative professionalism into a singleindex To be consistent with the existing literature we began with thisapproach We employed the widely used trichotomous categorizationdeveloped by the National Conference of State Legislatures (NCSL)as well as Squirersquos (1992) continuous index Both measures are based


upon the length of time that legislators spend in session the amount oftheir total compensation and the number of legislative staff membersBecause levels of state professionalism changed dramatically from the1960s through the 1980s but were relatively stable during our periodof study both of our measures are constant in each state across timeWe recast the NCSLrsquos red white and blue categories as ldquoprofessionalrdquoldquosemiprofessionalrdquo and ldquocitizenrdquo As a report by the NCSL (2005)details lawmakers in professionalized bodies dedicate much more timeto legislative service earn about four times as much and work witheight times as many staff members as their counterparts in citizenlegislatures

We began our analysis by examining for each type of legislaturethe bivariate relationship between the governorrsquos proposed budget andthe enacted budget in all states with three exclusions We excludedAlaska and Wyoming because they both rely heavily upon severancetaxes on natural resources The use of severance taxes results in fairlydramatic year-to-year variation in tax revenues and thus expendituresThese variations are driven largely by the global commodities marketas opposed to the budgetary choices of legislators and governors(Matsusaka 2004) We also excluded Nebraska because of its non-partisan legislature

Our preliminary results reported in Table 1 are entirely consistentwith the hypotheses derived from the staring-match model Across allthree categories of legislatures the coefficient on the variable thatmeasures the size of the gubernatorial budget proposal is positive andstatistically significant at the 99 level This result provides supportfor our contention that governors are consistently powerful in thebudgetary arena (Hypothesis 1) As predicted in Hypothesis 2 themagnitude of the effect is inversely related to legislativeprofessionalization Among citizen bodies this coefficient is 086Substantively this finding indicates that when a governor negotiatingwith a citizen legislature proposes increasing the budget by 1 thefinal enacted budget should increase by 086 A proposed cut inspendingmdashalthough empirically much rarermdashwould bring ananalogous decrease in the size of the budget When a governornegotiates with a more-professional legislature however the governorrsquospower to dictate fiscal outcomes declines A proposed change of 1leads to an enacted increase of only 073 when the governor negotiateswith a semiprofessional legislature and a 046 change when thegovernor faces a professional house These differences we will latershow are statistically meaningful and still present when other factorsare held constant Furthermore among states with citizen legislatures

69Who Blinks First

TABLE 1Governorrsquos Influence by Type of Legislature

Citizen Semiprofessional Professional AllVariables Legislatures Legislatures Legislatures Legislatures

Governorrsquos Proposal 86 73 46 90(05) (08) (07) (07)

Squirersquos Index of __ __ __ 01 Professionalism (1992) (168)

Governorrsquos Proposal times Squirersquos __ __ __ ndash83 Index of Professionalism (020)

Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25

Note Table entries are regression coefficients In the ldquoAll Legislaturesrdquo model the estimatedcoefficient for a governorrsquos proposal represents the effect that this variable would have in ahypothetical legislature with a ldquo0rdquo level of professionalism measured on Squirersquos scalep lt 01

the governorrsquos budgetary proposal alone explains almost half of thevariation in outcomes but among states with more-professionalizedlegislative bodies the governorrsquos proposal accounts for less than 20of the variation in outcomes

The last column in Table 1 combines data from all three types oflegislatures with an interaction testing the hypothesis that a governorrsquosproposal will have a smaller effect on the final budget outcome whenthe legislature is more professional according to Squirersquos (1992)continuous measure This interaction effect is strongly significant inthe expected direction To interpret it we obtained the effect of agovernorrsquos proposal when the legislature has a given level of profes-sionalism we added the product of that given level and the interactioncoefficient (ndash083) to the estimated coefficient of a governorrsquos proposalalone (090) The results indicate that a governorrsquos proposed 1increase in the size of the budget should translate into a 088 increasein spending when the governor negotiates with a legislature like Utahrsquoswhich is one standard deviation below the mean level of professionalismbut only a 062 increase when the statersquos legislative professionalismregisters one standard deviation above the mean as in Pennsylvania


Altogether these results provide preliminary evidence that governorswhile powerful are less influential in the face of legislatures that meetin long sessions pay high salaries and provide staff support

The results reported in Table 1 may of course reflect the influenceof omitted variables We addressed this problem by conducting amultiple regression analysis that included a number of potentiallyinfluential political and economic variables The first of these variablesis the partisan composition of the legislature Existing research in statepolitics has found evidence albeit weak and oftentimes conditionalthat Democratic control of the legislature leads to a larger state publicsector and larger year-to-year increases in expenditures or revenues(Alt and Lowry 1994 2000 Phillips 2005) To allow for this possibilitywe employed a continuous measure of the legislative strength of theDemocratic Party in each state calculated as the weighted percentageof Democrats serving in both the statersquos lower and upper legislativechambers recorded from appropriate editions of the Council of StateGovernmentsrsquo Book of the States Smith (1997) has identified thisapproach as the most appropriate method for capturing the partisanmakeup of state government Additionally we accounted for cross-sectional variations in the timing of state budget processes to ensurethat our measure accurately reflects the partisan composition of thelegislature at the time the budget was passed and signed into lawBecause empirical exploration indicated that the power of governorswas not contingent upon the presence of divided government15 we didnot include any measures of divided government in the modelspresented Similarly although the results of an unreported analysisshowed that governors who possess more institutional budget powersaccording to Beylersquos (2004) index exert more influence over fiscalchanges we omitted this test from our final models because the effectfell short of statistical significance and would have required three-way interactions We also omitted from the final reported modelspotential control factors such as whether a state had an annual orbiennial budget and whether or not it allowed continuing resolutionsThese factors (either entered alone or in interaction with a governorrsquosproposal) did not have large or statistically significant effects andtheir exclusion did not substantially change the estimated effects ofour key variables of interest

Previous research has also shown that economic factors areimportant determinants of state budgetary policy (Dye 1966 1984Winters 1976) We allowed for these influences by utilizing as inde-pendent variables per capita income (measured in thousands of dollars)and the state-level unemployment rate (both taken from the US Census

71Who Blinks First

Bureaursquos Statistical Abstract of the United States) To control for thepossibility that state expenditures increase during election years (thatis the presence of a political business cycle) we included a dummyvariable for years in which lawmakers must run for reelection asreported in the Book of the States Finally following Phillipsrsquos (2005)work we also included the previous yearrsquos per capita expenditures asreported in NASBO records to control for status quo fiscal policy anda lagged measure of the statersquos budget surplus or deficit (measured asa share of the total budget)

All of our econometric estimations also utilize year and statefixed effects The year fixed effects control for common shocks thataffect all states in a given year such as changes in the national orglobal economy or changes in the national political environment Thestate fixed effects capture all relevant variables that are idiosyncraticto individual states or that remain unchanged over the time period ofour analysis such as culture voter ideology and political institutionsOf particular relevance fixed effects control for other features of thebargaining environment that may enhance a governorrsquos ability to prevailin battles over the size of the state budget such as the item veto andgubernatorial impoundment powers

The first of our multiple regression results are reported in Table 2Model 1 is a baseline estimation that includes all states but does notaccount for cross-sectional variation in legislative professionalizationAs in Table 1 the coefficient on the governorrsquos proposal is positiveand statistically significant indicating that governors are powerfulactors in the budgetary arena even after one controls for a number ofpotentially confounding influences Model 2 is a direct test of oursecond hypothesis Here the governorrsquos budgetary proposal is inter-acted with two dummy variables one for the existence of a semi-professional legislature and the other for a citizen body The referencecategory in this regression is professional legislatures We did notinclude separate dummy variables for each legislative type in theequation because we used fixed effects which account for theindependent effect of professionalization16

This new estimation provides the strongest evidence yet for thestaring-match model Once again the size of the governorrsquos proposedbudget has a significant and positive effect on the size of the budgetultimately adopted by the legislature and signed into law Mostimportant the coefficients on the interaction terms are also positiveand significant at the 99 level the effect of the gubernatorial proposalon the final budget increases in a statistically meaningful fashion asthe professionalization of the legislature declines When the governor



Full Multiple Regression

Model 1 Model 2 Model 3a Model 3bVariables (Squire 1992) (Squire 2007)


Governorrsquos Proposal times __ 25 __ __ Semiprofessional Legislature (10)

Governorrsquos Proposal times __ 49 __ __ Citizen Legislature (11)

Governorrsquos Proposal times Squirersquos __ __ ndash96 ndash105 Index of Professionalism (21) (23)

Squirersquos Index of Professionalism __ __ __ 532(820)

Democrat Legislature ndash06 ndash07 ndash07 ndash08(05) (05) (05) (05)

Lagged Expenditures per Capita ndash01 ndash01 ndash01 ndash00(002) (002) (002) (00)

Legislative Election Year 138 156 158 191(77) (76) (76) (83)

Unemployment Rate ndash48 ndash56 ndash58 ndash66(36) (36) (36) (40)

Personal Income per Capita 101 102 101 63(37) (40) (37) (38)

Lagged Surplus 07 03 03 03(06) (06) (06) (07)

Constant ndash596 ndash402 ndash315 ndash715(949) (939) (939) (1031)

N 787 787 787 787

R2 30 32 32 30

Note Estimated using state and year fixed effectsp lt 01 p lt 05

73Who Blinks First

negotiates with a professional legislature an executive proposal foran additional 1 increase in state spending per capita leads to only a029 change in the legislaturersquos enacted budget all else being equalWhen the legislature is semiprofessional this marginal effect growsto 054 It rises to 078 when the governor negotiates with citizenlegislators

Tests also show that the difference between the coefficients onour two interaction terms is itself statistically significant at the 95level not only does the governor get significantly more of what she orhe wants when the chamber shifts from professional (our baselinecategory) to citizen or semiprofessional but the governor also getssignificantly more of what she or he wants when the chamber shiftsfrom professional to semiprofessional These findings are consistentwith the bivariate results shown in Table 1 as well as with the resultsof Models 3a and 3b which use Squirersquos (1992) continuous measureof legislative professionalism and Squirersquos (2007) updated dynamicmeasure17 rather than the trichotomous categorization Again thestatistically significant coefficient on the interaction between agovernorrsquos proposal and the level of professionalism suggests that chiefexecutives have less power when bargaining with full-time legislatures

Thus far we have discussed a measure of legislativeprofessionalization that aggregates the various components of thisconceptmdashsession length compensation and staffmdashinto a single indi-cator While there are strong theoretical and empirical reasons for thisaggregation the staring-match model makes a prediction about whichof these components matter In particular it suggests that session lengthis the primary factor affecting the legislaturersquos patience and thus thegovernorrsquos power By contrast our application of the staring-matchlogic does not predict that increased staffing will affect the balance ofpower between the branches because it affects a legislaturersquos informa-tional capacity rather than its patience High salaries which can freelegislators from other obligations might also affect patience butmembers of houses that regularly meet for full-time sessions shouldexhibit the highest levels of patience To examine this claim and tofurther explore the relationship between legislative structure andbargaining outcomes we replaced our summary measures with theseparate components of professionalism The first measure records alegislaturersquos level of compensation including both base salary and perdiem expenses The second measures session length in legislative daysper biennium and the third reflects the ratio of staff per legislatorBecause staff salary and session length are not perfectly collinear18

we estimated their separate effects


The results reported in Table 3 explore the effects of each ofthese components of legislative professionalism on gubernatorialpower Models 4 5 and 6 estimate the influence of each componentseparately and Model 7 tests their combined effects Whether analyzedseparately or together the story is the same session length providesthe link between professionalism and executive power with thegovernor exerting less control over the budget when a house meets forlonger and longer sessions The interaction between total session daysand a governorrsquos proposal is statistically significant in the expecteddirection but changes in the other two components do not signifi-cantly alter the effect of executive proposals To gauge the scale of theeffect of session length we calculated the effect according to the resultsin our fully specified Model 7 of shifting this variable from onestandard deviation below its mean to one deviation above its meanConsider two legislatures both of which pay the average salary in oursample (about $25000 a year) provide average staffing levels (35assistants per legislator) and exhibit similar values on all of the controlvariables If one of these legislatures meets for 66 days over a two-year period (a typical session for North Dakotarsquos legislature) thenevery extra percentage-point increase in spending proposed by thegovernor translates into a 078 increase in the enacted budget If ahouse is similar in all other respects but meets for 263 days perbiennium as Wisconsin did for 1997 and 1998 then a change of 1 inthe governorrsquos proposal yields only an estimated 059 change in thebudget that the legislature finally enacts This finding is consistentwith our conjecture that a full-time house has the patience to strengthenits bargaining position against the governor thus supporting the logicof the staring-match model

The results presented up to this point suggest that interbranchnegotiations over the size of the state budget are better conceptualizedusing a staring-match model than with a setter model Neverthelessone potential counterclaim is that the setter model may still be appro-priate for the handful of states in which continuing resolutions areallowed We do not see however why this would be true Rememberthat continuing resolutions in those states that allow them are (at best)short-term solutionsmdashnone can become permanent Furthermore theiruse does not insulate the governor and legislature from the high politicaland personal costs associated with a late budget Continuing resolutionsoperate quite differently at the national level Continuing resolutionsare used by Congress and the president on well over half of all budgetbills and they can be utilized for many months or longer (Meyers1997) In fact President Clinton and the Republican-controlled

75Who Blinks First

TABLE 3Governorrsquos Influence by Each Component

of Legislative Professionalism

Variables Model 4 Model 5 Model 6 Model 7


Salary (in $1000s) 02 __ __ 04(04) (04)

Governorrsquos Proposal times Salary ndash001 __ __ 005(002) (003)

Session Length (hundreds of days) __ ndash002 __ ndash06(53) (53)

Governorrsquos Proposal times Session Length __ ndash12 __ ndash13(02) (02)

Staff per Member __ __ ndash19 02(43) (45)

Governorrsquos Proposal times Staff per Member __ __ ndash01 ndash017(01) (014)





Personal Income per Capita (in $1000s) 104 97 103 101(38) (36) (37) (37)

Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34

Note Estimated using state and year fixed effects

p lt 01 p lt 05


Congress unable to reach a budget agreement in fiscal year 1999funded the federal government for the entire year using a series ofseven continuing resolutions (Davidson Oleszek and Lee 2007)

Still we empirically tested whether or not the availability ofcontinuing resolutions affects gubernatorial power in Model 8 To doso we reestimated Model 5 this time including an interaction betweenthe governorrsquos budgetary proposal and a dummy for whether or notcontinuing resolutions are expressly allowed The results are presentedin Table 4 We found no meaningful effect The coefficient on the newinteraction terms is negative and small in magnitude and it fails toeven approach statistical significance Because some states have nolegal provision regarding the use of continuing resolutions and no testcase (because of the lack of late budgets) we also made use of severalalternative codings Specifically we created dummy variables for stateswhere continuing resolutions may be allowed (those where continuingresolutions are expressly authorized and those where there is noprovision regarding their use) states where a late budget triggers afull government shutdown and states where a late budget triggers eithera full or partial shutdown19 These dummy variables are used in Models9 through 11 with the results also shown in Table 4 Again we foundno effect In each of these estimations the interaction between guber-natorial budget proposals and legislative session length (the mosttheoretically relevant component of legislative professionalization)remains negative and statistically significant providing additionalevidence that staring-match models are preferable to setter models forconceptualizing state budget bargaining

The Potential Endogeneity in Gubernatorial Budget Requests

It is also important to consider the extent to which state chiefexecutives have an incentive to misrepresent their preferences whenthey submit their proposed budgets and whether or not this misrepre-sentation would bias our econometric results Governors may for in-stance foresee legislative strength and adjust their budgetary proposalsaccordingly According to this logic governors facing professionallegislatures would weaken their initial offers moving closer to theirlegislaturesrsquo ideal points Governors negotiating with citizenlegislatures would have no incentive to adjust their proposals sincethey should be able prevail in budgetary negotiations given the insti-tutional weakness of citizen legislatures

We doubt however that governors frequently ldquogamerdquo theirbudgetary proposals in this manner When governors present their

77Who Blinks First

TABLE 4Governorrsquos Influence by Legislative Professionalization

and Status Quo Point



Session Length (hundreds of days) ndash02 ndash03 ndash02 ndash01(53) (53) (53) (53)

Governorrsquos Proposal times Session Length ndash10 ndash10 ndash12 ndash12(03) (02) (02) (02)

Governorrsquos Proposal times Continuing ndash14 __ __ __ Resolutions Allowed (14)

Governorrsquos Proposal times Continuing __ ndash15 __ __ Resolutions May Be Allowed (10)

Full Shutdown __ __ 002 __(10)

Full or Partial Shutdown __ __ __ 03(09)



Legislative Election Year 163 164 165 ndash164(76) (76) (76) (76)



Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33



budgets they send a signal to voters interest groups and campaigncontributors about their governing philosophy and legislative prioritiesSurely governors realize the signaling role of their actions and makeproposals accordingly Moreover the public is not likely to under-stand or appreciate complicated strategies and officials may not beable to explain them effectively Such considerations should attenuateany impulse the governor may have to game the proposed budget(Denzau Riker and Shepsle 1985)

If however governors facing highly professionalized legislaturesdo systematically move their initial budget proposals closer to theirlegislaturesrsquo ideal points then the observed effect should be a strongerrelationship between executive proposals and enacted budgets In otherwords the possibility of strategic misrepresentation should bias ourresults against finding that governors are less powerful in states withprofessionalized legislatures This result is of course the opposite ofwhat our econometric estimations actually reveal We have strongevidence that governors are least powerful in states with theselegislatures Thus we are even more confident that the insight providedby the staring-match model is correct and that initial gubernatorialbudget proposals are sincere

To further satisfy skeptical readers we empirically tested for thepossibility that gubernatorial budgetary proposals are gamed Theseresults appear in the Appendix available on the LSQ website lthttpwwwuiowaedu~lsqKousserPhillips_Appendixgt We examinedwhether or not Democratic and Republican governors systematicallyalter the size of their proposals when facing professional (that is strong)legislatures controlled by the opposition party We did not uncoverany evidence suggesting that governorsrsquo initial offers are shaped bythe strategic situations they face Thus we are confident that executivebudgetary proposals are not endogenous to statesrsquo institutionalconfigurations

Conclusion and Implications

Attempts to assess the budgetary influence of state legislatorsand governors have traditionally relied upon spatial or setter modelsof policymaking imported from studies of the US Congress In thesemodels legislators through their monopoly on proposal power andtheir ability to credibly threaten to keep expenditures at the status quolevel have substantially greater influence on budget making thangovernors wield This result contradicts numerous qualitative analysesin the state politics literature that find that governors are the chief

79Who Blinks First

legislators in the budgetary arena We proposed and tested an alternativesimplification of state budgeting modeled on the games developed byRubinstein (1982 1985) and Osborne and Rubinstein (1990) andapplied to states by Kousser (2005) Governors are quite potent in thisstaring-match model and the power of legislators declines when shortersessions or lower salaries make them impatient and willing to make adeal

Using an original dataset of gubernatorial budget proposals andlegislatively enacted state budgets we explored the modelrsquos predictionsOverall we found striking evidence of gubernatorial influence Oureconometric estimations show that across all types of legislatures thechief executiversquos proposed budget has a positive and statisticallysignificant effect on the budget that is ultimately passed and signedinto law Most important however the influence of legislators is closelylinked to levels of legislative professionalism Legislators can drive aparticularly hard bargain when they typically hold long sessions sessionlength is the component of professionalism most closely linked to thetheoretical concept of bargaining patience This empirical relation-ship may be driven by something other than variation in patience full-time legislators might acquire more policy knowledge or politicalacumen during their longer sessions or full-time work might attract adifferent type of legislator Still this finding is consistent with ourtheoretical model and shows the importance of separating the sessionlength component from the staff and salary components Just as Altand Lowryrsquos (1994 2000) work demonstrated the centrality of stateinstitutions and party control in determining fiscal outcomes ourempirical analysis reveals the power of governors and the importantmediating effect of legislative professionalism

We believe that these findings yield three more-general lessonsfor the study of bargaining between governmental branches First whenresearchers apply formal models of bargaining one size does not fitall legislatures Although setter models may capture the key dynamicsof federal budget bargaining in the US Congress where a continuingresolution is a realistic reversionary outcome these models do notappear to fit well with states that demand that a new budget be passedevery year Second while variation in legislative professionalismclearly determines legislative power it is session lengthmdashmore thansalary or staffmdashthat appears to drive this trend Isolating the theoreti-cally distinct components of legislative professionalism can yield newinsights into this variablersquos importance Finally directly measuringgovernorsrsquo preferences rather than inferring them from party affilia-tions demonstrates the significant influence that these preferences exert


over state policy Overall by closely examining the way that institu-tional contexts shape the strategies available to political actors wecan uncover links between rules political reforms and bargainingoutcomes that may have implications for broader comparative studies

Thad Kousser lttkousserucsdedugt is Associate Professor ofPolitical Science 9500 Gilman Drive 0521 SSB 369 University ofCalifornia San Diego La Jolla CA 92093-0521 Justin H Phillipsltjhp2121columbiaedugt is Assistant Professor of Political ScienceIAB 7th Floor 420 West 118th Street Columbia University New YorkNY 10027

NOTES

1 The time series of legislative and gubernatorial approval in California reportedby the Field Poll reveals how severe these penalties can be In the first two years ofGovernor Gray Davisrsquos administration 1999 and 2000 the branches reached budgetdeals before the start of the new fiscal year During Davisrsquos last two years 2002 and2003 negotiations dragged into September and August (Wilson and Ebbert 2006 276)In 1999 and 2000 the governorrsquos and the legislaturersquos approval ratings remainedessentially constant over the summer But the legislaturersquos approval ratings droppedfrom 45 to 35 from July to September of 2002 and from 31 to 19 from April toJuly of 2003 (Field Poll 2004 2) Davisrsquos already-low ratings edged downward as wellin each of those summers (Field Poll 2003 3)

2 When the 2001 budget deal in New York was delayed 84 of survey respon-dents were ldquovery concernedrdquo or ldquosomewhat concernedrdquo about the budget and 63blamed both Governor Pataki and the state legislature (Quinnipiac 2001) In 200481 of polled New Yorkers voiced concern over the statersquos late budget and 46 saidthat it made them more willing to vote out incumbents (Caruso 2004)

3 When New Mexicorsquos budget was delayed in 2000 Governor Gary Johnsonand the legislative leaders all polled poorly and ldquoNew Mexico voters faulted Johnsonand lawmakers almost equally for their failure to reach agreement during the sessionon a $3 billion budgetrdquo (ldquoVoters Unimpressed with Johnson Lawmakersrdquo AlbuquerqueJournal 19 March 2000 A1)

4 Unified government does not guarantee executive-legislative agreement overthe budget In Massachusetts for instance Democratic governor Michael Dukakisconsistently had his budget rewritten by the legislature which was overwhelminglycontrolled by his own party (Beyle 2004 Rosenthal 1990)

5 Our application of this model operates as a metaphor for the informal nego-tiations between the governor and legislative leadersmdashsuch as Californiarsquos ldquoBig Fiverdquoor Illinoisrsquos ldquoFour Topsrdquomdashthat take place behind closed doors and allow either side toinitiate negotiations or to make detailed counteroffers We believe that ignoring thesenegotiations and focusing solely on budget bills and vetoes neglects the key dynamicsthat drive interbranch bargaining

81Who Blinks First

6 By contrast Alt and Lowry (1994 813) specify the reversion point of theirmodel as follows ldquoIf the legislature fails to make a proposal or if a proposal is vetoedand the veto is sustained then we assume that the budget is set equal to its reversionlevel which is the previous yearrsquos expected levels plus any persistent effects of theunforeseen shockrdquo Indeed as we note nine states explicitly permit continuing resolu-tions that would preserve the status quo in this way But we argue even in these statespreserving the status quo is no more attractive to the legislature than it is to the governorWhatever happens when a budget is late whether it is a shutdown or a continuingresolution imposes high costs on both branches of government The payoff that bothbranches receive from the ultimate deal is eroded preventing legislators from crediblythreatening to live with the status quo if the governor rejects their offer

7 Since the Nash prediction is quite vague in this casemdashany division of thedollar can be reached in the first round in equilibrium because players can make threatsthat are not crediblemdashOsborne and Rubinstein (1990) employed Seltenrsquos (1975) notionof a subgame perfect equilibrium which requires that best responses be played atevery point in the game that begins a subgame (see Morrow 1994) Subgame perfec-tion generally refines the set of acceptable equilibrium strategies and in this casegenerates a unique prediction

8 Governors can exert influence over the size of government even in settermodels as Alt and Lowry describe ldquoDivided cases produce target levels somewherebetween those of the partiesrdquo (1994 820) and ldquoIn neither case does the legislature getall it wants as it must consider the threat of a vetordquo (2000 1042) Nevertheless thepower that governors have in a setter model is primarily negative the veto gives themthe ability to put a break on high-spending legislatures In our staring-match modelthe governorrsquos ability to make an offer to the legislature regarding the size of govern-ment gives the governor positive power to shape spending levels

9 Legislatures in 30 states have the authority to call their own special sessions(Council of State Governments 2000) but they are often forced to do so by a governorrsquosveto Although special sessions are not often called to resolve legislative-executiveconflicts the threat of a special session is not unimportant Delayed bargains are offthe equilibrium path of Rubinsteinrsquos basic model but they are weapons that do notneed to be unsheathed to be powerful

10 Even the lowest-paid governor Mainersquos chief executive earns $70000 ayear (Council of State Governments 2005)

11 Interview by Thad Kousser Salem Oregon July 8 200112 More than a month after New Mexicorsquos one-month session came to a close

without a budget deal the governor called the statersquos citizen legislators back to SantaFe for a special session One legislator opined that such meetings ldquo[C]ertainly are notspecial They are absolutely routine and in my opinion very annoyingrdquo (ldquoOnly ThingSpecial about These Sessions Are Lessonsrdquo Albuquerque Tribune 4 April 2000 C-2)In addition to exacting private costs the session cost the legislature $45000 a day torun and generated political controversy One legislator said ldquoI think it would behooveall of us to be out of here by Saturday I can just see a lot of really ugly newspaperstories if wersquore still in session on April Foolrsquosrdquo (Mark Hummels ldquoTheyrsquore BackrdquoSanta Fe New Mexican 28 March 2000 A-1) Perhaps because of their hurry thelegislators passed a budget that was a ldquopolitical home runrdquo for the governor (ldquoVetoesEnact Tax Reductionrdquo Albuquerque Journal 22 April 2000 E-3)


13 In 2007 five of the six states in which a budget standoff dragged on past thebeginning of the next fiscal yearmdashCalifornia Illinois Michigan Pennsylvania andWisconsinmdashhad professional legislatures that typically met at least 20 months in atwo-year biennium (personal communication between the authors and Arturo Perez ofthe National Conference of State Legislatures) Historically New York and Californiaboth of which have highly professionalized legislatures have been plagued by latebudgets As of fiscal year 2005 20 of the last 21 budgets in New York were adoptedwell after the legal deadline (McMahon 2005) Similarly the governor and legislaturein California have failed to adopt a budget on time since fiscal year 1987 (CaliforniaDepartment of Finance 2007)

14 We tested a parallel set of models that used changes in real per capita spendingrather than changes as a percent of past spending to measure both the governorrsquosproposals and final outcomes The models yielded substantively identical results tothose presented here

15 In a separate analysis we interacted the governorrsquos proposal with the presenceof divided government measured first by whether or not the party opposing the governorcontrolled both houses of the legislature and second by whether or not the opposingparty controlled one legislative house Neither interaction was statistically significantSee analysis by Bowling and Ferguson (2001) and Ferguson (2003) for explorations ofthe contingent effects of divided government on state legislative gridlock and the fateof gubernatorial proposals

16 Because we used fixed effects we identified the effect of professionalizationin Models 2 and 3 by the variation in the size of the governorrsquos proposal

17 Model 3b uses the dynamic professionalism scores reported by Squire (2007220ndash21) For observations up to fiscal year 1996 we used Squirersquos 1986 scores Forfiscal years 1996 through 2003 we used Squirersquos 1996 scores For the remaining yearswe used Squirersquos 2003 scores We also included professionalism by itself in this modelsince as a dynamic measure it was no longer correlated with each statersquos fixed effectsBecause these results were nearly identical to the ones obtained using the scores reportedby Squire (1992) and used most often in the literature elsewhere in our analysis weused only the Squire (1992) scores

18 The two-way correlations in our dataset are 051 between salary and sessionlength 067 between salary and staff size and 049 between session length and staff size

19 These categories represent the possible processes identified in the onlineversion of Grooters and Ecklrsquos (1998) table available at the NCSL website lthttpwwwncslorgprogramsfiscallbptablslbpc6t4htmgt We found no consistent relation-ship between legislative professionalization and these variables

REFERENCES

Abney Glenn and Thomas P Lauth 1987 ldquoPerceptions of the Impact of Governorsand Legislatures in the State Appropriations Processrdquo Western Political Quarterly40 335ndash42

Alt James E and Robert C Lowry 1994 ldquoDivided Government Fiscal Institutionsand Budget Deficits Evidence from the Statesrdquo American Political ScienceReview 88 811ndash28

83Who Blinks First

Alt James E and Robert C Lowry 2000 ldquoA Dynamic Model of State Budget Out-comes under Divided Partisan Governmentrdquo Journal of Politics 62 1035ndash70

Banks Jeffrey S and John Duggan 2006 ldquoA General Bargaining Model of Legisla-tive Policy-makingrdquo Quarterly Journal of Political Science 1 49ndash85

Bernick E Lee and Charles W Wiggins 1991 ldquoLegislative-executive Relations TheGovernorrsquos Role as Chief Legislatorrdquo In Gubernatorial Leadership and StatePolicy ed Eric B Herzik and Brent W Brown New York Greenwood

Berry William D Michael B Berkman and Stuart Schneiderman 2000 ldquoLegislativeProfessionalism and Incumbent Reelection The Development of InstitutionalBoundariesrdquo American Political Science Review 94 859ndash74

Beyle Thad 2004 ldquoThe Governorsrdquo In Politics in the American States ed VirginiaGray and Russell L Hanson Washington DC CQ Press

Bowling Cynthia J and Margaret Robertson Ferguson 2001 ldquoDivided GovernmentInterest Representation and Policy Differences Competing Explanations ofGridlock in the Fifty Statesrdquo Journal of Politics 63 182ndash206

California Department of Finance 2007 ldquoCalifornia Budget Frequently Asked QuestionsrdquolthttpwwwdofcagovBudgetingBudgetFAQsBudget_FAQsphp16gt(Accessed June 2008)

Cameron Charles M 2000 Veto Bargaining Presidents and the Politics of NegativePower Cambridge UK Cambridge University Press

Canes-Wrone Brandice 2001 ldquoThe Presidentrsquos Legislative Influence from PublicAppealsrdquo American Journal of Political Science 45 313ndash29

Carey John M Gary F Moncrief Richard G Niemi and Lynda W Powell 2003ldquoTerm Limits in the State Legislatures Results from a New Survey of the 50Statesrdquo Presented at the annual meeting of the American Political ScienceAssociation Philadelphia PA

Caruso Joe 2004 ldquoReflections on a Late Budgetrdquo Loudonville NY Sienna ResearchInstitute

Clark Wes 1998 ldquoDivided Government and Budget Conflict in the US StatesrdquoLegislative Studies Quarterly 23 5ndash22

Council of State Governments 2000 The Book of the States 2000ndash2001 edition vol33 Lexington KY Council of State Governments


Davidson Roger Walter J Oleszek and Frances E Lee 2007 Congress and ItsMembers Washington DC CQ Press

Denzau Arthur William Riker and Kenneth Shepsle 1985 ldquoFarquharson and FennoSophisticated Voting and Home Stylerdquo American Political Science Review 791117ndash35

Dye Thomas R 1966 Politics Economics and the Public Political Outcomes in theAmerican States Chicago IL Rand McNally

Dye Thomas R 1984 ldquoParty and Policy in the Statesrdquo Journal of Politics 46 1097ndash1116

Fenno Richard F 1966 The Power of the Purse Appropriations Politics in CongressBoston MA Little Brown

Ferguson Margaret Robertson 2003 ldquoChief Executive Success in the LegislativeArenardquo State Politics and Policy Quarterly 3 158ndash82


Field Poll 2003 Release 2081 San Francisco CA Field Research CorporationField Poll 2004 Release 2143 San Francisco CA Field Research CorporationFiorina Morris 1994 ldquoDivided Government in the American States A Byproduct of

Legislative Professionalismrdquo American Political Science Review 88 304ndash16Francis Wayne L 1989 The Legislative Committee Game A Comparative Analysis

of Fifty States Columbus Ohio State University PressGarand James C 1988 ldquoExplaining Government Growth in the US Statesrdquo American

Political Science Review 82 837ndash49Garand James C and Kyle Baudoin 2004 ldquoFiscal Policy in the American Statesrdquo In

Politics in the American States ed Virginia Gray and Russell L HansonWashington DC CQ Press

Grooters Jennifer and Corina Eckl 1998 Legislative Budget Procedures A Guide toAppropriations and Budget Processes in the States Commonwealths andTerritories Washington DC National Conference of State Legislatures

Grooters Jennifer and Corina Eckl 2008 ldquoTable 6-4 Procedures When the Appro-priations Act is Not Passed by the Beginning of the Fiscal Yearrdquo httpwwwncslorgprogramsfiscallbptablslbpc6t4htm (Accessed June 2008)

Gross Donald 1991 ldquoThe Policy Role of Governorsrdquo In Gubernatorial Leadershipand State Policy ed Eric B Herzik and Brent W Brown New York Greenwood

Hamm Keith E and Gary F Moncrief 2004 ldquoLegislative Politics in the Statesrdquo InPolitics in the American States ed Virginia Gray and Russell L HansonWashington DC CQ Press

Hofferbert Richard I 1966 ldquoThe Relation between Public Policy and Some Struc-tural and Environmental Variables in the United Statesrdquo American PoliticalScience Review 60 73ndash82

Karnig Albert K and Lee Sigelman 1975 ldquoState Legislative Reform and PublicPolicy Another Lookrdquo Western Political Quarterly 28 548ndash53

Kiewiet D Roderick and Mathew D McCubbins 1988 ldquoPresidential Influence onCongressional Appropriationsrdquo American Journal of Political Science 32 713ndash36

King James D 2000 ldquoChanges in Professionalism in US State Legislaturesrdquo Legis-lative Studies Quarterly 25 327ndash43

Kousser Thad 2002 ldquoThe Politics of Discretionary Medicaid Spending 1980ndash1993rdquoJournal of Health Politics Policy and Law 27 639ndash71

Kousser Thad 2005 Term Limits and the Dismantling of State Legislative Profes-sionalism Cambridge UK Cambridge University Press

Lowry Robert C James E Alt and Karen E Ferree 1998 ldquoFiscal Policy Outcomesand Electoral Accountability in the American Statesrdquo American Political ScienceReview 2 759ndash74

Matsusaka John G 2004 For the Many or the Few The Initiative Process PublicPolicy and American Democracy Chicago IL University of Chicago Press

McAtee Andrea Susan Webb Yackee and David Lowery 2003 ldquoReexamining theDynamic Model of Divided Partisan Governmentrdquo Journal of Politics 65 477ndash90

McCarty Nolan M and Keith T Poole 1995 ldquoVeto Power and Legislation AnEmpirical Analysis of Executive and Legislative Bargaining from 1961 to 1986rdquoJournal of Law Economics and Organization 11 282ndash311

85Who Blinks First

McMahon EJ 2005 Breaking the Budget in New York Albany NY Empire Centerfor New York State Policy

Meyers Roy T 1997 ldquoLate Appropriations and Government Shutdowns FrequencyCauses Consensus and Remediesrdquo Public Budgeting and Finance 17 25ndash38

Morrow James D 1994 Game Theory for Political Scientists Princeton NJ PrincetonUniversity Press

National Association of State Budget Officers Various years The Fiscal Survey ofStates various editions Washington DC NASBO

National Association of State Budget Officers 2002 Budget Processes in the StatesWashington DC NASBO

National Conference of State Legislatures 2005 NCSL Backgrounder Full-Time andPart-Time Legislatures Washington DC National Conference of State Legislatures

Osborne Martin J and Ariel Rubinstein 1990 Bargaining and Markets San DiegoCA Academic Press

Patashnik Eric M 1999 ldquoIdeas Inheritance and the Dynamics of Budgetary ChangerdquoGovernance An International Journal of Policy and Administration 12 147ndash74

Phillips Justin 2005 The Political Economy of State Tax Policy The Effects of ElectoralOutcomes Market Competition and Political Institutions PhD diss Universityof California San Diego

Primo David 2002 ldquoRethinking Political Bargaining Policymaking with a SingleProposerrdquo Journal of Law Economics and Organization 18 411ndash27

Quinnipiac University 2001 ldquoSenator Clintonrsquos Approval Tops 50 for First TimeQuinnipiac University Poll Findsrdquo Hamden CT Quinnipiac University

Roeder Phillip W 1979 ldquoState Legislative Reform Determinants and Policy Conse-quencesrdquo American Politics Quarterly 7 51ndash70

Romer Thomas and Howard Rosenthal 1978 ldquoPolitical Resource AllocationControlled Agendas and the Status Quordquo Public Choice 33 27ndash43

Rosenthal Alan 1990 Governors and Legislators Contending Powers WashingtonDC CQ Press

Rosenthal Alan 1998 The Decline of Representative Democracy Process Partici-pation and Power in State Legislatures Washington DC CQ Press

Rosenthal Alan 2004 Heavy Lifting The Job of the American Legislature WashingtonDC CQ Press

Rubinstein Ariel 1982 ldquoPerfect Equilibrium in a Bargaining Modelrdquo Econometrica50 97ndash109

Rubinstein Ariel 1985 ldquoA Bargaining Model with Incomplete Information about TimePreferencesrdquo Econometrica 53 1151ndash72

Selten Reinhart 1975 ldquoReexamination of the Perfectness Concept for EquilibriumPoints in Extensive Gamesrdquo International Journal of Game Theory 4 25ndash55

Smith Mark A 1997 ldquoThe Nature of Party Governance Connecting Conceptualizationand Measurementrdquo American Journal of Political Science 41 1042ndash56

Squire Peverill 1992 ldquoLegislative Professionalization and Membership Diversity inState Legislaturesrdquo Legislative Studies Quarterly 17 69ndash79

Squire Peverill 2007 ldquoMeasuring State Legislative Professionalism The Squire IndexRevisitedrdquo State Politics and Policy Quarterly 7 211ndash27


Squire Peverill and Keith E Hamm 2005 101 Chambers Congress State Legisla-tures and the Future of Legislative Studies Columbus Ohio State UniversityPress

Thompson Joel A 1986 ldquoState Legislative Reform Another Look One More TimeAgainrdquo Polity 19 27ndash41

Wilson E Dotson and Brian S Ebbert 2006 Californiarsquos Legislature SacramentoCA Office of the Assembly Chief Clerk

Winters Richard 1976 ldquoParty Control and Policy Changerdquo American Journal ofPolitical Science 20 597ndash636


Typically the legislature is treated as a monopoly proposer submittingldquotake it or leave itrdquo offers to an executive who possesses an absoluteveto The executive is then forced to choose between the appropriationsfigures contained in the bill and the reversionary or status quo pointThis reversion is almost always assumed to be the previous yearrsquosspending plan maintained in the absence of legislative-executive agree-ment on a new budget through a continuing resolution

In spatial models the legislaturersquos proposal power combined withits ability to credibly threaten to keep expenditures at the status quolevel gives the legislature substantially greater influence overbudgetary outcomes than the executive holds For instance using aspatial model of presidential-congressional bargaining Kiewiet andMcCubbins (1988) have shown that when the president prefers smallerexpenditures than Congress proposesmdashthe circumstance most favor-able to the presidentmdashthe president exerts only a limited influenceover budgetary outcomes When the president prefers a higher level ofexpenditures the president has no influence at all Kiewiet andMcCubbinsrsquos insights have received additional support from a subse-quent investigation by McCarty and Poole (1995)

In the study of American states applications of setter modelsalso predict legislative dominance In their influential analyses of statebudgeting under divided government Alt and Lowry (1994 2000)amended the spatial model developed by Kiewiet and McCubbins toaccount for the balanced-budget requirements that exist in most statesIn Alt and Lowryrsquos model the legislature and governor must reachagreement on fiscal balance (whether there is a surplus deficit orbalanced budget) in addition to fiscal scale Alt and Lowry also addedan assumption backed by Lowry Alt and Ferreersquos (1998) empiricalwork that fiscal imbalance results in significant electoral losses forthe governorrsquos copartisans in the legislature

Alt and Lowryrsquos model like the Kiewiet and McCubbins modelsuggests executive weakness When there is interbranch disagreementover the size of the budget the legislature can use its monopoly proposalpower to threaten the governor with fiscal imbalance by passing acontinuing resolution rather than a new budget Since in this modeldeficits or surpluses put the governorrsquos copartisans in the legislatureat risk the governor will be forced to make significant concessions tothe legislature on fiscal scale in return for a balanced budget Afterreviewing the empirical predictions of their model under different fiscalcontexts and configurations of party control Alt and Lowry concludedthat state legislatures are even stronger than Kiewiet and McCubbins(1988) predicted Congress to be Although governors can achieve some

57Who Blinks First









59Who Blinks First












61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























57Who Blinks First









59Who Blinks First












61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First
































59Who Blinks First












61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























59Who Blinks First












61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First



































61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























61Who Blinks First









1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First































1( ) ( )

1 1L GX Xδ

δ δ=

+ + 1( ) ( )

1 1L GY Yδ

δ δ=

+ +


63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























63Who Blinks First












1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First
































1 (1 )

( ) ( )1 1

G G LL G

G L G L

X Xδ δ δ


=minus minus

(1 ) 1( ) ( )

1 1L G L

L GG L G L

Y Yδ δ δ


=minus minus

65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























65Who Blinks First



0

01

02

03

04

05

06

010203040506070809





ollar










67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First


































67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























67Who Blinks First









69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First
































69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























69Who Blinks First






Constant 134 198 157 162(29) (44) (56) (046)

N 286 370 178 852R2 48 18 19 25








71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First
































71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























71Who Blinks First



















Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First










































Lagged Surplus 07 03 03 03(06) (06) (06) (07)


N 787 787 787 787

R2 30 32 32 30


73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























73Who Blinks First








75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First































75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























75Who Blinks First





Salary (in $1000s) 02 __ __ 04(04) (04)











Lagged Surplus 06 02 06 01(64) (06) (06) (06)


N 787 787 787 787R2 30 33 30 34


p lt 01 p lt 05







77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First


































77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























77Who Blinks First
















Lagged Surplus 02 02 02 02(06) (06) (06) (06)


N 787 787 787 787R2 33 33 33 33








79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First


































79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























79Who Blinks First







NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First































NOTES






81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




























81Who Blinks First
















REFERENCES



83Who Blinks First









































85Who Blinks First




































REFERENCES



83Who Blinks First









































85Who Blinks First




























83Who Blinks First









































85Who Blinks First
















































85Who Blinks First




























85Who Blinks First

































Documents

Who Blinks First? Legislative Patience and Bargaining with ...jhp2121/publications/WhoBlinksFirst.pdfLegislative Patience and Bargaining with Governors When legislators and governors