Government Growth in the Twenty-First Century Author(s): Randall G. Holcombe Reviewed work(s): Source: Public Choice, Vol.

124, No. 1/2, Policy Challenges and Political Responses: Public Choice Perspectives on the Post-9/11 World (Jul., 2005), pp. 95-114 Published by: Springer Stable URL: http://www.jstor.org/stable/30026705 . Accessed: 06/12/2011 11:34
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Public Choice.

http://www.jstor.org

Public Choice (2005) 124: 95-114 DOI: 10.1007/s 11127-005-4748-x

C Springer2005

Government growth in the twenty-first century

RANDALL G. HOLCOMBE
Departmentof Economics,Florida State University,Tallahassee,FL 32306-2045 USA (E-mail: holcombe@ garnet.acns.fsu.edu)

Accepted March2005 Abstract. Public choice explanationsof governmentgrowth fall into three main categories: theories, rational-choicemodels, and path-dependentmodels like the budget-maximization "ratchet hypothesis".The strengthsand weaknesses of these theoriesas explanationsfor governmentgrowthareconsideredalong with some facts aboutthe actualgrowthof government to aboutits trajectory the twenty-firstcentury.Government in size seems to have been conjecture constrainedin the pastprimarily its abilityto raiserevenue.Growthratesin the new century by thus appearto depend on factorsconstraininggovernment'sability to continue to expandthe tax base.

1. Introduction
The massive worldwide growthof governmentin the twentiethcenturywas a truly remarkable phenomenon,and one that has interestedthe community of public choice scholars since the discipline was in its infancy. Looking ahead to the twenty-firstcentury,will this growth continue, or is the era of big governmentover, as Bill Clintonsaid in 1995?1This paperexamines this issue by looking at the theories that public choice analysis has offered to explain governmentgrowth,along with some facts about the past growthof government,to try to gain some insight on what lies ahead. The task wouldbe more straightforward therewere one well-established if of governmentgrowth,but public choice scholarshave offered many theory hypotheses.After reviewingthe existing theoriesalong with the facts of governmentgrowth, this paper suggests that there is a single factor underlying governmentgrowth:growthin the availabilityof revenues.Governmentwill spend as much as it can collect fromits citizens so, as Brennanand Buchanan the (1980) suggest, the key to understanding future course of government is to develop an understanding the constraintsthat governments of growth face in raising revenue. That the course of twenty-firstcenturygovernment growth will depend primarilyon the ability of governmentsto extract revenues from theircitizens may be a controversial conclusion, so the paperwill on proceedby examiningthe existingpublicchoice literature governmentsize and governmentgrowthalong with some historicalfacts to indicatewhy this conclusion appearsconsistentwith both the facts and the literature.

96

2. The TwoDimensionsof the Literature

Explanationsfor the growth,or size, of governmentin the public choice literaturediffer on two differentdimensions.First, some of the literature focuses on the size of government,while some looks at the growth in government. Second, thereare,broadly,threedifferentcategoriesof explanationsfor government'sgrowthor size: budget-maximization modmodels, rational-choice models. Looking at the firstdimension,in one sense els, and path-dependent and it may appearthattheoriesaboutthe growthof government theoriesabout the size of governmentamountto the same thing,becausegovernment growth is the firstderivativeof governmentsize. An explanationfor one is therefore an explanationfor the other.However,if thereis a pathdependencyin government growth, models designed to explain governmentsize may not capture the elements that lead to governmentgrowth.Thus, it becomes importantto separatethe sometimes subtle differences between theories of government size and theoriesof governmentgrowth. Looking at the second dimension, many public choice explanationsof governmentgrowth rest on a model that depicts the size of governmentas a collective choice of its citizens. While there is an element of truthto this - in that collective choice mechanisms can act as a constrainton government action - the depiction of governmentsize as an outcome of collective or choice stands at odds with the literatureon revenue-maximizing budgetwheregovernmentspendsas muchas it can, subject maximizinggovernment, to certainconstraints.These two explanationsbecome more consistent with each other if collective choice mechanismsare viewed as constraintson the activities of governmentratherthan as mechanismsthat cause outcomes to reflect constituents'preferences.The thirdmajorcategory of explanationis that there is a path dependencyin governmentspending,which would make any staticmodel of collective choice misleadingin the analysisof government growth.Because there is no one establishedpublic choice theory of government growth, any single explanationis likely to be controversial,including the explanationoffered here. It is offered not with any hope that it is the last word on the subject,but ratherwith the hope that it will stimulateinterestin furtherresearchon the topic.

3. BudgetMaximization
Before Niskanen's (1971) book, Bureaucracyand RepresentativeGovernment,public choice almostexclusively examinedthe demandside of the public sector, neglecting the supply side. Demand-sidemodels were dominated by the medianvoter model, which as Holcombe (1989) explains, was almost always taken to imply that the public sector is a reflection of the median Niskanencreateda minorrevolutionwithinpublicchoice voter'spreference.2

97 that with his assumptionof budget-maximizing governmentbureaucracy led directlyto a numberof studieson government growth(Borcherding,1977). In the context of governmentgrowth,Niskanen'smodel of budget-maximizing is explains why government excessively large,but not why it has bureaucracy grown. Perhapsits greatestimpacton the literatureon governmentgrowthis the assumption- persuasivelyjustified, but an assumptionnevertheless- of budgetmaximization. While Niskanen's model was seen by some as a critiqueor refutationof the median voter model, it is in fact built on the median voter model, and Niskanen (1971, pp. 139 & 143) refers to the demanderin his model as the median voter in a number of places. The median voter's demand still determines the budget size, even though the outcome is higher government spending than the median voter prefers. Similarly, Romer and Rosenthal (1978) clearly depict the median voter's preferenceas determiningthe level of governmentspending,even thoughgovernmentspends more thanthe median voter prefers. In both cases, the institutionalstructuredepicted by the models acts merely as a constrainton the actions of expenditure-maximizing politicians. This budget-maximizing assumptionmanifestsitself in a slightly different form in BrennanandBuchanan(1980), who depict governmentas a revenuemaximizing Leviathanthat may be limited by constitutionalconstraintson its behavior.As with Niskanen's bureaucracy model, this is a model of big governmentbut not growing government,and again the driving force behind the model is an assumptionof revenuemaximization.In this literature, constitutionsand democraticinstitutionsserve only the functionof imposing constraintson the size of a budget-or revenue-maximizing government,and without these constraintsthe models give no indicationas to how large governmentcould actuallygrow.Of course, governmentsize will alwaysbe constrainedby the maximumamountof revenuethat governmentcould possibly collect - the Laffercurve3- but the point is thatin these models, government tries to spend as much as it possibly can, and constitutions,democraticinstitutions,and so forth do not determinethe optimal amountof government spending, they only constrainexcessively large governmentfrom spending even more. If the constraintis loosened, governmentgrows; if the constraint is tightened,governmentshrinks,but unlike the medianvoter model, collective decision-makingdoes not produce what citizens most prefer - it only constrainsexcessively large governmentto some degree. Along these same lines, Becker andMulligan(2003) note the constraining effect of the deadweightloss associatedwith taxes. If taxes arecollected more efficiently,deadweightlosses go down, which reduces the political pressure againsttaxes and causes taxes to rise and governmentto get larger.Similarly, Holcombe and Mills (1995) argue that deficit finance is constrainedby the political oppositionit generates,andBuchananandWagner(1977) arguethat

98 the acceptance of Keynesian economic policy measuresrelaxed a political constrainton deficit finance, allowing governmentto spend more than had that constraintremained in place. What these works have in common is that the size of governmentis constrainedby political opposition. If factors change to lessen the political opposition to governmentrevenue generation, governmentgrows. Kau and Rubin (1981) find that virtuallyall of the increasein the size of the United States governmentfrom 1929 to 1970 can be explainedby reductions in the cost to governmentof collecting taxes. Similarly,DavidFriedman (1977) arguesthatthe size and shapeof nationsis designedto maximize governmentrevenue. Budget maximizationor revenue maximizationhas been well-establishedas a workinghypothesisin a numberof public choice models for decades, althoughnot everyoneaccepts the assumption'svalidity.The main alternative, least as presentedin the publicchoice literature, to view at is actual outcomes in the public sector as a reflectionof citizen preferencesas expressedthroughthe political process.4

4. Rational-Choice Size Modelsof Government

The median voter model has alreadybeen noted as an example of a model that suggests that the size of governmentreflects the preferencesof citizens as expressedthrougha collective decision-makingprocedure,but thereare a numberof othermodelspointingto the sameconclusion.Forexample,Becker institu(1983) andWittman(1989, 1995) describecollective decision-making in tions as creatinga kind of political marketplace which competingdemands of variousinterestsareweighed againsteach otherandthe resultis an optimal allocationof resourcesin the public sector,analogousto the way that a competitive marketwould allocate resourcesin the privatesector.These models were not intendedto explain governmentgrowth,but if governmentspending optimally reflects the preferencesof its citizens, the implicationis that governmentgrows because people have a preferencefor largergovernment. MeltzerandRichard(1981) developa model explicitlybuilton the median voter model, and arguethat governmenthas grown because of extensions of the franchisethat have changed the position of the median voter. Similarly, Lott and Kenny (1999) arguethat governmentgrew in the twentiethcentury because women were extendedthe vote andwomen preferlargergovernment than men. Peltzman (1980) argues that governmentgrows because people vote themselves more redistributive benefits, and suggests that a more equal distribution income has producedmore of this rent-seekingbehavior,leadof ing to largergovernment.Models of governmentgrowthsuch as these do not argue that governmentsize is optimal, but ratherthat it is the outcome of a collective choice process that reflects the preferencesof those who choose.

99 In these cases, the set of people who could choose was expandedto include or people who wantedbiggergovernment, in Peltzman'scase, the distribution of income changedto makethe existing electoratefavormore governmentredistribution,and governmentgrew. These models are similar to the median voter model in thatthe size of governmentis a result of a democraticchoice, but they take into account changes in the group of people who are able to express theirpreferencesthroughvoting. Baumol(1967) offersanother for explanation government growth:because natureof governmentservices, productivitywill grow of the labor-intensive more in the private sector than in the public sector, so utility-maximizing individuals will want to shift their consumption away from private sector consumptiontowardpublic sectorconsumption.Baumolwas arguingthatthis shift has not happenedto the extent that would be optimal,so governmentis too small;nonetheless,if differencesin relativeproductivitygrowthbetween the public sector and private sector pointed toward a shift in production towardthe public sector,this would resultin individualscollectively choosing a largerpublic sector.Downs (1960) likewise argueswithinhis medianvoter frameworkthat voters tend to choose inefficientlylow levels of government expenditures. Institutional differences may lead some types of political decisionmaking to result in higher governmentalexpendituresthan others. Persson and Tabellini (1999, 2003) and Mueller (2002) explain that interest groups tend to be more successful in lobbyingEuropeanparliamentary governments than the more decentralizedgovernmentin the United States, resulting in largertransferexpendituresin Europeangovernments.Institutionsmay vary, resultingin different-sizedgovernments,but the underlyingsize of government is a result of a collective choice mechanism that makes government activities a productof grouppreferences. The variousexplanationsfor governmentgrowth,or governmentsize, offered in this section differ in a numberof ways, but they have the common element that they consider the size of governmentto be a reflection of the preferencesof its citizens aggregatedthrougha collective choice mechanism. As in muchof the publicchoice literature, collective decision-makingprothe cess in these models is depicted as a methodof transforming preferences the of a groupof people into an outcome thatreflects the group'spreferences.In many of the models the outcome is not necessarilyefficient because institutions do not necessarily aggregatepreferencesto produce the optimal level of output,but in models like those of Becker (1983) and Wittman(1995), the political system also weighs the intensitiesof preferencesin orderto generate an optimal allocation of resources. In either case, governmentgrowth is a result of a collective preferencefor largergovernment,as citizen preferences are aggregatedthrougha rationalchoice mechanism.

100 5. Path Dependency the "Ratchet and Hypothesis"

One of the earlier theories of governmentgrowth is the ratchet hypothesis, first put forward by Peacock and Wiseman (1961) and supportedby Rassler and Thompson (1985), Higgs (1987) and others. The theory is that governmentresponds to crises like wars and depressions by ratchetingup expenditures,and then after the crises pass, expendituresfall somewhatbut remain above their pre-crisis level. Thus, governmentgrowth is a series of ratchetsupwardin governmentspendingin response to crisis. While it does appearthat spendingratchetsup aftercrises and remainsabove its pre-crisis level, Holcombe (1993) notes thatin an empiricalexaminationof expenditure levels, the trendgrowthof governmentexpendituresin the twentiethcentury has been so substantialthat any ratchets are completely swamped by the trend,and do not show up as statisticallysignificantincreases. Yet as Higgs (1991) notes, the ratchetsmay still be there, because governmentcan grow in dimensionsotherthanexpenditures. The ratchethypothesisappearsat odds with the rationalchoice models of governmentgrowth(or governmentsize), unless a past crisis causes people to can believe thatcontinuedhighergovernment expenditures help averta future crisis. As Holcombe (1996, 2002, ch. 9) shows, after the ratchet upward increasein the growthof nonfollowing WorldWarI therewas a substantial federal governmentspendingand regulationin the 1920s, and after military rose from7.8%of federalgovernment WorldWarII non-military expenditures GDP to 10%,following militaryspendingdecreasesin each case. Clearly,the increasesafterthese majorwarswere not a rationalnon-military expenditure choice responseto the crisis thathad passed. Rather,as militaryexpenditures receded after the wars, the increased revenue used to finance the wars left money available for an increase in non-militaryexpenditures.Both World WarI and WorldWarII were financedin partby majorincome tax increases, and while afterthe wars ratesfell some from theirwartimehighs, they never fell to theirpre-warlevels. Priorto WorldWarI the highest marginalincome tax rate was 7%, and it went to 77% in 1918. The top marginalrate fell back to 25% in the 1920s, which was below the wartime high but more than three times higher than the pre-war rate. Similarly, from 1936 to 1939 the top marginal income tax rate was 79%, levied on incomes above $5 million.5 It was raised to 94% of incomes above $200,000 during World War II, and the rate came down only to 91% of incomes above $200,000 in the 1950s.6 Rates that would not have been toleratedprior to the war were accepted as a part of the war effort and, once in place, the federal governmentwas able to maintain much of the tax increase after the crisis had passed. Governmentgrew because revenue was availableafter the wars that could not have been collected prior to them. Citizens would not have accepted these tax increases

101 during normal times, but were willing to accept them as a response to crises. Another factor that aided governmentrevenue collection during World WarII was the implementationof income tax withholdingon wage income. Again, it is unlikely that citizens would have accepted withholdingwithout the crisis, but once implementedcitizens did not rebel againstthe statusquo. Note, however,thatin the 1980s therewere severalattemptsmade to institute withholdingon dividend and interestincome, but without a crisis to justify it, they were rejected because of public opposition. One would think that because interest and dividend withholdingwould have its largest impact on the rich, whereas wage withholding affects the average (median?) citizen much more, interestand dividendwithholdingwould meet with less political oppositionthanwage withholding.7Two observationsarerelevanthere:first, because wage withholding was implementedin response to a crisis, it was more acceptablepolitically,bolsteringthe ratchethypothesis;second, once a change has been implementedandbecomes partof the statusquo, it becomes difficultto reverse. Both the ratchethypothesis and the evidence just reviewed suggest that thereis a pathdependencyin the level of government that expenditures cannot be capturedin a comparative-statics framework. Institutionsand citizen fully preferencesact as constraintson the level of governmentexpenditures,but if a constraintis temporarily relaxed,resultingin an increase in government thatconstraintwill not fully be reimposedto cause government expenditures, to expenditures fall backto theirformerlevel. Olson (1982) presentsa similar theory of path dependency,arguingthat as political systems mature,interest and groupsbecome morefirmlyentrenched areableto diverteverlargershares of nationalincome away from expendituresin the public interestto support their own privateinterests,leading to the decline of nations. If these theories of path-dependentgovernmentgrowth are descriptive, they call into question models suggesting that the level of governmentexpenditures is optimal by some measure. These theories suggest that the level of governmentexpendituresis not the result of some process that responds only to currentconditions, but ratheris dependentupon historical circumstancesthat have allowed governmentexpendituresto rise. While a crisis may have some impact on what citizens believe to be the optimal level of governmentexpenditures,it is implausible to think, for example, that because of WorldWarsI and II citizens believed that non-militaryexpendituresshould be higher after the war than before. More likely, a constraint on government revenues that could not have been relaxed during peacetime was relaxed because of the war, and was not able to be reimposed after the war. This suggests a revenue-maximizinggovernmentcoupled with institutionalrigidities and path-dependentinstitutionaldevelopments along the lines described by Olson (1982), and not a process that

102 somehow reflects the static preferencesof citizens at thatparticular point in time. Path-dependencytheories of governmentgrowth do not have as solid a behavioralfoundationas the rationalchoice models discussedin the previous section. Yet this may reflect as much on the way that models of individual rationalchoice areappliedto collective choice mechanisms thanon the ratchet itself. Brennanand Lomasky (1993) suggest thatrationalchoice hypothesis models do not apply all that closely to political choice, and Caplan (2001, 2003) has arguedthat voters not only are rationallyignorantbut also ratioRationalchoice models may not applyvery well to collective nally irrational. choice mechanismswhere any one individualhas a negligible impacton the overallcollective choice. Thereis a substantial literature incorporates that psyinto economics, discussed by Kahneman(2003), which chological findings suggests that the axioms of neoclassical utility maximizationdo not hold so strongly in real-worldchoices. Kahneman,Knetsch, and Thaler (1991) discuss the endowmenteffect, loss aversion,andthe statusquo bias, all of which are relevantto the ratchethypothesis. People may be reluctantto tradetheir freedoms or their incomes for governmentalgrowth, but once those things have been ceded and become a partof the statusquo, they are equally reluctant to sacrifice the benefits of the governmentprogramsthat have resulted from governmentgrowth. The status-quo bias may be especially powerfulwith regardto government for Interestgroupslobby government expenditure proexpenditure programs. gramsthatbenefitthem, and along the lines of Becker (1983), the legislature weighs the interests for and against programsto decide whether to implement them, or how much should be spent on them. Once the programsare established, a new interest group is created in the form of the government that will administerthe programs,adding to the strengthof inbureaucracy as terests that favor the expenditure.Furthermore, Niskanen (1971) argues, becomes the legislature'sexperts on the program,adding that bureaucracy to the bias in favor of keeping the program.As Tullock (1982) observed, many new governmentprogramsare started,but few existing programsare terminated. This status-quobias, well-establishedin the economics literature both inside and outside of public choice, provides at least a partialfoundation to supportthe ratchethypothesis and path dependencyin government expenditures. theoriesunderscorethe differencesbetween theories of Path-dependency governmentgrowth and theories of governmentsize. Theories of government size suggest that certainfactors determinethe size of government,and that if those factorschange, the size of governmentwill change in response. As it happens, the theories of governmentsize cited above all hypothesize that changes occurredto make governmentlarger,but presumablyif those changes were reversed,governmentwould shrink.Theories of government

103 size treat governmentgrowth as the first derivativeof government size. If there is a path dependency,however, the theories discussed in this section suggest that once governmenthas grown that growth is difficult to reverse. In this path-dependent environment,the size of governmentis not simply a functionof currentconditions,but is a resultof events thatmay be well past. It is plausibleto argue,for example, thathad WorldWarII not occurred,political forces would have preventedincome tax withholding,and as a result the federal governmentin the United States would be substantiallysmaller thanit is now. Events thatoccurredsixty years ago have left theirmarkon the level of current of government expenditures completelyindependently current conditions.

6. Non-Expenditure Growth
Most of the literatureon government growth has focused on expenditure growth, perhapsbecause expendituregrowth is relatively easy to quantify, and this paperfollows the literature thatregard.However,governmentcan in grow in other dimensions too, and these other dimensions may be at least as significantas expendituregrowth.Higgs (1987, 1991) emphasizesregulatory growth,the power to confiscate property,and the ability of government officials to act unilaterally,without any checks on their behavior, as other dimensionsin which governmenthas grown.Holcombe (2002) discusses the erosionof individualrightswhich have been replacedby the powerof the majority,andPosner(1971) describesregulationas a type of taxation.Gwartney and Lawson (2003) have quantifiedeconomic freedom across countries,and Gastil (1978) has quantified politicalrightsandcivil liberties,which are other dimensions along which governmentcan grow beyond simply the level of expenditures. Growthin these non-expenditure dimensionsof governmentis relevantfor two reasons.First,it may haveimpactsas significant,or more significant,than expendituregrowth. For example, if we accept Posner's (1971) description of regulationas a form of taxation,researcherswho focus only on the buddimensionof govgetaryactions of governmentare leaving out an important ernmentgrowth. Second, these non-expenditure dimensions may be related to the underlyingcauses of governmentexpendituregrowth, in which case one could not fully understand expendituregrowthwithoutan understanding of these other dimensions of governmentgrowth. Expendituregrowth may have been facilitatedby the increasing scope of governmentin other areas. This argumentshould resonate with public choice scholars who may look for the roots of governmentgrowthin political institutions.While recognizdimening the potentialimportanceof governmentgrowthin non-budgetary this paperprimarilyfollows the literatureby focusing on expenditure sions, growth.

104
7. Public Choice - or Compulsion? Public choice, as an area of inquiry, has tended to describe public sector outcomes as a reflectionof grouppreferences.The public sectoris a resultof a collective decision-makingprocess that produces what the group prefers. The analogy of a political marketplaceholds up to a degree, but fails to be a complete analogy because transactionsin the private marketplaceare undertaken voluntarilywhereascompulsionunderliesall governmentactivity. As Yeager(2001, p. 234) puts it, not even a democraticstateis a mechanismvoluntarilyoperatedby its citizens to attendto theircommon concerns.Even underpolitical democracy the essence of the stateis compulsion.Any call for a particular government activity is a call for supportingit, if ultimatelynecessary,by force. As applied to the topic of governmentgrowth,governmenthas been able to grow because it has been able to force citizens, throughtaxation,to pay for its ever-increasingsize. Whereas public choice has often viewed governmentas the product of collectively expressed preferences,the literatureon governmentgrowth has deviatedfrom this approachto a degree. Both the budgetmaximizationliteratureand the pathdependency-ratchet hypothesisliterature depicts a government that has interestsat odds with its citizens. This literaturesuggests that the suppliersof governmentprograms- politicians and bureaucrats want to enlarge their domains, and the electoral process - the demandside of the does not try market- has acted only as an ineffective constraint.Government to reflect the preferencesof its citizens, following this line of reasoning,but tries to overridethem to commandmore resourcesfor itself. In this context, the public choice literatureoffers two starklycontrasting views of the public sector. In one view, the public sector is a result of a collective decision-makingprocess in which various interests interactin a throughvoting, lobbying, and otherpolitical activities, political marketplace and the political marketplaceallocates resources in a manneranalogous to more ordinarycompetitivemarkets.In the other view, public officials try to maximize governmental revenuesand expenditures,and the political process and constitutionalrules act as constraintson the government'ssize. Even this second view, however, rests on an ultimate foundation of agreement, albeit with one side having an advantageous bargainingposition. Niskanen's bureauscan only produceas much as their sponsorswill budget-maximizing agendasetterscan only agree to; Romerand Rosenthal'sbudget-maximizing get as much as the voters will agree to; Brennanand Buchanan'srevenuemaximizingLeviathancan only tax as much as the constitutionallows. Even here the analogy seems to allude more to a reluctantcustomerdealing with a

105 monopoly supplierthan to a citizenry that is forced to comply whetherthey consent or not. Ironically,this element of compulsionis more a partof neoclassicalpublic finance than it is a part of public choice. For example, optimal tax theory, based on DiamondandMirrlees(1971) andMirrlees(1971), depictsa welfare (not budget) maximizing governmentthat redistributesincome by forcibly takingit from some people and giving it to others.While one might make the excuse that this is unobjectionablebecause the governmenthas the welfare of its citizens in mind, what the model of optimal taxation argues is that it is optimal for the governmentto take income from some people forcibly in order to furtherits objectives. In contrast,public choice analysis has often depictedsuch a governmentactionas the resultof a collective agreement.In a of democracy,giving people the rightto vote gives the appearance legitimacy to governmentaction, as Edelman(1964) notes, obscuringthe coercion that underlieseverythinggovernmentdoes. If coercion were not necessary,then there would be no reasonfor governmentto get involved. It may be comforting to think that in optimal tax models governmentis forciblytakingmoney from some to give to othersto increasethe utilityof the recipients,but normallyeconomists assume thatpeople act to maximize their own utility, not the utility of others.Thus, Niskanen's (1971) argumentas to why those in governmentmaximize theirutility by maximizingtheirbudgets and supportsthe Leviathanmodel of government suggeststhatmoney forcibly takenfromtaxpayersis done not to maximize social welfare,butto maximize the utility of those who are takingit. Taxationis, as Usher (1992) explains, a form of predation. The implicationsof this line of reasoningfor public choice theoryin generalare substantial, for presentpurposestwo ideas will be carriedforward. but First, the model of revenue and expendituremaximizing governmentmakes sense as a descriptionof governmentalbehavior,and is in many ways more descriptivethan a model of a political marketplacein which resources are allocated throughbargainingamong various interests. Yes, the demands of variousinterestsmust be accountedfor if political leaders want to survive in office, but, as Bueno de Mequitaet al. (2003) argue, this is differentfrom a marketplacewhere people enter into agreementsfor mutuallyadvantageous trades.Second, one might question the effectiveness of electoral and constitutional constraintson government,because governmentcan implement its policies by force andin generaldoes not need the agreementof the electorate. The first idea - of budget-maximizinggovernment- is well-established in public choice. The second idea - about the ineffectiveness of electoral and constitutionalconstraintsin the face of governmentcoercion - is not as wellestablished. If anything, public choice has served to reinforce the idea of the effectiveness of electoral constraintson government.Democracy does constraingovernmentaction to a degree, undoubtedly, to the degree that but

106 political competitionis limited by barriersto entry,which is an old idea in publicchoice (Tullock,1965), incumbentsdo not face a truecompetitivemarket, and are in a position to use the force of governmentto furthertheir own ends. One gets a differentpictureof the causesof government growthif one looks at the size of governmentas a result of the rationalchoice of its citizens as opposed to viewing it as the result of a revenue-maximizing governmentthat is able to extractrevenuesfrom its citizens by force. Bringingin this element of compulsionmay lend moreinsightbothto the causes of government growth and to public choice analysis in general.8

8. Government Growthin the NineteenthCentury

Any complete theoryof governmentgrowthmust be able to explain not only why governmenthas grown rapidly in the twentieth century,but also why at other times there was relatively little governmentgrowth. Governments aroundthe world grew substantiallyin the twentiethcentury,but grew much more slowly in the nineteenth,and even retrenched some measures.Real by federal governmentexpendituresin the United States (in 1990 per capita dollars) were $79.76 in 1870, and were $79.56 in 1895, twenty-five years later (Holcombe, 2002, p. 140). Real per capita federal expenditureswere of unchangedfor a quarter a century,and when one considersthe substantial real income growthduringthe period, governmentexpendituresactuallyfell as a percentageof income. When looking at nationalgovernments,one explanationfor the relatively in slow growthof government the nineteenthcenturywas the increasingopenness of theexpandingworldeconomy,allowingcapitalto shiftbeyondnational bordersmore easily and therebylimiting the ability of nationalgovernments with to tax it. In the UnitedStates,the ease of westwardmovementaccelerated the growthof the railroads,allowing people to move to the frontierto escape the reach of government.By the end of the nineteenthcenturythe continent was populatedfrom coast to coast, closing the frontier,9and the twentieth century was dominatedby two world wars and a cold war that divided the worldandmadeinternational capitalmovementsmuchmorerisky.This offers a tentativeexplanationfor why the growth of nationalgovernmentsslowed considerablyor even stopped in the nineteenthcentury,but acceleratedin the twentieth, which is consistent with the budget-maximizingassumption. Taxableresourcesbecame moremobile in the nineteenthcentury,butbecame less mobile in the twentieth. While nationalgovernmentgrowthwas limited in the nineteenthcentury, Holcombe and local governmentgrowthin the United States was substantial. Lacombe(2004) show thatwhile therewas minimalstateandfederalgovernment growthin the nineteenthcentury,per capitalocal governmentrevenues

107
Table 1. Federal, state, and local governmentexpenditures as a shareof total governmentexpenditures Year 1870 1913 1922 Federal(%) 43.4 23.3 35.5 State (%) 13.7 12.7 15.1 Local (%) 42.9 64.0 49.4

Source:Holcombe and Lacombe(2001, p. 188).

the and expendituresgrew substantiallythroughout century,especially so after 1870. Local governmentsspent about $28 per resident in 1820, which increasedto $122 per residentin 1850 and $253 per residentin 1870 (all in In were slightlyless than 1990 dollars).10 1870, local government expenditures in federalgovernmentexpenditures the UnitedStates,with local expenditures makingup 42.9% of total governmentexpendituresand federalexpenditures making up 43.4%. By 1913 local governmentexpenditureswere 64% of the total while federal expenditureshad fallen to 23.3% of total governmentexpenditures,as shownin Table1. So while the nineteenthcenturywas a century of minimal federal governmentgrowth, it was also a centuryof substantial local governmentgrowth. This growthin 19thcenturylocal governmentsmight be explainedin part demandsas cities were expandingand the economy by rising infrastructure was being transformed from a ruralone based on agriculture an urbanone to based on manufacturing, this explanationdoes not fit many of the facts of but local governmentgrowth.As Holcombe and Lacombe(2004) note, expendituregrowthwas not concentrated infrastructure, on all types of public on but andnon-publicgoods. In comparisonsof two cities thatwere similarin many ways - Boston and Baltimore- growth rates were found to be more highly correlatedwith increases in tax bases than with the demand for public services. The evidence suggests that nineteenth-century governmentsgrew city because agglomerationeconomies attractedeconomic activity to cities, and cities could use propertytaxes to raiserevenuefrom a relativelyimmobiletax base. Thus, city governmentgrowthresultedfrom access to a largertax base thatcould not escape city taxes, ratherthanto demand-sidefactors. when analyzing the factors underlying 19th century local Furthermore, governmentgrowth, such as an increasingly urbanizedpopulation and infrastructure demands,it is worthnoting that these factors continuedthrough the 20th century,suggesting that local governmentexpendituresshould have continuedgrowing fasterthan federal expenditures.Yet the opposite is true: had by 1922 federalgovernmentexpenditures risen to 35.5% of total governmentexpenditures, while local expenditures fallen to less thanhalf of total had as Table 1 shows. What happenedin that decade governmentexpenditures,

108 from 1913 to 1922? The two biggest factorsaffectingthis change were World WarI and the creationof the federal income tax in 1913. As explanationsof federalgovernmentgrowth,WorldWarI fits the ratchethypothesiswhile the creationof the federal income tax fits the explanationthat it is the supply of funds availableto a budget-maximizing government ratherthanthe demand for governmentexpenditures- that drivesgovernmentgrowth. In the nineteenthcentury,increasedcompetitionat the nationallevel and increasedinternational mobility reducedthe ability of nationalgovernments to raise tax revenues,whereasreducedfactormobility at the local level due to agglomerationeconomies in cities increasedthe ability of local governments to raise revenues." In the twentiethcentury,greatermobility due to advances in transportation technology reduced some of the locational advantagesthat cities had, while the income tax increasedthe federal government'stax base and international hostilities inhibitedinternational capitalmovements,which of nationalgovernments.These stylincreasedthe revenue-generating ability ized facts regardingfederalversuslocal governmentgrowthin the nineteenth and twentiethcenturiesare more consistentwith the hypothesis that government is a budget-maximizing institutionthatspendsas muchrevenueas it can raise thanwith the hypothesisthatgovernmentspendingreflectsthe demands of its citizens.

9. An Explanationfor Nineteenthand TwentiethCenturyGovernment Growth

The facts of the preceding two centuries are consistent with the hypothesis that governmentgrowthhas occurredbecause revenue-and expenditureshares havebeen ableto appropriate ever-increasing maximizinggovernments of privatesector production.This explanationimplies that collective choice do constraints not determinethe size or growth mechanismsandconstitutional rateof government, along the lines of BrennanandBuchanan(1980), they but, act as constraintsthatkeep the governmentfrom getting larger.Those in govon ernmentfindways to erodethe constraints spendingthatthey face, causing to grow, but there is only the loosest connectionbetween group government growth. growthandthe actualpathof government preferencesfor government constraintson government'spowerto tax are easier to maintainthan Existing are new constraints to implement,or thanold constraintsareto reimposeonce they have been removed. This status quo bias creates a path dependencyin governmentgrowth and suggests that given the natureof the constraintson government'ssize, it is easier to remove them to allow governmentto grow thanit is to impose them to force governmentto shrink. This theoryof governmentgrowthhas as its foundationthe assumptionof and and expenditure revenuemaximizationon the partof those in government, while this budget-maximization hypothesisis well-establishedandfrequently

109 Even if it is accepted,the employed in public choice, it is far from proven.12 budget-maximization propositionas appliedto government growthis ambiguous. As Buchananand Lee (1982), Levi (1988) and Holcombe (1994) note, differentpolicies are implied dependingupon the time horizonof those who have governmentpower, and a budget maximizer with a long time horizon might prefer governmentto take a smaller share of income in order to encourage income growth. Nevertheless, the budget-maximization hypothesis standsin starkcontrastto theorieswhich suggest thatgovernmentgrowthis a reflectionof citizen preferences,andthe facts behindtwo centuriesof government growth align more closely with the hypothesis of budget-maximizing Leviathangovernmentthan with the hypothesis that the size of government is a reflectionof citizen preferences.Citizenpreferencesact to constraingovernment,but there is a path dependencyin governmentgrowth so that any static theory of governmentsize based on currentconditions leaves out an important partof the story. Both economic and political constraintscan limit the size of government. Economic constraintsrepresentan absolute limit to the amount of revenue the governmentcan extract.Revenue-maximizingtax rates depicted by the Laffercurve representan economic constraint.Political constraintscan keep limits andthroughpolitigovernmentsmallerthanthis throughconstitutional cal mechanisms,but as Peacock andWiseman(1961), BuchananandWagner (1977), Olson (1982), Higgs (1987), Holcombe (2002), and othershave sugare gested,once thesepoliticalconstraints relaxed,it is difficultto reimplement them.

10. The Role of Ideas

The closest the public choice literaturecomes to discussing the role of ideas on the size and growth of governmentis in models that depict government size as a function of citizen preferences. But those preferences tend to be modeled as static in nature(e.g., Republicansversus Democrats)ratherthan of being analyzedas a dynamicinteraction ideas andthe scope of government. (1936, p. 383) famouslysaidthat"theideas of economistsandpolitical Keynes philosophers,both when they are right and when they are wrong, are more powerful than is commonly understood.Indeed the world is ruled by little else." Perhapsthe growth of governmentin the twentieth century was due partly to the ideas of Marx (1906), and others. But in the early twenty-first centurythe ideas of Friedman(1962) and Hayek (1944) appearto have more popularsupportthanthose of Marx,and theirideas have been popularizedby Thatcher. politicians like RonaldReagan and Margaret Higgs (1987) attributesthe growth of twentiethcenturygovernmentto a changein ideology aroundthe beginningof thatcentury.If ideas have as much influenceas Keynes and Higgs suggest, the resurgenceof interestin classical

110
liberal ideas may point towarda retrenchment the size of governmentin in the twenty-firstcentury.But accordingto most of the theoriesreviewedhere, the appetiteof Leviathanis not swayed by ideas, and narrowspecial interests appearto dominatethe generalpublic interestin the political process.

11. Government Growthand PublicChoiceTheory

Not surprisingly, public choice literature governmentgrowthhas much the on in common with public choice scholarshipmore generally.Yet there are at least two facets of the literature governmentgrowththatdeviatesomewhat on from the medianof the literature,and those two facets might shed additional on light on otherpublic choice issues. First,a significantpartof the literature has focused on the coercive natureof governmentaction, governmentgrowth Leviathanrestrained varidepictinggovernmentas a revenue-maximizing by ous constitutionaland electoralconstraints.Following this line of reasoning, government'sactions are not the result of agreement,which is what the term public choice appearsto imply, but ratherof coercion. Some people are able to use the coercive power of governmentto force others to comply. Second, a significantpart of the literaturehas broken out of the comparativestatic muchof economics andmuchof publicchoice methodologythatcharacterizes to develop models of path dependency,where what happensin the presentis influencedby hisnotjust a functionof presentconditionsbutis substantially toricaldevelopments.If these theoriesare descriptiveof governmentgrowth, then they also must be descriptiveof the political process underlyinggovernon ment growth.Thus, this literature governmentgrowthsuggests thatpublic choice more generally could benefit from taking seriously the coercive nature of governmentand the path dependenciesthat may influence political action.

12. LookingAhead
Some of the factors underlyinggovernmentgrowth in the twentiethcentury have played themselves out, or even reversed.Income taxation, along with the institutionalfeaturesthat allow it to be collected - such as wage workers employed by corporationsand withholdingto extract taxes from those workers- helped governmentgrow in the twentiethcentury,but those past changes will not cause governmentto continue to grow in the twenty-first century.With the end of the Cold War,and with advancesin transportation and communicationtechnology, the world economy promises to be more open in the twenty-firstcenturythan in the twentieth,and resourcemobility takes away some of government'spower to tax. Of course, new wars, including the "Waron Terror", might intervene to restrictinternationalresource mobility.

111 What aboutthe huge unfundedliabilities - social securityand healthcare entitlements- thatregularlymake the news in the early twenty-firstcentury? If the size of government determined the amountof revenuesgovernment is by will can extractfromits citizens, thentheseprograms haveto be restructured to fit withinthe government's Thatconstraint budgetconstraint. maybe loosened somewhatthroughthe political process as the medianvoterages, butit seems recipientswill be able to extractenoughfromproductive unlikelythattransfer citizens to make good on all of government'scurrententitlementpromises. Some citizens may be disappointedwhen they get less from governmentthan they believe they were promised- but this would not be the first time that governmentdisappointedsomeone. The factors that propelled twentiethcentury governmentgrowth are unlikely to generate more growth in the twenty-firstcentury,but government growth is difficult to reverse. Taking all factors into account, the best forecast for the twenty-firstcentury may be for stabilized government,but not shrinkinggovernment.The era of big governmentwill remainwith us. Notes
1. PresidentClinton declared in his 1995 State of the Union address that "The era of big 27, governmentis over."He refinedthis idea furtherin a January 1996 radioaddress,when he said, "The era of big governmentis over, but we can't go back to a time when our citizens were just left to fend for themselves." 2. For example, Barlow (1970), Borcherding and Deacon (1972), and Bergstrom and Goodman(1973) are threepapersin topjournalswrittenaroundthe time of the publication of Niskanen's(1971) book thatused the medianvoter model to justify an assumptionthat the public sector produceswhat the medianvoterprefers. 3. As Buchananand Lee (1982) explain, revenue-maximizing behaviorby governmentmay level in the long run. actually set taxes higherthan the revenue-maximizing 4. For an interestingexchange on this topic, see Wittman(2002) and Niskanen (2002). 5. Even this rate was as high as it was in response to the crisis of the GreatDepression.The highest marginalincome tax rate was increasedfrom 25% to 63% (for incomes above $1 million) in 1932, and then increasedto 79% (for incomes above $5 million) in 1936. 6. These figures are from the US Census Bureau (US Departmentof Commerce, 1975, p. 1095). 7. This is discussed in more detail in Holcombe (2002, pp. 224-228). 8. For example, McChesney(1987, 1997) incorporates this element of compulsion into the rent-seekingliteratureto arguethatrent-seekingis even more perniciousthanpreviously recognized. 9. This was the famous thesis of Turner(1896). 10. These figuresare takenfrom Table 3 of Holcombe and Lacombe (2001, p. 187), adjusted to 1990 prices. 11. This story would have been differentif the US federal governmenthad been able to levy propertytaxes, but political institutionspreventedthat, allowing the local governments' tax bases to grow more rapidlythanthe federalgovernment's. 12. Niskanen(1975) has himself backedaway from the purebudget-maximization hypothesis that serves as the foundationof his 1971 book.

112

References
Barlow, R. (1970). Efficiency aspects of local school finance. Journal of Political Economy, 78(5), 1028-1040. Baumol, W. J. (1967). Macroeconomicsof unbalancedgrowth:The anatomyof urbancrisis. AmericanEconomicReview,57(3), 415-426. Becker, G. S. (1983). A theory of competitionamong pressuregroups for political influence. QuarterlyJournalof Economics, 98(3), 371-400. Journal Becker,G. S., & Mulligan,C. B. (2003). Deadweightcosts andthe size of government. of Law and Economics,46(2), 293-340. T. Bergstrom, C., & Goodman,R. (1973). Privatedemandforpublicgoods.AmericanEconomic Review,63(3), 280-296. T. Growth. Borcherding, E. (Ed.).(1977). BudgetsandBureaucrats:TheSourcesofGovernment Duke UniversityPress:Durham,NC. Borcherding,T. E., & Deacon, R. T. (1972). The demandfor services of non-federalgovernments.AmericanEconomicReview,62(5), 891-901. Brennan,G., & Buchanan,J. M. (1980). ThePower to Tax:AnalyticalFoundationsof a Fiscal Constitution.CambridgeUniversityPress:Cambridge. Brennan,G., & Lomasky,L. (1993). Democracyand Decision: ThePure Theoryof Electoral Preference.CambridgeUniversityPress:Cambridge. Buchanan,J. M., & Lee, D. R. (1982). Politics, time, and the laffercurve.Journalof Political Economy,90(4), 816-819. Buchanan,J. M., & Wagner,R. E. (1977). Democracyin Deficit: ThePolitical Legacy of Lord Keynes.Academic Press:New York. Bueno de Mesquita, B., Smith, A., Siverson, R. M., & Morrow,J. D. (2003). The Logic of Political Survival.MIT Press:Cambridge,MA. and of Caplan,B. (2001). Rationalirrationality the microfoundations political failure.Public Choice, 107(3-4), 311-331. Caplan,B. (2003). The logic of collective belief. Rationalityand Society, 15(2), 218-242. Diamond, P. A., & Mirrlees,J. A. (1971). Optimaltaxationand public production:I and II. AmericanEconomicReview,81(March),8-27, (June),261-278. Downs, A. (1960). Why the governmentbudget is too small in a democracy.WorldPolitics, 12(4), 541-564. Edelman,M. (1964). TheSymbolicUses of Politics. Universityof Illinois Press: Urbana. Friedman,D. (1977). A theoryof the size and shape of nations.Journalof Political Economy, 85(1), 59-77. Friedman,M. (1962). Capitalismand Fredom.Universityof Chicago Press: Chicago. PoliticalRightsand CivilLiberties,1978. Freedom Gastil,R. D. (1978). Freedomin the World: House: New York. Gwartney,J., & Lawson, R. (2003). Economic Freedom of the World,2003 Annual Report. FraserInstitute:Vancouver,B.C. Hayek, F. A. (1944). TheRoad to Serfdom.Universityof Chicago Press:Chicago. Higgs, R. (1987). Crisis and Leviathan:CriticalEpisodes in the Growthof AmericanGovernment. OxfordUniversityPress:Oxfordand New York. Higgs, R. (1991). Eighteen problematicpropositionsin the analysis of the growthof government. ReviewofAustrian Economics, 5(1), 3-40. Holcombe, R. G. (1989). The median voter model in public choice theory. Public Choice, 61(2), 115-125. Holcombe, R. G. (1993). Are there ratchetsin the growth of federal governmentspending? Public Finance Quarterly,21(1), 33-47.

113
Holcombe, R. G. (1994). The Economic Foundationsof Government.New York University Press:New York. Holcombe, R. G. (1996). The growth of the federal governmentin the 1920s. Cato Journal, 16(2), 175-199. Holcombe, R. G. (2002). FromLibertyto Democracy: The Transformation American Govof ernment.Universityof MichiganPress:Ann Arbor. Holcombe, R. G., & Lacombe, D. J. (2001). The growth of local governmentin the united states from 1820 to 1870. Journalof EconomicHistory, 61(1), 184-189. Holcombe,R. G., & Lacombe,D. J. (2004). Factorsunderlyingthe growthof local government in the 19thcenturyunited states. Public Choice, 120(3-4), 359-377. Holcombe,R. G., & Mills, J. A. (1995). Politics anddeficit finance.Public Finance Quarterly, 23(4), 448-466. Kahneman,D. (2003). Maps of bounded rationality:Psychology for behavioraleconomics. AmericanEconomicReview,93(5), 1449-1475. Kahneman, Knetsch,J. L., & Thaler,R. H. (1991). Anomalies:The endowmenteffect, loss D., aversion,and statusquo bias. Journalof EconomicPerspectives,5(1), 193-206. Kau, J. B., & Rubin,P. H. (1981). The size of government.Public Choice, 37(2), 261-274. Keynes, J. M. (1936). The General Theory of Employment,Interest,and Money. Harcourt, Brace & Company:New York. Levi, M. (1988). Of Rule and Revenue.Universityof CaliforniaPress: Berkeley. Lott, J. R., Jr., & Kenny,L. W. (1999). Did women's suffragechange the size and scope of Journalof Political Economy,107(6, part 1), 1163-1198. government? K. (1906). Capital. CharlesH. Kerr& Company:Chicago. Marx, McChesney,F. S. (1987). Rent seeking and rentcreationin the economic theoryof regulation. Journalof Legal Studies, 16(1), 101-118. McChesney,F S. (1997). Moneyfor Nothing: Politicians, Rent Extraction,and Political Extortion.Harvard UniversityPress:Cambridge,MA. A. H., & Richard,S. F. (1981). A rationaltheoryof the size of government.Journal Meltzer, of Political Economy,89(5), 914-927. Mirrlees,J. A. (1971). An explorationin the theory of optimumincome taxation.Review of EconomicStudies,38(2), 175-208. and the size of government.In Winer Mueller, D. C. (2002). Interestgroups, redistribution S. L. & ShibataH. (Eds.), Political Economy and Public Finance: The Role of Political Economy in the Theoryand Practice of Public Economics (pp. 123-144). EdwardElgar: Cheltenham,UK and Northampton, MA, USA. Niskanen, W. A. (1971). Bureaucracy and RepresentativeGovernment.Aldine-Atherton: Chicago. and Niskanen, W. A. (1975). Bureaucrats politicians.Journal of Law and Economics, 18(2), 617-643. Niskanen, W. A. (2002). On the origin and identificationof governmentfailures. In Winer S. L. & ShibataH. (Eds.), Political Economy and Public Finance: The Role of Political Economy in the Theoryand Practice of Public Economics (pp. 111-116). EdwardElgar: Cheltenham,UK and Northampton, MA, USA. Olson, M., Jr.(1982). TheRise and Decline ofNations. YaleUniversityPress:New Haven,CT. in Peacock, A. T., & Wiseman,J. (1961). The Growthof Government Expenditures the United Kingdom.PrincetonUniversityPress:Princeton,NJ. Peltzman,S. (1980). The growthof government.Journalof Law and Economics,23(2), 209287. Persson, T., & Tabellini,G. (1999). The size and scope of government:Comparative politics with rationalpoliticians.EuropeanEconomicReview,43(4-6), 699-735.

114
MITPress:Cambridge, Persson,T., & Tabellini,G. (2003). EconomicEffectsof Constitutions. MA. Posner, R. A. (1971). Taxationby regulation.Bell Journal of Economics and Management Science, 2(1), 22-50. Rassler,K. A., & Thompson,W. R. (1985). Warmakingandstatemaking:Government expenditures,tax revenues,andglobal wars.AmericanPolitical Science Review,79(2), 491-507. Romer,T., & Rosenthal,H. (1978). Political resourceallocation,controlledagendas,and the statusquo. Public Choice, 33(4), 27-43. Tullock, G. (1965). Entrybarriersin politics. AmericanEconomicReview,55(2), 458-466. Tullock, G. (1982). Why so much stability?Public Choice, 37(2), 189-202. F. Turner, J. (1896). The problemof the west. AtlanticMonthly,78(September),289-297. of US Department Commerce,Bureauof the Census.(1975). HistoricalStatisticsof the United Statesfrom Colonial Timesto 1970. Bureauof the Census:Washington,DC. Usher, D. (1992). The WelfareEconomics of Markets, Voting,and Predation. University of MichiganPress:Ann Arbor. Wittman,D. A. (1989). Why democraciesproduceefficientresults.Journalof Political Economy, 97(6), 1395-1424. D. Wittman, A. (1995). TheMythofDemocratic Failure.Universityof ChicagoPress:Chicago. Wittman,D. A. (2002). Normativepublicfinancewithoutguilt:Why normative public finance is positivepublicfinance.In WinerS. L. & ShibataH. (Eds.), Political Economyand Public Finance: The Role of Political Economyin the Theoryand Practice of Public Economics MA, USA. (pp. 101-110). EdwardElgar:Cheltenham,UK and Northampton, L. B. (2001). Ethics as a Social Science. Edward Elgar: Cheltenham, UK and Yeager, MA, USA. Northampton,