Fiscal Policy Cycles and Public Expenditure in Developing Countries Author(s): Ludger Schuknecht Source: Public Choice, Vol.

102, No. 1/2 (2000), pp. 115-130 Published by: Springer Stable URL: http://www.jstor.org/stable/30026139 . Accessed: 01/06/2011 05:50
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=springer. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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 102: 115-130, 2000. 2000 KluwerAcademicPublishers. Printed in the Netherlands.

115

Fiscal policy cycles and public expenditure in developing countries *

LUDGERSCHUKNECHT
Rue de Lausanne154, CH-1206 Geneva,Switzerland;e-mail WTO, ludgerSchuknecht @wto.org Accepted 26 May 1998 Abstract. The paper studiesempiricallythe fiscal policy instruments which governments by try to influence election outcomes in 24 developing countriesfor the 1973-1992 period. The fiscal policies aroundelections is increasing study findsthatthe mainvehicle for expansionary public expenditureratherthanlowering taxes, and public investmentcycles seem particularly prominent.Institutionalmechanisms which constraindiscretionaryexpenditurepolicies and which strengthenfiscal control are thereforeworthwhileconsideringto preventopportunistic policy makingaroundelections.

1. Introduction Numerousstudies have analyzedwhetherdemocraticgovernmentsin industrial countries adjust their macroeconomicpolicy mix aroundthe election data to enhance their re-electionprospects.Very few studies, however,have examined the incidence of policy cycles in developing countries, or have looked in moredetail at the fiscal policy instruments with which governments try to influencethe election outcome.' The purposeof this paperis to look in detail at the fiscal policy instruments used by developing country governmentsto enhance their prospects of reelection. The study is the first to look empirically at the choice of policy instrumentsfor a relativelylarge sample of developing countries(i.e., 24 in total). It argues that public spending increases are the preferredvehicle for beforeelections becausethey typically policy makersto boost theirpopularity have a very direct and immediate impact on voters' welfare. The findings largely supportthis hypothesis,especially for public investmentcycles.

* Comments from Ke-young Chu, Peter Moser, and an anonymous referee are highly The views expressedare the authorsand not necessarilythose of the WTO. appreciated.

116 2. Fiscal policy cycles and public expenditure policies in the context of elections 2.1. The underlyingmodel and the literature The study applies the so-called Nordhaus-approach analyze fiscal policies to around elections. According to this approach, governments are assumed to stimulatetheir economies with expansionaryfiscal or monetarypolicies before elections. Resulting employment gains or wealth transfersincrease the governments'popularity.After the elections, governmentsstabilize the economy with restrictivepolicies again. The earlier literatureinitiated by Nordhaus(1975) assumedadaptiveexpectations.More recently,Perssonand Tabelini(1990) showedthatNordhaus-type cycles can also emergein models with rationalexpectations.Votersareassumednot to know the policy makers' and competence(asymmetricinformation), they have to infer it from observable economic data. The governmentthen pursues expansionarymonetary policies before elections to raise outputand lower unemployment,which in turn signals to voters (who have no other source of information)that the governmentis very competent.Restrictivepolicies afterthe election then lead to a full cycle in economic activity.2Rogoff (1990) designs a model in which information aboutthe government's asymmetric competencecan lead to fiscal and monetarypolicy cycles. The other main strandof the literaturelooks at so-called Partisancycles which resultfrom ideological differencesin the preferencesfor inflationand between political parties(Hibbs, 1977). Left-wingpartiesasunemployment sign greatervalue to low unemployment,while right-wingpartiesvalue low inflationmore highly. As a result, these models predictthatright-winggovernmentsconsistentlygeneratelower inflationandhigherunemployment than do leftist governments.In more recent models, cycles are caused by longterm labor contractswhich preventan immediateadjustmentof real wages to an unanticipatedmonetaryexpansion or contractiondue to a change in governmentafterelections (Alesina, 1987). Empiricalstudies in the 1970s until the early 1990s focussed almost exclusively on industrialcountries,and there are by now numeroustheoretical More recently,however,interest and empiricalstudies of both approaches.3 has been extended also to developing countries. Schuknecht(1996) finds fiscal policy cycles of the "Nordhaus"-type a panel of 35 developing for countrieswhere governmentspursueexpansionarypolicies before elections and fiscal austerityafterwards. Regardingpolicy instruments,industrialcountry studies mainly look at transferpayments. In developing countries, by contrast, evidence is very scarceandlargelyanecdotal.Bates (1988) discussespublicinvestmentcycles

117 in Zambia in the 1960s and Krueger and Turan (1993) find such cycles in Turkey for the 1950 to 1980 period. Calvo (1995) looks at the Mexican crisis, and argues that this was triggered partly by a significant increase in the quasifiscal deficit associated with the extension of credit through development banks before the elections in late 1994. Regarding the two main approaches discussed above, the Nordhausapproach seems more suitable for developing countries. The distinction between political parties frequently do not exhibit the typical Western left-right pattern which is crucial for applying the Partisan-approach. The Nordhaus-approach is also supported by the above-mentioned studies by Bates, Calvo, Krueger and Turan, or Schuknecht. 2.2. Hypotheses on fiscal policies and public expenditure Fiscal policies can be an effective means of influencing government popularity. Expansionary fiscal policies allow significant wealth transfers in order to gain votes. Increases in cash transfers to households, employment and profit opportunities from public investment projects, reductions in tax rates and the delayed collection of taxes are examples of such wealth transfers. They either increase public expenditure or reduce revenue. In both cases, we would expect an increase in the fiscal deficit before elections and fiscal consolidation thereafter. In addition, we would expect that much of the election-oriented policy making in developing countries affects the expenditure side: Hypothesis 1. Elections are preceded by expenditure increases rather than lower revenues Tufte (1978) argues that "[measures] must be easy to start up quickly and must yield clear and immediate economic benefits to a large number of voters". In industrial countries with a broad tax base, tax cuts can significantly enhance the government's popularity. They are highly visible and the benefits can accrue very quickly. This was the case, for example, in the United States during the Reagan era. In developing countries, however, the tax base, particularly for income taxes, is small, and tax cuts would therefore not enhance broad government support. In addition, the indirect effect of a tax cut on economic performance is not very direct, predictable,

and immediate, which makes it difficult for the government to reap the
political benefits from the resulting economic stimulation. Tax-related measures also result in long-term problems such as an eroded tax base, and tax morale may be difficult to reconstruct after elections against the resistance of special interests.

118 Expenditurepolicies can be an effective instrumentfor governmentsto increase their popularity,for example, through the distributionof free or subsidized food, or through temporaryemployment generation in public works programs.The effect of expenditureincreaseson employmentcan be easily observed and governmentscan claim direct credit for it. The same holds for increases in disposableincomes throughcash or in-kindtransfers. For these reasons, expenditureincreases are likely to be more important in pre-election fiscal expansion than tax reductions.However, expenditure increases can also have importantdraw-backsfor governments.Similar to tax-related measures, expenditureincreases can be difficult to reverse if permanentcivil service posts and entitlementsratherthan temporarypublic works or transfersare introduced. Hypothesis 2. Election-orientedpolicies raise capital expenditures more stronglythancurrentexpenditure The governmentcan change not only the level of expendituresbut also the composition. Increasing current expenditures,such as food subsidies or increasedgovernmentemploymentis frequentlyreportedto increase the welfare of voters and the popularityof governmentsbefore elections. Higher on expenditure capitalprojects,for example,throughpublic works programs has the same effects. Therefore we expect spending on both expenditure categoriesto increase. However, there is reason to argue that capital expenditureis likely to be more strongly affected. Public investmentspendingthrough,e.g., public re-election can worksprograms be fine-tunedmoreeasily to the government's objectives. Public works programs(such as improvinga road) are relatively easy to start up or acceleratebefore an election, and they can be set up in well-defined geographicalareas and sectors. They are also easy to stop or slow-downafterthe election is over.Governments may thereforebe reluctant to engage in commitmentswhich are more difficultto reverseif alternative spendingprogramscan be fine-tunedto its political objectives more easily. The studiesby Bates or KruegerandTuranshow thatpublicinvestment cycles have indeed been very popularin a numberof developingcountries. reason why we may not observe the full There may also be "statistical" of increasein currentexpenditure beforeelections. The distribution free food sources of the in urbanareasor villages is often financedby extra-budgetary governingparties.

119 Hypothesis3. The governmentwage bill is likely to increasebefore elections The limited availability of data restricts the more detailed analyses of the policy instruments by which expansionary expenditure policies are transmitted.We would expect that expenditureon transfers are strongly affectedby elections but a reasonabledataset on this expenditure categoryis not available.The government wage bill is one detailedeconomic expenditure category on which most countriesreportto the IMF's GovernmentFinance Statistics (1970-95). As mentioned, increasing government employment before elections is very visible and can have a direct effect on the wealth of many people as in many developing countries a large numberof family membersis typically supported throughone civil service post. Within the public sector work force, there are typically two types of employment. While governmentswill be happy to expand the number of contractsbefore elections, they may be temporaryemployees on short-term more reluctantto take on significant numbersof new (typically life-time) civil servants.Ideally,therefore,the government wage bill shouldbe analyzed for these sub-categories,but such detailed data are not available. separately versus permanent Althoughthe relativeimportanceof temporary employees shareof the wage bill is on permanent differs,the predominant civil servants, and we would expect only moderateelection-cycles in the wage bill.

3. Empirical study 3.1. Methodology 3.1.1. Technique Regressionswith annualpanel datafor the period 1973-1992 arerunon a 24countrysample with a fixed-effectsmodel. Unit root tests were conductedon all independentvariablesand the occurrenceof cointegration be rejected can at the 95% level for almost all variables.The following panel regressionis tested:
FisVarit = aConstit + /Electit

+ YjitVarjit ei +

where "FisVarit"standsfor variousdependentvariablesfor i countriesand t periods, "Elect"representsthe variablecapturingthe influence of elections, and "Varj"stands for the impact of other factors such as terms of trade of changes, the trade-orientedness a country,the real effective exchangerate, the exchangeregime, catastrophes IMF-supported or programs.

120 3.1.2. Dependentvariables In a first estimation, the overall fiscal balance of central government,expressed as a share of GDP, serves as dependent variable for the analysis of overall fiscal policies aroundelections as it reflects the combined effect of expenditureand revenue measures.4In a second step, the hypotheses on public revenueand expenditureare tested with the help of total revenueand as expenditure a shareof GDP as dependentvariables.Finally,the hypotheses on the behaviorof certainexpenditure policy variablesareexaminedwith the and as help of current capitalexpenditure a shareof GDP andthe publicsector as wage bill as a shareof total expenditures dependentvariables. 3.1.3. Election variable Most importantfor our analysis, a dummy variableis introducedto reflect the effect of elections. As mentioned,we only considercountry-widegeneral, legislative or presidentialelections, dependingon the political system of the sample countries,as indicatingin the EuropaWorldYearbook(1995). The election variable takes the value of one in the period when expansionary policies are expected and minus one when we anticipatethe post election contraction.In all otherperiodsit is set as zero. In many countriesthe fiscal year and the calendaryear do not coincide. Therefore,the variablevalue is set relative to its position during the fiscal year to be consistentwith the other fiscal variablesit supposedlyinfluences. Elections sometimes take place at the beginning,duringor at the end of the fiscal or calendaryear.Therefore,the periodduringwhich we expect expansionarypolicies and duringwhich a contractionis expectedmust be defined, takinginto accountthe fact thatmost stimuliaffect the economy and thereby popularityonly with a shortlag. in The specificationis based on the following argumentation: years when the election is during the first two months of the fiscal year, expansionary policies are expected for the fiscal year precedingthe election. Because the elections are very early in the fiscal year, the contractionstartsalreadylater during the election year; hence, the variabletakes the value of one for the pre-electionfiscal year,and minus one for the election year. If the election is in the thirdor fourthmonthof the fiscal year,expansionary policies are expectedto startduringthe fiscal year precedingthe election. They will continue in the election year, but more or less at the same level. the Consequently, election variabletakesthe value of one for the pre-election measfiscal year and zero for the fiscal year with the election. Contractionary for which the ures are only introducedin the fiscal year after the election, variabletakes the value of minus one. This specificationtakes into account the fact thatgovernmentsafterelections have to rewardtheir"supporters".

121 If the election takes place duringor afterthe fifth monthof the fiscal year, all expansionarypolicies are expected to take place duringthe election year. There is also some time left after the election to reward "friends".Fiscal consolidationthen startswith the post-electionfiscal year. The coefficient of this variable is expected to have a negative sign in estimations of the fiscal balance, and fiscal revenue and a positive sign in all estimations of expenditurevariables, because elections shall lead to a worseningof the fiscal balance,lower revenueand higherpublic expenditure. 3.1.4. Otherindependent variables 3.1.4.1. Lagged dependentvariable. All estimationsinclude a lagged deadminispendentvariablewith an expectedpositive coefficient.Government trationsare constrained budgetsand the current by budgetlargelydetermines the next period's appropriations (Niskanen, 1971). This "inertia"provides fiscal revenue,spending,and deficit patterns. stabilityand predetermines 3.1.4.2. Trade-orientation.A variable which reflects a country's tradeorientation definedas the ratioof the sum of importsandexportsover GDP. is It is expectedto be correlated with an improvedfiscal balance.Expansionary fiscal policies in trade-oriented economies lead to more leakage of demand abroadthan in less trade-oriented economies. The higher importdemand,in can eitherresult in externalpaymentdifficultiesin countrieswith fixed turn, exchange regimes or in an exchange rate devaluationand importedinflation in countries with flexible exchange regimes. Both outcomes - balance of payment problems or inflation - could be very damaging for government popularitybefore elections. The higherlikelihood of such outcomes in more trade-oriented countriesmake expansionarypolicies and high fiscal deficits before elections less attractive (Lindbeck,1976). 3.1.4.3. Terms trade. Effects of changes in the termsof tradeare also exof amined.If the externalshock is positive (i.e., countriesexperiencean increase in the relativeprice of exportables),outputis likely to increasewhich should raise fiscal revenue and improvethe fiscal balance. Declines in the terms of trade,on the otherhand,shouldlower revenueand worsenthe fiscal balance. This translatesinto an expected positive sign of the coefficient for the fiscal balanceand governmentrevenue. The effect of changes in the terms of trade on public expenditureis not quite as unambiguous.Improvingterms of trade and the resulting relative decline in the price of inputs leads to lower public expenditure,especially if importsare a large share of expenditure.This is probablythe case in many developing countries.The output growth resulting from improvedterms of trademay also reducethe need for social assistancefrom governmentwhich,

122 in turn, should lower currentexpenditure.A decline in the terms of trade, on the other hand, could requirehigher currentexpenditureis, for example, public enterprisesare not allowed to adjusttheirpricing policies to changes in export and import prices and require more support,or if social assistance needs rise. These argumentssuggest lower spending and deficits are correlatedwith positive termsof tradedevelopments. However,the growthin revenuefrom a positive shock may also soften the government'sbudgetconstraintand allow the governmentto increasepublic spending,especially if the shock is of considerablemagnitudeand duration. In fact, Bevan, Collier, and Gunning (1990) and Schuknecht(1997) report thatthe coffee andcocoa boom initiallyreducedfiscal deficitsin severalLatin Americanand African countriesin the late 1970s. Over time, however, and well before the end of the boom, expenditureobligations startedto increase - often more rapidlythan revenue- in many of the examined countries.In variableis likely to have a significant therefore,the terms-of-trade summary, positive effect on revenue and a negative effect on currentexpenditure,but the aggregateeffect on expenditureand the fiscal balanceis more uncertain. 3.1.4.4. Real effectiveexchangerate. The real effective exchangerateis an When the real effective exchange indicatorof a country'scompetitiveness.5 rate depreciatesand the economy's competitivenessimproves, government revenuesfrom profits and export taxes rise. A negative sign of this variable can, therefore,be expected for revenues. 3.1.4.5. Exchangeregime. An exchangeregimevariableis derivedfromthe Financial Statistics (IFS) (IMF, 1970-95). A dummy variable International takes the value of one for a relatively fixed exchange regime and 0 otherwise. Fixed exchange rates frequentlyresult in periods of overvaluationof the domestic currencywhich, in turn,depressesdomestic economic activity and the competitivenessof producers.Governmentsmay respond in many ways, but from a fiscal perspective,the effect of an overvaluedexchangerate are on the governmentwage bill is likely to be most important. Governments which results from the recesfrequentlyexpected to mop up unemployment sionaryeffect of an overvaluedexchange rate. Fixed exchange rate regimes then are likely to resultin higherpublic sector wage bills, and a positive sign of the coefficientcan be expected in the estimationof the wage bill. The aggregateeffect on overallexpenditure (andthe fiscal balance)is less obvious as fixed and overvaluedexchange rates may raise wage costs but lower the costs of importedproducts. 3.1.4.6. Catastrophes. Shocks are incorporatedas dummy variables taking the value of one during periods of major catastrophes,such as floods,

123 or earthquakes, the eruptionof volcanos, as indicatedin the EuropaWorld Yearbook(1995).6 If the governmentcontributesto catastrophe-relief reor constructionmeasures throughthe budget, governmentexpenditureshould rise which worsens the fiscal balance. The variable is expected to have a negative coefficient in estimationsof the fiscal balance and a positive sign for expenditure. 3.1.4.7. IMF-supported programs. An analysis of fiscal policy in developing countries also needs to take into considerationprogramswith international financial institutionslike the International MonetaryFund (IMF). These institutionsusually provideaccess to more international financingand containconditionalitywhich stresseseconomic stabilization,includingfiscal consolidation.Programssupportedby the IMF should, therefore,"harden" the governments'budgetconstraint resultin smallerfiscal deficits,higher and revenueand lower public expenditure.7 The estimations thereforealso include three types of programssupported by the IMF. Dummy variablesstand for structural adjustmentprograms and (SAF/ESAF),stand-by(SB) arrangements, extendedfund facility (EFF) The variablestake the value of one when the respectiveproarrangements. gramis in force for at least six monthsof the respectivefiscal year.It becomes one half when the particular of programcovers at least a full quarter the fiscal SAF/ESAF arrangements assumed to have a lagged effect on the are year. fiscal balance, because structuralmeasures take some time to be effective and may initially even result in net costs for the budget which is then financed with additionaldonor assistance.For these countrieswith SAF/ESAF it arrangements is in fact tested whetherthe fiscal situationimprovesin the second year of the program.A "generalprogram"dummy variablereflects all periodswhen a countryhad any kind of these threearrangements with the IMF.For all IMF-program variables,the expectedeffect on the fiscal balance, and revenuesis positive while spendingon the variouscategoriesis expected to decline. 3.2. Results The results largely confirmthe hypotheses on the fiscal policy instruments by which governmentstry to influencetheirpopularityaroundelections, and evidence is strongestin supportof public investmentcycles. The results also indicate that a numberof other domestic and externalfactors show the predicted effect on fiscal policies. However,for some variables- for which the aggregateeffects were sometimescomplex and difficultto predictin the first place - we could only find weak if any support.

124 Hypothesis 1 on the relative importance of revenue versus expenditure measures to enhance popularitybefore elections can not be rejected. Column 1 in Table 1 indicates that most governmentsuse fiscal policies to enhance their popularitybefore elections. Columns 2 and 3 suggest that, on balance, expendituremeasuresare more important achieving this obfor jective. Two thirds of the increase in fiscal deficits of about 0.7% of GDP aroundelections is explainedby an increasein public expenditure.The fact that the coefficient is only marginallysignificantat the 10%level, however, suggests that there is considerablevariationin the expenditurepolicies of the sample countries.Furthermore, some governmentsmay also have used at times, but the coefficient of the election variablein the revenue policies estimationand estimationof total revenueis smallerthanin the expenditure it is not statisticallysignificant. The resultsfor otherindependent variablesare also worthwhilereviewing. The role of the lagged dependentvariableas reflectingbudgetaryinertia is confirmed.The effect of catastrophesand trade-orientedness the overall on deficit is more significant than on its components, i.e., revenue and exin penditure.The estimationresultsalso suggest thatimprovements the fiscal balance throughpositive developmentsof the terms of tradearise from revenue increases. Loss of competitivenessby producers,as measuredby the real effective exchange rate, has a significant adverse effect on total revEFF programsare shown to have a enue. AmongstIMF-supported programs, effect on public expenditure. significantconstraining Regardingthe second hypothesis on the relativedevelopmentof current and capital expenditurearoundelections, the estimationresults supportthe claim of public investmentcycles (Table 2). Governmentsexpand capital spendingbefore elections, e.g., on public works programsto enhance their political support.The coefficientsof the two estimationsin columns 1 and 2 accountfor an approximately show thatcurrentandcapitalexpenditure equal share of the overall expenditureincrease but the significance level is only As satisfactoryfor capitalexpenditure. predictedby the thirdhypothesis,the increasein currentexpenditureseems to be at least partlydue to an increase in the wage bill butthe significanceof the coefficientof the wage bill variable is not very satisfactory. affect current Otherfindingsinclude thatnaturalcatastrophes expenditure on emergency relief. Improvesignificantly,probablythroughexpenditure ments in the terms of trade correlates with lower currentexpenditure,as social assistanceexpenditure costs of importsdecline. The same variable, and and however,does not show a significanteffect on capitalexpenditure overall expenditure,probablyas higher revenue and "softer"budget constraintsalto commitments.As projected,the low governments take on new expenditure

125
Table1. Estimationresults, 1973-1992 period Dependentvariables Overall fiscal balance variablesa Independent Elections Total revenue Total expenditure

-0.66 (-2.85)**

-0.23 (-1.15)

0.39 (1.78)

Otherinfluences Lagged dependentvariable Naturalcatastrophes Termsof trade Termsof trade(one period lag) Termsof trade(two periods lag) Trade-orientednessb Real effective exchangerate SAF/ESAF arrangement (one period lag)b EFF arrangement Stand-byarrangement 4.86 (1.99) 0.68 (1.25) 0.31 (0.88) Numberof observations R2 adjusted 376 0.62 0.76 (24.1)** -7.19 (-4.74)** 0.015 (2.00)* -0.006 (-0.68) -0.009 (-1.34) 0.639 (1.74) 1.65 (1.09) -0.13 (-1.73) 0.69 (0.37) 0.28 (0.72) 0.16 (0.33) 246 0.96 379 0.91 1.96 (1.50) 0.70 (16.4)** -0.95 (-0.85) 0.02 (3.33)** 0.64 (15.4)** 2.41 (1.51) 0.01 (1.12)

-0.02 (-1.12) -1.08 (-1.95)

Sourceof data IMF: Government Finance Statistics(1970-95), Transactionsof the Fund (1995), International FinancialStatistics (1970-95), Directionsof TradeStatistics(1995); Worldbank, WorldTables; EuropaWorldYearbook(1995). * = significantat 5% level; ** = significantat 1%level. alndividualcountryinterceptsnot presented. statisticsjust below criticalvalue of 95% significance. bDickey-Fuller-test

126
Table2. Estimationresults, 1973-1992 period Dependentvariable Current expenditure variablesa Independent Elections Capital expenditure Wagebill

0.24 (1.49)

0.25 (2.44)*

0.56 (1.54)

Otherinfluences Lagged dependentvariable Naturalcatastrophes Termsof trade Trade-orientednessb Exchangearrangements Fund-supported program -0.78 (-0.28) Numberof observations R2 adjusted 350 0.92 -0.20 (-1.18) 379 0.81 0.72 0.61 0.83

(18.7)**
2.74

(14.0)*
-0.92

(28.2)**

(2.02)*
-0.01 (-2.46)* 0.50

(-1.07)
0.04 (1.51) 0.59

(0.52)

(0.98)
2.24

(2.70)**
-0.40 (-0.66) 359 0.90

Source of data IMF: GovernmentFinance Statistics (1970-95), Transactionsof the Fund (1995), International Financial Statistics (1970-95), Directions of Trade Statistics (1995); Worldbank, WorldTables;EuropaWorldYearbook(1995). * = significantat 5% level; ** = significantat 1%level. aIndividual countryinterceptsnot presented. statisticsjust below criticalvalue of 95% significance. bDickey-Fuller-test

coefficient of the exchange arrangement variableshows that fixed exchange to regimes contribute a rising public sector wage bill. These resultsraise a numberof questionsfor further researchwhich could also lead to interestingpolicy implications.A numberof authors(Buchanan in many of his writings, and, more recently Von Hagen and Harden, 1994; Alesina, Hausmann,Hommes, and Stein, 1995, Camposand Pradhan,1996; or Tanziand Schuknecht,1997) have arguedthatthe institutionalframework

127 is key to maintainingfiscal discipline in many industrialand LatinAmerican countrieswhich have featuredweak fiscal institutionsin the past. Emphasis on macroeconomicdiscipline in the budget formulationstage, for example, througha strong Ministry of Finance can help to contain fiscal expansion. Constraintson the legislaturein introducingnew spendingproposals in the budget approvalstage are a second importantelement for maintainingfiscal of discipline.Third,strictimplementation the budgetis requiredto eliminate for overspendingat this stage, which could be abused,for exopportunities ample, for election-orientedspending.Fiscal rules such as constitutionalor legislative deficit limits can complementthe otherfiscal institutions. A number of the sample countries of this study belong to the group of countries with relatively weak fiscal institutions studied by the abovementioned authors but for many of the countries, the quality of fiscal institutions has not yet been established. As weak institutions may have facilitatedelection-orientedfiscal policy making, it would be interestingto see whetherfiscal cycles are correlatedwith the qualityof fiscal institutions. From a normativeperspective,institutionalmechanismswhich are adapted to each country'sinstitutionalframeworkand which constraindiscretionary government policies, mightbe worthwhileconsideringif expansionary policy aroundelections is perceivedto be undesirable.8 making

4. Conclusion The study of fiscal policy cycles in 24 developing countries for the 197392 period supportsthe claim that election-orientedfiscal policy making in developing countriesmainly affects public expenditure.Empiricalevidence stronglysuggests the prevalenceof public investmentcycles. The weak institutionalstructuresin many developing countries may facilitate election-orientedfiscal policies. It is possible for this reason that the empiricalstudies on industrialcountries (where political discretionis much lower than in developing countries)tend to show less significantresults for Nordhaus-type cycles thanthe studies of developingcountries.The strengthening of fiscal rules and institutionscould be a key element towardsreducing policy volatility,for example, aroundelections.

Notes
1. For surveys of the political business cycles literaturesee, for example, Alesina (1988), Nordhaus (1989), Willet (1989), Gaertner(1993, 1994), or Frey (1996). Schuknecht (1996) analyses fiscal policy cycles and their effect on deficits in developing countries.

128
counEmpiricalstudies of the choice of policy instrumentsare limited to industrialized tries (e.g., Tufte, 1978; or Frey and Schneider, 1978); Bates (1988) or Krueger and Turan(1993) provide case studies of public investmentcycles in Zambia and Turkey, on respectively.There is also a considerableliterature the political economy of developing countries which discuss the importanceof the policy regime (e.g., Krueger,1992), micro- and macroeconomicpolicy failures (e.g., Krueger, 1993), domestic institutional factors (e.g., Borner,Brunetti, and Weder, 1996) or internationalinfluences (Frey and Eichenberger,1994). While the "traditional" is Nordhaus-approach often criticizedfor its reliance on adaptive model hinges crucially on the assumptionthat voters expectations,the Persson/Tabellini can be influencedrepeatedly, thatdo not learnfrompast experiencesanddo not adjust i.e., theirbehaviorover subsequentbusiness cycles. For details, see the above-mentionedsurveys. The most detailed and institutionallyrich study on political cycles in the United States and in otherdemocraciesis Tufte (1978). A combinationof the two principalapproachesis analyzedby Frey and Schneider(1978). This estimationin fact reiteratesthe findingsby Schuknecht(1996) on overall fiscal deficits. The datadoes not include spendingby regions or local governments,since the study analyses election-orientedpolicies at the national level, and the governmentonly controls directlythe centralgovernmentbudget.The definitionof the fiscal balanceexcludes from abroad,therebyexcluding"noise"fromfluctuations grants,i.e., foreign aid transfers in grantsand reflectingmore clearly changes in the fiscal position arisingfrom domestic policy choices only. The real effective exchange rate representsthe ratio of the averageexchange rate of the country in question to a weighted geometric average of exchange rates for the currencies of selected other countries,adjustedfor differencesin the relativeinflationbetween these countries.There could be correlationbetween the terms of trade variable and the real effective exchange rate variableas terms of tradedevelopmentsare to some extent correlatedwith inflationor exchange rate changes. Here, the correlationcoefficient for these two variablesof 0.23, however,is rathersmall. Thereare no samplecountrieswhich experiencedmajorwarsover the observationperiod, except maybe for Guatemalawhich experiencedprolongedunrest. financialinstitutionsrelieve inIt has been arguedthat programswith the international dustrialcountry governmentsof the unpleasanttask to impose policy conditionalityon developing countrieswhich might be unpopularwith their domestic electorate (Vaubel, 1986). Developing country governments,by the same token, can attributethe negative effect of austerity measures to the internationalfinancial institutions and protect their political support (Frey and Eichenberger,1994). The importanceof revenue and expendituremeasuresin IMF-supported programhas been assessed in IMF (1997). The above-mentionedstudies provide some evidence that institutionalproblems are on average more severe in developing than in industrialcountries.While developing countries can not always copy institutionswhich have evolved in differentcontexts, they can strengthentheir institutionsif they follow the above-mentioned principlesin theirreform efforts.

2.

3.

4.

5.

6. 7.

8.

129 References

Alesina, A. (1987). Macroeconomicpolicy in a two-partsystem as a repeatedgame. Quarterly Journalof Economics 102: 641-678. Alesina, A. (1988). Macroeconomics and politics. NBER MacroeconomicsAnnual, 13-52. MIT Press. Cambridge: Alesina, A., Hausmann,R., Hommes, R. and Stein, E. (1995). Budget institutionsand fiscal in performance LatinAmerica.Mimeo. marketsin Africa. In R. Bates (Ed.), Toward Bates, R. (1988). Governmentsand agricultural a political economy of development: rational choice perspective. Berkeley:University A of CaliforniaPress. Tradeshock: The Bevan, D., Collier,P. and Gulling, J. (1990). Fiscal responseto a temporary aftermath the Kenyancoffy boom. The World of BankEconomicReview 3: 359-378. Borner,S., Brunetti,E. and Weder,B. (1996). Political credibilityand economic development. New York:St. Martin'sPress, 1995. Calvo, G. (1995). Capital flow and macroeconomic management:Tequila lessons. Paper presentedat the Seminar on Implicationsof International Capital Flows for Macroeconomic and FinancialPolicies at the IME Washington,December. Campos, E. and Pradhan,S. (1996). The impact of budgetaryinstitutions on expenditure outcomes. World BankPRWPaper No. 1646. WashingtonDC. Frey,B. (1996). Political businesscycles. Aldershot:EdgarElgar. R. Frey, B. and Eichenberger, (1994). The political economy of stabilizationprogrammesin developingcountries.EuropeanJournalof Political Economy 10:169-190. Frey, B. and Schneider,F. (1978). An empiricalstudy of politico-economicinteractionin the United States. TheReviewof Economicsand Statistics60: 174-183. Gaertner,M. (1993). Macroeconomics underflexible exchange rates. Brighton: Harvester Wheatsheaf. Gaertner,M. (1994). Democracy, elections and macroeconomic policy: Two decades of progress.EuropeanJournalof Political Economy 10: 85-110. Hibbs, D. (1977). Politicalpartiesandmacroeconomicpolicy. TheAmericanPolitical Science Review7: 1467-1487. Hibbs, D. (1987). TheAmericanpolitical economy.Cambridge:Harvard UniversityPress. International Monetary Fund (IMF). (1995). Direction of trade statistics. Washingon,DC: MonetaryFund. International MonetaryFund (IMF). (1970-95). Government finance statistics. Washington, International DC: International MonetaryFund. International financial statistics.Washington, MonetaryFund(IMF). (1970-95). International DC: International MonetaryFund. InternationalMonetary Fund (IMF). (1995). Transactionsof the fund. Washington, DC: International MonetaryFund. MonetaryFund(IMF). (1997). Reducingunproductive International expendituresis important for fiscal adjustment. IMD Survey49-51. Krueger,A.O. (1992). Economicpolicy reformin developingcountries.Oxford:Blackwell. A.O. (1993). Political economyof policy reformin developingcountries.Cambridge: Krueger, MIT Press. Krueger,A.O. and Turan,I. (1993). The politics and economics in Turkishpolicy reformin the 1980s. In R. Bates and A.O. Krueger(Eds.), Political and economic interactionsin economicpolicy reform:Evidencefrom eight countries.Oxford:Blackwell.

130
Lindbeck,A. (1976). Stabilizationpolicies in open economies with endogenous politicians. AmericanEconomicReview66: 1-19. Niskanen,W. (1971). Bureaucracyin representative government.Chicago:RandMcNelly. Nordhaus,W. (1975). The political business cycle. Reviewof EconomicStudies42: 169-190. Nordhaus,W. (1989). Alternativemodels to political business cycles. BrookingsPapers on EconomicActivity2: 1-50. The BrookingsInstitution. Persson, T. and Tabellini, G. (1990). Macroeconomicpolicy, credibilityand politics. Chur: HarwoodAcademic Publishers. Rogoff, K. (1990). Equilibriumpolitical business cycles. Review of Economic Studies 55: 1-16. Schuknecht,L. (1996). Political business cycles and fiscal policies in developing countries. Kyklos49: 155-170. Schuknecht,L. (1997). Tying governments' hands in commodity taxation. Geneva: World TradeOrganization Staff WorkingPaperERAD-97-006. Tanzi, V. and Schuknecht,L. (1997). Governmentreform:The recent experience. European Journalof Political Economy 13: 395-417. Tufte, E. (1978). Political controlof the economy.Princeton:PrincetonUniversityPress. Vaubel, R. (1986). The political economy of international organizations.Public Choice 51: 39-57. Von Hagen, J. and Harden,I. (1994). National budget processes and fiscal performance.In GreaterFiscal Discipline".Brussels. EuropeanUnionReportsand Studies "Towards Willett, T. (Ed.). (1989). Political business cycles: Thepolitical economyofmoney, inflation, and unemployment. Durham:Duke UniversityPress. WorldBank. (1970-93). Worldtables. Washington,DC: WorldBank.