EDWARD M. GRAMLICH* Brookings Institution HARVEY GALPER* UrbanInstitute State and Local Fiscal Behavior and Federal Grant Policy PURCHASES BY STATE AND LOCAL GOVERNMENTS havelong beenthe most rapidlyrising componentof aggregatedemand. While real consumption and investmentexpenditures have both doubled since the end of the Koreanwar, and federalpurchases have increasedbarely at all, state and local purchasesof goods and serviceshave almost tripled. They have grown at an annualaverage rate of 5.5 percentand now accountfor over 10 per- cent of real gross nationalproduct (GNP). For much of this period,the budgetarysurplus for state and local gov- ernmentshovered very close to zero, being negativeas often as positive and neveramounting to more than $2 billion. Recently,however, the sur- plus has grown at a remarkablerate. It was only $0.7 billion as late as * We have benefitedfrom the comments of severalmembers of the Brookingspanel, and also from discussions with Frank de Leeuw, Robert W. Hartman, Larry L. Orr, George E. Peterson,and Robert D. Reischauer.Charles A. Waite has suppliedus with unpublished national income accounts data, Joseph Valenza ably assisted with the computer work, Su Nokkeo with data preparation,and Kathryn Breen and Patricia Sachs with the typing. Some of the results in the paper are taken from an Urban Insti- tute study, financed by the Office of Economic Opportunity, on the distributional aspects of urban fiscal behavior. A preliminaryreport on this project can be found in Harvey Galper, Edward Gramlich, Claudia Scott, and Hartojo Wignjowijoto, "A Model of Central City Fiscal Behavior," forthcomingin the Proceedingsof the 28th Congress of the InternationalInstitute of Public Finance, 1972. 15 16 Brookings Papers on Economic Activity, 1:1973 1969but then began a rapidexpansion, reaching $4.8 billion in 1971 and $12.3billion in 1972-when it attainedan annualrate of almost$20 billion in the fourthquarter. Though special factors have accountedfor some of this rise,a 1972report on the fiscalpolicies of PresidentNixon and Senator George McGovernpredicted that the state and local surpluswould rise even higherunder both sets of proposals.' Grantsto stateand local governmentsfrom the federalgovernment were undoubtedlyresponsible for much of the increasein expenditures,and possiblythe budgetsurplus as well.Whereas in 1954these grants amounted to only $2.9 billion, by 1974 they are expectedto reach $41.6 billion, a thirteen-foldexpansion.2 And grants are of currentinterest not only be- cause of their sheer growth. The recent enactmentof general revenue sharing,the administrationproposal to convertexisting categorical grants to specialrevenue sharing, and numerousother plans to federalizewelfare paymentsor to provideproperty tax reliefor incometax creditsto stateand local taxpayers-all indicatethat fundamentalchanges are occurringin the form of federalassistance to states and localities. The increasingimportance of the state and local sectorand the changing role of federalgrants point to the needfor a morethorough understanding of the budgetarybehavior of state and local governments,particularly the way in which it is influencedby intergovernmentaltransfers. To explore this topic, we firstestimate a model of state and local fiscalbehavior and then use it to examinethese policy questions. We beginby discussingdifferent forms of grantassistance and how they mightbe expectedto affectthe budgetarybehavior of statesand localities differently.These ideas underlieour theory of state and local fiscal be- havior,from whichwe derivea consistentset of estimatingequations for state and local expenditures,revenues, and the budgetsurplus. The inde- pendentvariables in these equationsare federalgrants of varioustypes, income, relative prices, previous stocks of financialassets, and demo- graphicvariables. The model is estimatedwith two separatebodies of data: (1) quarterly time seriesobservations on the entirestate and local sectorin the national income accountsfor the period 1954-72; (2) annualbudgetary observa- tions for a sampleof ten urbangovernments for the period1962-70. While 1. See David J. Ott and others, Nixon, McGovern,and the FederalBudget (Washing- ton: AmericanEnterprise Institute for Public Policy Research, 1972). 2. Special Analyses,Budget of the UnitedStates Government,Fiscal Year1974, p. 8. Edward M. Gramlich and Harvey Galper 17 the coverageof the time seriessample is more comprehensive,the pooled cross-sectiondata contain informationon a wider range of intergovern- mentalassistance that is more finelydisaggregated by function,and there- fore helps to sharpenthe estimatesfor certaincritical parameters. These empiricalresults are then used to judge how the state and local sector respondsto federal aid of various sorts, especiallygeneral and special revenuesharing, and to interpretthe recent spectacularincreases of the state and local budgetsurplus. FederalGrants Althoughthere is a relativelywell-developed theory of the roles played by differenttypes of intergovernmentaltransfers, most empiricalstudies in this area have paid little attentionto it. They have usuallyassumed that grants of whateverkind affectedstate and local governmentbehavior in muchthe sameway, disregardinga theorythat postulatedthey wouldnot. This approachmay have yieldedacceptable predictions of the growth of state-localspending as long as grantsincreased in volumewithout changing in structure,but it is clearlyinappropriate now when grantpolicy is under- going such a radicalrestructuring. The theory of intergovernmentaltransfers suggests that grants from higher to lower levels of governmentcan be classifiedinto three broad types. CaseA. Open-endmatching grants, under which the higherlevel of gov- ernmentpays some portion of the cost of certainstate or local expendi- tures, thus effectivelyreducing their price, and the lower governmentis freeto take as much of the grantmoney as it wantsat this new priceratio. Federalgrants of this type have all been in the welfarearea-for public assistance,Medicaid, and social services.The responseof expendituresby lower governmentsto these grantsdepends on the price elasticityof de- mand for the relevantgood or service:The lowerlevel of governmentin- creasestotal spending(from its own resourcesplus the grant)by morethan the grant-and reducesit on all other goods-if demandis elastic, and increasesspending by less thanthe grant-and raisesit on all othergoods- if demandis inelastic. Case B. Closed-end lwnp-sum transfers, under which the higher level of governmentmerely transfers a fixed amountof money to a lower gov- 18 Brookings Papers on Economic Activity, 1:1973 ernment without any effective restrictions on its use or any change in relative prices. The recently enacted general revenue-sharingprogram con- stitutes the first important federal policy of this type, though such programs have existed on the state level. The response of expendituresto these grants depends simply on the lower government's propensity to spend budgetary resources rather than to reduce taxes or add to its surplus. In the normal case, in which both public and private goods have positive income elas- ticities, lump-sum transfers will stimulate some increased public spending and some tax reduction. Case B grants will also stimulate less total spend- ing per dollar of grant than case A grants if the demand for expenditures is at all sensitive to price changes. Case C. Closed-endcategorical grants, through which the higher level of government transfers a limited amount of money to be used for a specific program. These grants are a hybrid of case A and B grants in that the higher government lowers the price of the aided activity but limits the size of the grant. All important grant programs at the federal level except those already noted have been of this type. Without any other restrictions, case C grants can be shown to have expenditure effects somewhere between those of case A and case B grants-less than open-end matching grants because the limitation on funds diminishes the impact of the price reduction, and more than lump-sum transfers because at least some price reduction occurs for the specified activity. If limited to incremental expenditures above some base amount, however, case C categorical grants can increase expenditures by more than open-end case A grants that are not subject to such restric- tions.3 These straightforwardideas become difficult to apply once diverse types of expenditures are aggregated into functional categories for empirical study. For any one type of expenditure, open-end case A grants will stimu- late more spending per dollar of grant than closed-end categorical case C grants without other restrictions, which themselves stimulate more expen- ditures than lump-sum B transfers. But case C grants could be observed to have a larger impact per dollar on spending than case A grants if existing C grant programs make more use of effort maintenance requirements that confine aid to incremental expenditures in a certain category, or if C grants have been more extensively used to support activities that have not 3. These propositions are describedmore completely in James A. Wilde, "Grants- in-Aid: The Analytics of Design and Response," National Tax Journal,Vol. 24 (June 1971), pp. 143-55. Edward M. Gramlich and Harvey Galper 19 previously been carried out at the state-local level and are therefore incre- mental expendituresin effect. The situation is complicated even further be- cause grants have different legal and administrative