public infrastructure and regional economic development

Public Infrastructureand Regional EconomicDevelopmentby Randall W. Berts Randal W. B s i s is an assistant

vice president and economist at theFederal Reserve Bank of Cleveland.The author is pa teM to MichaelBe». Douglas Dalertwg. KevinDuffy-Oeno. Michael Fogarty. WU-liam Fox, and C W Sco Pa* , for els-cussens that were helpful in fonru-lating some of the ideas presented inthis paper.

Introduction

Recent attention given to the serious deteriorationof the nation's public infrastructure raises thequestion of whether public capital significantlyaffects economic development. Local policy-makers and researchers concerned with regionalissues have claimed for years that public infra-structure investment is one of the primary meansto implement a strategy of regional growth. Infact, one of the ways local governments competefor new firms is through investing in varioustypes of public facilities.

Yet, very little is known about even the mostbasic relationships between public and privateinvestment, such as the propensity to substitutebetween public and private capital, the relativetiming of public and private investment, and theeffect of public investment on firm and house-hold decisions, to mention a few. Recent researchby Aschauer (19H9) and Munnell (1990) reportsa positive correlation between public infrastruc-ture and productivity aggregated to the nationallevel. However, this research has not identifiedempirically the linkages by which public infra-structure affects productivity by addressing ques-tions such as the ones posed above. From aresearch standpoint, one of the benefits of exam-

ining the effect of infrastructure at the regionallevel as opposed to the national level is that thelinkages between physical infrastructure andthose that use it are more direct when the analy-sis focuses on smaller geographical areas.

The purpose of this paper is to summarizeprevious work that has examined the effect ofpublic infrastructure on various types of eco-nomic activity at the state and local levels. Sec-tion I defines public infrastructure and discussesvarious ways to measure it that are useful foranalytical purposes. Section II examines theeffects of public infrastructure on regionalgrowth by first reviewing regional growth theoryand then presenting empirical evidence of thisrelationship. Section III raises the issue ofwhether the observed effect of public infrastruc-ture on regional growth results from its effect onproductivity or from its effect on factors of pro-duction. The subsequent discussion focuses onpublic infrastructure as an input into the firmsproduction process. Section IV briefly examinesthe effect of public infrastructure on householdlocation decisions. Finally, the "causation" ofpublic and private investment is discussed insection V. The paper concludes with an overallassessment of the relationship between publicinfrastructure and regional growth.

http://clevelandfed.org/research/review/1990 Q 1

Best available copy

I. Definition andMeasurement of PublicInfrastructure

Definition

This paper focuses on the public works compo-nent of public infrastructure. This categoryincludes roads, streets, bridges, water treatmentand distribution systems, irrigation, waterways,airports, and mass transit—installations and facil-ities that are basic to the growth and functioningof an economy. The term public infrastructureincludes a range of investments broader thanpublic works investment. To distinguish betweenthe various functions of different types of infra-structure, several definitions and classificationsare used throughout the literature.

For example, Hansen (1965), in looking at therole of public investment in economic develop-ment, divides public infrastructure into two cate-gories.- economic overhead capital (EOC) andsocial overhead capital (SOC). EOC is orientedprimarily toward the direct support of productiveactivities or toward the movement of economicgoods and includes most of the public worksprojects listed above. SOC is designed to enhancehuman capital and consists of social servicessuch as education, public health facilities, fireand police protection, and homes for the aged.

Other classifications of public infrastructureinclude investments by the private sector. Mera(1973), examining the economic effects of publicinfrastructure in Japan, extends Hansen's defini-tion of EOC to include communication systems,railroads, and pollution-abatement equipment.Mera also expands the SOC list of investments toinclude administrative systems. In some studies,the term infrastructure also includes the spatialconcentration of specific sets of economic activi-ties, similar to what urban and regional econo-mists refer to as agglomeration economies.

Common to all of these classifications of publicinfrastructure are two characteristics that distin-guish them from other types of investment. First,public infrastructure provides the basic founda-tion for economic activity. Second, it generatespositive spillovers: that is, its sixial benefits farexceed what any individual would be willing topay for its services. These positive spilloversoccur for at least three reasons. First, some com-ponents'of public infrastructure, such as roudsand waterways, are nonexcludable services.Users can share these facilities up to a pointwithout decreasing the benefits received byother users. Second, some infrastructure invest-ments, exemplified by water treatment facilitiesand pollution-abatement equipment, reduce

negative externalities (for example, pollution)generated by the private sector. Third, manyinfrastructure projects, such as power-generatingfacilities, communication networks, sewer sys-tems, and highways, exhibit economies of scale.Because the large costs of these investments canbe spread among many users, the unit cost ofproduction continually falls as more users gainaccess to the system.

For the purposes of this paper, the scope ofpublic infrastructure is limited to public worksinvestment. Public works projects, in addition toexhibiting many of the public infrastructurecharacteristics listed above, are also under directgovernment control and thus can be effectivepublic policy instruments in promoting eco-nomic development.

Measurement

One reason for the lack of empirical work on theeffect of public infrastructure on economicdevelopment is the paucity of consistent andaccurate measures of infrastructure that are suit-able for empirical analysis. Unlike measures ofprivate input usage in manufacturing, there areno reliable and consistent government sourcesof information on public infrastructure, particu-larly for individual states and metropolitan areas.

Two basic approaches have been suggestedfor measuring public infrastructure. One methodis to measure physical capital in monetary termsby adding up past investment. An alternativeapproach is to use physical measures by takinginventory of the quantity and quality of all perti-nent structures and facilities. Each approach hasits advantages and disadvantages.

The standard method of measuring privatecapital stock is to use the monetary approach,often referred to as the perpetual inventory tech-nique. The measure of capital under this methodis the sum of the value of past capital purchasesadjusted for depreciation and discard. Twoassumptions are essential in using this scheme.First, the purchase price of a unit of capital,which is used to weight each unit of capital,reflects the discounted value of its present andfuture marginal products. Second, a constantproportion of investment in each period is usedto replace old capital (depreciation). The firstassumption is met if a perfectly competitive capi-tal market exists. The second assumption is ful-filled if accurate estimates of the asset's averageservice life, discard rate, and depreciation func-tion are available.


Best available copy

L m l s of Public Capital Stock parCapita and Rankings by Total PublicCapital Stock for 40 Saltctad SMSAsIn 1985

SMSA

New York CityBuffaloSan FranciscoSeattleMemphisMilwaukeeClevelandLos AngelesBaltimoreDetroitPittsburghMinneapolisRochesterChicagoKansas CityCincinnatiJersey CityNew OrleansPhiladelphiaPortlandAtlantaAkronLouisvilleNewarkDaytonToledoGrand RapidsDenverIndianapolisRichmondColumbusYoungstownHoustonDallasBirminghamSt. LouisSan DiegoReadingCantonErie

Capital Stockper Capita

$1,216.0871.7871.5858.7842.4823.9762.8753-0716.6714.6713-7687.1663.8661.7649.6613.4610.1592.3584.0

563-1561.9552.6546.4529.4517.2500.94935492.-485.1484.14^5.4467.446-.O

446.3443.2443.04O6.43-6.1330.2322-

Ranking byTotal Capital

Stock

118

412231513

375

101126

420223424

6201433291730313619rr

352837

98

321621393«40

NOTE: Size UIKI rankings of trnal public capiul stuck :nv iiK-isun\l in 196"dollars.

SOI KCK: Amli. >r's c-.ik ulaiii MIS.

A frequent criticism of the |x.T|x*tual inventoryappnxich for public capiuil stock is that thegwernment is not subject to competitivemarkets, and public gtxxls are not allocatedthrough a price mechanism. In some cases, usercharges such as gasoline taxes finance local publie infrastructure investment, but this reflectsaverage costs more than marginal costs.

A considerable portion of the analysis relatedto economic development is based on a neoclas-sical production function in which inputs areused up to the point where the value of theirmarginal product is equal to their cost of use. Insuch a context, current input capital should bemeasured as the maximum potential flow of serv-ices available from the measured stock. Such ameasure of capital can be constructed with theperpetual inventory technique by using a depre-ciation function that reflects the decline in theasset's ability to produce as much output aswhen it was originally purchased. This approachis used by the Bureau of Economic Analysis(BEA) for national-level estimates of both privateand government assets and in many national andregional studies of total factor productivity.

This approach has been used recently byEberts, Fogarty, and Garofalo (see Eberts, Dalen-berg, and Park [ 1986] for details) to constructestimates of five functional types of public infra-structure for 40 metropolitan areas from 1958 to1985. Public outlays for each city since 1904were obtained from City Finances and other U.S.Bureau of the Ceasus publications, and wereaggregated using average asset lives, deprecia-tion, and discard functions used by the BEA andother sources to obtain capital stock measures.The size and rankings of total public capitalstock for each standard metropolitan statisticalarea (SMSA) in 1985 (measured in 1967 dollars)are presented in table 1 as an illustration of theestimates such a method would yield. The percapita estimates of public capital stock reveal awide \-ariation across SMSAs in the amount ofcapital invested and presumably in the amountof infrastructure sen-ices offered within theseareas. In addition, the growth rates of public andprivate capital stock for the 40 SMSAs are shownin table 2. These estimates illustrate the generalslowdown in public capiuil stock accumulation.The notable exceptions to this trend are in thefaster-growing regions of the country.

Capital stock estimates have also been con-structed for other levels of aggregation. Costa,Ellson. and Martin (198"7) use similar techniquesto construct public capital stock for states,although with a much shorter time period. Bos-kin. Robinson, and Huber ( W ) estimate capital


Best available copy

Pimntagi Change in Public Capital forSalietad SMSAs. 1955-1985

SMSA

New York CityLos AngelesChicagoSan FranciscoDetroitPhiladelphiaBaltimoreDallasHoustonPittsburghMinneapolisSeattleClevelandAtlantaMilwaukeeSt. LouisNewarkBuffaloDenverKansas CitySan DiegoCincinnatiMemphisNew OrleansPortlandRochesterIndianapolisColumbusLouisvilleDaytonToledoBirminghamAkronJersey CityRichmondGrand RapidsYoungstownCantonReadingErie

1955-1965

22.531.122.334714.812.919.710.251.4-2.033.510.410.344.731.89.7

11.921.127.510.389.25.9

54.244.39.0

10.326.417.933.227.64.8

11.115.614.99.83-0

28.82.2

-5.10.5

1965-1975

13.015.98.529.513-611.425.035.940.6-3.935.322.17.5

59.316.95.52.19.8

37.121.931.79.2

35.915.520.418.824.625.223-420.88.8

25.015.7-9.924.523.5-2.87.27.71.9

1975-1985

-4.41.65.18.81.00.5

17.646.763.9-1.822.28.70.8

68.04.91.5

-3.410.347.013.127.32.69.06.4 -

21.02.7

14.015.22.9

10.04.9

10.88.2

-10.215.128.9-7.48.26.8

-8.9

NOTE SMSAs are listed from largest to smallest public capital stix-k.SOURCE: Author's calculalioas.

stock series aggregated across all state and kxalgovernments, to critique the BEA's methodologyin constructing its state and local public capitalstock series aggregated to the national level.

Leven, Legler, and Shapiro (1970) advocateusing physical measures of public infrastructureto avoid problems related to the use of prices inthe monetary approach. In order to account fordifferences in capacity and quality, as the priceand depreciation measures do to some extent inthe first method, they propose to collect informa-tion on the physical characteristics of these assetsthat reflect capacity and quality. In the case ofhighways, for example, they cite a study thatconverts physical characteristics of highways to .estimates of the traffic flow capacity.

Although their approach avoids the issue ofasset prices, there are problems with thisapproach as well. One issue is the monumentaltask of collecting adequate measures of the phys-ical size and quality of each type of public infra-structure. For the private sector, it would be vir-tually impossible because of the diverse types ofcapital in use. For the public sector, the task issomewhat less formidable because public capi-tal, as Leven, Legler, and Shapiro suggest, can beclassified into a few dozen basic types.

Another issue is how to enter these variousmeasures into a regression analysis that relatespublic infrastructure to economic activity. Enter-ing more than a half dozen public infrastructuremeasures simultaneously into a regression equa-tion would introduce a number of estimationproblems, including multicollineariry. Further-more, how would one interpret the separateeffects of miles of roads versus cargo capacity ofports, for example? In addition, it may prove use-ful at some point to construct broader classifica-tions of infrastructure, for example combiningroads, highways, and bridges into a transporta-tion network, which would be difficult to dounder this approach. Also, it would be moreconvenient in regression analysis to have qualitydifferences incorporated within a single meas-ure. Both requests would require some type ofaggregation scheme, perhaps more arbitrary thanusing prices or user costs.

One alternative is to develop a hybrid approach.The monetary estimates of public capital couldbe benchmarked by using the physical quantityand quality measures of public infrastructure.This approach would improve the accuracy ofcomparisons across metropolitan areas and overtime by essentially adjusting the price of capitalfor differences in quality and quantity.


Best available copy

I I . Public Infrastructureand Regional Economic'

Economic development depends primarily onlocational advantage, whether it is between cit-ies, states, or countries. Firms seek areas thatoffer greater opportunities for economic profit.Public infrastructure can enhance these oppor-tunities either .by increasing productivity or byreducing factor costs; that is, by augmenting theefficiency of private inputs employed by firms orby providing an attractive environment withinwhich households are willing to accept lowerwages in order to reside.

Regional Models

Regional and national economic growthdepends on processes that are more complexthan simply the aggregation of independentdecisions of firms and households. The deci-sions of economic agents are inextricably inter-twined, and this interdependent must be takeninto account in order to explain the process ofdevelopment. The traditional, neoclassical viewof regional development ignores this interde-pendence and relies heavily on the notion thatcapital is perfectly mobile between regions. Asdescribed by Romans (1965), capital tends toflow toward those regions offering the highestprice and away from regions offering the lowestprice, maintaining at all times an equilibrium ofprice equality after subtracting transport costs.The price of capital is determined by supply anddemand. The supply in a region continuallyadjusts via imports and exports to changes inregional demand so as to maintain interregionalprice equality.

Richardson (1973) and other regional econo-mists dismiss this framework as too simplistic.Instead, they maintain that regional investmentdecisions are characterized by the durability ofcapital, the sequential and interdependent natureof spatial investment decisions, the importanceof indivisibilities in the regional economy, spa-tial frictioas on interregional capital flows, andthe distinction between private-sector capital andpublic infrastructure. The interdependencebetween public-sector investment and private-sector investment is paramount to understandingthe regional development prcx'ess and for pre-scribing regional economic development policy.

I^ven, Legler, and Shapiro (1970) provide asimple picture of the feedback relationshipsbetween public and private investment decisions.

Their model recognizes that an important shareof the regional capital stock consists of socialand public capital and that the scale and spatialdistribution of public capital may have a significant impact on subsequent private investmentdecisions and on the location decisions made byfirms and households. Since the initial size anddistribution of the public capital stock is at leastpartly predetermined by the prior spatial distri-bution of households and economic activities inthe region, an interdependent system emerges.

Once growth in such a system is under way,the process can easily become self-sustainingand cumulative. However, if the initial popula-tion and level of activity are small, and their spa-tial distribution costly and inefficient, a regionmay remain in a low-level equilibrium trap(Murphy, Shleifer, and Vishny [ 1989]). In such acase, attempts to promote regional growth mayneed the exogenous injection of public andsocial capital expenditures to generate an expan-sion rather than merely as a response to changesin the level and spatial distribution of populationand economic activity. The difficulty with thisapproach, as Richardson points out, is that weknow very little about the generative impact ofvarious types of public infrastructure on privateinvestment decisions. Furthermore, we know lit-tle about the effect of a region's economic con-ditions on infrastructure's contribution to output.

Hansen (1965) theorizes that the potentialeffectiveness of economic overhead capital willvary across three broad categories of regions:congested, intermediate, and lagging. Congestedregions are characterized by very high concentra-tions of population, industrial and commercialactivities, and public infrastructure. Any marginalsocial benefits that might accrue from furtherinvestment would be outweighed by the margin-al social costs of pollution and congestion result-ing from increased economic activity. Interme-diate regions are characterized by an environmentconducive to further activity—an abundance ofwell-trained labor, cheap power, and raw mate-rials. Here, increased economic activity resultingfrom infrastructure investment would lead tomarginal social benefits exceeding marginalsocial costs. Lagging regions are characterized bya low standard of living due to small-scale agri-culture or stagnant or declining industries. Theeconomic situation offers little attraction to firms,and public infrastructure investment would havelittle impact.

A number of policy implications emerge fromthis regional growth theory. The most obviouspolicy conclusion is that subsidies for infrastruc-ture investment are more likely to pay off in thelong run than investment incentives to firms and


Best available copy

other subsidies to privute capital. Furthermore,following Hansen (1965) and Hirsch'man (1958),the main task of infrastructure subsidies forunderdeveloped areas is to generate the minimumcritical size of urbanization that can serve as acore for economic development. For these lag-ging regions, however, infrastructure may not beenough to attract firms; additional means such aswage subsidies may be necessary. Finally, amajor outcome of a spatial approach to regionalgrowth analysis is the need for more coordina-tion between government agencies at all levelsand for the integration of all infrastructure deci-sions in an overall regional development strategy.

Before the wisdom of such policies can beassessed, a number of questions must be an-swered For example, how do we identify themechanisms by which infrastructure investmentgenerates regional growth? What types of infra-structure investment are crucial for promotingregional growth? Partial answers are found in theliterature.

Empirical Findings

A direct test of Hansen's hypotheses about theeffects of public infrastructure on regionaldevelopment is provided by Looney and Freder-iksen (1981). Unfortunately from the perspectiveof U.S. policy, they examine economic develop-ment in Mexico. Their findings support Hansen'sintuition, however: economic overhead capital •lias a significant effect on gross domestic productfor intermediate regions but not for laggingregions; social overhead capital exhibits theopposite effect, as Hansen predicted.

Costa et al. (1987) support Hansen's hypothesisof differential impacts of infrastructure on regionalgrowth using VS. data. They find that the largerthe stock of public capital relative to private capi-tal within a state, and the larger the stock of pub-lic capital per capita, the smaller the impact ofpublic capital stock on manufacturing production.Eberts (1986) also finds regional differentials inthe effectiveness of public capital on manufactur-ing output. He reports that public capital wasmore effective in SMSAs in the South than in theNorth and in SMSAs with a lower amount of pub-lic capital relative to private capital and labor.

Duffy-Deno and Eberts (1989). examining 28metropolitan areas from 1980 through 1984, findthat public capital stock has positive and statisti-cally significant effects on per capita personalincome. The effects come through two channels:first, through the actual construction of the publiccapital stock: and second, through public capital

stock as an unpaid factor in the production proc-ess and a consumption good of households. Thissecond effect is twice as large as the first effectusing ordinary least squares (OLS) estimation,but the relative magnitudes of the two effects areroughly reversed using two-stage least squares(2SLS).

Other evidence of the differential effects ofpublic infrastructure among regions comes fromanalysis of the operation of U.S. federal regionaldevelopment programs on the growth rates of per-sonal income for different categories of distressedareas. Martin (1979) finds that investment in pub-lic capital yields few gains for low-income areas,but that business development and planning/technical assistance are more effective in high-unemployment areas.

Mera (1975) provides one of the most compre-hensive analyses of the effect of public infrastruc-ture on regional economic growth for the UnitedStates. He hypothesizes that the growth of regionaleconomic activity is determined primarily by thegrowth of public infrastructure and technicalprogress in the region. The growth of labor andprivate capital, which are allocated through pricedifferentials, responds to growth differentials insocial capital and technical progress. He exam-ines the growth characteristics of the nine US.census regions from 1947 to 1963- Mera con-cludes that more-developed regions are growingbecause of the growth of public infrastructure,while less-developed regions are growing pri-marily because of the growth of technology.

Garcia-Mila and McGuire (1987) estimate thecontribution of state educational and highwayexpenditures to gross state product. Usingpooled cross-section time-series data from 1970to 1983, they estimate a Cobb-Douglas produc-tion function with these two public inputs alongwith manufacturing capital stock and productionemployees as the private inputs. They find thathighway capital stock and educational expendi-tures have a positive and significant effect ongross state product, with educational expendi-tures having the larger impact.

Other studies support these findings. Forexample. Helms (1985) shows that governmentexpenditures on highways, local schools, andhigher education positively and significantlyaffect state personal income. A study of theeffects of public investment in rural areas by theCONSAD Research Corporation (1969) attemptsto assess the effect of public works investmenton the growth of real income in 195 small Mis-souri municipalities. This study finds that publicworks infrastructure accounted for 30 percent ofthe gain in real income between 1963 and 1966.


Best available copy

Of the major investment projects considered,federal highways, barge docks, vocationalschools, and recreational facilities contributedthe most to income growth.

III. Public Infrwtructunand Hrm

Is the effect of public infrastructure on regionalgrowth a result of an overall increase in firm-level productivity or an increase in the region'sattractiveness to labor and capital? Hulten andSchwab's (1984) research on regional productivitydifferentials lends some insight into this distinc-tion. They test the hypothesis that the economicdecline of the Snowbelt was due to differencesin economic efficiency relative to the Sunbelt, bycalculating regional differences in total factorproductivity (TFP). They find little support forthis hypothesis, determining instead that theseinterregional differences are largely a result ofdifferences in the growth rate of capital and labor.Thus, the implication from these findings is thatregional differences in the quality and quantityof public infrastructure may have a greater effecton the migration decisions of factors of produc-tion than on productivity differentials.

There is another reason to look at factors ofproduction rather than at Hicks-neutral produc-tivity changes in analyzing the effect of publicinfrastructure. If public infrastructure is indeedan input (as will be discussed later in this sec-tion), then relating public infrastructure to ameasure of TFP, which includes only labor andprivate capital as inputs, may be a misspecifica-tion of the relationship. Munnell (1990) raisesthis issue for explaining TFP at the national level.When public capital is entered into the TFP cal-culations as a third input, she finds that the varia-tion in TFP over time reflects more a change inpublic infrastructure than a change in technolog-ical innovation.

Very little attention has been given to thetechnological relationships between publicinfrastructure and other inputs in a firm's pro-duction process. The extant literature addressesthis issue primarily from a theoretical stand-point. Three basic questions are considered:1) How does public infrastructure enter the pro-

duction process: as a factor-augmentingatmosphere-type input or as an unpaid input?

2) What implications do these two types of pub-lic inputs have on the efficient allocation of'resources?

3) What effect do public inputs have on a firm'sprofits, and thus on an urea's locationaladvantage?

Infrastructurei t i Public Input

The basic premise of the theoretical literature isthat public infrastructure may increase firm pro-ductivity either through increasing the efficiencyof private inputs employed by firms or throughits own direct contribution to production as aninput into the production process. Economistshave taken both approaches. Meade's (1952)classification of external economies distin-guishes between these two approaches. Meaderefers to the first type of public input as the crea-tion of atmosphere. It is analogous to Samuel-son's pure public good and is exemplified byfree information or technology. In this case, anincrease in the level of public inputs results inincreased output for all firms through neutralincreases in the efficiency with which the privateinputs are used. Any firm entering a regionimmediately benefits from the existing level ofpublic input without affecting the benefits fromthe public input received by other firms.

In more formal terms, public inputs are consid-ered to enter the production function as factorsthat augment the productivity of each of the pri-vate inputs. If a firm is assumed to operate in aperfectly competitive environment, then eachprivate factor of production receives a payment,equal to the value of its contribution to output.Factors, whose productivity has been enhancedby public inputs, receive compensation higherthan they would receive in the absence of publicinputs. For example, suppose that government-supported research and worker training pro-grams targeted at the electronics industryincrease the productivity of labor and capitalemployed by an electronics firm. Workers andowners of capital receive higher compensationbecause of increased productivity. However,since the firm's entire revenue has been distrib-uted among the private factors of production, norevenue is left to pay for public inputs. Thus,public inputs will not be supplied withoutgovernment intervention.

Meade refers to the second type of public inputas an unpaid factor. An example is free accessroads. This input has private-good characteristics,except that it is not provided through a marketpnxess and thus is not paid for on a per-unitbasis and does not have a market-determinedprice. Its private-good characteristics generallyresult from congestion. In the case of highways,as the number of firms in a region expands,increased use of the highways results in conges-tion, which effectively reduces the total amountof highway .services available to each firm. Thus,


Best available copy

from the firm's perspective, the level of public-input is fixed, unless the facility is continuallyunderutilized.

Having many characteristics of a private input,the unpaid-factor type of public input is enteredinto the production process in the same way asprivate inputs. Unlike the first case, the publicinput does not augment the productivity of theprivate inputs. Rather, it contributes independ-ently to the firm's output. Because firms, by defi-nition, do not pay directly for the public input,they initially earn profits or rents according tothe value of the marginal product of the publicinput It is usually assumed that the rent accruesto some ownership factor such as capital or entre-preneurship. As with a private unpriced input,these profits from the public good will attractother firms into the industry (or area). As addi-tional firms enter the industry, the per-firm usageof the public input declines relative to otherinputs. Before and after entry into an industry,capital or the factor collecting the accrued rent ispaid the value of its marginal product plus therent to the unpriced factor. The influx of firmsincreases the ratio of private inputs to publicinputs, causing the marginal product of public-inputs to rise relative to private inputs. Localgovernments, acting as agents for these firms,increase the allocation of public investment rela-tive to the private inputs because of its highmarginal productivity. Additional firms move intothe region until the rents are dissipated and capi-tal earns a competitive rate of return.

Optimal Allocationof Public Inputs

The two types of public inputs have differentimplications concerning the efficient allocationof resources, the level of provision of the publicinputs, and the appropriate financing arrange-ments. For the first case, Samuelson's conditionsfor the allocation of pure public goods apply.Kaizuka (1965) and Sandmo (1972) show thatresources are allocated efficiently when the totalsavings of all firms brought about by substitutinga public input for a single private input are equalto the resource cost of using that pri\-ate input toproduce the public input. However, because ofthe free-rider problem, government must supplythe intermediate input. The revenue necessaryfor government to provide the service must beraised by some form of taxation or user charges.

Negishi (1973) demonstrates that for a purepublic input, an optimal level of public g(xxJ willbe produced if the government supplies a levelof public inputs that maximizes the joint net

profit of industries. The firm's ability to deter-mine the allocation of public investments isdefended by Downs (1957), who argues that afirm's lobbying activities sufficiently influencegovernment decisions. However, when householdpreferences for public expenditures are repre-sented by majority rule voting, public goods arein general not optimally supplied. Pestieau(1976) shows that only under very restrictiveassumptions will majority voting lead to anoptimal supply of the public input. In mostcases, it will oversupph/ public inputs.

The same optimality conditions hold for thecase of an unpaid factor of production, but thelevel of provision is different. Negishi shows thatfor an unpaid factor, the public good most likelywill be oversupplied when the government triesto maximize the joint net profit of firms in thelong run. He offers the following explanation.Since returns to public goods are imputed tocapital in the case of the unpaid factor, capitaltends to concentrate excessively in industriesthat can enjoy more gains from public expendi-tures than other industries. Unless public goodsand capital are perfect substitutes, the capitalintensity of the industry raises the productivity ofpublic goods, which implies that more of thepublic good is required to maximize trie-Indus-try's profits. Thus, allocation is inefficient evenwithout the additional complication of house-hold preferences and majority rule voting.

Financing PublicInfn structure

These theoretical results highlight the impor-tance of the total fiscal package, not simply taxesor public investment, in firm location decisions.As previously mentioned, firms with access topublic infrastructure earn rents according to thevalue of the contribution of public infrastructureto production. In the unpaid factor case, a por-tion of these rents (if not all rents) may be taxedor paid out as user charges in order to financepublic infrastructure. The amount of rentsremaining with the firm as a result of public-unpaid factors depends on the taxing schemeadopted and on properties of the productionprocess or utilization of public inputs.

For any given level of public investment, theamount of rents accruing to firms depends onthe sharing arrangements between taxpayersinside and outside a local jurisdiction. Forexample, if public infrastructure is financedentirely by individuals outside the area (throughfederal grants, for example), then a firm receives


Best available copy

the entire rent, which in turn creates a greaterincentive for that firm and others to locate in thearea. On the other hand, if the entire burden offinancing the public infrastructure investmentfalls on individuals within the local area, thenprofits would be much smaller, creating less ofan incentive for firms to locate or remain there.

Another arrangement is for households toassume a larger proportion of the financing costsof public investment than warranted by thedirect benefits they receive. Some communitiespursue this approach through tax moratoriumsand lower tax rates for firms, with the idea thatthe benefits to the community from creating newjobs outweigh the increased burden of financingthe investment

An additional feature of the fiscal package isthat taxes need not equal the total rents accruingto firms (and even to households). Benefits frompublic investment projects characterized by econ-omies of scale and sharing properties will exceedthe cost of the project Since many componentsof public infrastructure, such as highways andwater distribution and treatment facilities, exhibitthese properties, it is reasonable to assume thatpublic investment may have a net positive effecton firm productivity and thus on firm location.

Empirical Findings

A number of basic questions emerge from thetheoretical foundations of the relationshipbetween infrastructure and firm-level behavior.1) How does public capital enter into the pro-

duction process?2) What effect does public infrastructure have on

a firm's productivity? How does this vary withthe type of firm and type of infrastructure?

3) Are private and public capital related as sub-stitutes or complements?

4) What effect does public infrastructure have onfirm location decisions?Only recently have researchers estimated the

technical relationships between public infrastruc-ture and other production inputs. Previously, theliterature looked primarily at peripheral issuessuch as the effects of federal programs on eco-nomic growth in distressed areas or the effects ofvarious government expenditures on firm loca-tion. These are undoubtedly important questions,but their particular focus does not provide muchinsight into the technical relationships outlinedabove. Another problem with the earlier studiesis that, with the exception of Mer.i (19"3, 1975),they use public expenditures or the number ofgovernment employees as proxies for publicinfrastructure.

More recent studies have tried to address theseissues directly by estimating production functionswith public capital-stock estimates included asinputs. Eberts (1986) estimates the direct effectof public capital stock on manufacturing outputand the technical relationships between publiccapital and the other production inputs. Publiccapital stock is estimated using the perpetualinventory technique, described in section I, foreach of 38 U.S. metropolitan areas between 1958and 1978. With this method, capital is measuredas the sum of the value of past investmentsadjusted for depreciation and discard. Public cap-ital stock includes highways, sewage treatmentfacilities, and water distribution facilities withinthe SMSA. He estimates a translog productionfunction with value added as output, hours ofproduction and nonproduction workers as thelabor input, and a value measure of privatemanufacturing capital stock as private capital.

Eberts finds that public capital stock makes apositive and statistically significant contributionto manufacturing output, supporting the conceptof public capital stock as an unpaid factor ofproduction. Its output elasticity of .03 is smallrelative to the magnitudes of the other inputs:0.7 for labor and 0.3 for private capital. It followsthat the magnitude of the marginal product ofpublic capital is also relatively small.

The small estimated contribution of publiccapital may be \iewed in two ways. If one con-siders public capital stock to-be a pure publicgood, then the marginal product of public capi:

tal stock reveals the manufacturing sector's valua-tion of the total stock of public investment inplace in the SMSA. If local governments allocatepublic funds in response to the preferences ofthe local voters, then the marginal valuationshould be equal to their tax share. Thus, it is notunreasonable that a typical firm pays 4 percent ofits total value added to local taxes—a value closeto the marginal product of public capital.

Another way to interpret the results is toassume that the manufacturing sector uses only aspecific portion of the stock. For instance, firmsmay be spatially concentrated in one area of themetropolitan area and thus intensively use onlythe roads and sewer systems in that part of theregion. If one assumes that the per-unit cost ofconstructing one unit of private capital is the.same as the per-unit cost of constructing one unitof public capital, then the marginal products ofthe mi) capital inputs should be equal. Estimatesshow, however, that the marginal product of pri-vate capital is seven times that of the marginalproduct of public capital. This difference may


Best available copy

result from assuming that the manufacturing sec-tor uses the total capital stock instead of someportion of it. If one assumes that the use of thetotal public capital sux^k by manufacturing firmsis approximately proportional to manufacturingemployment's share of the total population, thenthe use of the public capital stock is overesti-mated by roughly seven times. Multiplying themarginal product of public capital stock by sevenbrings it in line with the marginal product of pri-vate capital.

With respect to technical relationships, Ebertsfinds that public capital and private capital arecomplements, while the private capital/laborpair and the public capital/labor pair are substi-tutes. Public and private capital are interpreted tobe complements when an increase in the levelof public capital reduces the price of private cap-ital by increasing its relative abundance. Dalen-berg (1987), using the same data as Eberts butestimating a cast function, also finds public capi-tal and private capital to be complements.

Deno (1986) also estimates technical relation-ships, but uses investment data instead of capitalstock data. Using pooled data for U.S. metropoli-tan areas from 1972 to 1978, he estimates laborand private investment demand equationsderived from a Cobb-Douglas production func-tion. He finds that local public investment andprivate capital are complements. In addition, hefinds that a 1 percent increase in public invest-ment is associated with a 0.01 percent increasein net private investment in the short run and a0.2 percent increase in the long run. Further-more, he concludes that public investment has asignificantly greater positive effect on net privatecapital formation in distressed cities than ingrowth cities. In subsequent work. Deno (1988)finds the output elasticities of water, sewer, andhighway infrastructure for the full sample of 36SMSAs are 0.08, 0.30, and 0.31. respectively.These estimates were obtained using a profitfunction approach for the period 19~0 to 1978.

At the state level, Costa et al. (198") estimatethe contribution of public capital stock to manu-facturing output by estimating a translog produc-tion function. Their analysis differs from that ofEberts in two key ways, in addition to the unit ofanalysis. First, Costa et al. estimate the prcxluc-tion function using cross-sectional data for 1972.while F,berts combines cross-sectional and timeseries data in his estimation. Second. Costa et al.distribute the BF.A estimate of capital amongstates in proportion to the gross book value offixed assets at year-end 19~1. The private capitalstixk used by Eherts. on the other hand, is basedon the same perpetual inventory technique usedto construct the public capita! stex'k.

Costa et al. also find that public capital stockmakes a statistically significant contribution tomanufacturing output. However, the magnitudesof their public capital elasticities are higher tlianwhat Eberts found, which may be partly explainedby their inclusion of more categories of publicinvestment. Another difference between theresults of these studies is that Costa et al. findprivate and public capital to be substitutes andpublic capital and labor to be complements,while Eberts and Deno find the opposite. Oneexplanation for the difference may be in the cal-culation of these relationships. Costa et al. usethe log form of the production function to derivethe cross-partial derivative, while Eberts convertsback to the original production relationship tocompute the technical relationships.

Mera (1973) estimates the technical relation-ships between various types of infrastructure andother inputs for Japan. Using pooled data of nineregions in 10 years from 1954 to 1963, he esti-mates a Cobb-Douglas production function foreach of three major economic sectors and fourtypes of infrastructure. He reports the followingfindings: (1) when the infrastructure variable isentered as a separate factor of production, itsproduction elasticity ranges from 0.1 to 0.5, mostfrequently around 0.2; (2) the transportation andcommunication infrastructure appears to have asizable effect on mining, manufacturing, andconstruction-, (3) in most cases, the rates ofreturn from infrastructure are similar to those ofprivate capital; but (4) the elasticity of substitu-tion between private capital and infrastructure isundetermined in this stud)'.

Studies of the determinants of firm locationusually concentrate more on the effect of taxesthan on the effect of expenditures on locationdecisions. However, those studies that haveincluded various measures of public infrastruc-ture have found that certain forms of infrastruc-ture are attractive to firms. Some of the strongestresults were reported by Fox and Murray (1988),who found that the presence of interstate high-way systems had a positive and highly significanteffect on the location of individual establish-ments in the state of Tennessee. Banik (1985),using a national sample, also found that thenumber of new branch plants was higher withinstates with more miles of roads. Eberts (1990)offers evidence that public infrastructure posi-tively affects the number of firm openings inmetropolitan areas.


Best available copy

IV. Public Infrastructureand Households

Public infrastructure may also affect the migra-tion decisions of households by enhancing thearea's amenities. However, the existing literaturerelated to household location decisions does notfocus much on public infrastructure. Labor migra-tion studies tend to concentrate primarily ondemographic characteristics and wage differentialsto explain migration flows. Urban quality-of-lifecomparisons, which deal with the same underly-ing decision process, come closer to addressingthis issue, but their major focus is on attributessuch as air quality, climate, and so forth.

One exception is the migration study by Fox,Herzog, and Schlonmann (1989). They estimatethe effect of local fiscal expenditures and revenueon household decisions to migrate across met-ropolitan areas. Using Public Use MicrodataSamples, which record a household's place ofresidence in 1975 and 1980, they determine thatfiscal variables are more important factors inpushing people from an area than in attractingthem toward one. They explain this result interms of information. Information on fiscal struc-ture is more readily available in an area where aperson has been living than for areas under con-sideration as migration destinations.

V. "Cwst l " RelationshipsBetween Public andPrivate Investment

Most of the studies that address the stimulativeeffect of public investment presume that public-investment "causes" or precedes private capital.Yet, scant attention has been given to testing thisrelationship. Eberts and Fogam1 (1987) explorethe causal relationship between public and pri-vate investment. Their premise, following thecumulative model of regional growth, is that thetiming of investment indicates the role of publicinvestment in promoting local economic devel-opment. If public investment precedes privateinvestment, then it would appear that kxal areasactively use public outlays as instruments todirect local development. On the other hand, ifthe sequence of events occurs in the oppositedirection, it would appear that lcx'al officialsmerely respond to private investment decisions.

Using data on public capital outlays and manu-facturing investment from 190-i to 1978 for 40U.S. cities, Eberts and Fogarry find a significantcausal relationship between public outlays andprivate investment in 33 of the -lO metropolitan

areas examined. The direction of causation goesboth ways. Private investment is more likely toinfluence public outlays in cities located in theSouth and in cities that have experienced tre-mendous growth after 1950. Public outlays aremore likely to influence private investment incities that experienced much of their growthbefore 1950.

Looney and Frederiksen (1981), in their studyof Mexico, support the findings of Eberts andFogarty for older VS. cities—that public investment appears to be the initiating factor in thedevelopment process, rather than a passive oraccommodating factor. They do not attempt todetermine whether causal directions differ acrosstypes of regions, however.

VI. Overall Assessment

The importance of public infrastructure in pro-moting economic development has been widely-recognized among policymakers. Economistshave only recently begun to assess the effects ofinfrastructure on regional economic develop-ment beyond simply a stimulus of constructionactivity. The consensus among economists is thatpublic infrastructure stimulates economic activ-ity, either by augmenting the productivity of pri-vate inputs or through its direct contribution tooutput. Furthermore, by enhancing a region'samenities, public infrastructure may also attracthouseholds and firms, which further contributesto an area's growth.

Results show that public capital stock signifi-cantly affects economic activity. The magnitudesof the effects for public capital are much lessthan for private capital, however. Results alsoshow, with some exception, that public capitaland private capital are complements, not substi-tutes. This relationship may be interpreted tomean that the existence of public infrastructureis a necessary precondition for economic growth.

Evidence suggests that the effect of publicinfrastructure on regional development dependson the type of investment and on the economicconditions of the region. Studies of Japan andMexico, in particular, show that investment incommunications and transportation appears tohave the most significant impact on regionalgrowth. In the United States, public investmentappears to have a greater effect on economicactivity in distressed cities than in growth cities,in Sunbelt cities than in Northern cities, and inthose areas with less public capital stock relativeto private capital and population.


Best available copy

The critical question is at what point, rf any,does an additional increase in public infrastruc-ture cease to have any effect on economic-development? Alder (1965) sums up the effect oftransportation on economic development: "It isfrequently assumed that all transport improve-ments stimulate economic growth. The sad truthis that some do, and some do not...." In abroader context, it can be concluded that sometypes of infrastructure investment will have sig-nificant effects, while others will not.

Many local and state governments in the UnitedStates are faced with the monumental task ofreplacing and upgrading their present publiccapital stock. But the challenge is more thansimply maintaining existing structures. The chal-lenge facing these governments is to meet thefuture infrastructure needs of a US, economythat is undergoing dramatic changes with therestructuring of both manufacturing and serviceindustries and the spatial redistribution of theseactivities. Innovations in areas such as telecom-munications and computer automation, to men-tion only two, are changing the way businessesoperate, and infrastructure investment mustadapt to this changing technology. Results fromstudies reported in this paper underline theimportance of maintaining, improving, andexpanding public capital stock in order to sup-port future economic growth.

Rafirencas

Alder, Hans A. "Economic Evaluation of Transport Projects," in Gary Fromm, ed., TransportInvestment and Economic Development.Washington, D.G: Brookings Institution, 1965,pp. 170-94.

Aschaucr, David A. "Is Public Expenditure Pro-ductive?" Journal of Monetary Economics,vol. 23, no. 2 (March 1989), pp. 177-200.

Bartik, Timothy J. "Business Location Decisionsin the United States: Estimates of the Effects ofUnionization, Taxes, and Other Characteristicsof States," Journal of Business and EconomicStatistics, vol. 3, no. 1 (January 1985), pp."14-22.

Boskin, Michael J., Marc S. Robinson, and AlanM. Huber. "New Estimates of State and LocalGovernment Tangible Capital and NetInvestment," National Bureau of EconomicResearch Working Paper Series No. 2131,Cambridge, Mass., January 1987.

CONSAD Research Corporation. A Study of theEffects of Public Investment, Prepared forOffice of Economic Research, EconomicDevelopment Administration, Washington,D.C., 1969.

Costa, Jose da Silva, Richard Ellson, and Ran-dolph C. Martin. "Public Capital, RegionalOutput, and Development: Some EmpiricalEvidence," Journal of Regional Science, vol.27, no. 3 (August 1987), pp. 419-37.

Dalenberg, Douglas. "Estimates of Elasticities ofSubstitution Between Public and PrivateInputs in the Manufacturing Sector of Metro-politan Areas," Ph.D. dissertation, Eugene,Ore.: University of Oregon, 1987.

Deno, Kevin T. "The Short Run RelationshipBetween Investment in Public Infrastructureand the Formation of Private Capital," Ph.D.dissertation, Eugene, Ore.: University ofOregon, 1986.

"The Effect of Public Capital on U.S.Manufacturing Activity: 1970 to 1978," SouthernEconomic Journal, vol. 55, no. 2 (October1988), pp. 400-11.

Downs, Anthony. An Economic Theory ofDemocracy. New York: Harper and Row, 195".

Dufly-Deno, Kevin T., and Randall W. Eberts."Public Infrastructure and Regional EconomicDevelopment: A Simultaneous EquationsApproach," Working Paper 8909, FederalReserve Bank of Cleveland, August 1989.


Best available copy

Eberts, Randall W. "Estimating the Contributionof Urban Public Infrastructure to RegionalGrowth," Working Paper 8610, FederalReserve Bank of Cleveland, December 1986.

. "Some Empirical Evidence on the link-age between Public Infrastructure and LocalEconomic Development," in Henry W. Herzog,Jr. and Alan Schlottmann, eds., Industry Loca-tion and Public Policy. Knoxville, Tenn.: Uni-versity of Tennessee Press, forthcoming 1990.

, Douglas Dalenberg, and Chul SooPark. "Public Infrastructure Data Developmentfor NSF," mimeo, Eugene, Ore.: University ofOregon, May 1986.

Eberts, Randall W., and Michael S. Fogarty."Estimating the Relationship Between LocalPublic and Private Investment," WorkingPaper 8703, Federal Reserve Bank of Cleve-land, May 1987.

Fox, William F., and Matthew N. Murray. "LocalPublic Policies and Interregional BusinessDevelopment," mimeo, Knoxville, Tenn.:University of Tennessee, June 1988.

Fox, William F., Henry W. Herzog, Jr., and AlanM. Schlottmann. "Metropolitan Fiscal Structureand Migration," Journal of Regional Science,vol. 29, no. 4 (November 1989), pp. 523-36.

Garcia-Mila, Teresa, and Thcrese J. McGuire."The Contribution of Publicly Provided Inputsto States' Economies," Research Paper No.292, State University of New York at StonyBrook, April 1987.

Hansen, Niles M. "Unbalanced Growth andRegional Development," Western EconomicJournal, vol. 4 (Fall 1965), pp. 3-14.

Helms, L Jay. "The Effect of State and LocalTaxes on Economic Growth: A Time SeriesCross Section Approach," Review oj Econom-ics and Statistics, vol. 67, no. 4 (November1985), pp. 574-82.

Hirschman, Albert O. The Strategy of EconomicDevelopment. New Haven, Conn.: Yale Uni-versity Press, 1958.

Hulten, Charles R, and Robert M. Schwab."Regional Productivity Growth in U.S. Manu-facturing: 1951-78," American EconomicReview, vol. 74, no. 1 (March 1984), pp.152-62.

Kaizuka, Keimei. "Public Gcxxis and Decentrali-zation of Production," Review of Economicsand Statistics, vol. -T (1965), pp. 118-20.

Leven, Charles, John Leglcr, and Perry Shapiro.An Analytical Framework for Regional Devel-opment Policy. Cambridge. Mass.: The MITPress, 1970.

Looney, Robert, and Peter Frederiksen. "TheRegional Impact of Infrastructure in Mexico,"Regional Studies, vol. 15, no. 4 (1981), pp.285-96.

Martin, Randolph C "Federal Regional Devel-opment Programs and US. Problem Areas,"Journal of Regional Science, vol. 19, no. 2(May 1979), pp. 157-70.

Meade, J.E. "External Economies and Disecon-omies in a Competitive Situation," EconomicJournal, vol. 62 (March 1952), pp. 54-67.

Mera, Koichi. "Regional Production Functionsand Social Overhead Capital: An Analysis ofthe Japanese Case," Regional and Urban Eco-nomics, vol. 3, no. 2 (May 1973), pp. 157-85.

' . Income Distribution and RegionalDevelopment. Tokyo: University of TokyoPress, 1975.

Munnell, Alicia. "Why Has Productivity GrowthDeclined? Productivity and Public Investment,"New England Economic Review, FederalReserve Bank of Boston, January/February1990, pp. 3-22.

Murphy, Kevin M., Andrei Shleifer, and RobertW. Vishny. "Industrialization and the BigPush," Journal of Political Economy, vol. 97,no. 5 (October 1989), pp. 1003-26.

Negishi, Takashi. "The Excess of Public Expendi-tures on Industries ," Journal of Public Eco-nomics, vol. 2, no. 3 (July 1973), pp. 231-40.

Pesticau, Pierre. "Public Intermediate Goodsand Majority Voting," Public Finance, vol. 31,no. 2 (1976), pp. 209-17.

Richardson, Harry W. Regional Growth Tloeory.London: The MacMillan Press Ltd., 1973.

Romans, J. Thomas. Capital Exports and GrowthAmong US. Regions. Middletown, Conn.:Wesleyan University Press, 1965.

Sandmo, Agnar. "Optimality Rules for the Provi-sion of Collective Factors of Production,"Journal of Public Economics, vol. 1 (1972),pp. 149-57.


Best available copy

public infrastructure and regional economic development

Documents