analysing institutional arrangements for sustaining rural infrastructure indeveloping countries

Analyzing the Performance ofAlternative Institutional Arrangements

forSustaining Rural Infrastructure in

Developing Countries1

Elinor OstromIndiana University

Larry SchroederSyracuse UniversitySusan Wynne

Indiana University

'Research for this paper wassupported by the Decentralization:Finance and Management Project(DFM) Project sponsored by theOffice of Rural and InstitutionalDevelopment of the Bureau forScience and Technology of theVS. Agency for InternationalDevelopment (Contract DHR-5546-Z-00-7D33-00). The viewsand interpretations here are thoseof the authors and should not beattributed to the U.S. Agency forInternational Development Thepaper draws heavily on EOstrom, Schroeder, and Wynne(forthcoming).

J-PAKT, 3(1993):l:ll-45

ABSTRACT

This paper, which draws on principles from the new institution-al economics and institutional analysis, argues that a principalunderlying cause of the lack of sustained rural infrastructure invest-ments in developing countries is the set or sets of perverse incentivesfacing participants in the development and operation of such facili-ties. The analysis examines the intermediate transaction costs ofproviding and producing public goods under six different institution-al arrangements, each of which creates a different set of incentives forthe many actors involved in the design, construction, finance, opera-tion, maintenance, and use of rural infrastructure. By considering afull set of intermediate transaction costs and a wider array of institu-tional arrangements, analysts can become aware of the tradeoffsinvolved and may more likely identify those alternatives that yieldgenuine net cost reductions.

One of the more perplexing problems facing developingcountries and international development agencies is sustaininginvestments, particularly infrastructure investments, longenough to allow the facilities to at least repay the cost of theirdevelopment.2 It is depressingly common to find that manycapital infrastructure facilities such as irrigation systems, roads,bridges, and water systems fail to generate net benefits despiteex ante projections to the contrary. Even facilities that yieldpositive net benefits frequently deteriorate rapidly due toinsufficient maintenance.

11/'Journal of Public Administration Research and Theory

by guest on Septem

ber 11, 2010jpart.oxfordjournals.org

Dow

nloaded from

http://jpart.oxfordjournals.org/

Analyzing the Performance

'Although the subject of this paperis the problem of developing sus-tainable physical infrastructure inrural areas in developing coun-tries, much of the analysis is equ-ally relevant to social forms ofinfrastructure such as trainingcenters and agricultural extensionservices, investments in urbanareas, and the problems of infra-structure sustainability in devel-oped countries.

Why is it so difficult to sustain rural infrastructure indeveloping countries? Although the lack of appropriate main-tenance is often a root cause of deteriorating infrastructure,maintenance cannot be adequately considered apart from theother activities involved in the wider process of supplyinginfrastructure benefits. The type and level of maintenancerequired is intimately related to how a facility is designed,financed, constructed, operated, and used (Uphoff 1986, 63-70).Unless facilities are designed with appropriate combinations oftechnical knowledge and information about local conditions,they face a greater likelihood of failing even with maintenanceinputs. Infrastructure cannot be sustained without sufficientfinancial resources to operate and maintain it. The quality ofconstruction also affects sustainability; facilities may be sopoorly built that no level of maintenance effort is capable ofovercoming the initial deficiencies. For some facilities, opera-tional inputs in addition to maintenance are also required.Finally, inappropriate use patterns can have a major impact onsustainability (e.g., heavily loaded vehicles that ruin otherwisewell-designed and constructed roads).

The multiplicity of factors associated with the sustainab-ility of rural infrastructure means that no solitary, easily identi-fiable cause exists for failures. We argue, however, that thereis one underlying analytic cause of all these infrastructuremaladies: the set or sets of perverse incentives facing partici-pants in the design, construction, finance, operation, mainte-nance, and use of infrastructure facilities. Rather than presumethat the individuals involved intend to develop unsustainableinfrastructure, we assume that they are rational decisionmakerstrying to generate net benefits. These individuals, however,face considerable uncertainty, and some behave opportunistica-lly. Thus they may make decisions that inadvertently orpurposely either harm some members of a community or leaveothers much better off. When incentives are better matched tothe situation, individuals make decisions that yield outcomesthat are both personally and socially more rewarding. Theadmittedly difficult task, therefore, is to design the institutionalarrangements that will produce such incentives.

Designing appropriate institutions requires the ability toanalyze alternative institutional arrangements for infrastructuredevelopment. The purpose of this article is to demonstratehow such arrangements can be analyzed using conceptualtools developed by scholars working in the fields of the newinstitutional economics and institutional analysis (V. Ostrom,Feeny, and Picht 1988; Adelman and Thorbecke 1989;Nicholson and Connerley 1989). Because these concepts haveseldom been applied directly to the problems of rural infra-

U/J-PART, Januaiy 1993

by guest on Septem


Dow

nloaded from



structure development, particularly in less developed coun-tries, we devote the next section to the special challenges thatsuch an environment creates. Development administrationanalysts commonly presume that central governments mustprovide certain types of rural infrastructure. This is not partic-ularly surprising when the analysis is limited to a comparisonof two idealized forms of organization: a simple market and asingle governmental hierarchy. We argue in Section 2 that it ispreferable to analyze a fuller set of institutional arrangements,using a more extensive set of intermediate and overall perfor-mance criteria. Section 3 uses this more comprehensive set ofperformance criteria to evaluate six types of institutional ar-rangements for the provision and production of rural infra-structure. Because much rural infrastructure development indeveloping countries is characterized by donor interventions,Section 4 briefly considers how donors can influence the incen-tive structures affecting the sustainability of capital invest-ments. A brief summary concludes the discussion.

PUBLIC SECTOR INTERVENTIONS RELYING ON MULTIPLEACTORS

That infrastructure development requires a set of interre-lated activities does not alone explain the complexity of thesustainability problem. Designing, constructing, financing,operating, maintaining, and ensuring proper use of infrastruc-ture facilities are complicated by the fact that generally theseactivities are (1) provided for by the public sector where (2)numerous decisionmakers facing a variety of incentives mustact in an uncertain environment.

Rural infrastructure facilities such as roads, irrigation, andvillage water systems are all distinguished by characteristicsthat make them unsuitable for wide-scale provision by individ-uals acting alone. These facilities are used by many individu-als or firms and are costly to "fence off" so as to precludepotential beneficiaries from enjoying benefits without paying.Efficiency can thus be attained only through some joint provi-sion arrangement. One obvious solution is to rely on thepublic sector.

Although public sector involvement in rural infrastructuredevelopment is often necessary, distinguishing between theprovision and production of these facilities helps to clarify thevariety of ways in which public and private sector activitiescan be linked (Musgrave 1959; V. Ostrom, Tiebout, and Warren1961; Advisory Commission on Intergovernmental Relations(ACIR) 1987, 1988; V. Ostrom, Bish, and E. Ostrom 1988). Inthe public sector, provision refers to making decisions through

13//-PART, January 1993

by guest on Septem


Dow

nloaded from



collective choice mechanisms concerning the kinds, quantity,and quality of services to be provided, arranging for theirproduction and finance, and monitoring or regulating theirproduction and use. Production, on the other hand, refers to"the more technical process of transforming inputs into out-puts-making a product, or, in many cases, rendering a service"(AOR 1987, 7). Once a public body has decided to "provide" aparticular infrastructure facility, it must then decide whether toproduce the facility itself, mandate others to produce it, en-courage its production by giving financial incentives to otherunits of government, or contract for its production by privateor other public agencies.

The nonexcludability and joint use attributes of muchrural infrastructure suggest that the public sector should beresponsible for provision. Sometimes infrastructure facilitiesare also expected to exhibit production economies of scale,particularly during the construction phase of development.Together, these attributes not only suggest that governmenttake the lead in providing the infrastructure, they also consti-tute strong arguments in favor of a central government bureautaking the initiative.

A third reason often dted for highly centralized provisionof rural infrastructure facilities is that this arrangement ensuresthat the latest scientific information will be utilized in design-ing and building large-scale capital structures such as dams,road networks, power plants, and so forth. In some isolatedinstances, local residents have such knowledge, but expertise ismore commonly available to relatively few persons in develop-ing countries, most of whom are situated in central ministries.

In summary, the conclusion that central governmentsmust provide certain types of rural infrastructure is based onan analysis of why markets would be expected to fail to gener-ate sufficient investment in infrastructure. Such an analysisfocuses exclusively on a limited set of three problems to besolved-realizing economies of scale, utilizing modem technol-ogy, and reducing free riding-then compares two idealizedforms of organization: a simple market and a single govern-mental hierarchy. Because markets cannot solve these prob-lems for many types of rural infrastructure, it is presumed thata central government can. Policy prescriptions generated bysuch truncated analyses assign sole responsibility to a centralgovernment for providing and producing rural infrastructurefacilities.

14 /J-PART, January 1993

by guest on Septem


Dow

nloaded from



The work of institutional econo-mists such as Coase and Com-mons can be considered the foun-dation for the comparative analy-sis of markets and firms and thework of V. Ostrom and Tteboutthe foundation for comparativeanalysis in the public sector.

The results of these efforts to centralize the public sectorhave been disappointing. With a few striking exceptions inEast Asia, the economic health of most developing countrieseven after years of assistance is still fragile. Extensive relianceon central government ministries has generally stymied, in-stead of stimulated, efforts to find creative solutions to publicproblems including infrastructure development (Coward 1980;Chambers 1983; Cernea 1985; Uphoff 1986; Rondinelli 1987;Wunsch and Olowu 1990).

Analyses that conclude that only centralized governmentagencies can achieve economies of scale, utilize modern scien-tific knowledge, and solve the free-rider problem have ignoredimportant intermediate and overall performance criteria andinstitutional arrangements other than markets and states. Weconsider the significance of additional performance criteria inthe next section and then turn to a systematic analysis ofalternative institutional arrangements.

INTERMEDIATE AND OVERALL PERFORMANCE CRITERIA

Systematic comparisons of the performance of institutionalarrangements for providing and producing goods and serviceswith diverse characteristics is a recently developed and stillevolving field of inquiry.3 In the zero transaction cost environ-ment of neoclassical economics, open and competitive marketspush producers to seek combinations of land, labor, and capi-tal that produce private goods at their lowest possible cost perunit Similarly, consumers obtain and pay for the best mix-ture of goods and services obtainable given the resourcesavailable to them. Thus an open competitive market keepsproduction costs at as low a level as is feasible and distributesgoods to those who have the highest marginal value for them.Economic efficiency is achieved. Although markets generatesubstantial incentives to create wealth, markets do not auto-matically redistribute current resources from wealthier topoorer individuals.

Inclusion of transaction costs associated with the provisionand production of private goods has rendered comparativeevaluations of institutional arrangements a more complexundertaking. Coordination, information, and strategic costsneed to be considered, in addition to production costs. As aresult of including transaction costs in the analysis, someinstitutional arrangements that had previously been misunder-stood or considered inefficient have been evaluated morepositively (Williamson 1985). When transaction cost analysis isapplied to goods and services normally provided by the publicsector, the task of analyzing alternative institutional arrange-ments becomes still more complex.

15//-PAKT, January 1993

by guest on Septem


Dow

nloaded from


'Because we need to examine boththe production and provision ofrural infrastructure, we use theterm transformation cost for whatare most frequently referred to asproduction costs to avoid the awk-ward terminology of the produc-tion costs of production and pro-vision activities. North (1990) us-es similar language.


In Section 3 we compare several alternative institutionalarrangements. Our evaluations of the different arrangementsrely on the traditional criteria of efficiency, equity, and ac-countability as ultimate performance criteria. Assessing effi-ciency, accountability, and the two aspects of equity—fiscalequivalence and income redistribution—requires, however, afull examination of both benefits and costs. Only very detailedstudies can derive approximate measures of these performancecriteria. We believe, therefore, that it is more useful to exam-ine a set of intermediate performance criteria, which whentaken together constitute the transformation (production)4 andtransaction costs associated with the provision and productionof facilities with public goods characteristics, such as ruralinfrastructure.

On the production side of infrastructure development(including design, construction, operation, and maintenance),we identify the following types of costs:

TRANSFORMATION COSTS-the costs of transforminginputs into outputs (including the design, construction,operation, and maintenance of an infrastructure facility).These costs are affected directly by the characteristics ofthe good or service involved, the scale of production, andthe type of technology adopted, including its susceptibili-ty to breakdown.

TRANSACTION COSTS-increases in transformation costsassociated with (1) coordination, (2) gathering infor-mation, and (3) strategic behavior. These costs are affect-ed directly by the characteristics of the good or serviceinvolved, the scale of production, the technology used,and the particular rules that govern transactions.

Coordination costs are the sum of the costs of the time,capital, and personnel utilized in negotiating, monitoring, andenforcing agreements among actors. Information costs are thesum of the costs of searching for and organizing informationand the costs of errors resulting from a lack of or an ineffectiveblend of knowledge about:

1) time and place variables, and2) general scientific principles.

Strategic costs are the increased transformation costs thatare produced when individuals use asymmetric distributionsof information, power, or other resources to obtain benefits at


by guest on Septem


Dow

nloaded from



the cost of other individuals. The most common kinds ofstrategic costs related to production activities are:

1) shirking,2) adverse selection,3) moral hazard, and4) corruption (or fraud).

Although analytically separate, transaction costs arenormally recorded simply as part of the transformation orproduction costs of a firm. The time and personnel allocatedto transacting within a production firm or bureau and acrossproduction enterprises to organize production can vary sub-stantially from producer to producer—even those producing thesame mix of outputs and using similar technologies.

Providing goods and services also involves transformationand transaction costs (Silverman 1990). These costs are ig-nored frequently in private sector institutional arrangementsbecause they are borne largely by those who consume goodsand services. Provision costs in the public sector are, on theother hand, substantial even though they are recorded in manydifferent locations and are extremely difficult to identify andmeasure. The costs of provision include the following:

TRANSFORMATION COSTS-the costs involved in: (1)transforming citizen preferences and their willingness topay into articulated demands for packages of publiclyprovided goods and services, (2) arranging to finance andproduce these packages, (3) monitoring producers' perfor-mance, (4) regulating consumer use patterns, and (5)enforcing compliance with taxation and other resourcemobilization measures. Transformation costs are affecteddirectly by the characteristics of the goods and servicesinvolved, the scale of the provision unit, the technologiesused in aggregating interests, arranging financing andproduction, monitoring producers, regulating users, andenforcing compliance.

TRANSACTION COSTS-increases in transformation costsassociated with (1) coordination, (2) gathering informa-tion, and (3) strategic behavior. Transaction costs areaffected directly by the characteristics of the goods andservices involved, the scale of provision units, and thetechnologies used for interest aggregation, monitoring,regulating and policing, and the particular rules used togovern transactions.

W/J-PART, January 1993

by guest on Septem


Dow

nloaded from



Coordination costs are the sum of the costs of the time,capital, and personnel invested in negotiating, monitoring, andenforcing agreements about provision among actors.

Information costs are the sum of the costs of searchingfor and organizing information and the costs of errors result-ing from a lack of or an ineffective blend of knowledge about

1) time and place variables, and2) general scientific principles.

Strategic costs are the increased transformation costs thatare produced when individuals use asymmetric distribution ofinformation, power, or other resources to obtain benefits at thecost of other individuals. The most common kinds of strategiccosts related to provision activities are:

1) free riding,2) rent seeking, and3) corruption (or fraud).

Tradeoffs must be made between most of these costs andeven within broad categories of costs. For example, the totalcost of time and place information is composed of two compo-nents: the cost of errors due to insufficient local informationand the cost of acquisition and use of this information (cost ofsearch). In order to reduce the cost of errors resulting from alack of information about time and place variables, resourcesmust be invested in obtaining better information about localconditions. If one could accurately estimate both the cost oferror and the cost of search, it would be possible to make anoptima] investment in information search (i.e., to acquireinformation until the marginal cost just equaled the marginalbenefits of reduced error). Because participants in infrastruc-ture development rarely have access to such complete costinformation, we cannot assume that optimal investments aremade. Instead, we assume that the level and shape of the totalcosts of time and place information vary across physical do-mains and institutional arrangements. Thus farmers withmany years of experience in irrigating lands from a riveracquire intimate knowledge about variations in river flow andabout the soil types present in their fields. This time and placeinformation is acquired as a by-product of other activities.When these farmers undertake design, construction, and main-tenance activities, the total cost of time and place informationis low because both the cost of error due to the lack of place-specific information and the cost of information acquisition arerelatively low.

On the other hand, it is far more costly for employees ofnational ministries to acquire time and place information. For

WJ-PART, January 1993

by guest on Septem


Dow

nloaded from



them this type of information cannot be obtained as a by-product of day-to-day activities unless these officials are as-signed to a particular location for a long period of time (some-thing that rarely happens) and are highly motivated to seek itout. Consequently, we can assume that the costs of error dueto a lack of time and place information are higher. We antici-pate that the cost of acquiring relevant scientific informationwould be lower in a professionally staffed government bureau-cracy than in a farmer-managed irrigation system. Total infor-mation costs in any particular situation may or may not begreater for the bureaucracy; it will depend on the specificcircumstances.

The presumption that tradeoffs between intermediatecosts are required is an essential component of comparativeinstitutional analysis. It is never possible to reduce all costs tozero. If one institutional arrangement has lower costs withregard to one intermediate criteria, it is apt to have highercosts elsewhere. The key question is whether the differencessimply offset one another or whether a net gain is achieved.

ANALYZING INSTITUTIONAL ARRANGEMENTS

Although all societies make use of a rich array of institu-tional arrangements, an extremely sparse vocabulary is avail-able to describe these arrangements. As soon as one departsfrom the usual market versus state or centralization versusdecentralization dichotomies, one finds few well-acceptedterms to describe the variety of institutions that currentlyoperate—sometimes quite successfully-to sustain rural infra-structure. Considering such a limited set of institutions severe-ly handicaps the analyses of social problems as well as effortsto design and redesign institutions.

In analyzing an institutional arrangement, one mustinvestigate who is involved, what their stakes and resourcesare, and how they are linked to one another and to outcomes.Specifically, the types of actions that actors can take, the typeof information available to them, how actions lead to out-comes, and how rewards and punishments are allocated inlight of the outcomes achieved and the actions taken all re-quire identification. Then the analyst predicts the actions andaggregated outcomes that are most likely given the structure ofthe incentives. When the predicted actions and outcomes areverified in an empirical setting, the analyst has provided aninitial explanation for what is being observed. The welfareoutcomes can then be evaluated using various intermediateand overall performance criteria (Kiser and E. Ostrom 1982; E.Ostrom 1986).

19/J-PART, January 1993

by guest on Septem


Dow

nloaded from



In the remainder of this section, we illustrate how sixinstitutional arrangements-a simple market, a differentiatedmarket, a user group, a centralized government bureau, adecentralized government bureau, and a polycentric governingstructure—would be expected to perform in the provision andproduction of specific types of private and public infrastruc-ture.

Simple Versus Differentiated Markets

In order to demonstrate the application of the analysis ofintermediate transaction costs and overall performance criteria,it is useful to compare two types of markets for private goods.We call these simple and differentiated markets (Malone 1987).Exhibit 1 illustrates a simple market institutional arrangementfor the processes of providing and producing infrastructure.Each individual citizen consumer is responsible for provision,and each must seek out the producers (designers, builders,operators, and maintainers) necessary for each infrastructureproject he or she wants undertaken. Infrastructure financing isaccomplished through a series of quid pro quo exchangesbetween citizen consumers and the various artisans concerned.Similarly, the demand for various types of infrastructuredesign, construction, operation, and maintenance activity isarticulated directly by citizen consumers to the producerinvolved.

Exhibit 1A Simple Market for Infrastructure Development

Key: c = citizen-consumersd = designersb = builderso = operators-maintainers


by guest on Septem


Dow

nloaded from



The rules underlying such a market are quite simple. Allartisans and citizen consumers are permitted to be buyers orsellers at their own initiative. No one is allowed to take theproperty of others without mutual agreement on acceptableterms of exchange. Monitors to enforce property rights andarenas in which conflicts about property rights can be resolvedare essential to the operation of the market and to holding thevarious parties accountable.

If the only institutional arrangement available were asimple market, information asymmetries would reduce thenumber of beneficial trades actually transacted to provide andproduce a private investment. In a differentiated marketinstead of each consumer negotiating personally with numer-ous artisans, consumers may choose from a smaller number ofconstruction firms that employ artisans on longer-term con-tracts, monitor their performances, and attempt to build repu-tations for high performance. Some consumers may be able tochoose from a full array of potential producers and financialinstitutions; others may face a more limited set. We assumesufficient competition among producers and financial institu-tions such that no one has a monopoly position. In addition tofirms, some independent artisans continue to offer services aswell. Some consumers also may obtain loans from financialinstitutions after they have established credit- worthiness andsufficient property to secure the loan.

Exhibit 2A Differentiated Market for Infrastructure Development

$

m

Key: c =$ =mF =db =0 =

F $

d b o

= citizen-consumers= financial institution*= quality monitor= construction firms= designers= builders

1

/

F d b o\

Ar i ^

d b o

} (banks, credit rings, etc.)

= operators-maintainers

21 //-PART, January 1993

by guest on Septem


Dow

nloaded from



The financial institutions, in turn, have strong incentives toinspect the construction to ensure that quality standards aremet. In any particular setting, numerous rules will specify theconditions under which individuals can secure credit, create anew firm, monitor the performance of contractors, and soforth. Here (see Exhibit 2) we present highly simplified situa-tions and rule structures in order to illustrate a mode of analy-sis rather than the more detailed assessment necessary toanalyze a specific problem.

Transformation costs involved in a differentiated marketare likely to be lower than those in a simple market becausefirms can capture economies of scale unavailable to individualcraftsmen. Coordination costs are likely to be lower in thedifferentiated market because the number of potential linkagesinvolved in completing a project is lower. Individuals in-volved in either a simple or a differentiated market wouldhave access to accurate time and place information conveyedthrough the medium of prices. One could surmise, however,that artisans operating in a simple market are less likely tohave access to the latest scientific information than are firmsoperating in a differentiated market because firms can moreeasily afford to invest in the acquisition of technical informa-tion. Costs associated with adverse selection, moral hazard,producer fraud, and shirking problems are apt to be higher inthe simple market than in the differentiated market wherecounteracting institutions have been devised specifically tocope with these problems.

On the provision side for a purely private good, transfor-mation costs are low for both simple and differentiated mar-kets because each decision unit decides for itself what type andhow much of a good is desired. Coordination costs will belower in a differentiated market because the number of pro-ducers with whom one would need to communicate has beenreduced. Obtaining information about specific pre-ferences involves few costs in either case. Without intermedi-ate organizations on the provision side, such as consumers'unions, individuals may not have access to scientific informa-tion in either case. Free riding, rent seeking, and public cor-ruption are not problems in either a simple or a differentiatedmarket for a purely private good.

The total level of benefits achieved in the simple market islikely to be less than in the differentiated market becausemany potentially beneficial transactions would not be complet-ed without the help of the counteracting institutions involvedin the differentiated market. Thus the differentiated market fora private good is likely to be more efficient. In both types of


by guest on Septem


Dow

nloaded from



markets (for private goods), those who pay the costs are theprimary beneficiaries of the investment and no redistribution islikely to occur. The addition of banks, firms, and monitors ina differentiated market is apt to increase the overall account-ability of all participants in provision and production transac-tions.

Nongovernmental Institutional Arrangements ForInfrastructure Development

Now let us consider how three alternative non- govern-mental institutional arrangements would perform in the provi-sion and production of a jointly consumed infrastructure. Forthe purpose of exposition, we discuss the example of a small-scale irrigation system that benefits a group of citizen consum-ers. We contrast the simple market and the differentiatedmarket with a user group to which all using the irrigationsystem belong. A simplified representation of such an institu-tional arrangement is presented in Exhibit 3.

ExhibitsUser Group Organization for Infrastructure Development

Key: c = citizen-consumersUG = user group$ = financial institutions (banks, credit rings, etc.)m = quality monitorF = construction firmsd = designersb = builderso = operators-maintainers


by guest on Septem


Dow

nloaded from



This description of a user grouprule configuration is a simplifiedversion of the rules used in manysuch groups located in Nepal andthe Philippines. See Sly 0982),Martin and Voder (1983),Bagadion and Korten (1985), Cow-ard (1985), E Ostrom (1992), andTang (1992).

The production side of the user group structure remainsessentially the same as for the differentiated market, assumingthat user groups are not limited to a simple market structure.The major change between Exhibits 2 and 3 is the organizationof the citizen consumers themselves (the provision side). Thefollowing type of rule structure is assumed to underlie thissituation.5 A group of farmers decides to construct an irriga-tion system that will serve only those citizen consumers whoinitially buy shares in the enterprise or those who later pur-chase shares offered by the user group. Shareowners areassigned a weighted vote equal to the number of shares theyown. User group officials are chosen from among the mem-bers of the user group using some voting rule. When water isobtained it is distributed according to the number of sharesowned. Each shareholder must contribute a proportionateshare of resources, commodities, and/or funds, to the usergroup each year. These resources are used to pay persons tooperate the control works and to guard the canal so that wateris distributed according to an agreed-upon formula. Eachshareholder is also responsible for contributing a defined shareof labor when the user group collectively cleans out the canalsand undertakes any routine or emergency repair work on theirrigation system.

Assuming that the construction and operation of a smallirrigation system are within the technical competence of localdesigners, builders, and operators, any of the three sets ofinstitutional arrangements described above could be used toprovide and produce such a system. In general, the differ-entiated market performs better than the simple market, andthe user group structure built on top of the differentiatedmarket has the potential to perform better still. All of theintermediate costs on the production side remain the same fora differentiated market and for such a user organization.

The principal advantage of the user group arrangement isa reduction in the costs of free riding. If the user group is ableto exclude noncontributors from using irrigation water and isable to monitor and enforce the required contribution of monetary, commodity, and labor inputs in conformance with therules, the user group arrangement will be able to prevent freeriding; the two types of market arrangements will not. Theincentive to free ride, particularly in the provision of mainte-nance activities, is weak when a user group is effectivelyorganized. The members of a user group are able to monitoreach other's activities closely as a by-product of other routineactivities. The absence of a member who fails to join the


by guest on Septem


Dow

nloaded from



others in cleaning out the channels is easily noticed. His orher reputation as a reliable member of the community is affect-ed adversely. Although overt sanctions are used in usergroups to reduce free riding, the reduction in this behaviorlargely results from the low cost to all members of informationabout each other's activities and the importance of a goodpersonal reputation in such settings.

Transformation costs on the provision side must increasefor a user group as contrasted to a simple or differentiatedmarket. Farmers now must discuss their preferences and reacha common agreement about design and operating character-istics and maintenance strategies. If the farmers have similarinterests, coordination costs will remain relatively low. Coor-dination costs always will be higher where consumers actjointly than where they act individually. How much higherthe costs are depends on the homogeneity of the communityinvolved and the rules it uses to aggregate preferences.

Both rent seeking and corruption should be low when auser group organizes provision. The resources used to buildthe facility come from the group that benefits rather than froma public treasury to which others contribute. Given this con-gruence between those who own the input resources and thosewho enjoy the benefits of a facility, it is unlikely that a groupwould overinvest in new facilities. In fact, underinvestmentdue to uncertainty about benefits and costs is more likely thanoverinvestment. Underinvestment may occur also if the re-sources directly available to the user group are insufficient toundertake the investment and credit is not available.

Once collective provision is organized, some individualsmay invest in rent seeking activities to obtain a disproportion-ate share of benefits. If leaders have a disproportionate say inhow benefits are allocated, even local institutions may offeropportunities for rent seeking. Corruption is reduced substan-tially where user groups do not mobilize large amounts ofcash. A farmer has more control over how group resourcesare used if the farmer personally contributes required mainte-nance labor rather than giving an official cash to hire laborersto maintain the facility. Although a farmer can use crops tobribe irrigation officials to deliver more than the authorizedamount of water to his gate, an illegal payoff in crops is farmore likely to be noticed, and therefore less likely to happen,where operation and maintenance crews are composed of localfarmers,


by guest on Septem


Dow

nloaded from



Exhibit 4 contains a summary of these intermediate costsas well as our assessment of how these three arrangementscompare on overall performance criteria. In the upper panel ofthe exhibit we show how we would expect the intermediatecosts of production and provision to vary under the threeinstitutional arrangements. An L, M, or H in each row indi-cates whether the associated costs or performance levels wouldbe low, medium, or high. In this and in subsequent exhibits,an L should be interpreted as positive because it representsour judgment that an intermediate cost will be comparativelylow. As suggested above, the differentiated market providesseveral important advantages over a simple market Specifical-ly, transformation, coordination, and scientific informationcosts, on the production side, as well as costs due to adverseselection, moral hazard, fraud, or shirking should all be lessfor the differentiated market than for the simple market.

Exhibit 4A Centralized National Government forInfrastructure Development

DB CB OB DB CB OB DB CB OB

A A h\ / / \ A / \ \ A A kddd bbb ooo ddd bbb ooo ddd bbb ooo

Key: c = citizen-consumersG = central governmentSi = sectoral ministry (e.g., irrigation, agriculture)DB, CB, OB = design, construction, and operating

bureaus within each sector ministryd = designersb = builderso = operators-maintainers


by guest on Septem


Dow

nloaded from



ifany indigenous institutions likele zanjeres of the Philippinesive extensive internal mechani-ns to assure accountability (Siy>82; Coward 1980). When userroups are created by external au-lorities and rely primarily onlonetary resources rather than bi-nd resources, achieving high le-ds of accountability can be diffi-dt

We anticipate no difference in production costs betweenthe differentiated market and the user group. On the provi-sion side, however, transformation and coordination costs willbe greater for the user group than for either of the two privatemarket arrangements. The principal gain on the provision sideis the lower cost of combating free riding realized by the usergroup. The user group, however, generates higher transforma-tion costs and coordination costs than the differentiated mar-ket.

In the lower panel of Exhibit 4, we use four overall per-formance criteria-efficiency, fiscal equivalence, income redistri-bution, and accountability-to evaluate the expected outcomesunder the three institutional arrangements. An entry of "1"denotes the highest level of performance, whereas a "3" de-notes the lowest. Important differences between the usergroup and either type of market arrangement emerge from theanalysis. Given the potential for free riding where facilitieswith public goods characteristics are involved, an adequatesupply of potential benefits cannot be achieved until someform of collective provision is established with effective sanc-tions against those who do not contribute. Where free ridingoccurs, fiscal equivalence has not been achieved; instead, somesort of redistribution is taking place. It is the free riders, notnecessarily the poorest members of a community, who are thebeneficiaries of the redistribution.

Most user groups make some provision for temporarily reduc-ing the responsibilities for operating and maintaining a systemof households in distress. Differentiated market institutionsincrease accountability on the production side. User groupsthat maintain open records concerning who contributes howmuch labor and other resources and how monetary resourcesare spent increase accountability on the provision side. Theycan take advantage also of the more accountable productionside arrangements of the differentiated market'

Centralized Government Hierarchy ForInfrastructure Development

The inability of individuals relying strictly on marketinstitutions to overcome free riding has led many analyststorecommend that a central government bureau take responsi-bility for the provision of jointly used infrastructure facilities.This policy prescription frequently ismade without distinguish-ing between large- and small- scale facilities. Let us nowconsider the problem of designing, constructing, operating,maintaining, and using large-scale infrastructure facilities byexamining a fourth type of simplified institutional arrange-ment: that of a central government bureau. A diagram of thistype of institutional arrangement is shown in Exhibit 5.


by guest on Septem


Dow

nloaded from



Exhibit 5A Polycentric Arrangement for Infrastructure Development

DB

Addd

c

/S i

CB

Abbb

\C

: c

S2 S3

OB

/ \ooo

C (

c

: c

7/K

d b o

c <

\

: c

7//

$ F

Vm \\w

d b o

Key: c = dtizen-consumersG = national governmentSi = sectoral ministry (e.g., irrigation, transportation)DB, CB, OB = design, construction, and operating

bureaus within each sectoral ministryd = designersb = builderso = operators-maintainers$ = financial institutions (banks, credit rings, etc.)m = quality monitorF = construction firms

The organization of both sides of the provision-productionnexus is radically different from that of the three institutionalarrangements already considered. On the provision side,instead of a small set of citizen consumers acting independent-ly or collectively, numerous dtizen consumers residingthroughout an entire nation face a constrained set of choices.At periodic intervals, these citizen consumers select a set offull-time officials. Between elections, groups of citizen con-sumers can try to influence those officials to use their authorityto obtain desired benefits for their supporters. On the produc-tion side, another set of full-time officials is organized in aspecialized production department within a ministry organizedalong sectoral lines.

Both sets of officials are employed full time, and theirfuture careers depend on pleasing those who help them retainand /or advance their positions. The incentives facing bothsets of officials in a central government bureau can lead to a


by guest on Septem


Dow

nloaded from



Exhibit 6The Comparative Performance of Three Institutional Arrangements Related to theProvision and Production of Small-Scale Irrigation Systems

Intermediate Production CostsTransformation costsTransaction costs

Coordination costsInformation costs

Time and placeScientific

Strategic costsShirkingAdverse SelectionMoral HazardCorruption

Intermediate Provision CostsTransformation costsTransaction costs



Strategic costsFree RidingRent SeekingCorruption

Overall Performance CriteriaEfficiencyFiscal EquivalenceRedistributionAccountability

SimpleMarket

M*

H

LH

MMMM

L

H

MH

HLL

3**3

2 tol2

DifferentiatedMarket

L

M

LM

LLLL

L

M

LH

HLL

33

2 tol1

UserGroup

L

M

LM

LLLL

L+

L+ or M

LH

LLL

2 or 1131

Key: *L = low, M = medium, H = high (L is most desirable)**1 = high, 2 = medium, 3 = low (1 is most desirable)


by guest on Septem


Dow

nloaded from



The recorded production costs ofmany large-scale irrigation pro-jects are extremely high. Ratherthan challenge the well-acceptedpresumption that a centralizedgovernment bureau will captureeconomies of scale, we presumethat these recorded costs arelargely attributable to the highercoordination, information, andstrategic costs associated with thisarrangement

compounding of perverse consequences regarding decisionsabout the infrastructure facilities to be designed and construct-ed and the investments to be made in operation and mainte-nance activities. To focus our discussion, we will examine theincentives involved in the construction and the operation andmaintenance (O&M) of a large-scale irrigation system (Coward1980; Harriss 1984; Wade 1984; Ascher and Healy 1990).Exhibit 6 contains our anticipated performance results for acentral government bureau engaged in both the constructionand the O&M of such a system.

Anticipated Performance On The Production Side

Whether a national bureau contracts with private enterprises toconstruct large-scale irrigation systems or constructs them withits own personnel, we assume that it is able to capture econo-mies of scale. Thus we presume that the transformation costsof construction will be low.7 It is more difficult to justify theassumption that the transformation costs of operation andmaintenance will, for the same reason, also be low. Neverthe-less, we adopt this second assumption as well in order to focusattention on transaction costs.

We expect the cost of coordinating production activities tobe higher than in our earlier examples because lower-levelbureaucrats in a centralized agency must obtain the permissionof more senior officers before proceeding with many tasks.Efforts to enhance the accountability of government agencies,which frequently require that all stages of an infrastructureproject be reviewed by various officials, also increase coordina-tion costs substantially.

The cost of securing the relevant time and place informa-tion would be high in most central bureaus for both construc-tion and O&M activities. Design engineers would find itdifficult to acquire accurate information about all pertinentlocal conditions from available maps and other data. Obtain-ing local information relevant to maintenance is even moredifficult for a central bureau. Minor failures in an irrigationsystem can occur anywhere, at any time; maintenance person-nel located at some distance from a site may not know when afailure occurs. By the time information about a problem isobtained by repair crews, a minor repair job may have growninto a major reconstruction project. As hypothesized above,the cost of securing scientific information should be relativelylow for employees of a government bureau.

We expect the problem of shirking to be higher in agovernment bureau than in private markets or user groups.


by guest on Septem


Dow

nloaded from



Even with the best intentions, civil servants frequently findthemselves in situations where demands on their time andenergy force them to restrict severely the time they spendsupervising projects. One way to reduce project monitoringtime is to undertake a smaller number of larger projects, whe-ther or not this results in higher or lower costs of producinginfrastructure facilities. Regardless of whether the bias towardsupervising a few large projects is attributed to shirking, highcoordination costs, or a combination of both, overall efficiencycan be reduced when small, productive projects are rejected infavor of large projects.

We also expect shirking to be higher in operations andmaintenance divisions than in design and construction divi-sions. All employees are likely to find some tasks more fulfill-ing than others; most engineers gain greater satisfaction fromdesigning and constructing systems than from overseeing theirmaintenance. Similarly, bureaucratic rewards may favor thosewho engage in highly visible design and construction projectsrather than those who organize effective routine maintenanceregimes. Finally, because it is difficult to monitor how mainte-nance engineers or their staffs spend their time, fewer bureau-cratic punishments can be administered to those who spendtime on personal tasks rather than on maintenance. Thus,although internal and external incentives can counteract shirk-ing in construction, few internal or external incentives counter-act shirking in O&M.

Both construction and O&M activities provide opportuni-ties for corruption. In general, the amount of money in illegalside-payments that is available to an official is determined byhow much money is spent through any given post as well ashow much can be raised from outside the budget, often direct-ly from farmers (Wade 1984,1985; Jagannathan 1987). Personswho hold positions in the higher reaches of the ministry hier-archy, where responsibility for high cost construction projectsis often lodged, tend to profit more from construction projectsthan do persons at lower levels. Lower level officials may,however, profit from the letting of contracts paid for throughthe O&M budget and by extracting resources directly fromfarmers. The difference in the quantity and quality of cropsthat can be grown with the aid of a regular supply of waterand those that can be grown with an irregular supply meansthat many farmers are perfectly willing to pay a regular fee toa lower-level irrigation official to ensure the predictable deliv-ery of water to their fields.

31 /J-PART, January 1993

by guest on Septem


Dow

nloaded from



The incentives that public officials face must be under-stood in the context of the generally low salaries they receive,their limited career advancement opportunities, their poorworking conditions, and the opportunities (both legal andillegal) available to them in different types of work assign-ments. Most government employees prefer to live in thecapital of their country, where they and their families canobtain better health services and education. Their own careeropportunities may be enhanced by working where they can beobserved by superiors, rather than by working in the country-side. This leads to a strong preference for work associatedwith the design and construction of large infrastructure sys-tems that generate many employment positions in the centralbureau. Indeed, they are once assigned to the countryside,many officials continue to maintain a household in the capitalcity for school-age children and devote considerable timetrying to get reassigned to the capital.

The working conditions of officials assigned to an O&Mdivision in the countryside are also more difficult than those ofofficials working in the capital. The commitment of govern-ments of many developing countries to use public funds toreduce unemployment, combined with the extreme budgetaryconstraints of these countries, has resulted in bureau budgetsthat are almost totally committed to salaries. It is hardlysurprising that public officials responsible for infrastructureO&M are unable to supervise these processes properly wherethey lack the funds necessary to enable them simply to visitfacilities in their jurisdiction on a regular basis.

Given low salaries and the high cost of keeping twohouseholds, the incentive to search out opportunities to earnextra funds while in the countryside is also understandable.Relatively few opportunities arise from investments of timeand energy to maintain an existing system. The opportunitiesfor additional income arise where goods and services can bewithheld from potential beneficiaries unless side payments aremade, or where the processes of awarding contracts and/orcertifying contractor performance are organized so that officialscan demand side payments from contractors (Jagannathan1987; Loft 1989).

This is not to say that all administrative officials in centralgovernment bureaus are corrupt shirkers. Many devotedpublic officials everywhere resist the temptations afforded bythe systems in which they find themselves. This analysis doessuggest, however, that the bureaucratic arrangements found inhighly centralized governments produce incentives for officialsthat enhance the opportunities for corruption and shirking; at


by guest on Septem


Dow

nloaded from



"Whether the central governmentis elected or not is not crudal tothe analysis and does not affectour conclusions. The officialswithin a military government, orany other form of nonelectedgovernment, are still faced withthe problem of allocating scarceresources to different districts andgroups in a society. WhetheT theystay in office and improve theirpositions depends on satisfyingthese relevant groups.

the same time they generate few counteracting pressures thatdiscourage these practices. Furthermore, once shirking andcorruption are well established, those who attempt to fight theproblem from within are vulnerable to retaliation by theirpeers and their superiors. Consequently, corrupt behavioroccurs frequently in both construction and O&M.

Anticipated Performance On The Provision Side

Here we examine the transformation of preferences andwillingness to pay into specific activities involved in infrastruc-ture development and the monitoring and enforcement ofprocedures regulating operation and use. Assuming a compet-itive, democratic selection process, each adult citizen consumerparticipates in the election of national officials who run onplatforms representing their promised positions regardingfuture investments of public funds. One cannot presume,however, that a national electoral outcome is a clear articula-tion of majority preferences about the proposed investmentsand allocations to be made from a central treasury. Morerealistically, the outcome is the selection of a set of actors whowill each try to obtain as much as possible for the group he orshe must please in order to stay in office. Even if this behav-ior were not endemic, the problems of aggregating the prefer-ences of heterogeneous populations mean that the outcome ofa national election cannot provide a reliable indication of themost preferred set of public policies to be pursued.*

The costs of obtaining time and place and scientific infor-mation are high when provision is organized by one very largeunit. Citizens located in one setting cannot know much aboutwhat is needed elsewhere, and officials know little about thepreferences of citizens other than those who are highly moti-vated and organized to obtain disproportionate benefits.Moreover, gaining reliable, scientifically grounded informationabout the effectiveness of various policies is very costly whenpolicies must be adopted for an entire country before anyexperimentation in smaller jurisdictions has occurred.

Many taxes levied in developing countries are collected bythe central government and placed in a general fund fromwhich most expenditures are allocated. The general fundbecomes a "common pool" for all central government officials.Because the funds are seldom earmarked for particular publicsector activities, all elected officials are motivated to obtain asmuch of the general fund as possible for projects that benefit


by guest on Septem


Dow

nloaded from



The National Irrigation Agenciesin South Korea and Taiwan areamong the better performingcentral government agencies withregard to both construction andO&M. USAID investments inKorean irrigation projects haveresulted in well-engineered pro-jects, completed close to scheduleand resulting in predicted increas-es in crop yields. Steinberg et al.(1980) argue, however, that theinvestment in large-scale irriga-tion, as compared to otheT poten-tial investment, cannot be consid-ered efficient given the highlysubsidized price for rice. InTaiwan, fanners are well orga-nized at several levels and aremuch more involved as keyparticipants in managing irriga-tion systems than they are in mostcountries where a centralizedgovernment agency plays a majorrole. See Levine (1980).

their constituents. When selecting projects to be funded, eachofficial weighs the observable benefits of specific projects morethan the costs. Many benefits are visible and can be directedtoward specific sets of supporters; costs are relatively hiddenand can be spread across all taxpayers. Some form of logroll-ing among elected officials will occur depending on the specif-ic set of rules used to make central government decisions.Absent strong institutional constraints on logrolling, it is high-ly likely that this activity will lead to substantialoverinvestments in some types of public projects and majorunderinvestments in others (Weingast, Shepsle, and Johnson1981).

The greatest advantage of governmental provision overprivate provision is presumed to be a government's capacity toprevent free riding. Yet even this is not always the case. Forexample, the actual revenue collected from farmers in manydeveloping countries falls far short of covering the O&M costsof government-run irrigation systems, let alone contributing tocapital costs. In Bangladesh, farmers contribute only 13 per-cent of the costs of O&M; in Thailand, only 26 percent (Repetto1986, 5). It is difficult to interpret the repeated findings thatfarmers do not contribute required fees. The likelihood thatfarmers will comply with government imposed taxes or feesdepends on many factors, including whether perceived benefitsare at least as great as the required assessment. Obviously,governments in many developing countries have managed tosolve neither shirking problems on the production side norfree riding problems on the provision side.

Anticipated Overall Performance

In regard to our overall performance criteria, efficiencyvaries from low to high across projects, and fiscal equivalenceis rarely achieved. Given the difficulties outlined above in re-gard to intermediate criteria, it is not surprising that centralgovernment agencies in most countries cannot be expected toperform at a high level with regard to any of the four overallperformance criteria arrayed in the lower panel of Exhibit 6.'Whether the poor are subsidized to some extent by thewealthy is almost impossible to determine without detailedanalysis. But all too frequently, careful analyses have insteadshown redistribution to be in the opposite direction (Akin,Birdsall, and de Ferranti 1987, 27).


by guest on Septem


Dow

nloaded from



DOES ADMINISTRATIVE DECENTRALIZATION IMPROVETHE PERFORMANCE OF CENTRALIZED GOVERNMENTS?

The problems discussed above have inspired policies torelocate ministry personnel to the field and to deconcentratededsionmaldng authority in order to give employees workingin regional or subregional offices a greater role in deciding thecharacter of infrastructure development and maintenance. Insome instances, local representative councils also have beencreated to serve as advisory bodies for ministry employees. Inthose cases in which law-making authority is exercised bycouncils, ministries often retain administrative veto power overlegislation adopted. Because local councils are not intended tobe independent governments with law-making or law-enforc-ing authority, structural changes associated with administrativedecentralization are essentially changes in the production sideof the original, centralized government bureau with littlechange in provision arrangements. Employees are merelyshifted geographically from headquarter locations to fieldlocations.

The dispersal of administrative offices inevitably places alarger number of officials in closer physical contact with theresidents of communities where infrastructure developmentswill occur and with members of local councils. The time andeffort required of officials who wish to leam more about localcircumstances is thereby reduced. Administrative reorganiza-tion could be expected to improve the performance of anofficial who is assigned to the same area in which he wasraised and in which the principal economic interests of theofficial or his immediate family are located. It would, in thiscase, be in the interest of the official to provide infrastructurefacilities that benefit him. Such facilities, however, might ormight not be considered appropriate by the membership of alocal advisory council or by most residents of the jurisdiction.In fact, many developing countries routinely post employeesoutside their home areas in order to deter officials from usingtheir positions to promote their own economic interests orthose of local, traditional leaders within their jurisdictions.

Administrative reorganization does not, however, substan-tially alter the nature of the career tracks within the ministries.Lower level officials are still concerned about how their perfor-mances are assessed by senior officers responsible for promo-tions and disciplinary action. Reorganization does not alter thefact that the best schools and hospitals are located in the capi-tal city. It also may not alter the information that officialsbring to bear on their decisionmaking.


by guest on Septem


Dow

nloaded from



Although the village where a new road or irrigationsystem is to be built is located nearby, an official still mustmake some effort to learn something about the area. If theofficial is convinced that local people know nothing of conse-quence about economic development or knows that seniorofficials strongly prefer a particular course of action, the prox-imity of a village may not result in improved informationabout village circumstances. Members of a local council withconnections to senior officials or political leaders are in aposition to make local level public service employees take theirviews into account. They may not, however, have the interestsof others at heart when they exert their influence but, rather,act to advance their own interests. Thus patron client linksbetween public officials at different levels and between localand national political leaders, and not a local consensus, maydetermine the allocation of funds among projects.

Although some decentralization projects have involvedtemporary shifts of some joint authority to local level officialsand to the citizens involved in a project—often a project fundedby an international donor—few such shifts have been retainedafter the end of the projects. Most decentralization efforts haveinvolved little more than shifting personnel from headquartersto field locations (Cheema and Rondinelli 1983; Mawhood1983; Olowu 1990). If time and place information is not takenmore seriously than before decentralization, the policy isunlikely to affect the intermediate costs involved in either theconstruction or maintenance of a large-scale irrigation system.The performance of decentralized agencies is thus likely to besimilar to that of centralized agencies, unless major structuralchanges are made to assure considerable joint autonomy andresponsibility for lower-level officials and the citizens theyserve.

Polycentric Institutional Arrangements

Although they were designed to reduce the costs of ac-quiring time and place information or the costs of errors dueto a lack of this information, most administrative decentraliza-tion efforts have failed to accomplish these objectives. Admin-istrative decentralization has failed also to reduce the highcosts of strategic behavior-shirking, corruption, and rentseeking. An analysis of the structure of authority relationshipsmakes it clear that centralized and administratively decentral-ized arrangements are virtually identical. Both are unicentricsystems (i.e., they rely on long hierarchical chains of commandwith a single, ultimate center of authority).


by guest on Septem


Dow

nloaded from



A substantial reduction in time and place information andstrategic costs requires a change in the basic structure of au-thority. Polycentric systems would appear to provide this typeof change (V. Ostrom 1989). A polycentric governing structureoffers citizens the opportunity to organize not one but manygoverning authorities, each of which operates independentlywithin a circumscribed scope of authority. The scope of au-thority of different governments in a polycentric system canvary enormously. Some could be general purpose authoritiesproviding a wide array of public services whereas others maybe special purpose authorities that provide for, operate, andmaintain a single facility. Every government is a provisionunit Each must decide whether to organize its own produc-tion bureaus or contract with other public bureaus or privatefirms that produce a particular good or service. Exhibit 7illustrates some of the potential diversity of relationshipsbetween citizens and governments and between governmentsin such a system.

Periodic elections provide voters with the opportunity toselect governing officials they believe will supply them withappropriate goods and services. The fact that principal offi-cials are elected means that an official in one jurisdiction doesnot function as a superior to officials in other jurisdictions andthus does not control their career paths. Hierarchical relation-ships will exist within any one jurisdiction, but the long chainsof such relationships associated with loss of information andcontrol will not Disputes between locally elected officialsmust be settled in courts or other institutional arenas thatenjoy substantial independence, not by a senior bureaucrat.(For a classic analysis of nonbureaucratic control seeTocqueville 1945,1, ch. 5.)

Anticipated Performance On Intermediate Criteria

Polycentric institutional arrangements sacrifice few of theadvantages of centralized governing systems and generateadditional advantages of their own. Governing authorities canstill penalize free riders. The opportunity to contract withlarger jurisdictions or with private firms provides a convenientway of purchasing scientific expertise as it is needed. Theelections by which local authorities are selected, if competitive,provide a means of aggregating time and place informationthat is imperfect but superior to a reliance on advisory coun-cils. Once elected, officials face strong incentives to keep atleast a majority of their constituents happy in order to securetheir reelection. Retaining popularity requires that officialskeep themselves well informed about changing preferences.Local candidates obviously also possess considerable localinformation.


by guest on Septem


Dow

nloaded from



Exhibit 7The Performance of a Central Government Bureau Related to the Construction andO&M of a Large-Scale Irrigation System

Intermediate Production CostsTransformation costsTransaction costs



Strategic costsShirkingAdverse SelectionMoral HazardCorruption

Intermediate Provision CostsTransformation costsTransaction costs



Strategic costsFree RidingRent SeekingCorruption

Overall Performance CriteriaEfficiencyFiscal EquivalenceRedistributionAccountability

Construction

L*

MtoH

HL

MLLH

M

M

HH

LtoMHH

3 t o l "3?

3 to 2

Operation &Maintenance

L

MtoH

HL

HLLH

M

M

HH

LtoHLL

3 to 23?

3 to 2

Key: *L = low, M = medium, H = high (L is most desirable)**1 = high, 2 = medium, 3 = low (1 is most desirable)


by guest on Septem


Dow

nloaded from



The fact that local officials are elected also reduces strate-gic costs because officials know that citizens who wish toremove corrupt or lazy officials need not rely on the coopera-tion of senior administrators. In addition, larger numbers ofofficials operating independently in a larger number of juris-dictions reduce the likelihood that any one of them can gainmonopoly control over important public goods or services.

Coordination costs among authorities in different jurisdic-tions could increase because authorities can refuse to go alongwith proposals made by others. Conflict over different policiesbetween independent units is, however, likely to be more openthan conflict that regularly takes place within bureaus. Howmuch coordination costs actually rise depends on whetherrelationships are structured so as to encourage mutually pro-ductive or zero sum interaction. Furthermore, if coordinationcosts rise considerably but result in the refinement of a policythat in turn produces an improved outcome, the increasedcoordination costs may still be well worth the price.

Anticipated Overall Performance

If local jurisdictions are able to exert more control overthe design of their own infrastructure investments, one wouldexpect the level of benefits generated by these investments torise. The overall efficiency of polycentric arrangements wouldthen depend on the ability of local jurisdictions to containcosts. For that reason, special purpose and general purposejurisdictions must have substantial independence over mattersthat determine administrative costs. Redistribution decisionsmust be considered carefully so that resource transfers do notdiscourage local efforts to economize.

The existence of multiple governments with some inde-pendent taxing authority makes fiscal equivalence much easierto achieve in a polycentric system. Redistribution of resourcesamong jurisdictions can be achieved through the agency ofoverarching jurisdictions where public support for redistribu-tion exists. Citizens aware of the threat to productive relation-ships that severe inequalities pose could be expected to viewsome types of redistribution as good investments. The ac-countability of elected officials to majority opinion can beexpected to be high, due to the ease of removing unresponsiveofficials. The accountability of officials to minorities woulddepend heavily on the skill with which jurisdictional bound-aries are drawn, how the system of judicial appeal is constitut-ed, and the accessibility of judicial remedies. Indeed, themaintenance of a polycentric arrangement requires a high levelof investment in adjudication services and in the education ofa citizenry in the use of these services for the purpose ofdefending rights.


by guest on Septem


Dow

nloaded from



THE ROLE OF DONOR AGENCIES

An analysis of institutional arrangements for infrastruc-ture development in the developing world would be incom-plete without considering the impact of donor agencies onincentives. Anyone observing donor-financed infrastructureprojects in operation is struck by the number of extremely hardworking, highly motivated individuals, employed in both hostgovernments and donor agencies, whose principal goal isclearly to improve the wellbeing of poor people. Yet assess-ments of projects designed by donor and host government staffrepeatedly show that many of the projects increased or rein-forced the already high concentration of public authority in therecipient country, were poorly designed given local circum-stances, generated few economic benefits for the target popula-tion, and produced inappropriately large debt burdens (Asher1970; Hayter 1971; Wall 1973). How is it possible for highlymotivated, hard-working people, who sincerely want to im-prove conditions in the recipient countries, to be involvedrepeatedly in the design and implementation of projects thatdo not accomplish these goals?

This question has been the subject of several carefulstudies (Nelson 1968; Tendler 1975), and we cannot fullyaddress it here. By identifying the incentives facing actors inthe donor agencies as they relate to public officials and privateenterprises in the host countries, one can suggest how theseconsequences could be generated. Some of these incentives arepresent in most large-scale bureaucratic agencies. Others areunique to the foreign aid domain or even to particular donoragencies, such as the U.S. Agency for International Develop-ment (USAID).

USAID, like other bilateral donors, faces the apparentlycontradictory problems of creating a reliable constituency insupport of foreign assistance, while seeking appropriate waysto spend the large amount of foreign aid money mandated byCongress. The legislative mandate to allocate a proportion offoreign assistance to the purchases of equipment made in theUnited States was an important initial source of considerablebias toward very large and capital-intensive projects. Thenotorious need to "move money" faced by all governmentagencies in which next year's funding depends on the agency'sability to spend this year's budget has reinforced this bias anddiscouraged labor intensive projects and those using small-scale and locally manufactured equipment. The difficulty ofcoping with a gushing "pipeline" has recently been exacerbatedby policies designed to reduce agency staff. Fewer people arenow available to spend more money.

AO/J-PART, January 1993

by guest on Septem


Dow

nloaded from


"Far major exceptions, see Ascherand Healy (1990), Chambers(1988), and Repetto (1986).


Fear of external criticism about lack of control and corruptpractices also leads donor agencies to favor project designs thatappear to give project monitors greater control over what ishappening. This usually means funding a few large projectsrather than several small ones. The concern for monitoringperformance, particularly to reduce corruption, requires projectperformance criteria that can be measured easily. Concernabout accountability, added to the universal difficulty of mea-suring public service output, is likely to lead to a focus onproject inputs, rather than on outputs (Connerley et al. 1989;Fitzcharles and Connerley 1990).

Similar preferences of both donor and host governmentagencies for projects that are easy to monitor and equipmentintensive generate incentives for personnel that are similar andprobably reinforcing. The types of projects that receive themost funding by external donors can thus be explained by thenature of the incentives inside donor agencies and host gov-ernments without resorting to conspiracy theories.

SUMMARY AND CONCLUSIONS

In this article we have illustrated how theoretical princi-ples drawn from the new institutional economics and institu-tional analysis can be used to analyze the incentives generatedby alternative institutional arrangements for the supply ofrural infrastructure in developing countries. The new institu-tional economists have demonstrated the value of systematical-ly examining the intermediate costs of transacting in an uncer-tain environment. However, most of them have analyzed onlytransactions in private goods.

Our comparative institutional analysis examined theintermediate transaction costs of providing and producingpublic goods. We considered the costs of coordination, infor-mation gathering (both scientific and time and place informa-tion), and various strategic behaviors especially associated withpublic institutional arrangements. This required an assessmentof the costs of the types of strategic behavior potentially in-volved in both public good provision—free riding, rent seeking,and corruption—and public good production—shirking, adverseselection, moral hazard, and corruption. Most of these costshave received very little attention in current analyses of infra-structure deterioration (U.S. General Accounting Office 1983).

We then used relative measures of intermediate trans-action costs to compare the performance of six institutional

4\/J-PART, January 1993

by guest on Septem


Dow

nloaded from



arrangements for the supply of public infrastructure. In lightof the more complete set of intermediate performance mea-sures utilized here, the unintended consequences of currentinstitutional arrangements for infrastructure development canbe seen to be expected outcomes given the incentives of thevarious participants. We conclude that if greater attention isnot given to a fuller range of transaction costs and a morediverse set of institutional arrangements, projects will continueto be planned, financed, and constructed in ways that generateunintended consequences such as high levels of corruption,overinvestment in large-scale projects, and underinvestment insmaller projects and in operations and maintenance.

Recognition of a wider array of transaction costs does notimply that there are cost free ways of reducing them, however.In order to reduce the cost of errors due to a lack of informa-tion, costs must be incurred to obtain more information.Institutional reforms aimed at reducing the costs of one type ofstrategic behavior, such as free riding, can create improvedopportunities for other types of strategic behavior, such as rentseeking or fraud. When searching for a superior institutionaldesign, the analyst attempts to guard against excessive costs ofone type without expending more resources or opening thepossibility for even worse problems of a different order.

Given the constrained sets of both performance criteria(production economies, access to technical knowledge, andcontrol of free riding) and institutional arrangements (marketsand central government bureaucracies, which characterizemuch current analysis), the tendency to presume that the needfor public sector involvement requires a reliance on centralgovernment bureaus is understandable. Widening the set ofintermediate performance criteria to be examined, however,reveals the necessary tradeoffs that a reliance on this particulararrangement entails. And enlarging the set of institutionalarrangements to be considered increases the probability ofidentifying alternative arrangements that yield genuine net costreductions.

Adelman, lima, and Thorbecke,Erlk,eds.1989 The Roie of Institutions

in Economic Development.Oxford: Pergamon Press.

Advisory Commission on Intergov-ernmental Relations.1987 The Organization of Load

Public Economies. Washing-ton, D.C: Advisory Commis-sion on IntergovernmentalRelations, A-109.

REFERENCES

1988 Metropolitan Organiza-tion: The St. Louis Case.Washington, D.C: AdvisoryCommission cm Intergovern-mental Relations, M-158.

Akin, John; Birdsall, Nancy; and deFerranti, David.1987 Financing Health Services

in Developing Countries.Washington, D.C: Tne WorldBank.

Ascher, W., and Healy, R.1990 Natural Resource

Polkymakatg: A Frameworkfor Developing Countries.Durham, N . O Duke Univer-sity Press.


by guest on Septem


Dow

nloaded from



Asher, Robert E1970 Development Assistance in

. the Seventies: Alterna-tives for the UnitedStates. Washington, D.C:Department of State.

Bagadion, Benjamin U., and Korten,Frances F.1985 "Developing Irrigator's Orga-

nizations: A Learning ProcessApproach to a ParticipatoryIrrigation Program." In Mi-chael Cernea, e d New YorlcOxford University Press, 52-90.

Cernea, Michael M., ed.1985 Putting People First:

Sociological Variables inRural Development. NewYork; Oxford UniversityPress.

Chambers, Robert.1983 Kura/ Development: Putt-

ing the Last First. London:Longman.

1988 Managing CanalIrrigation: PracticalAnalysis from South Asia.Cambridge: Cambridge Uni-versity Press.

Cheema, G. Shabbir, and Rondinelli,Dennis A., eds.1983 Decentralization and De-

velopment: Policy Imple-mentation in DevelopingCountries. Beverly Hills,Calif.: Sage.

Connerley, Ed, and associates.1989 Bangladesh Rural and

Feeder Roads Sector As-sessment. Prepared for theU.S. Agency for InternationalDevelopment/Dhaka.Burlington, Vt.: Associates inRural Development.

Coward, E Walter, ed.1980 Irrigation and Agricul-

tural Development in Asia.Ithaca, N.Y.: Cornell Universi-ty Press.

1985 Technical and Social Changein Currently IrrigatedRegions: Rules, Redes, andRehabilitation." In MichaelCemea, ed., 27-52.

Fitzcharles, Ann, and Connerley, Ed.1990 Review of the USAID/CARE

IFFW Program. Prepared forthe Office of Project Develop-ment and Engineering, UnitedStates Agency for Internation-al Development, Dhaka, Ban-gladesh. Burlington, Vt: Asso-ciates in Rural Development

Harriss, J. C1984 "Social Organization and

Irrigation: Ideology, Planningand Practice in Sri Lanka'sSettlement Schemes." In T. P.Bayliss-Smith and S.WanmaH, eds. Understand-ing Green Revolutions.Cambridge: Cambridge Uni-versity Press, 315-38.

Hayter, Teresa.1971 Aid as Imperialism. Balti-

more: Penguin Books.

Jagannathan, N. Vijay.1987 Informal Markets in De-

veloping Countries. NewYork: Oxford UniversityPress.

Kiser, Larry L, and Ostrom, Elinor.1982 The Three Worlds of Action:

A Metatheoretical Synthesis ofInstitutional Approaches." InElinor Ostrom, ed. Strate-gies of Political Inquiry.Beverly Hills, Calif.: Sage,179-222.

Levine, Gilbert1980 The Relationship of Design,

Operation and Management"In E Walter Coward, ed.Irrigation and Agricul-tural Development inAsia: Perspectives fromthe Social Sciences. Ithaca,N.Y.: Cornell UniversityPress, 51-62.

Loft, Mkhael V.1989 Road Contractor Study: A

Rapid Institutional As-sessment of Rural andFeeder Road Constructionin Bangladesh. Prepared forthe U.S. Agency for Interna-tional Development/Dhaka.Columbia, S.C: Wilbur SmithAssociates.

Malone, Thomas W.1987 "Modeling Coordination in

Organizations and Markets."Management Science33:10:1317-33Z

Martin, Edward, and Yoder, Robert.1983 "Review of Farmer-Managed

Irrigation in Nepal." In WaterManagement in Nepal: Pro-ceedings of the Seminar onWater Management Issues,July 31-August 2.Katmandu: Ministry of Agri-culture, Agricultural ProjectsServices Centre, and the Agri-cultural Development Coun-cil, 82-91.

Mawhood, Philip, ed.1983 "Decentralization: The Con-

cept and the Practice." InLocal Government in theThird World, New York:John Wiley and Sons, 1-24.

Musgrave, Richard A.1959 The Theory of Public Fi-

nance. New York: McGraw-Hill.


by guest on Septem


Dow

nloaded from



Nelson, John M.1968 Aid Influence and Foreign

Policy. New York:Macmillan.

Nicholson, N. K., and Connerley,E.F.1989 T h e Impending Crisis in

Development Administra-tion." International Jour-nal of Public Administra-tion 1£3:38S425.

North, Douglass C.1990 Institutions, Institu-

tional Change and Econom-ic Performance. New York;Cambridge University Press.

Oiowu, Dele.1990 "The Failure of Current De-

centralization Programs inAfrica." In James S. Wunschand Dele Oiowu, eds., 74-99.

Ostrom, Elinor.1986 "A Method of Institutional

Analysis." In Franz-XaverKaufmann, GiandomenicoMajone, and Vincent Ostrom,eds. Guidance, Control,and Evaluation in thePublic Sector. New York:Walter de GruyteT, 459-75.

1992 Crafting Institutions forSelf-Goveming Irriga-tion Systems. San Francis-co: ICS Press.

Ostrom, Elinor; Schroeder,LarTy; and Wynne, Susan.Institutional Incentivesand Sustainable Develop-ment. Boulder, Colo.:Westview Press. Forthcom-ing.

Ostrom, Vincent.1989 The Intellectual Crisis

in American Public Admin-istration, 2d ed.Tuscaloosa: University ofAlabama Press.

Ostrom, Vincent; Bish, Robert; andOstrom, Elinor.1988 Local Government m the

United States. San Francis-co: ICS Press.

Ostrom, Vincent; Feeny, David; andPicht, Hartmut, eds.1988 Rethinking Institutional

Analysis and Development:Issues, Alternatives, endChoices. San Francisco: ICSPress.

Ostrom, Vincent; Tlebout, CharlesM.; and Warren, Robert.1961 The Organization of Govern-

ment in Metropolitan Areas:A Theoretical Inquiry." Amer-ican Political ScienceReview 55:4:831-42.

Repetto, Robert.1986 Skimming the Water: Rent-

Seeking and the Perfor-mance of Public Irriga-tion Systems. Research Re-port No. 41. Washington,D.C.: World Resources Insti-tute.

Rondinelli, Dennis A.1987 Development Administra-

tion and US. Foreign AidPolicy. Boulder, Cok>.:Lynne Rienner.

Silverman, Jerry M.1990 Publk Sector Decentral-

ization: Economic PolicyReform and Sector Invest-ment Programs. DivisionStudy Paper No. 1. Washing-ton, D.C: The World Bank.

Siy, Robert Y.1982 Community Resource Man-

agement: Lessons from theZanjera. Quezon dry: Uni-versity of Philippines Press.

Steinberg, David I.; Morrow, RobertB.; Palmer, Ingrid; and Dong-U,Kim.1980 Korean Irrigation. Project

Impact Evaluation No. 12.Washington, D.C: VS. Agen-cy for International Develop-ment.

Tang, Shui Yan.1992 Institutions and Collec-

tive Action: Self-Gover-nance in Irrigation. SanFrancisco: ICS Press.

Tendler, Judith.1975 Inside Foreign Aid. Balti-

more: Johns Hopkins Univer-sity Press.

Tocqueville, Alexis de.1945 Democracy in America, vol.

I. Edited by Philip Bradley.New York: Alfred A. Knopf.


by guest on Septem


Dow

nloaded from


Uphoff, Norman T., ed.1986 Local Institutional De-

velopment: An AnalyticalSourcebook with Coses.West Hartford, Conn.:Kumarian Press.

U.S. General Accounting Office.1983 Irrigation Assistance to

Developing CountriesShould Require StrongerCommitments to Operationand Maintenance, Washing-ton, D.C: General AccountingOffice.


Wade, Robert1984 "The System of Administra-

tive and Political Corruption:Canal Irrigation in SouthIndia." Journal of Develop-ment Studies 183287-328.

1985 The Market for Public Office:Why the Indian State is NotBetter at Development"World Development 13:4:467-97.

Wall, David.1973 77K Charity of Nations:

The Political Economy ofForeign Aid. New York:Basic Books.

Welngast, Barry; Shepsle, Kenneth;and Johnson, Chalmers.1981 "The Political Economy of

Benefits and Costs: A Neo-classical Approach to Distrib-utive Politics." Journal ofPolitical Economy8934:642-64.

Williamson, Oliver E1985 The Economic Institutions

of Capitalism: Firms,Markets, Relational Con-tracting. New York: FreePress.

Wunsch, James S., and CHowu, Dele,eds.1990 The Failure of the Cen-

tralized State: Institu-tions and Self-Governancein Africa. Boulder, Cola:Westview Press.

45/J-PART, Januaiy 1993

by guest on Septem


Dow

nloaded from


analysing institutional arrangements for sustaining rural infrastructure indeveloping countries

Documents

infrastructure investments

theoffice of rural

institutional analysis

yieldpositive net benefits

toinsufficient maintenance

different institution

developed countries

public goods