_____s_w last

POLICY RESEARCH WORKING PAPER 125 8

Decentralizing Infrastructure Decentralization - or ratherthe realization that the

optimal decisionmaking

For Good or for Ill? structure in the public sector

is almost certainly

Richard Bird noncentralized (polycentric)-- may in principle yield a

more efficient and equitable

pattern of infrastructure

investment and use than the

overcent,alized and

Background paper for World Development Report 1994 unresponsive public sectorfound in many developing

countries. But it will do so in

practice only if it is properly

impiemented along the lines

sketched in this paper.

The World BankOffice of the Vice President

Development EconotmicsFebruary 1994

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

POLICY RESEARCH WORKING PAPER 1258

Summary findingsBird examines the many faces of infrastructure informed about the consequences of all decisions.decentralization: the costs and benefits, the government Making politicians bear the consequences of their ownstructure (constraint or variable?), the "polycentric" mistak-s is as close as one can get to a "hard" politicalapproachi, and how to make decentralization work (for budget constraint.whom?). He proposes basic principles and guidelines for * Economically, it must be difficult for local residentspolicy design, for both small projects and large. to shift costs to nonresidents who do not receive benefits

Broadly, these guidelines are summed up in a few and to make Iccal decisionmakers ful!v responsible topropositions: their citizens for the use they make of revenues collected

* In all countries, some critical infrastructure is from them (through local taxes), to users ofprovided through a decentralized political structure. infrastructure (local or otherwise) for the use made ofCuirant trends make that likely to be more true in the the reventies they contribute (through user charges offutire. various sorts), and to taxpayers in general for the use

* Decentralization, however defined, in and of itself made of any transfers (or subsidized loans) they receive.has no necessary implications for good or evil so far as * Administratively, what such a syscem requires is ainfrastructure is concerned: its effects depend on the clear set of "framework" laws (on local budgeting,incentives various decisionmakers face. financial reporting, taxation. contracting, dispute

* The key to ensuring that these incentives are settlement, rules to be followed in designing userconducive to "good" decisions (about design, siting, charges, and so on), as well as adequate institutionaltiming, finance, pricing, operation, maintenance, and use support for localities to operate in this environment.of infrastructure) is to ensure that those who made the To tne extent that these conditions are not met, thedecisions bear the financial (and political) consequences, perverse incenti'.es that too often exist because of theas much as possible. strucolre and finance of the public sector in many

* Politically, this means that political leaders at all countries will probably be exacerbated by the currentlevels should be responsive and responsible to their tendency to decentralize more and more decisions in theconstituents, and that those constituents are fully public sector.

This paper - a product of the Office of the Vice President, Development Econcnics - is ont in a series of background papersprepared for WorldDevelopment Report 1994 onI infrastructure. Copies of the paper are available free from the World Bank, 1818H Street NW, Washington, DC 20433. Pltase contact the World Development Report office, room T7-1 01, extenision 31393 (34pages). February 1994.

The Policy Resea ch Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas aboutdevelopment issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. Thepapers carry the names of the authors and should be used and cited accordingly. The findings, ;nterpretations, and conclusions are theauthors' oun and should not be atf-ibuted to the World Bank, its Executive Board of Directors, or any of its member countries.

Produced by the Policy Research Dissemination Center

Background Paper for the 1994 World Development Report (forthcoming)

DECENTRALIZING INFRASTRUCTURE:

For good or for ill?

Richard M. BirdUniversity of Toronto

DxlcentrizIng Infrastructure: For good or for M1?

Introducdon

Te may facs of decentralizationThe benefits and costs of decentralizationGovernment structure: constraint or variable?The opolycentric" approach

Infrastructure for whom?How to make decentralization work

Guidelines for policy designThe basic principlesSmall projectsLarge projecu

Conclusion

References

3

List of Boxes

1. Altenative ways to provide local infrastructure

2. Infrastructure for whom?

3. The informal financing of infrastructure

4. Decentralizing infrastructure: the problem of technical capacity

S. Giving the wrong signals to local govermment

6. Local policies can make a difference

7. Accommodating local diversity

8. Enforcing local accountability

9. User-Inanced infrastructure

10. Earmarling: good or bad?

11. The virtues and pitfalls of matching grants

12. Local borrowing

13. The recurrent cost problem

14. Evaluation and accountability

15. Local government 'clubs'

4

Investment in physical infrastructure (e.g. water, sewerage, roads, transit, power,telecommunications) has long been considered to be important for economic growth.Indeed, some of the earlier writings on economi( growth, like some of the early prctices ofagencies dispensing foreign aid, appear simply to have assumed that investment ininfastructure was not only a necesary but almost a sufficient condition for growth.Although subsequent analysis and experience has undermined this simple belief, severalrecet studies have again awakened interest in the connection between public sectorinvestment in infrastructure and private investment and economic growth.'

Whatever the stren of the connection between infstructure and economic growth,infrastructure investments almost invariably constitute the core of both national and regionaldevelopmet policy in most countries. Infhstructure investment projects, and the servicesthey provide, are inherently located,in, and serve, particular areas and customers . A roadbuilt in the middle of an uninhabited desert is not the same as a road built in a major urbanarea, even if the physical characteristics of the two projects - the thickness of the pavement,the width of the roadway, etc. - are identical. The productivity (in terms of growth) of aninfrastructire project thus depends largely upon its environment - upon where it is located.Moreover, location also determines to a considerable extent who benefits from infrastrucureinvestment: water supply systems with pipes that serve only the rich do not do much for thepoor. The efficiency and equity of any particular investment thus depends in part - often inlarge part - upon its physical location. Infrastructure investment is thus invariably location-or site-specific.

Infrastructure investment is also jurisdiction-specific. In principle, there are manyalternative ways in which ocal infrastructure may be provided (Box 1). Is the projectdesigned, financed, regulated, operated and/or maintained by the national government? Aregional government? A local government? A state enterprise or other agency reporting toany of these governments? Some sort of public-private interaction (e.g.BOI?2 How doesthe government or agency responsible for these different aspects of the investment finmce itsresponsibilities? To whom is it accountable, in what sense, and how is this accountabilityexercised?

Seo, notably, Aschauer (1989), and, for an application to developing countries, World Bank (1993). As a aidepoint, it appears that moat such studies assume that infrastructure is provided to the private sector free of cose thismay be an ccurate depiction of reality, but it is not a particularly desirable state of affairs.

2 There am many possible forms of public-private interction in the provision of infrastructur: for a recentdiscussion in the context of a developed country, see Bird and Slack (1993), chap. 7, and also Kitchen (1993).

E S ~~~~~~~~~~~~~~~~~~~~~

... ~ ~~ ~ ....... :

u : a@. . . .~~~~~~......

. . .................... E.

gU S"iN~~~~~~~~.... . . .; s , N ~~~~~~~ ~~~... ......- . .... ;.s...

: ss egioaa :,:::::: ..........

"''''' answeirssoustionsnsucf ahf e'as s.a

The~ ~ ~ ~ ~~~~~~c answers tolosn sc stee a fet somtes dreamatically buothth

6

net social benefit of a particular investment and the distribution of its benefits a±id costs.Which public infrastructure r .cjects are built, wher^ they are built, when they are built, andhow they are operated, mnaij.tained, and utilized invariably depends largely upon the way inwhich the various public sector institutions involved in the process are organized andfinancet (Bo0x 2).

lx E harutEtr fo W.hom?,...,',,,'gg m* e='. Euelle Bm.= u=r= inatcu prjetsIst6ri,Xn 0a

g E~ a|su E lag ._ i huh, wn tEibo .~Iu4omalta F&df vt ~ iW o b b poumalz4~cvea( 'Mntd3vc

0... .

_g.~~p* W=-~ct.

*u Thatlothi iupstreis obvi~ousct to e anone fam3iiar wit the sad ~r lfiitany of las taed~( lprdutvel p.uinfrsrcua __ofren in all to maycu traes-ortad htht leadnowhereuninsed hosptal an sumch oolh price s w undterrnain.yataiedw andodtberioaredatin wate ansde

sewerage sytems. ngsm counttries ather psamet messarag e Onis cionvye Pbojenct.h(i othaloaive andQr distributive terams)- pulcrinvstedwntoae withn expentive pubilictiesfrvihe fea-wh

7

as a rule do not pay for them - waile at the same time the many pay a large share of theirsmall incomes for less adequate (ad generally unsubsidized) versions of the same service.3

To mention only two common exampls found in many developing countries: "free,university education (albeit usually in undermaintaLvied buildings) for the few and not evenprimary schooling for the rest; highly-srbsidized (but unreliable) supplies of piped-water andelectricity foi the affluent suburbe v' ,AJ the poor pay ten times as much for even worseservice al the other end of town, often1 provided through the. so-called 'informal' sector.

Such common outcomes in many countries reflect neither inevitable fate ("the poorare always with us") nor dire necessity but rather the structure of institutional incentives towhich decision-makers at all levels are reacting. Inappropite incentives as codified in theorgnizational and financial structure of the public sector, largely explain why the "wrong"project seems so often to be built in the 'wrong" place (often at the 'wrong' time), soon todeterio-u;e owing to lack of adequate upkeep, while at the same time other projects thatpeople are both wiling and able to pay for do not get built - or at least not bv the "official'public sector (Box 3).

....

.e e a et v d i c e .a a e .t a o i: e

ahio ved * Iwpnawg sm s oia4 f uoe and getin by sy-ncm ommunity in aki ustng" mo xnov... atructde facilitdes.:

?v finaced e g by sye f :.........~ ~ ~~~~~~~. . .... : .....

How and to s/hat exetent the public sector is decentrasezed, and what exactly is meantby hdecentralization,l are thus critical factors detenrining the productivity of infhastructureinvesthrent in any country. What infrmastucture gets built, where, wen, to what

'For various examples, see the papers in Bird and Horton (1989).

8

specifications, and how it Is maintained and utilized, depends largely upon who isresponsible for these decisions, and what "responsibility means in this context. The nexttwo sections briefly introduce some of the many dimensions of decentrlization: what is it(many things), why do so many people advomate it (for many different and often incompadblereasons), and what are its imrlications for infrastructure (it depends)? Tae fourth sectioa ofthe paper bnefly sketches a framework for considenng infrastructure investments in adecentralized setting and, as summarzed in a brief concluding section, suggests some ~,uidingprinciples for dealing with the many and divere issues that may arise in practice.Throughout, the discussion in the body of the paper is .ipported by a number of "boxes thatdevelop a few aspects of the argumnent in more detail and illustrate some of the points madein the text by brief case studies.

The Many Faces of Decentralization

'Decentralization' is a slippery term. One way to pin its mear;ng down is todistinguish three varieties of decentralization as measured by the degree of independentdecision-making exercised at the local leve. peco,en means the dispersion ofresponsibilities within a central government to regional branch offices: while clzarly relevantwith respect to some infrastructural investments, this variety of decentralization is notfmrther discussed here. In contrast, delem tin refers to a situation in which localgovernments act as agents for the central government, executing certain functions on itsbehalf, while devoluton refers to a situation in which not only implementation bnt also theauthority to decide what is done is in the hands of local governments. How one views theeffects of decentralization on infrastructurm depends in part upon whether one is thinng ofdelegation or devolution. Unless exactly the same people face cxactly the same incentives inboth situations - which iS hardly possible - the outcomes of the two varieties ofdecentralization are likely to differ shar' y.

The Benefits and Costs of Decentalization

How one evaluates such differing outcomes depends both upon the nature of theinvestment in question and upon whether one focuses on the intnnsic or the instrumentalaspects of dcxentralization. The literature is replete with passages praising the virtues ofdecentralization.4 Not only will it produce more efficient and equitable service deliverythrough making better use of local knowledge, but it will also lead to greater participationand democracy resulting in more popular consent to government and hence improvedpolitical stability. When to these good qualities are added such further ascribed virtues asincreased resource mobilization and reduced strain on central finances, greater accountability,and more responsive and responsible govemment in general, it is not surprising thiat somehave seen decentalization in and of itself to be intrinsically valuable.

' The particular list that follows is based on the discussion in four reent intems. Bank documz a (on Aex.zo,Nigeria, Indonesia, and Venezuela, respectively) but it could be replicated many times from many sources.

9

Whatever the precise outcomes that may emerge from a decentralized (in the sense ofdevolved) system of decision-making, from one pentpective such outcomes must oi presumedto be satisfactory simply because the process itself :s desirable. Loc people may maLe"wrong decisions from the perspective of the central government or of an outside observer,but if gh= make them, the decisions rtust, by definition, be assumed to be right' for them.From this perspective, decentaization is intrin2ically good because it institutionalizes the

participation of those affected by local decisions, and the results of this good process mustthemselves be good.

Under certain coneitions, this argument is persuasive. The conditions for successfuldecentralizaUon, however, are seldom sufficiently emphasized. In their absencedecentraliation may not only fail to improve local service delivery, but it may carry risks upto the level of national destabiIiza.¢,on.' If more expenditure responsibilitia than revenueresources are decentralized, service levels will liklly fall; if more revenues than expendituresare decentralized, local revenue mobilization will likely decline; even if both sides of thebudget are decentalizee' in a balanced fashion, local governments may not have adequateadmrn!istative or technical capacity to carry out their new functions in a satisfactory fashion.These problems may be particularly marked with respect to infrastructure owing to the 'largeticket nature of many projects and the degree of technical competence required to carrythem out (Box 4).

Three conditions that are particularly important; for successful decentralization in thiscontext are:

[1] that the local decision process is fully democratic in the sense that the costs andbenefits of decisions are transparent and that everyone affected has an equalopportunity to influence the decision;'

(2] that the costs of local de.sions are fiuly borne by those who make thedecisions, i.e., there is no 'tax exporting' and no funding at the margin fromtrnsfers from other levels of government; and

[3] that the benefits (like the costs) do not 'spiIl over' jurisdictional boundaies.

s A experience in a number of ILa Ameica coun;Aes has sugesteod, this risk is greatest wh revenues aredecentrlized without adeipate steps to enure that local revenue mobilization is maintainod and thst local authoritiesare cpable of carzring out the correponding expeaditr responsibilities. Argentina in the 1980. offers perhapsthe clearest exam;ple, but others may be found in e.g. the Philippines and in many of the transitional economies ofeastern and central Europe (see Bird and Wallich, 1993).

' Given the inevitable imperfection of democrtic institudons, and the ability of the rich and powerful to comeout on top in most systems, this is obviously a counsel of perfection.

10

OX :4. Deiit iai Tnftastrwctury 'no Problein o Tfhnical C b :iti :

I 97CDOlo bia devolvedite prm r u osbby(rtepoii of d nking water

ce. * , . ',' .. ,.....

sadwziatia aceuto ocl gvormena. h. wo ~tOnalOerMOw agemcin previ=wli.rl"4d *ith Jrvdn h. vost nn 10 wlllca gvenens n to imnaf towns

sadonefo naalaiss wee isbndd ad cedby dvsor E(th Minsty,of Public; Worki~~Whentb the c o awestr tisi .detting is end rinec whtadrd view d i nt

itrinsically uWhatnen (thee. ar UMe A ndid not affect the.35 aui asEven . whal ner .. serage of ther.se co anodthe 237 ioesiii .l.. g.ov.n. o

im lementation threspo sibltes olca ois acu aeintueta es poie

that~~~~~~~~ th netveaiglc3aleisionmaer are)Qh pro perl~ y sitrua actued, that s, stuctretprdc the diferent deeare bye theae lceelrof government (inisalactla ee endnth

.......... a..leisths rocm...e...rt ts fr.xapV

p a ti on as attb. whole).7 t derthae asen, however, thwe efc

uu'e How this ob doney hpasa boe c oa disur etenively velstewhe prjee.g. Br 1

These needs were met in different ways, and to different degrees~~~~~................

nalidee, In ~-~ instances,certain requrements for taining or adice wo.e.co.tacted..u....

ajnd these conditionsawe seagifed facilutieoi diserentspare, owhethetry vioe wed iosrumenotayer

intrinsically. When they are not, it may not be.

Even when one or more of these conditions does not hold, the delegatin ofimplementation responsibilities to local bodies may still make instrumental sense providedthat the incentives facing local decision-makers are properly structured, that is, structured toproduce the resl!lts desired by the central govemnment (in its capacity as representing th-epopulatioin as a whole).' In the absence of the right incentive structure, however, the effects

7 How ths may be done has been discussed extensively eLsewhere e.g. Bird (1993).

11

of either delegadon or devulution on the efficiency and equity of resource allocadon may bemuch less beneficial than often alleged (Box 5).

~~~~~~~~.. . .................. .......................

... ,5. ..... s .e .W., .ia .o ,ol vee.. .:..:::.::

Oni.: ::a: ay w y a te Aossible effec e pulc e iie *{.e.a; ... ,.:ez .s.out th poor.. .... ss., ,ult of. :o:a .govrnmt adwinis:::a:i: .n :::deeop:o:i: KW::l thit n in ot.u 800 rinaoXlIra j: fo uc onsn oacoadeul.etnt a m cousntryn :

.... v.s : . : use b. f ... : : disc...... - : s :. :: : :',ss.e,:: :,: ::....: .ss; ::: :: ....:.s .:..:s:s:, ,. .s:s::. s

6efo yan v o pt m h nits o d s os p i .b4 w p a diefcAm o a

goi'Knmints.Oue nbiwer toW 6hs nabem bviul t lte toe poa~ tholtnes ton keiss :: :' "ssk ''UIeffls gm 'tZ Aia:"° s. s:' .. : n:d.:s' e,.''=

s~~~~~~~~~~~~~Uo' .....Ns

poss ib ibandtctvo wr ho ,: - ed peopl e to a aaeei:::n lobt tgovernt sen an..

se~~~:C ;S ss Ak SP.M .1: ]p: .ec 9V> 1

.. .. .... .. e.. ..

:l::::s:oMA: o ~O 1dyAla .endrs w t :: :. :::

stnwmm if there was.~ ~ ~ ~ ~ ~ ~ sliu fohr. i4~c

The tlsti perbap trho o:adt o o n e: ets t .1o te d m of o verm

who hey izi,whattheypay, whee th buss t~ etc~ ad eAve nto redomou of ateioran lsloa

izatzve:: T thaese countries, local citizens look: to the national govrnement to fixpotholes on thair stree

adtys,: th t do .o

W le tea eomor~ *Duppst ilo mdWricnro isIeginngt ee il nfobr)

G ofve ounties aS u re: of tinppor ariabdle?i sae For ezamp1e;:a number of

To ansioa point, iehs b.esaen impdiciyralssumed thav t tihesrcueo local governmmesim n nationa

reDvenues a^s well asre8ponaIibiity frimortaint pub1i~ service functions, but without stng up an::

adeliyquratle: nthatnto whocr doeswhae tatd tnder what rulds, may bein proerly sdeterined bay

::::alntaliimakimm anrsom insevicesi hedt o heastreh. of pb :: d

Whehe ::::s:over- or::imde-nrollod, local governmet inall :too: aY conties receivepvea ~n t~ aa jenmes in nume o way In. ome m n=s' for exampe th:a mount of natina fundin deeved depeds pon the aizof:lhe local buget deficit: a sperversei:

no cs btter r no Fund foprton an wnteac thscmo ytmi evs easi pay localtes not to mantin stin fclties (whc tey woud have :to do out of tei o: fud)

:: n ordeir t:o strenten escr appaent ned for:new ones (which th eoa government will pay for).:-i:::

Source World Bankc (1993.), bird and Wallch (1993) ::: ::

Government Structure: Constraint or Vaniable?

To this point, it has been ixnpliciily assumed that the structure of local government isa policy variable: that who does what, and under what rules, may be freely determined bycentral policy-makers. In some instances, however, the structure of public sector decision-

12

maldng is not a variable but a gij=. Perhaps the most important such case arises in federalstates, in which two levels of government (national and state) have specific and independentpowers assigned by a constitution which is not easy to change. Even apart from formalfederations, however, it is often difficult to alter governmental structures quickly,particularly when, as is usually the case, the existing structure reflects historically-determinedpatterns of power-sharing.'

In such circumstances, the existing degree and pattern of decentralization in effectbecomes part of the environment or context within which infrastructure decisions must bemade. When, as is often the case, some of the incentives resulting from the prevailing formof decentralization are perverse, careful attention has to be paid to the extent to which thoseincentives may be altered at the margin without undertaling major (difficult, costly, tme-consuming) constitutional (or, perhaps better in formally unitary states, quasi-constitutional') reforms. Even within a given constitutional structure, however - even onewith many "wrong" signals being given (Box 5) - there may often be a surprising amount oflocal discretion in both policy design and policy delivery, for good or for ill (Box 6).

Decentralizing infrastructure decisions in an efficient and effective manner may thusrequire very differenat strategies in different circumstances depending both upon why thisapproach is being taken - for intrinsic (because it is good) or instrumental (because itproduces good results) reasons - and upon the perceived rigidity (or flexibility, as the casemay be) of the formal government structure.

Many other factors must also be taken into account in assessing the likely effects of aparticular form of decentralization. Each type of infrastructure investment, for instance, hasits own specific characteristics (economies of scale, t&e identfiability and spatialconcentration of beneficiaries, etc.) that may affect the desirable degree and nature ofdecentralization at each stage of the process - design, finance, construction, operation,maintenance, and utlization. Moreover, since, as was emphaized earlier, infrastructureinvestment is inevitably location-specifc, the local environment may differ in many relevantways: rural vs. urban; large metropolitan vs. small city; the relative importance of formaland informal markets; historical and political background; and so on. The potential uses,effects, and limitations of different varieties of decentralization inevimably differ in responseto these, and other, factors. It is thus singularly difficult to generalize on this subject, evenwithin a single country (Box 7), let alone across countries.

I For a recent discussion of the importance of such factors, see e.g. Bird (1993a).

13

4O .LclPlce a Make a Differenc

ggi:s:: : . . - .. - . - >i ...... ,,...._ ....W:'::~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. ... ... ..., ........r:...."

By. hn th neUe &Brn loal gvrm.t ua ~oghedie (e Box 5), mimetD rU o 6f t.e, .. .p.., .s.iwid t.e wd . ,.ht iocul polcies can ru .a ,eal .in.. ,... .. , .ecet ;.. ...

of theia e Sv reflet or a y fee m..,..... ,. -.. .. .S ....... ... ..

.f thio and Des-GItan1, 192'h X,

.ith aim= a. .=-

Ss~ ~ Obu~i izoa gvseram locl gvemin inst ,,, 1ea e.;sX do,.,,m,u,,hthe utes td4 outlgd wotn ao n etnnt dow. Or it'S < sy becauise S

The wPolycentric' Aipoach

Matters may often be even more complex than this because the optimal strategy inmany cases may not be wdecentralizationu in the sense of establishing a two- or three-levelhierarchy of general-purpose governments, but ratlher a noncentda or "polycentricwinstitutional arrangement in which there is no single, ultimate center of authority but rather anumber of independent bodies each excercising authority circumscribed by rules (as opposedto superior authority).' The private market alone is unlikely to provide adequateinfrastructure in part because the phenomenon of *free-ridingw makes it impossible to realizeeconomies of scale. Centralized provision may overcome these problems but it also oftenresults in misplaced and misspecified investments owing to its inability to take adequately

' This concept is most fully developed in Ostrom, Schroeder, and Wynne (1993), chapter 9.

14

... BaOX:7,Accommodat Local D iveity . .... .. ....

.~~~~~~~~~~~VA .

'Local:: goveriamen'15te that co n Wide r:n:gof reali wimost Cities:of 0 mllon eope,villages .Witlh 200bihabftantS, esl ouae Auran and rba ara s,sa

: , ~~~~~~~~~~~~~~~~~~...... . .. .. PA ...

.ocalities mrs rich, 'ome are poor somebavo strocj local comimmity spiit, s Ome v n Some.,., ~ ~ ~ ::1 . :.:s:.::.::.:.::-: S,-7 S,.S: ::::.. ::,::: .,..:::..,.. .,... .,.. .. ,

azu m by weflIntmb ioned wel-ftrane people thrb ncompetnt n orptofcas

%.'Sg ,s5; , S 1ffi~~~~~~...... .. .. ............. , ....... S ...... govaenment Sz>Gu::lu .:.overni .. ocu .:v n -:. v t i .: t,

(Box 4)s,up and including accommodating ~'home-brewed informal-sectoz oluion.s t .. ..i

'S'e G ' . ,dcto and ,.lt ar d,atmna ,epnilte 'n Coom, altoug .s of .t,h::'e@m', ' 'policy in: teru iof building roads, schools, etc. InE somc eorteX oilprdcn regons the tat oil.

* w2S ' w :~. G':.:.. .......... . ..... . ............ ........ ......... :.:: :.a . ...........,,,, .e. ''

''''" o* ' ~ o G,pniblt G.o:': man, ,.:.:;::':::: ics :n :s i as som of th.os poics

........ 6e 3.. offis of Gl '

* =. = of ipl e l G, :: .:::::,.:*' ............................................ " ' ' .:: : . ' ...... ...... .

, ., ,. , .... ..................... G, G,:::f:-w es~~~~~~~~~~~~~~~~~~~~~~~~~~~~.. ... .. ..* ,~~~~~~~~~.i: gso.:.,: :*::j.., ..... .. ...... ................ .G .:::::','

..... z. o ' .E ........ ~~~~~~~~~~~~~~~~~~.. ... . . . . . .. . . . .. . . . . .,..,.... .,,... _X0

into account place-specific information.d Decentc alized provision intended to overcomethese problems in turn has seldom succeeded and has given rise to its own problems -corruption, mismanagement, rent-seeldng, and so on. Indeed, there are probably as manypapers in the literature decrying the limitations of fiscal and administrative decentralization as

"I See Box 2. This argument assumes that it is more economical (requires lower tansactions costs) to aggregateinformation on local conditions and preferences at local than at central levels of government. This proposition maybe correct, although it is (so far s I low) unteste d, but it should be noted that it implies that ch informatiocould be gathered at the center if sufficient effort was made to do so. Moreover, if the relative costs of acquiringinformation at different scales change over time - as seems likely - then the comparative advantage of carrying outdifferent activities at different levels of government will also change (as emphasized by e.g. Breton, 1989).

15

a solution to the problems of developing countries as there are papers propounding itsvirtues.'1

In contrast to this linear view of the possibilities - private, decentralized public, orcentralized public - what may really be ideal might be, in effect, a different govemment'for every relevant public', that is, for every group affected by a common problem. Insome instances, the best solution may be to leave the problem to the market; in others, toform some sort of 'club', or some form of joint public-private organization; in still othersto create a single-purpose jurisdiction, or to form an association of different general-purposejurisdictions (Box 1).

Of course, the more jurisdictions there are, the higher the tansactions costs will beand, all too often, the more obscure the lines of accountability. Single-purpose jurisdictions(such as water districts and school boards), for example, may be appealing on efficiencygrounds in terms of providing the specific service with which they are concerned. But at thesame time, they may undesirably weaken general-purpose local governments both in terms ofcompetition for resources and reduced political accountability and hence hamper the efficientprovision of other public services.12

When this rich palette of institutional possibilities is applied to the diverse settingsalready mentioned, the optimal results, although in theory presumably limited only by one'simagination, in practice will depend primarily upon three factors:

[1] the importance attached to various criteria (minimizing resourcecosts, economic growth, poverty alleviation, participation, etc.);

[2] the nature of the infrastructure investment (economies of scale, span ofbenefit, etc.); and

[3] the political, economic, and institutional environment of the country orregion in question.

It is obviously not possible in a short paper to consider all of the possibly relevantcombinations of these factors, so most of the following discussion will assume (1) that themajor concern is to promote economic growth (with some attention, however, to povertyalleviation and participation), and (2) that there are two classes of investment (small or local

" See, for example, the introductory overview in United Nations (1991).

12 Kitchen (1993), for example, in a rocent examination of Canadian experience with special-purpose districtsconcludes that (1) they make govemment more difficult for citzcns to understand; (2) they reduce the degree ofcontrol citizens have over government; (3) they reduce accountability and hence, probably, the overall efficiencyof resource allocation in the public sector; and (4) they appropriate significant proportions of revenue that wouldotherwise accrue to local governments.

16

and large, as defined below). In addition, the discussion could easily be extended toaccommodate two types of countries (federal and unitary) and two types of regions (urbanand rural), but this is not really essential: the federal/unitary distinction has already beennoted, and the urban/rural distinction may to some extent be subsumed under the small/largedistinction. ITis simple taxonomy is more than complex enough for this preliminary look atsome of the implications of decentralization for infrastructural investment.

Inrastructure For Whom?

The answer to the question in the title of this section may seem obvious: for the'people.' The problem, however, is that there are often two distinct and potentiallyrelevant groups of people - those who benefit and those who pay. Only when the twocompletely coincide is the answer obvious. At one extreme, those investments that benefitall national residents equally - that are truly Opublic goods - and are paid for out of nationaltaxes levied by a democratically-elected government that is fully accountable for its actionsshould clearly be provided by the national government." At the other extreme, when thepublic goods aspects of the provision (nonexcludabllity) and financing (user charges) ofinfrastructare are small, when there is little or no redistributional concern, and when thespatial clusteing of beneficiaries is marked, the investment should be provided by the mostefficient (least-cost) form of organization that can aggregate local preferences and collectlocal contributions - a voluntary club, a cooperative or non-governmental agency, a specialdistrict, a local or regional governrment, or some combination of the preceding, as the casemay be (Box 1).

Unfortunately for analytical simplicity, most infrastructure does not fall cleanly intoeither of these extreme categories. Distributing water to residences and businesses may be apurely local concern, but supplying the water to the distribution system is, except in smallrural communities, almost always a matter of regional concern. Similarly, maintaining theairways (air traffic control, etc.) may be a national problem, but regional and local interestsare greatly conceined with the location and characteristics of airports. Even when aninfrastructure activity appears to be obviously local - for example, the provision of seweragefacilities - there may be overriding concerns (e.g. public health) that in principle requirehigher-level governments to intervene in a regulatory or supervisory fashion.

For these reasons, the only possible answer to the question posed in the title of thissection is: it depends. It depends on what infrastructure, on why it is being provided (atwhose behest), on how it is being paid for, and, in some instances, on where it is located(where in the watershed or air basin the facility is located, for example).

" Possible examples might be a national telecommunications satellite or the maintence of the airways.

17

How to Make Decentralization Work

To the extent local people decide to carry out some activity through an organiizationalstructure that they form or that is under their control, annd they bear the fiull costs of theirdecision, they should be free, even encouraged, to do so. Of course, there may remainmany potential problems even with such small groups e.g. enforcing accountability,preventing shirkcing, and so on (Box 8). But on the whole it is probably not misleading tosay that the more decentralized the decision structure in such instances, the better theinfrastructure should be from the only point of view that matters - namely, that of the localbeneficiaries (and payors).

W_~~~~ . .. ... " -........-.......

,~~~~~~~~~~~~~~~. c. EX.t E.. E 4.44

XOm3g te aomah >E aD *.r F. '! __ .. . .. .

Vnfe~snae 3J J** 5* 3: 32 =3eaMDiNg oi aboe (Bo 5) ut lso E3 th proidm. adqzt inomtint local.anut 33X3

.-yCi ;2X W>3 N?;;'** 3_wa lg ee c~~~~~~~~~lotb deiDitio mus =e stitrdIl.iirhe an

toInrouo. -h Eoid~ i_ o ~4~o ot 4Sl~~~~~trpnvmit foma R,oven ta ogn.'

.....::..4..::..:U 73#,]WL3170 M~01~Q pi7d.UDs ,i incasls and sgs

infor;a4io Is d § ofte avilbe ilag nudnSlU what may e miuin e lainy a:tousI n.. the::B:invilg s|ded g.a iui; toinlil did= R.t Teugi hneb~m hr a owf R

.. ~ ~ ~ ~ ~ a ::i g :.::::.;,.:: :::.:j :.o,:::, '::,:::: jo e :::::::S:::':e:: : :: :.:

wometimav tuhe doer taiation hft~vlagescho bhee dcnsidrdan impossi ublc ha depamnd waoor

countries owing to the lack of experience at local levels and the difficulty of figuring out whothe likely beneficiaries of projects are going to be, let alone making them pay for what theyget. In fact, however, in many countries it appears that even the poorest people opeatinginformally, within the severely limited conditions open to them, can sometimes manage toprovide the local public services that they want and are willing to pay for but that theycannot obtain from the established (and usually much too centralized) provision system (Box

18

3)."4 Experience thus suggests that at least in some cases the obstacles can be overcome, atleast with respect to small-scale activities.'5

What is needed to get the right infrastructure for the right people in thesecircumstances, whether they are located in remote rumral areas or metropolitan slums, i-essentially to provide mechanisms by which their wishes can becomc reality. Three stepsseem essential in this regard:

[1] to ensure that fiscal, financial and political accountability is both transparent andenforced (Box 8);"6

[2] to provide a mechanism (e.g. a uousted court system) for resolving any disputesthat may arise within the local service unit (or, conceivably, when there are boundaryproblems, between units);17 and

[3] to provide adequate technical support (e.g. access to engineering and projectdesign and administration skills) to enable small units to carry out projectsefficiently."

None of these tasks is easy, and, as already suggested, they become even moredifficult when there are significant benefit or cost spillovers or when the redistributional

1" Perhaps the most detailed account of this process at work is de Soto (1989); see also Jekin (1988) andOstrom, Schroeder and Wynne (1993).

1 Of course, small-scale activities may add up to be big ones: Jenkins (1988, p.20), for example, reports thatsmall 'informal' suppliers provide 93 percent of urban mass transit facilities in Lima.

"I For a more extensive discussion of accountability in generl, se Paul (1991) and on fiscal accountability,see Bird (1993). See also the interesting discussion of the need to develop forml evaluation u a mean ofaccountability in Wiesner (1993). Although Wienr (1993) appear in part to see evaluation as an alternative todecentralization as a way of improving public sector efficiency, I would soe it more as a necessary complement (Box14). Effective decentralization depends upon improved accountability, and formal reporting and evaluation systemsconstitute essential components of any workable accountability system - whether to users, to local taxpayers, or tothe central govemment, depending upon the source of financing. In Wiesncr's words (1993, p. 16) 'the nature andsource of financing is the crucial detemina of efficiency' but, a he also emphaizes (p.18), a system ofindependent and well-publicized assssment is required not just for accountability but also, often more importantly,to help establish a 'public' to whom to be accountable.

17 A ll too often, a well-founded lack of tuSt in formal political (and judicial) institutions is one reason whycommunity 'slf-help' organization have developed to provide local public services, including infrstur: forexamples, see de Soto (1989) and Jenkins (1988).

" Note that ther is of course no presumption that small local agencies will themselves actually e.g. designwater systems; the point is rather that they must have access to specialized and knowledgable private (or public)agencies that can do so for them (Box 4). In this connection, it is of course especially important to keep the processas open and accountable as possible (see Wiesner (1993) and sources cited there).

19aspects of infrastructural investment are important. Again, however, the principles of goodsystem design are relatively obvious, and more or less identical to those in the 'smallW localinfrastructure ctse just discussed. B;sically, the idea is that while local residents shouldagain finance - through user clarge- (Box 9) or locally-borne taxes - the costs of designing,building, operating, and maftitaining facili .es tW the extert that they receive the benefits,regional or national governments (as the cse may be) should assume the balance of the cost- that is, 'their share' of the benefits spilling over. Locally, taxes, like transfers from thecentral government, are often earmarked for investment. As Box 10 notes, however, thiscommon practice is usually not a good idea.

v.sz; i .VYo fbendMt..aaion.= In Lat T , riL' :for example, atret mroveet,ws suply .and C-W :S S 116,2 'S * Si.......s s,

o w s. i :: :: :2: .:.2 ... . .>v::d:SmCI have boee financnd ys !tzd~ ii '1aui~ Inwbc

mstnci. Stes n .:..a j e :: ::a: ben mot ue... .. t tt c.. t. :.. socsi creu plDannin anf ecutin o project:s, aticpto of bee:. ads in othplanian

:.miilar lessons have emerged born experiece with an lternatve apprac clled '.3and reajstinenin Wos -- :::ich lrg land: pesC a consodad an deeopdb the loal gvrS":'*:t m".:>c:::::.:'d the. nS; zprofthe proper s reuned to the oignal menin proportionto thir ownhrsipwlle te blanlce::... ... .. ,. .:..-... .....:.-

::: xsold luy the government at minakt pnce6 mn order to :recoup deveopmet costs.: Again carefull::.'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ "...... .. .. s. ' *'

::.B..... .... ......

. ........c. esetexperences:dmostteatlocal goe Cmet me r a dev op ur:an.. asrucr i.nect by payigthe ole o developr.Recently azuother way inwc eficianri

o-alled 'exactions, wlot. evies', '.evelopnt: uarpa,: and.: mla.. . ..ee... ndrwhche p .i p ......... ................. costs w: ..develomn zwil imposeon.o due urba inratrutue :For example, tf l00 new resdences are to be:...e ..rced, and... e verage. .ost of add.ng th..mto.e u .wterand ge system u::$10, the.

~~~~~~~~~~~~~~~~~~~~~~~~. ; S ........:i uz::............................ ...... ..

.. ~~~~~~~~Otmmd o:uvstiaf the§

use .u:by. finacify-ressed rba govrnmnt .to: acommodate poplatidon:expansio without:.:

: 5 B y .- ...... ~~~~~... .......... ..: : Of COU: al fioal $sytems of sps ifsrt devlopuu ca ope successful

only mn theformal sector..::To thie cxet development takec nTace prmail i the informa sector -::.. uatter settlemetsan .the like -. les.sd foma ystems most 'oe use ifac thre to be anybeneficiary.:

Sowe Word Bak(1988); Slc u Bir (1991). .:

20

... .X ,. .a9~In~ .,od , r Dad?...............

~ o:th. moss puvuiv. fhhbRim of local jV.~minL finance in dAVIIOFUm............t.e

. p:.s;1uzc. :: : hiurk:: x:1 ::^:::: eat. :n ::d:a .: :zmp. ... o,.§: o:.:: .:a at .. t .ta..g . p .. .- . - ... i.w..f..!.oEX ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~... ...

tax Is a.~upb4 f1r urban locaI I!D gsv.muu. and o.of thia porcioX ~ in turn mu&sd .i.- MaV & - MIW*U '"' #S. :., -

~ A imilg........... , ,2spWmg ~ ~u already difficult: pmbl.rn o3fmhing S i... .... n..

t E iS X.Bi:XfiSB W B^<SB;>iz ,~~~~~~~~~~~~~~~~~~~~~ . ......

the ostlsplittimayb achieved inuvarious wa thr appd ropiataly dlesigead

servic chre tou andotn-residen beneicmarieghs, itndividuaUy oar poretisl, though matchinsgne

grants sBox11) thrug borrowing, phuuossibl axto slubsiied addties (Bo 1)ors thoughyi dlbec

gouvemmmn eienttomk f tily mbinvestmen and toliquirl the, liokal algov ernmen toe fuptnd

subsequetsnarninglcosts Epneren.Lkuc sugcest, hwev er, aprorstl thes commoneee sp uidtnis

danereousince ion thie thbsen,c nel of pnolermectwion. ?theines tmeinterepde eneo lofrca ngangovemmuents drecso a ndooEl tohe apathetall wimothn mspec to operat*iogn ainkmingtevene (Box

1-4XpIe lowsafren disribtoal concm , hemdon~m mtiance stouldnaecpnt lona b fitfas oppoused to 3bliy.

governmentnton make the aio nd eional nvelstmn and, u toi rqis ecodyisei the loapgvr msent tonfund

§Th le.s,s>salen ditiutoa concern.ss::::>:::suigSsi -s,< ts-e more reiac solbeptobnfias oose ts abi ;s ,ty-s

bae,iacn tteiai~Iad einllvl lo u hsi a scndr isssuein hepsntconet

21

13). Why should they fund the maintenance of something that all too often they did not wantin the first place? - especially since, 311 too often, they can count on the central governmentto come to the rescue agaLn if things get bad enough. Ausa,what is needed isto face allrelevant decision-makers with a sufficiently whard budget constraint. Provid .daccountability is again taunsparent (and there is an accepted dispute resolution mecbaism), ifthe prices facinlg decision-makers at all levels from the indivridual to the national governmentare rightw in this sense, the results should beas good as canbheachieved.

| I . | i1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.. .

~~~~d~tl argn | l |*

m -'t i' FS ~~~~ ~~~~~~~.4-,....

S ~~lsai d bmyfrte*Ie ,bcbvd.

;44 .'4 '.:.... " > .$~4 ... -..........

wa =geyb ihmgbaX .s. s th?

Ipmt ais iw of coure murvch cainer t~o state such tictangnrl trermsu thanearl to prescribe

preisey hatshol bent sudothu in and particular cae o actievtes effwicitent and eqiabclea d

resource allocation, let alone to work out what is likely to be adminlistratively and politicallyfeasible in the circumstances of a particular investment in a particular country. In federalcountries, for instance, the scope for creative institutional design may be much more

r

22

rUOX 12. Loa l o rowi n,

....:.';, ... .. :. : * . .... .:.. .::h . ,:: v .: .: . .. :.....:.

i .S . X .. = K l.. X . M | X,0 ,,.a,'m,,>....~~~~........

Whu the ........ ..mifr.tmc prq.oct . .... ovr~o otm.imyb both

lala tN xa..s frm.... ... .. y

mu aadmkr , alw n yeuc acces by oc l govemmet ihe loalos bonwmgis 4 R 8 eS ' . 5 : a : xi ,..,.:..8 , .... ..... ........

= : ',~~~~~...........:.

.'S StSsimi i > .S w ~~rite ...

s . . .. . . . . . . . +w::::d:4Swl - iS~~~~..... ..

'Si S*u ewfra pptia a.. :.: liev>a .ud1d ms aza local

j M j,............~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.... ..

restlicted with respect to federal-state interaction than at the state-local level, particularlywvhen states rmPre ont real difference in valuo and experiences (e.g. there are linguisticdifference-). Even with respect to the Esimplem cas of purely local infrastructure, thepossibility of building on indigenous infonmal organizations to provide local infrastructure inan efficient and qui.u ble E shimon my be much greater in a traditional rural commueity cntiin a dynamically changing metropolitan area.2o

Guidelines for Policy Design.

Thc essential frmework for policy design with respect to the decentrazed provisionof infastructure thus basically amounts to little more than the familar econor ic dictum "Getthe Pr,ices RighLt" In terms perhaps more familiar in the public sector, this may betanslated as 'Impose a Hard Budget Constaaint", i.e. face aU decision-makers at tlte margin

X It is perhaps not entirely due to the terms of reference of the research project on which it is based that thestimatin c boomr by Octrom, Schroeder, and Wynd e (1993) deals solely with nni are ls.

23

:O B=13. The Recurent Cost Problom . :. ::..

op:to abn d aitenac of :nrsicw Is tX aassup tha loa ovrmtsw, even whn

.:ual tecs ::t involedi th* seecio oreeuino roet,cnadwillo fe

: '" .:, .:l S E = S =

Ali tdepir,owee. .. ere are E :: f::E::: In :: L:~j > i

'',. ', ,, found_ ,W , tode rbli o innig h rcnetonns sr4o)ee nfatrobQ~S~ Ad jua.oa4= 0 I- EEwea,fsaul

withothe responibility eae to raising the funds and fepicitg rtaeuir chastitens sithe mothroa fil*(ikcemetal)ir cosAtiofted adoe)iin Tehe wielentn secion sespa forute inroe s itervice.dtilsm

sugges ~tios ese tohowtea istia tutionl-dfrned prlojet wth oe public r secor p wa o hae ntutue godahelp achieve. Whis e oaime.tal~nu~f~ti ecrpin asal rjcsd n

Tove bergin pith,ect hiswfli4 clear thtonbiewilnotfait all, ;hat iso, thart pr ueid whargeaand ashould the dorefrre antyparticula casedeead,t s upoil n amyrial pofec ploannl t factors (Bw7ndcann rotepriae stcribed rym ausr.Nonetheroms the basigpinncnplean to battem follt e in alltc I thses

with the rspoensibiliy of rumarisigtedfuns anf fcngtelonttunswihsh:fl

24

,1] Ensure to the extent possible that who benefits, pays;

',2] Do so in part by making the lines of accountability (the rules, or incentives) asransparent as possible; and

(3] Proiide some enforcement mechanism to ensure that the system workcs as it should(Box 14,.

S -X ' ,tisicm. g .tu a atc x S. ah S. §iE ; S 21 ,ffivffi = B %2s. = '

g a . 4, ; E== = i S =~..

s~~~~* 4 l1 oc-al; s= |:ua.*vI 4 (tuuaflj)ieka . EaUc~ aXuad m

Baidom to wel It It d, ho _ 1 = N~4 = = S~~~ . ,S S::S::.2 t= = ===

hive dectst,a&t. x..l 5 _;

ma s3 BtIOoal4alw*is=e 4*Z: l2~1 . . 3gmsbmz Mel$;u1 O4- BxVto ln4nce of clea3i.t loca:x:j:B ::::s::: s ydo cett .oove. capsolt t:.> ahae of dbI

Alhugh prciel how these principles may~ be best saZt1*isfie d , e of crourse upon (say

naur numero ynew.en, rvddthey structur of governen,andman oter luocal factoers theonz

nollowin brif the ommcznnts the strctue ofgoeir met aplcaind clanyoheeoclfatr.h

25

Purely Lccal Projects

So far as purely local infrastructure projects are concerned, they should clearly befinanced to the extent possible on benefit lines, that is, by appropriately-designed usercharges where suitable and otherwise by local taxes, where local taxes are understood to bethose borne by local residents (the presumed beneficiary group). Pgvided there is aresponsible and accountable local political structure (Box 8), the results in terms of both theefficiency and equity aspects of infrastructural decisons should be as economically andpolitically good as can be hoped for. Only those projects will be built that people are willingto pay for, and when they are built they will be adequately maintained so long as people arewilling to pay for it; moreover, if the correct prices are charged, the projects will beefficiently maintained and utilized.?

Of course, this simple picture conceals many potential problems. One concerns thedistributional aspect of infrstructure. Lally-&dered demands for distribution may readilybe accommodated through the siting of projects, their design, precisely how they arefiaced, etc. - but only to the extent such demands are effectively articulated through thelocal political structure. If that structure is captured by the better-off, as may often be thecase in practice, and if there is a national concern (articulated through the presumably moreheterogeneous national political structure) for a more redistributive delivery and/or finaicingsystem, the national government may attempt to satisfy that concern in a number of ways -e.g. by restricting the extent to which projects can be financed through user charges or bymandating certain service delivery patterns. Unless the national government pays for anyexcess costs its requirements place on local power-holders, however, the result of such

interference will be tJ reduce the efficiency with which the noncentralized public sectordecides on and delivers infrastructumal services.

Another concern is how to ensure the transparency and accountability of local politicalbodies. Local authorities should not, for example, be able to 'export any of the costs oftheir decisions to persons to whom they are not politically accountable. Their access to taxeson businesses that trade outside their jurisdiction should therefore be restricted wheneverpossible.' Political and administrative mechanisms for public accountability - e.g. makingthe books open or setting up watchdog agencies of various sorts - need to be in place (seeBox 14). Whenever possible, infrastructure projects should proceed only when there is aclear decision by a politically responsible body as to exactly how their construct!on and

1 For further discussion of the choice of local taxes, and of various vareties of user charge., see Bird (1993,pp. 212-14). Such recommendations are of course not uncommon in the literaturr: but what should perhaps beemphaizad is how very sldom they have been followed in practice.

For examples of the nks stressed in the text, see Brent (1993) and Box 13.

D Tbis common recommendation, like most suggestions for increased transparency, is morm honoured in thebreach than the observance: all the incentives for politicians, both local and national, ae to conceal what is goingon. Too often, obscurity, not trnsparency, is the key to political success

26

maintenance are to be financed. Where the projects are big enough, this requirement maycall, for example, for a spezial voting or other approval procedure.

Finally, anyone contemplating letting small (or even large) subnational agencies maketheir own investment decisions in the circumstances of most developing countries must alsobe prepared to invest substantial resources in 'backstopping' such agencies by giving themadequate access to e.g. technical skill in designing, financing, and managing the constructionand operation of infrastructure (Box 4). There is no need for such aid to be provided by aspecific central ofrregional agency, of course: local bodies can and should be encouraged tocontract such assistance privately, to hire it from other, more experienced public sectoragencies, or, if it is appropriate, to contract with a central or regional agency experienced inthe particular field (Box 7). But they should be to pepare projects in a professionalmanner, to report on their execution and operation in a transparent fashion (Box 14), and tobear the consequences (political and economic) of any mistakes they make in these respects.

It may of course be easier to enforce some of these requirements when some of thefmancing comes directly from a superior level of government - see below - but in generalwhat seems required for success in this approach to decentralizing local physicalinfrastructure is to create four pieces of institutional infrastructure:

[1] A local finance law that provides access to local (ngt non-local) revenues insufficient quantity and that also requires regular and uniform financial reporting bothto the local citizens and to independent auditing and evaluation authorities (Box 14);

[2] A political structure that provides for adequate local financial and politicalaccountability;

[3] Adequate institutional infrastructure to support and develop the capacity oflocalities to operate in the environment sketched above.

[4] Acceptance at the national (and international financial institution) level that whatlocal governments (or other local actors) choose to do with their own funds is theirown lookout: there should be no Father' who either knows best or rescues Juniorfrom the consequences of his (or her) own mistakes.

While it is unlikely that yaY country in the world today fully satisfies these conditions,if the general line of argument made here is accepted, it is these matters - buildinginstitutional capacity to support and operate responsive and responsible local governments,establishing adequate 'rules of the game3 for local finances, and ensuring that the desireddegrees of local participation in local political institutions and of the political responsibility ofthose institutions are achieved - that require attention if political, administrative, and fiscaldecentralization is to produce better and more sustainable decisions with respect toinfrastructure.

27

'Su2rLocale Proiects

Even if all of the above conditions are satisfied, when the benefits of infrastructuralinvestment spill over local boundaries, decentralized decision-making can produce the 'right'results in the absence of intervention from above only in very special circumstances. If, forexample, only two or three contiguous jurisdictions are involved, particularly if there is somereciprocity of benefit or cost flows, voluntary agreements may approximate to the efficientresult (Box 15).Y When more jurisdictions, or less clearcut reciprocal gains, are involved,however, the costs of reaching an acceptable agreement may often to outweigh the benefitsfrom doing so. The traditional solution to this problem is to call in a 'higher level ofgovernment. But how can one obtain the benefits of doing so (i.e., taldng extermalities intoaccount) without incurring the costs (i.e., imposing undesired uninformities)? The answer is,broadly, that an optimal result may, at least in principle, be achieved by designing anapprupriate system of national (or regional) matching grants (Box 11) and, within thisframework, allowing the non-central authorities to make the decisions.25

Once again, the first requirement for successfully Carrying out jurisdiction-specificinvestments with extrajurisdictional spillovers is to ensure that to the extent local benefits arereaped, local beneficiaries pay for them, whether through user charges or current or futurelocal taxes. Applying the same principle to those benefits that spill over, wherever possibleusers (mdividuals or communities as the case may be) should pay for what they get, baringsome overriding distributional reason - to be covered from general national revenues - as towhy they should be relieved of this burden. Where this is not possible, or it would not beefficient to do so - for example, because of problems in identifying and collecting frombeneficiaries - the 'nonlocal' portion of the project could be financed by matching grants.

To be effective 'incentive-revealing mechanisms' such grants (loans) should bedesigned and administered in accordance with the following principles:

[1] Recipients should be required to prepare adequate plans with respect to (a) thedesign of the project; (b) its subsequent operation and maintenance; and (c) itsfinancing (including, where appropriate, pricing policy).

(2] Assuming that the qualified projects eligible exceed the available financing,priorities should be assigned in accordance with the assessment by the financingauthority (which is of course responsible to its own taxpayers for the use made of

2 As Box 15 uggests, voluntay agreemts may also be used to s4divide local governments into relevantbenefit aras.

25 Of course, if the central govemment really wants something done (even something with purely local effects)it may either do it itself or, if it makes sense to delegate operational responsibiliy, it may pay the entire cost eitherdirectly or through a cost-reimbursement grant.

28

DOX 15. Local OovernmenX ......

....a s . d o the l g ts ed elo ,co. c1d

. .~~~~~~~~~~~...... .......... .. . ...,.....s ..........

M' sW&~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..i. ... .... c....

neatly~~~~x~ bet h ~iftx's ofteatvdgo oa omet~ n ayuoda ihitdclt

t ettt,gjjfi,gottSS...o.,W,.~~~~~~~~~~~~~~~~~~~~~~~~~... .....

thirW omonay.s of uh edfor tuhe cprojet ad o tuhe sclapoaciy ofethehrecipienttheyexayute

(or otherad appropin ate agny ) s hec uld tadredynanovdschtchial assistance fo ihrlvl fgvmt Scase may beaneedd i developin nes'tm ~ent plansz, arranging fwoinancin lans,~m~3*evpweconsrution,andman in tefclityabone contrnbuctd. e It ay ls

assist t no 'pprlocalsasein their ft=e Isialstructue needs n e ole ty helpn plavye the

Both nee and apacitbshuld b taken int ccount:t for uther iscussio of thme dsg ofmamin w andeaothe) uwfers theeir and aui4 h eer,1992 iadan aWatd) wMg.4asluis1

cononpob. = -~otn biu,ugme deiablt nUb irese =oca ptciaos i .pltiupocs ta zsle.O ci

tb zeutaoth onl nbidigppoeta ils ft&atemue& Noehees whte loa ulcsrie r rvddb opoi omn~ soibobypivat dvlpiet(a s nssingl :-:ie asei h S o aumle,siteeaom 5000reiats omnt wocatinwnwoea.,tene o uc mboa ciiissebru

miiial.vic.

their money) of the need for the project and of the capacity of the recipient to executeand finance itY2

t3] Information asymmetry works both ways: the central government does not know~bai to do, the local government does not kcnow hbe to do it. The granting authority(or other appropriate agency) should stand ready to provide such technical assistanceas may be needed in developing investment plans, arranging financing plans,managing construction, and maintaining the facility once constructed. It may alsoassist localities in assessing their future infrastructure needs e.g. by helping survey thecondition of existing infrastructure.

X Both need and capacity should ie takecn into accounte for further disusion Of the design of n2thing (andother) transfers: cec Bird and Wallich (1992).

29

(4] In the absence of shared goals, the granting authority should also monitor andevaluate the progress of projects, requiring progress reports, perhaps making fieldinspections, and conducting periodic evaluations of outcomes in order to improve itsprocedures, assist applicants, and better assess their capacity to carry out theirpromises and the extent to which they are carried out (Box 14). A credibleenforcement mechanism is needed to ensure that the contract explicit (or implicit) in amatching grant is carried out.

The similarity of some of these requirements to the sort of institutional support noted earlierto be essential for successful decentralizion of purely local infrastructure investmentdecions is, of course, not a coincidence.' The costs of monitoring intergovernmentaltnsactions is not small and must be explicitly taken into account in designing andimplementing decentralization policies.

The principles just staed apply to all intergovernmental financial assistance forspecific investment projects. What they do not and cannot indicate is just how muchassistance should be provided to whom for what. To make such principles operational in anyparticular country, a good deal of work would have to be done e.g. to identify as best aspossible the proportion of the cost of partcular projects that should be borne by other thanlocal residents. In addition, since both the willingness (price-elasticity) and the ability(income-eJasticity) of different communities to contribute the locally-financed proportion willvary with the nature of the project and the wealth and interests of the community, an'equalization component may be needed even in a strictly project-oriented matching grantprogram, thus further complicating the design of such programs.2 '

Conclusion

it is difficult to draw very sharp conclusions from the rather diffuse exposition of principlesand examples relating infrastructure and decentralization that has been presented in thispaper. On the whole, however, the moral of the story this paper tries to tell nay perhaps besummed up in a few short propositions, as follows:

z At this point - or sooner - some might say: if all this effort will be needed to make dectralization work,why bother? Why not simply eate a single national agency to, say, provide water or power, saff it withcompetent technicins, allow it to set its prices properly (providing from budgetary revenues any socially-requiredshortfall), and let it do the job? Not only will the danger of political interference and technical and financialmismanagement be reduced by this approach, but it is obviously much easier for an agency like the World Bankto dtal with a single professional instiution than with lOOs or 1000s of less-qualified local agencies. This line ofthought is tempting - but it is also of course precisely tlk: slippeay path of overcentralized, unresponsive decision-making from which so many countries are trying, however imperfectly, to redeem themselves by various forms ofdecentralization. For a detailed evaluation and critique of the 'professional' approach to infmstniture provisionand the ensuing ills of 'functional fragmentation,' see Bird (1980), pp. 28-32.

2 See Feldstein (1975) and Bird and Wallich (1992).

30

[1] In all countries, some critical infrastructure is provided through a decentralizedpolitical structure. Moreover, cunrent trends in many countries make it likely thatthis will be even more true in the future. It is therefore important to understand therelation between infrastructure and decentralization.

[2] The first thing to be understood about this relation is that decent;alization,however defined, in and of itself has no necessary implications for good or evil so faras infrastructure is concerned: its effects depend upon the incentives facing thevarious decision-makers in the decentralized structure.

[3] The key to ensuring that these incentives are conducive to 'good decisions withrepIect to the design, siting, timing, finance, pricing, opration, maintenance, andutilization of infrastructure is to ensure that to the greatest extent possible those whomake the decisions bear the financial (and political) consequences.

[4] Politically, what this means is that political leaders at all levels should beresponsive and responsible to their constituents, and that those constituents are fullyinformed about the consequences of their (and their leaders') decisions. Makingpoliticians bear the consequences of their own mistakes is as close as one can get to a'hard political budget constraint.'

[5] Economically, what is required is to make it difficult for local residents to shiftcosts to nonresidents who do not receive benefits and to make local decision-makersfully responsible to their citizens for the use they make of revenues collected fromthem (through local taxes), to users of infrastructure, local or otherwise for the usemade of the revenues they contribute (through user charges of various sorts), and totaxpayers in gener?l for the use made of any transfers (or subsidized loans) theyreceive.

[6] Administratively, what such a system requires is a clear set of 'framework' laws(on local budgeting, financial reporting, taxation, contracting, dispute settlement,rules to be followed in designing user charges, etc.), as well as adequate institutionalsupport for localities to operate in this environment.

It is of course much easier to lay down such general prescriptions than to satisfy themin the very diverse situations found in the real world. Nonetheless, to the extent that theseconditions are not met, the perverse incentives that too often already exist owing to thestructure and finance of the public sector in many countries seem all too likely to beexacerbated by the current tendency to decentralize more and more decisions in the publicsector. Decentralization - or, perhaps better, the realization that the optimal decision-makingstructure in the public sector is almost certainly polycentric (non-centralized) in nature - may

> For further discussion, see Ismel (1987) and Wiesner (1993).

31

in principle yield a more efficient and equitable pattern of infrastructure investment and usethan the overcentralized and unresponsive public sector found in many developing counties.But it will do so in practice only if it is properly implemented, along the lines sketched here.

I ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

32

REFERENCES

Aschauer, David (1989) *Is Public Expenditure Productive?' Journal of MonetarEconomics, 23: 177-200.

Bird, Richard M. (1980) Central-Local Fiscal Relations and the Provision of Urban PublicSevices (Canbra. Centre for Research on Federal Fimancial Relaions Australian NationalUniversity).

Bird, Richard AL (1984) Inteiovernment Fiscal Relations in Colombia (Cambridge, Mass.:Harvard Law School Intnatil Tax Program).

Bird, Richard M. (1993) 'Threding the Fiscal Labyrinth: Some Issues in FiscalDcentralization,' National Tax Journal, 46: 207-27.

Bird, Richard M. (1993a) 'A Comparative Perspective on Fedeal Finance," Paper preparedfor Conference on The Future of Fiscal Federalism, Queen's University, Kingston, Canada,November.

Bird, Richard M. and Susan Horton, eds. (1989) Government Policy and the Poor inYevel2Ding Countries (roronto: University of Toronto Press).

Bird, Richard M. and Enid Slack (1993) Urban Public Finance in Canada (2nd ed.; Toronto:John Wiley).

Bird, Richard M. and Christine Wallich (1992) 'Financing Local Government in Hungary,'WPS 869, World Bank, March.

Bird, Richard M. and Christine Wallich (1993) 'Fiscal Decentralization andIntergovernmental Relations in Transition Economies: Toward a Systemic Framework ofAnalysis,' WPS 1122, World Bank, March.

Brent, Robert J. (1993) 'Cost-Benefit Analysis, User Prices and State Expenditures inIndia,' Worldng Paper No. 5, National Institute of Public Finance and Policy, New Delhi,May.

Breton, Albert (1989) *The Growth of Competitive Governments,' Canadian Journal ofEconomics, 22: 717-50.

Campbell, Tim, Ceorge Peterson, and Jose Brkar (1991) oGovernment in LAC National Strategies and Local Response in Planning. Spending. andManagement, Regional Studies Program Report No. 5, Latin America and the CaribbeanTechnical Department, World Bank.

33

de Soto, Herando (1989) The Other Path: The Invisible Revolution in the Third World(New York: Harper and Row).

Dilger, Robert Jay (1993) 'Residential Community Associations: Their Impact on LocalGovenment Finance and Politics," Government Finance Review, 9: 7-10.

Dreze, Jean and Mrinalini Saran (1993) 'Primary Education and Economic Development inChina and Iduia Oveview and Two Case Studies,' No. 47, Development EconomicsResearch Programme, London School of Economics, September.

Feldtin, Martin (1975) Wealth Neutraity and Local Choice in Public Education,Americ Economic Review 65: 75-89.

'Financing Rural Road Maintenance in Indonesia' (1991) evenws (Metropolitan StudiesProgram, Syracuse University), Summer.

Isael, Aro (1987) Institutional Development: Incentives to Perfonmance (Btimore: JohnsHopkins University Press).

Jenkins, Jerry, ed. (1988) Beyond the Informal Sector Including the Excluded in DevelopingCountries (San Francisco: ICS Press).

Kitchen, Harry (1993) 'Efficient Delivery of Local Government Services,' Discussion Paper93-15, School of Policy Studies, Queen's University, Kingston.

Ostrom, Elinor, Larry Schroeder, and Susan Wynne (1993) Institutional Incentives andSustainable Development: Infastrcture Policies in Perspectve (Boulder: Westview Press).

Paul, Samuel (1991) Strengthening Public Service Accountability, World Bank DiscussionPaper 136.

Rao, M. Covinda (1991) *State Government Transfers to Urban Local Bodies, NationalInstitute of Public Finance and Policy, New Delhi.

Rao, M. Govinda and A. Das-Gupta (1993) 'Intergovemmental Transfers as an Instrument toAlleviate Poverty,' Economic Development Institute, World Bank.

Slack, Enid and Richard M. Bird (1991) 'Financing Urban Growth through DevelopmentCharges, Canadian Tax Journal, 39: 1288-1304.

Stackhouse, John (1993a) 'Underground Movement No Longer a Pipe Dream,' Te GIandMail (Toronto), September 29.

34

Stackhouse, John (1993b) "An Environment for Controversy," The Globe and Mail(Toronto), September 29.

United Nations Economic and Social Commission for Asia and the Pacific (1991) EiseDecentalization and the MobiliZdon and Use of National Resources for Development:Issues, Expence and Policies in the ESCAP Region (Bangkok).

Webber, Melvin et al. (1991) 'World Bank Urban Projects in Madras: A Review," INURDTechnical Paper, April.

Wiem D., Eduardo (1993) Prom M onomic Correction to Public Sector Reforrn TheCritical Role of Evluation World Bank Discussion Paper 214.

World Bank (1988) World Develgpment Report 1988 (Published for the World Bank byOxford University Press).

World Bank (1989) 'Colombia: Decetralizng Revenues and the Provision of Services - aReview of Recent Experience," Report No. 7870-CO, October.

World Bank 11993) 'The Impact of Instructure on Economic Development," LA4IE, May14.

World Bank (1993a) 'Municipal Development Sector Review", Draft paper.

Pollcy Research Working Paper Series

ContactTitle Author Date for paper

WPS1235 Eastern Europe's Experience with Alfredo Thorne De. aer 1993 N. JoseBanking Reform: Is There a Role for 33688Banks in the Transition?

WPS1236 The Impact of Two-Tier Producer Maurice Schiff December 1993 S. Fallonand Consumer Food Pricing in India 38009

WPS1237 Bank Performance and the Impact Yavuz Boray December 1993 C. Limof Financial Restructuring in a Hector Sierra 30864Macroeconomic Framework: A NewApplication

WPS1238 Kenya: Structural Adjustment in the Gurushri Swamy January 1994 '.. Saldanha1980s 35742

WPS1239 Principles of Regulatory Policy David E. M. Sappington January 1994 WDRDesign 31393

WPS124o Financing the Storm: Macroeconomic William Easterly January 1994 R. MartinCrisis in Russia, 1992-93 Paulo Vieira da Cunha 39026

WPS1241 Regulation, Institutions, and Pablo T. Spiller January 1994 B. MooreCommitment in the British Ingo Vogelsang 35261Telecommunications Sector

WPS1242 Financial Policies in Socialist Boris Pleskovic January 1994 M. JanduCountries in Transition 33103

WPS1243 Are Institutional Investors an Punam Chuhan January 1994 R. VoImportant Source of Portfolio Investment 31047in Emerging Markets?

WPS1244 Difficulties of Transferring Risk- Edward J. Kane January 1994 P. Sintim-AboagyeBased Capital Requirements to 38526Developing Countries

WPS1245 The Adding-Up Problem: Strategies Takamasa Akiyama January 1994 A. Kimfor Primary Commodity Exports Donald F. Larson 33715in Sub-Saharan Africa

WPS1246 Determinants of Cross-Country Branko Milanovic January 1994 R. MartinIncome Inequality: An 'Augmented" 39065Kuznets' Hypothesis

WPS1247 Complex Transactions Under Andrew Stone January 1994 P. Sintim-AboagyeUncertainty: Brazil's Machine Tool Industry 37644

WPS1248 Do Domestic Firms Benefit from Brian Aitken February 1994 D. BallantyneForeign Direct Investment? Evidence Ann Harrison 37947from Panel Data

Policy Research Working Paper Series

ContactThle Author Date for paper

WPS1249 Competitiveness and Environmental Piritta Sorsa February 1994 P. KokilaStandards: Some Exploratory Resu.s 33716

WPS1250 Explaining Miracles: Growth William Easterly February 1994 R. MartinRegressions Meet the Gang of Four 39026

WPS1251 Excise Taxes John F. Due February 1994 C. Jones37699

WPS1252 On the Dangers of Decentralization Remy Prud'homme February 1994 TWUTD31005

WPS1253 Can Conipetition Policy Control 301? J. Michael Finger February 1994 M. PatanaK. C. Fung 37947

WPS1254 What A a OECD Trade Preferences Alexander J. Yeats February 1994 J. JacobsonWorth .) Sub-Saharan Africa? 33710

WPS1255 Intrahousehold Resource Allocation: Lawrence Haddad February 1994 P. CookAn Overview John Hoddinott 33902

WPS1256 World Fossil Fuel Subsidies and Bjorn Larsen February 1994 C. JonesGlobal Carbon Emissions in a Model 37699wrth Interfuel Substitution

WPS1257 Old-Age Security in Transitional Louise Fox February 1994 E. VincentEconomies 82350

WPS1258 Decentralizing Infrastructure: Richard Bird February 1994 'NDRFor Good or for III? 31393

WPS1259 The Reform of Fiscal Systems in Robin Boadway February 1994 C. JonesDeveloping and Emerging Market Sandra Roberts 37754Economies: A Federalism Perspective Anwar Shah

WPS1260 When Is a Life Too Costly to Save? George L. Van Houtven February 1994 A. MaranonEvidence from U.S. Environmental Maureen L. Cropper 39(74Regulations