Devarajan & Reinikka (2004) Making Service Work for Poor People

8/13/2019 Devarajan & Reinikka (2004) Making Service Work for Poor People

1/25

Making Services Work for Poor People

Shantayanan Devarajan and Ritva Reinikka1

The World Bank Group

The weak link between public spending in health and education, and

health and education outcomes can be partially explained by the fact that

the delivery of services that are critical to human development health,

education, water and sanitation are widely failing poor people. The

money is often spent on private goods or on the non-poor; it often fails toreach the frontline service provider; incentives for service delivery by

providers are weak; and poor people sometimes fail to demand these

services. This paper examines the experience with alternative mechanisms

for service delivery contracting out to the private and NGO sectors,

community participation, co-financing by service beneficiaries and

shows that this, as well as the experience of more traditional public sector

provision, can be interpreted by looking at three principal-agent

relationships in the service-delivery chain: between policymakers and

providers; between clients and providers; and between clients (as citizens)

and policymakers. Weaknesses in one or more of these relationships can

lead to service-delivery failure, while attempts to strengthen one may not

always work because of deficiencies in other links in the chain.

1. Introduction

Over the last decade, a consensus has emerged that poverty ismulti-dimensional. Along with income, health, education, genderequality and a sustainable environment are now seen as integral

JOURNAL OFAFRICANECONOMIES, VOLUME13, AERC SUPPLEMENT1, PP. i142i166

Journal of African Economies Vol. 13 Suppl. 1, Centre for the Study ofAfrican Economies 2004; all rights reserved

1 The topic of this paper is the theme of the World Development Report 2004, ofwhich Devarajan is the Director and Reinikka the Co-Director. This paper presentssome of our preliminary ideas on the subject. They do not necessarily representthe views of the World Bank. We are grateful to Jeffrey Hammer, Emmanuel

Jimenez, Bernadette Kamgia, Michael Kremer, Magnus Lindelw, Nicholas Stern,an anonymous referee, and participants at the May 2002 AERC Plenary Session forhelpful comments and suggestions in developing some of the ideas in the paper.


2/25

elements of human well-being. Nowhere is this consensus betterreflected than in the Millennium Development Goals (MDGs), anunprecedented agreement by the development community about thegoals of poverty reduction (Table 1). The MDGs range from incomepoverty (living on less than US$1 a day) to child mortality and primaryeducation, to gender equality, maternal mortality and safe water andsanitation. That each goal has measurable targets to be reached by 2015also means that the MDGs enable the development community tomonitor its progress toward poverty reduction.

Unfortunately, without concerted action, many countries, especiallythose in low-income Africa and South Asia, will not reach these goals

by 2015 (Sahn and Stifel, 2002; World Bank, 2002). How then can we

accelerate progress toward the MDGs? At least two findings stand out:

Growth alone is not enough (Table 2). While per-capita GDP growthhas a significant effect on health and education outcomes, just as it

Table 1:Millennium Development Goals (Each goal is to be achieved by 2015,

compared with 1990 levels)

1. Eradicate extreme poverty and hunger Halve the proportion of people with less than one dollar a day. Halve the proportion of people who suffer from hunger.

2. Achieve universal primary education Ensure that boys and girls alike complete primary schooling.

3. Promote gender equality and empower women Eliminate gender disparity at all levels of education.

4. Reduce child mortality

Reduce by two thirds the under-five mortality rate.5. Improve maternal health

Reduce by three quarters the maternal mortality ratio.6. Combat HIV/AIDS, malaria and other diseases

Reverse the spread of HIV/AIDS.7. Ensure environmental sustainability

Integrate sustainable development into country policies and reverseloss of environmental resources.

Halve the proportion of people without access to potable water. Significantly improve the lives of at least 100 million slum dwellers.

8. Develop a global partnership for development Raise official development assistance. Expand market access.

Making Services Work for Poor Peoplei143


3/25

Table 2:Growth is not Enough

Annual averageGDP per capitagrowth, 200015(%per year)

$1 a day (%)

Target

2015growthalone

East Asia 5.4 14 4 Europe and Central Asia 3.6 1 1 Latin American and the Caribbean 1.8 8 8 Middle East and North Africa 1.4 1 1 South Asia 3.8 22 15 Africa 1.2 24 35

Sources: GDP growth projections fromGlobal Economic Prospects 2003; Devaraja


4/25

does on poverty reduction, the magnitude of the effect is muchsmaller for the human-development indicators. For instance, the

growth elasticity of poverty is between 1 and 2, whereas the growthelasticity of infant mortality is between 0.1 and 0.2 (Pritchett andSummers, 1996), and that of primary enrolment is around 0.4(Schultz, 1987).

Increasing public expenditures alone is not enough. Since growthalone may not be sufficient, many observers have called for anincrease in financial resources going to health and education, inorder to accelerate progress towards those MDGs. However, there isconsiderable evidence that increasing public expenditures in healthand education, if current patterns of allocation and use continue,will not on average improve outcomes in these sectors. For example,the cross-country evidence shows that public expenditures have nosignificant association with health and education outcomes, whenyou control for the countrys level of per-capita GDP and a smallnumber of socio-economic variables (Filmer et al., 2000; Filmer,2001). To be sure, these cross-country regressions at best only tell uswhat the average relationships are. There are individual countries,

and programmes within countries, where public spending has hada powerful effect on outcomes. Nevertheless, this second finding isparticularly troubling since the international community has alsocalled for an increase in domestic and external financing toaccelerate progress toward the MDGs (Devarajan et al., 2002). Ifcurrent patterns of public-expenditure productivity continue, thisadditional financing is unlikely to make a difference.

Why does public expenditure on average have such a limited

effect on health and education outcomes? We can break the problemdown into at least four components. First, governments may bespending on the wrong goods or the wrong people. A large portion ofpublic spending on health and education is devoted to private goods ones where government spending is likely to crowd out privatespending (Hammer et al., 1995). Furthermore, most studies of theincidence of public spending in health and education show that the

benefits accrue largely to the rich and middle-class; the share going tothe poorest 20% (where it can make a difference) is always less than20% (Table 3).

Secondly, even when governments spend on the right goods or theright people, the money fails to reach the frontline service provider. In



5/25

Uganda in the early 1990s, the share of nonwage recurrentexpenditures for primary education actually reaching the primaryschool was only 13% (Reinikka and Svensson, 2001). There wasconsiderable variation in grants received across schools but largerschools and schools with wealthier parents received a larger share ofthe intended funds (per student), while schools with a higher share ofunqualified teachers received less. These findings are comparable tosimilar studies in Ghana, Tanzania, and Zambia (for a review see Dehnet al., 2003).

Thirdly, even when the money reaches the primary school or healthclinic, the incentives to provide the service are often very weak. The

service providers may be poorly paid and hardly ever monitored. Theyrespond to incentives from the central-government bureaucracy, whichis mostly concerned with inputs rather than outputs. The clients,meanwhile, be they schoolchildren, parents, patients or expectantmothers, have limited knowledge and even less power to hold theservice provider accountable.

Fourthly, even if the services are effectively provided, householdsmay not take advantage of them. For economic and other reasons,parents pull their children out of school or fail to take them to theclinic. This demand-side failure often interacts with the supply-sidefailures to generate a low-level of public services and outcomes amongthe poor.

Table 3:Benefit Incidence of Public Spending

Health Education

Country/yearPoorest 20%of population

Richest 20%of population

Poorest 20%of population

Richest 20%of population

Cte dIvoire (1995) 11 32 13 35Ghana (1992) 12 33 16 21Guinea (1994) 4 48 5 44Kenya (1992) 14 24 17 21

Madagascar (1993) 12 30 8 41South Africa (1994) 16 17 14 35Tanzania (1992/93) 17 29 14 37

Source: Castro-Lealet al.(1999).

i146Shantayanan Devarajan and Ritva Reinikka


6/25

TheWorld Development Report 2004is aimed at making services workfor poor people, so that we can make greater progress toward the

MDGs. This paper is an introduction to that report. In Section 2, weexamine cases where policies have resulted in better service delivery,as well as the reasons why in other situations they have not. In Section3, we use these examples to develop a framework for identifyingpublic actions that could result in better service-delivery outcomes. Wedescribe what research has to say about the different elements of theframework. Finally, in Section 4, we highlight some of the critical,unanswered questions in this area. Throughout the paper (and thereport), our focus will be on health, education, water and sanitation

services, because they are critical to achieving the human-development MDGs. Many of the lessons obtained from frontlineservice delivery in health, education and water are, however,applicable to other services as well.

2. Some Examples of Making Services Work for Poor People

The four reasons why public expenditure has such a weak relationship

with human-development outcomes provides a framework forexamining the successful and unsuccessful attempts at improvingservice delivery. The first reason that governments may bemisallocating resources stems from the usual argument thatgovernments should intervene only if there is a market failure orredistributive rationale. Note that this is an argument for government

financingof these activities not necessarily for governmentprovision.We discuss below alternative types of service providers for publiclyfinanced goods and services.

What is worth noting here is that, despite the evidence onexternalities in primary education and health, there is substantialprivate financing of primary education and health in developingcountries (Figure 1). Clearly, there is private demand for these services,even (or maybe especially) among the poor. Furthermore, the presenceof government failure may undermine the classic arguments for publicfinancing based on market failure. In India where 80% of curativehealth expenditures are private the poor frequently bypass free,

public clinics to use fee-based private services because the latter areoften better equipped, including with medical personnel (World Bank,2001). A similar phenomenon has been observed when clinics orschools levy user fees for recurrent expenditures, such as textbooks or



7/25

medicines (Litvack and Bodart, 1993; Burnett and Kattan, 2002). InKenya, until very recently, all such recurrent expenditures in primaryschools were paid by the parents of schoolchildren. To be sure, these

user fees have in some cases prevented poor families from havingaccess to health care and education. But as the Litvack and Bodartstudy of Cameroon showed, the introduction of user feesincreaseduse

by poor people of the local clinic. Previously, they had to travel longdistances to the nearest town because the local clinic lacked essentialmedicines and instruments.

Support for user charges for social services comes also fromcross-sectional estimates showing small price elasticities for educationdemand (e.g., Jimenez, 1987). When there are large policy changes,

however, evidence from a number of cases where user fees wereabolished points to a different outcome. Perhaps the most remarkablefeature of the 1997 universal primary education initiative in Ugandawas the doubling of overall enrolments following the removal ofschool fees (Appleton, 2001). Similar enrolment surges occurred as aresult of comparable initiatives in Kenya and Tanzania in the 1970s,and more recently in Malawi in 1994, when enrolments rose by 50%following the abolition of primary school fees (Reddy andVandermoortele, 1996). Had we used data from Uganda before theremoval of school fees, we would have obtained small and indeedperverse estimates of the price elasticities for enrolments. In fact, theactual response of enrolments to the abolition of fees was very strong.

Figure 1:Private Share of Primary Enrolments (%)

Source: EdInvest (2000), World Bank (2002).



8/25

Similarly, randomised experiments from Kenya show that an NGOprogramme that provides school uniforms the most significant

private cost of education in Kenya increased years in school by 17%and grade attainment by 15% (Kremeret al., 2002). The history of userfees and gross primary enrolment in Kenya tells a similar story. Whenfree primary education was introduced in 1974, gross enrolment rosefrom 68 to 98%. When cost-sharing was reintroduced in the 1980s,gross enrolment fell from 116% in 1980 to 104% in 1986.

Payments by households or local communities can also addressthe second weak link in the chain, namely, the problem ofcentral-government allocations not reaching the service provider.The provision of health and education services, especially, involvesubstantial non-wage recurrent expenditures, such as textbooks, chalk,medicines, syringes, etc. Yet these are the very expenditures that leakfrom initial government allocation to delivery agency.

At least two other factors contribute to the underfinancing of theseexpenditures. The first factor is that much donor financing, includingall World Bank investment operations, only pay for capitalexpenditures. Recurrent expenditures are covered by domestic

resources. But governments rarely take into account their ability tocover the recurrent costs when deciding on a donor-financedinvestment project (World Bank, 1997). As a result, the Africanlandscape is littered with schools without chalk and textbooks, clinicswithout medicines. When, in Uganda, the World Bank and otherdonors switched their assistance to untied budgetary support, therecurrent budget was able to accommodate the increase in the teachingforce needed to lower the studentteacher ratio, which had reached100 to 1 in the lower grades of primary education.

The second factor that contributes to the underfinancing of non-wage recurrent expenditures is that, within recurrent expenditures, thelions share goes to wages and salaries. With their political power,teachers and doctors are able to protect their incomes when there ispressure for budget cuts. The only thing left to cut, therefore, is non-wage operations and maintenance expenditures. Many governmentshave responded by creating a second class of teachers who are outsidethe civil service, and are correspondingly paid less with fewer benefits.The experience in several West African countries shows that there aremany people willing to take these jobs (a recent announcement inSenegal generated 30,000 applicants for 1,000 positions); even if theyare less qualified, the evidence on student performance is mixed; and,



9/25

over time, these contractual workers have come to dominate the publicservice, as in Benin.

The political power of teachers unions or the local (bureaucratic orpolitical) capture of public funds can be countered if citizens have theinformation to lobby for a better mix of inputs or for making sure that

budget allocations are actually received. The Uganda case illustratesthe impact that collection and dissemination of quantitative data canhave as a tool to mobilise voice (Hirschman, 1970). Complaints aboutservices based on isolated experiences tend to be brushed aside asanecdotal or at best partial evidence. But when systematic comparativedata support public feedback, it is difficult to ignore and, as theUganda case shows, it can provide a spark for public action. Whenevidence of the degree of leakage became public knowledge inUganda, the central government enacted a number of changes: it

began publishing the monthly transfers of public funds to the districtsin newspapers, broadcasting information on the transfers on radio,and requiring primary schools to post information on inflows of funds.The objective of this information campaign was to promotetransparency and increase public sector accountability in two different

ways. First, the central government signaled that it had moreinformation on local governments actions than was the casepreviously. Secondly, by giving citizens access to information on thecapitation grant programme for primary schools, they empoweredschools and citizens to monitor and challenge abuses of the system.Initial (internal) assessments of these reforms a few years later showthat the flow of funds improved dramatically (Republic of Uganda,2000, 2001).

Regarding the third weak link, even if the mix of current and capital

spending is about right, because these services are transaction-intensive, it ispeoplewho have to make services work (Pritchett andWoolcock, 2002). Many of the problems confronting service deliveryhave to do with the incentives facing service providers. First, many ofthese services require skilled personnel qualified doctors, trainedteachers, etc. Not only are many countries short of these skills, but theremuneration and working conditions, especially in remote ruralareas, makes it difficult to get qualified people to work there. The

vacancy rate for clinics in West Papua, Indonesia is almost 80%,whereas in Bali or Jakarta it is less than 2%.More fundamentally, effective service delivery involves more than

just being able to hire skilled staff. It requires that they show up for



10/25

work, and perform their duties in a reasonable fashion. This isprobably the biggest constraint to making services work for poor

people. Since the delivery of the service is poorly monitored (if at all),the service provider does not feel accountable. Typically, he is anemployee of the central government, which sends him a paycheckregardless of whether he performs his functions. The symptoms of thisproblem are everywhere (absenteeism in schools, nurses hittingmothers during childbirth, drunk teachers, etc.). Many governmentshave tried to address this problem by contracting out the services tonon-governmental organisations (NGOs) or private organisations,who are then held accountable for performing the contract. Fourexamples illustrate this point.

In Cambodia, where 30 years of civil war had depleted the countrysphysical and human health infrastructure, the government introduceda programme of contracting with NGOs to manage district healthservices (Bhushan et al., 2002). They introduced two types of con-tracting: contracting out (CO), where the NGO can hire and firestaff, set wages and procure drugs on their own; and contracting in(CI), where the NGO managed the district from within the Ministry of

Health, they could not hire and fire staff (although they could transferthem), and they had to obtain drugs from the government. The twelvedistricts (with populations ranging from 100,000 to 180,000) wererandomly assigned to CO, CI or the standard government manage-ment, the control/comparison group (CC). A before-and-after surveyof both households and health facilities (carried out by a third party)revealed a clear pattern: districts that contracted out registered the

biggest improvement, with those contracting in next, and the controlgroup last (Figure 2).

The decade-long civil war in El Salvador left its education system,especially in the rural areas, in shambles. Recognising the difficulty ingetting urban teachers to serve in these fairly remote areas, thegovernment introduced the EDUCO programme, whereby localcommunity organisations, known as ACEs, could hire teachers on oneyear renewable contracts, pay the teachers and manage a small fundfor school supplies (Jimenez and Sawada, 1999). Each ACE would havea contract with the Ministry of Education that set personnel guidelines,including criteria for hiring teachers. The ACEs would meet with allthe parents once a month, and regularly visit the schools and teachers.The result has been that EDUCO is associated with an increase innet enrolment rates from 76% in the 1980s to 85% in 1995. More



11/25

importantly, student and teacher absenteeism was significantly lower,

and surveys attributed this to the greater community participation inthe schools. Finally, there is evidence that EDUCO was also associated

with an increase in student learning, with both math and language

scores higher (although only the former was statistically significant).In Burkina Faso, a seemingly dysfunctional health system and

declining health indicators were turned around by a donor-financed

project that required each health district to sign a contract with the

Ministry of Health, based on district health plans that were prepared

by district health teams and civil society (Vaillancourt, 2002). Eachdistrict set targets based on performance indicators. The financial

accounts of the districts were reviewed monthly, and the performance

targets were monitored quarterly. The bank accounts of each district

were replenished based on satisfactory reviews of their performance

on both technical and financial-management grounds. After 10 years

of declining health indicators, the first years review of the programme

showed improvements in vaccination rates, utilisation rates of critical

curative and preventive services (especially for mother and child

health) and of family planning services.The municipality-run water and sanitation system in Cartagena,

Colombia, was plagued with chronic inefficiency, excessive political

Figure 2:Relative Performance of NGO Contracting Schemes in Cambodia (Percentageincrease from baseline of different health indicators)



12/25

interference, poor maintenance, poor service delivery, and weakcommercial and financial management. To address these problems, the

government decided to liquidate the water authority, and offer acontract for assuming responsibility for operations and maintenanceas well as some of the new investment needed. A long-term (25 years)contract with performance criteria requiring improvements in servicewas signed with a Spanish water company. The result has been asubstantial increase in the quality of water and sanitation services inCartagena. All the performance indicators show a marked improve-ment. The number of water connections went up by almost 50%,continuity of service went from 7 hours per day to 24, and new

connections in poor areas went up by 98% (Libhaber, 2002a,b).Furthermore, customer waiting time at the service facility dropped byover half, from 30 minutes to 14 minutes. A similar concession inCochabamba, Bolivia, however, had less favourable results. Threemonths into the contract, farmers were informed that their water tariffswere being increased significantly. Riots broke out and two peoplewere killed by the police. The government cancelled the contract, andthe utility is now being managed by a users association.

Finally, many governments have experimented with demand-sideinterventions, to address the fourth weak link in the chain. Oneexample is the Female Secondary Scholarship Programme inBangladesh. Under this programme, a girl gets a stipend, depositeddirectly into a bank account in her name, provided she can show thatshe is attending secondary school, maintaining passing grades and isunmarried. The school in turn will receive a transfer proportional tothe number of girls enrolled. While the programme has yet to beevaluated rigorously, preliminary estimates indicate that it has had a

sizeable impact on secondary enrolment rates among Bangladeshigirls. Another, more celebrated, example is the PROGRESAprogramme in Mexico, which has been extensively analysed. Thisprogramme gave cash transfers to families provided they regularlyvisit the clinic and send their children to school. Most of the analysesshow significant improvements in health and education status of thesepoor families.

3. Elements of a Framework for Analysing Service Delivery

The examples outlined above illustrate both the potential and thechallenges in making services work for poor people. To understand the



13/25

problem better, we now propose a simple framework for examiningvarious ways of improving service delivery. Our starting point is that

the service delivery chain is a series of principal-agent relationships.The organisational structure of public sector agencies involvesmultiple tiers of management and frontline workers. In addition,public services have multiple stakeholders. Multiplicity is also acharacteristic of the tasks that need to be performed. Consequently, theoutput and actions of public services are often difficult to verify ormeasure. Although incentive theory having been developed mainlyto understand the firm does not specifically address issues relatedto public service delivery, it provides a useful framework and testablehypotheses.2

Dixit (2000) gives an example of multi-tasking and multiplestakeholders in the context of public education. The multiple tasksinclude providing literacy and numeracy and other direct skills,supporting the emotional and physical growth of children, providingvocational skills and preparing pupils for working life, providingskills in health and financial management, instilling citizenship,overcoming the disadvantages of home life, and ensuring children can

grow up in a violence-free environment. While these goals are notmutually exclusive, they compete for limited resources in the schoolsproduction process, and, as mentioned earlier, it is often difficult tomeasure the output of each of these tasks.

Similarly, the diverse body of stakeholders in public educationincludes parents and children, teachers and their unions, taxpayers,potential graduate employers, society as a whole, private schools, andvarious groups favouring or opposing specific components of thecurriculum. These principals have diverse preferences and objectives.Parents want good education and day care for their children, teachersand unions want higher pay, taxpayers want low costs, employerswant vocational skills, society wants good citizens, and private schoolscompete for pupils and some public funds. Again, many of theseobjectives are mutually conflicting for instance, taxpayers objectiveof low costsvis--visthe teachers and unions goal of having higherpay. And yet, teachers are often not interested in money alone but inchallenging work and career prospects.

The existence of multiple principals reduces the agents incentives,

2 In the last few years, applications of this theory to the public sector have begunto emerge in the literature (e.g., Burgess and Metcalfe, 1999; Dixit, 2000).



14/25

because activities desired by the principals to realise their respectivegoals are often substitutes for each other (Bernheim et al., 1986;

Holmstrm and Milgrom, 1988; Martimort, 1992; Dixit, 1996, 1997).Similarly, when some task outcomes are verifiable and other are not, itmay not be optimal to provide explicit incentives to any task as theagent would divert all effort from unverifiable to verifiable tasks. Ineducation, for example, exam results would be disproportionatelyemphasised. Incentive schemes are most suitable when tasks areclearly defined and unambiguous, and become weak when neitheroutcomes nor actions are verifiable, such as in a typical governmentministry. Public service providers also often lack competitors. While

the introduction of competition does not in itself guarantee betterperformance, the lack of it places greater emphasis on other manage-ment devices.

Standard principal-agent theory also provides a framework foranalysing incentives of frontline workers. Health workers, forexample, can be seen as agents for multiple principals, includingpatients, communities and government agencies with responsibilitiesfor the delivery of health services. In general, the challenge is to induce

health workers to exert effort in a range of different areas: clinical tasks(diagnosis, treatment, follow-up, outreach activities, etc.), psycho-social interaction with patients, and administration and maintenanceof hygienic standards. In addition, there is a need to restrainopportunistic behaviour (such as absenteeism, overcharging, andpetty corruption) by health workers. A sizeable literature has dealtwith the incentive issues that arise due to the asymmetry ofinformation between patients and providers (for an early contribution,see Arrow, 1963). A related literature has studied the effect of provider

payment systems on the incentives and behaviour of health workers.3Recently, the principles of agency theory have motivated reforms inpublic sector management, which emphasise performance measure-ment and incentives (e.g., Mills, 1997; Martinez and Martineau, 1998;Goddardet al., 2000). These efforts have tried to remedy a perceivedlack of incentives in the public sector by introducing systems ofrewards and sanctions in the form of performance managementsystems.

3 Gosdenet al.(1999) discuss the effect of different forms of physician payment capitation, shared financial risk, fee-for-service and salary on medical practice.Barnum et al. (1995) discuss the effect on payment systems on hospitalperformance.



15/25

In sum, the link between public spending and outcomes is a complexweb of multiple principal-agent relationships, with imperfect

monitoring and sometimes unclear objectives. To cut into this problem,we suggest below that it is useful to distinguish among threeimportant principal-agent relationships. These relationships areclosely linked, but separating them allows us to probe deeper intoeach, and understand individual experiments or innovations in termsof which relationship they strengthen.

3.1 The Policymaker and the Service Provider

Even when the policymaker is benevolent, she has the problem ofensuring that the service provider has the proper set of incentives toprovide the service effectively. In the case when the service can beeasily measured and therefore monitored, the policymaker maychoose to write a contract with the service provider. This is moretypically the case in water and sanitation, where you can specify thenumber of water connections and other monitorable performance

indicators. As a result, such contracts are more common in water andsanitation than in education, say. The Cartegena water concession is acase in point. (The unsuccessful Cochabamba privatisation failed forother reasons, not because the contracts were difficult to specify. Wewill turn to this case shortly.) For the same reason, we find fewerprivate-sector concessions in education. In health, the possibility forcontracting is always there, but it is often tailored to particularactivities. Recall that the Cambodia example of contracting healthservices to NGOs was fairly carefully circumscribed (for instance, the

contracting-in NGOs could not hire or fire workers).Workers do not respond only to externally-imposed rewards and

penalties; they are also driven by intrinsic motivation (Deci, 1975) inthe form of, for example, professional ethics or norms. Intrinsicmotivation may be strong enough to result in a self-enforcingcontract.4 Some contributors have suggested that intrinsic motivationis more likely to arise and be sustained in certain types oforganisations, so that NGOs or religious organisations, for example,

4 Kreps (1997) suggests that in many cases what is referred to as intrinsicincentives may in fact be workers response to fuzzy extrinsic motivators, such asfears of discharge or career concerns, but he also acknowledges that true intrinsicmotivation may be an important factor in many contexts.



16/25


17/25

evidence and lessons at the level of the service-providing unit, such asthe clinic or the school. A new body of micro-level evidence from

surveys of frontline service providers is underway and includespublic, private not-for-profit, and private-for-profit sectors. A surveyinstrument, the Quantitative Service Delivery Survey (QSDS), iscurrently being designed and tested for that purpose (details areavailable at http://www.publicspending.org). In the same way thatthe firm is the unit of observation in enterprise surveys and thehousehold in household surveys, the QSDS takes the frontline servicefacility or service provider, such as the health clinic or the school, asthe principal unit of observation. It provides comparable micro data on

the service delivery climate across countries. Furthermore, it isimportant to link the service-providing unit downstream to evidencefrom actual and potential users (through household surveys) andupstream to the public administration and political processes(through public officials surveys). This will allow us to analyse supplyand demand factors jointly, as well as to bring political economyfactors explicitly in the analysis.

3.2 The Service Provider and the Client

The second critical principal-agent relationship is that between theprovider and the client. As many of the examples illustrate, theoversight and accountability roles of the client can often improve thedelivery of the service. The EDUCO programme in El Salvadorshowed how parents associations or local communities, by visitingschools and monitoring teachers, could improve learning outcomes.

The many cases when the client pays to obtain the services (such asuser fees for certain health and education services) is a particularrelationship between provider and client, and one which we haveconsiderable experience in modeling. Demand-side interventions thatprovide cash transfers to clients to use in obtaining the service wouldalso fall under this category (although the decision of who gets thetransfer depends on the underlying political economy see below).In general, however, the relationship between service provider and

client does not always lend itself to economic modeling. Whilegame-theoretic models of community behaviour have been developed,some types of community action arise spontaneously. Even if wecannot model it, this phenomenon needs to be studied and



18/25

documented. To that end, these relationships are often examined in thecontext of experiments.

Policy experiments, such as the Cambodia example above, are aneffective way to obtain micro-level evidence on service delivery and itsimpact. Evaluating programmes by examining correlations of inputsand outcomes can sometimes be misleading. For instance, researchersmight observe that schools with more textbooks typically have bettereducated children. However, the greater educational achievementmight reflect other factors correlated with textbooks, such as incomeor parental interest in education, rather than being a direct causal effectof the textbooks. On the other hand, if compensatory programmes

provide textbooks to problem schools, then retrospective studies mayunderestimate the effect of these programmes. One way to addressthese concerns is to conduct randomised prospective evaluations. Suchprospective evaluations, with random assignment to treatment andcomparison groups, revolutionised medicine, and they could have asimilar impact in other fields. In several cases, field experiments haveproduced strikingly different results from retrospective econometricanalyses of inputs and outcomes (LaLonde, 1986; Miguel and Kremer,2001; Glewweet al., 2004).

In fields such as health and education, randomised evaluations areoften feasible. For example, programmes could be phased in over timewith the order of phase-in determined randomly among suitable sites.The randomisation would mean that areas treated early and latershould be comparable aside from the effect of the programme. Thenthe effects of the programme can be measured directly, and the resultswill be transparent to policymakers. One example of this is thePROGRESA programme in Mexico, where a subset of the villages was

selected randomly for delayed implementation of the programme. Thesuccess of PROGRESA is spawning similar programmes in othercountries. Other examples demonstrating the feasibility of conductingrandomised evaluations include a study of school-based dewormingin Kenya and an evaluation of a school voucher programme inColombia (Miguel and Kremer, 2001; Angristet al., 2002).

3.3 The Client and Policymaker

The third relationship is that between the client (or citizens in general)and the policymaker. The reason this relationship is important is

because it is not always the case that the policymaker acts in a



19/25

benevolent fashion. When policymakers maximise a different objectivefunction than the welfare of the citizens, the citizens or clients can

exercise various incentive mechanisms to change the policymakersbehaviour. This is the realm of political economy, a vast field that wewill not enter into here. However, one of the areas where thisrelationship becomes salient is when citizens or clients use voice toinfluence policymakers. The Uganda example of disseminating theinformation on leakage of funds had the effect of among other things,increasing citizens voice on public-expenditure decisions. Political-economy considerations are also important in understanding why theincidence of public spending is not always pro-poor. It is sometimes

necessary to marshall support from the middle class in order toprovide some support to the poor (Gelbach and Pritchett, 2000).

There is in fact a fourth relationship that we have alluded to before:that between aid donors and policymakers. Because they providefinancing for development, donors set certain requirements on howthe money is to be used. This is part of any contract.6 What is perhapsless appreciated is that the nature of the contract between donor andpolicymaker affects the delivery of services between providers andclients. Specifically, as mentioned earlier, the requirement by somedonors that they only finance capital investments results in under-financing of recurrent expenditures, especially non-wage operationsand maintenance, which in the health and education sectors, stronglyinfluences expenditure outcomes. A second phenomenon is thatdonors often insist that the recipient country follow the donorsprocedures in financial management and procurement. As a result,some countries have to comply with many different sets of procedures(typically one for each donor), using up their scarce administrative

talent. In some cases, donors set up their own project implementationunits, which draw away this scarce manpower from government,undermining financial management in the rest of the government. Animportant feature of the Burkina Faso district health programme wasthat donors pooled their resources into a common fund, so thatthe district health agencies only had to comply with a single set offinancial management procedures.

6 For another treatment of the donor-recipient relationship as a contract, see Killick(1998).



20/25

4. Concluding Remarks

The framework outlined here provides a way to examine options for

improving the outcome of public services for the poor. The differentexamples cited in Section 2 all addressed, to different extents, therelationship between the policymaker and the service provider; the

provider and the client; and the client and policymaker. In each case,

they were able to improve outcomes by addressing a particularproblem in the relationship, such as being able to monitor perform-ance. To be sure, the real world is much more complicated than these

simple examples and frameworks would suggest. Just as it may be

difficult to transfer lessons from the water sector to education, so willit be foolhardy to expect that an innovation that worked in El Salvadorwill also work in Ethiopia.

Furthermore, the examples and framework cited here only deal withthe problem of designing and implementing a programme to improve

service delivery. Difficult as that may be, it leaves out at least two other

important issues. The first is the transition to the new system. As theCochabamba and Cartegena examples show, more or less the same

programme to privatise water could have dramatically differentconsequences because of the way in which the change was introduced.In Cochabamba, tariffs were raised almost immediately, undermining

support for the reform. More generally, our framework does notaddress the question of how these innovations come about. In manycases, they were spontaneous, arising out of particularly difficult

situations or inspired by a visionary leader. They almost never

occurred as a result of a well-thought-out, long-term plan. We need to

understand better the process by which these changes happen if we areto help countries with dysfunctional service delivery systems improve.The second issue that has been left out is the sustainability of these

changes. An innovation such as contracting out health services toNGOs or community oversight of schools may yield significant resultsin the near-term. But can it be sustained in the long-run? Even if these

improvements help countries reach the MDGs, what happens after2015? For some of the innovations cited here, the question is already

being asked. For example, the Bangladesh secondary schoolscholarship programme for girls, while it appears to have helpedsecondary enrolment among girls, may also turn into an entitlement

programme that the government cannot afford. How will these



21/25

programmes be designed so that their effects will continue to be felt inthe long-run, without undermining the public budget?

While these questions may seem overwhelming, we cannot neglectthem. The world is in a period of unusual ambition and commitmenttoward poverty reduction. The ambition is reflected in the MDGs; thecommitment in the recent increase in foreign aid based on improvedpolicies and institutions in developing countries. For the ambition andcommitment to be fulfilled, we have to make exceptional progress,especially in the human-development areas, because we are evenfurther off on those targets than in some other areas. With additionalresources and reasonably good, growth-promoting policies in place,

we need to focus on the remaining major challenge making serviceswork for poor people.

References

Angrist, J.D., E. Bettinger, E. Bloom, E. King and M. Kremer (2002)Vouchers for Private Schooling in Colombia: Evidence from aRandomized Natural Experiment,American Economic Review, 92 (5):

153558.Appleton, S. (2001) What Can Expect from Universal Primary

Education? in R. Reinikka and P. Colliers (eds),Ugandas Recovery:The Role of Farms, Firms, and Government, Washington DC: WorldBank.

Arrow, K.J. (1963) Uncertainty and the Welfare Economics of MedicalCare,American Economic Review, 53 (3): 94173.

Barnum, H.N., J. Kutzin and H. Saxenian (1995) Incentives andProvider Payment Methods,International Journal of Health Planning

and Management, 10: 2345.Bennet, S., D. Gzirishvili and R. Kanfer (2000) An In-depth Analysis of

the Determinants and Consequences of Worker Motivation in TwoHospitals in Tbilisi, Georgia, Major Applied Research 5, WorkingPaper No. 9, Partnerships for Health Reform, pp. 162, available athttp://www.phrproject.com/publicat/arp/mar5wp9ab.htm

Bernheim, B.D. and M.D. Whinston (1986) Common Agency,Econometrica, 54 (4): 92342.

Bhushan, I., S. Keller and B. Schwartz (2002) Achieving the TwinObjectives of Efficiency and Equity: Contracting for Health Servicesin Cambodia, Policy Brief Series No. 6, pp. 110, Asian Develop-ment Bank.



22/25

Burgess, S. and P. Metcalfe (1999) Incentives in Organisations: A

Selective Overview of the Literature with Application to the Public

Sector, CMPO Working Paper No. 00/16, Bristol: LeverhulmeCenter for Market and Public Organisation, University of Bristol,

available at http://www.bris.ac.uk/cmpo/

Burnett, N. and R.B. Kattan (2002) User Fees in Primary Education,

processed.7

Castro-Leal, J., L. Dayton, L. Demery and K. Mehra (1999) Public

Spending in Africa: Do the Poor Benefit? World Bank Research

Observer, 14 (1): 4972.

Deci, E.L. (1975)Intrinsic Motivation, New York: Plenum Press.Dehn, J., R. Reinikka and J. Svensson (2003) Survey Tools for Assessing

Performance in Service Delivery, forthcoming in F. Bourguignon

and L. Pereira da Silva (eds),Evaluating the Poverty and Distributional

Impact of Economic Policies, Ch. 9, pp. 191212, Oxford: Oxford

University Press and the World Bank.

Devarajan, S. (2001) Growth Is Not Enough, Washington DC: World

Bank, processed.

Devarajan, S., M.J. Miller and E.L. Swanson (2002) Goals forDevelopment: History, Prospects, and Costs, Policy ResearchWorking Paper No. 2819, Washington DC: World Bank, available at

http://econ.worldbank.org/view.php?type=5&id=13269.

Dixit, A. (1996)The Making of Economic Policy: A Transaction-cost Politics

Perspective, Cambridge MA: MIT Press.

(1997) Power of Incentives in Public Versus Private Organizations,

American Economic Review Papers and Proceedings, 87(2): 37882.

(2000) Incentives and Organizations in the Public Sector: An

Interpretative Review, revised version of paper presented at the

National Academy of Sciences conference on Devising Incentives to

Promote Human Capital, Irvine CA, 1718 December, Princeton

University, processed.

Filmer, D. (2001) A Note on Public Spending and Health and Education

Outcomes, Development Research Group, Washington DC: World

Bank, processed.

Filmer, D., J.S. Hammer and L.H. Pritchett (2000) Weak Links in the

7 The word processed describes informally reproduced works that may not becommonly available through library systems.



23/25

Chain: A Diagnosis of Health Policy in Poor Countries,World BankResearch Observer, 15 (2): 199224.

Franco, L.M., S. Bennett and R. Kanfer (2002) Health Sector Reform andPublic Sector Health Worker Motivation: A Conceptual Framework,Social Science and Medicine, 54: 125566.

Gelbach, J.B. and L. Pritchett (2000) Indicator Targeting in a PoliticalEconomy: Leakier can be Better, Journal of Policy Reform, 4 (2):11345.

Glaeser, E.L. and A. Shleifer (2001) Not-for-profit Entrepreneurs,Journal of Public Economics, 81: 99115.

Glewwe, P., M. Kremer, S. Moulin and E. Zitzewitz (2004) Retrospectivevs. Prospective Analyses of School Inputs: The Case of Flip Charts inKenya,Journal of Development Economics, 74: 25168.

Goddard, M., R. Mannion and P. Smith (2000) Enhancing Performancein Health Care: Theoretical Perspectives on Agency and the Role ofInformation,Health Economics, 9: 95107.

Gosden, T., L. Pederson and D. Torgerson (1999) How Should We PayDoctors? A Systematic Review of Salary Payments and their Effecton Doctor Behaviour, Quarterly Journal of Medicine, 92: 4755.

Hammer, J., I. Nabi and J. Cercone (1995) Distributional Effects ofSocial Sector Expenditures in Malaysia, 1974 to 1989, in D. van deWalle and K. Nead (eds),Public Spending and the Poor, Baltimore MD:

Johns Hopkins University Press.Hirschman, A.O. (1970)Exit, Voice, and Loyalty: Response to Decline in

Firms, Organizations, and States. Cambridge MA: Harvard UniversityPress[OK?].

Holmstrm, B.R. and P.R. Milgrom (1988) Common Agency and

Exclusive Dealing, New Haven CT: Yale School of Management,processed.

Jimenez, E. (1987) Pricing Policy in the Social Sector: Cost Recovery forEducation and Health in Developing Countries, Baltimore MD: JohnsHopkins University Press.

Jimenez, E. and Y. Sawada (1999) Do Community-managed SchoolsWork? An Evaluation of El Salvadors EDUCO Program, World BankEconomic Review, 13 (3): 41542.

Killick, T. (1998)Aid and the Political Economy of Policy Change, London:ODI/Routledge.

Kremer, M., S. Moulin and R. Namunyu (2002) The Political Economyof School Finance in Kenya, Harvard University, processed.



24/25

Kreps, D.M. (1997) Intrinsic Motivation and Extrinsic Incentives,American Economic Review, 87 (2): 35964.

LaLonde, R.J. (1986) Evaluating the Econometric Evaluations ofTraining Programs with Experimental Data, American EconomicReview, 76 (4): 60420.

Libhaber, M. (2002a) PublicPrivate Partnership: InstitutionalTransformation Experience in the Water Supply Utility ofCartagena, prepared for Water Sector Board, December 2001, andpresented at World Bank Empowerment Retreat, 78 May,processed.

(2002b) Concession of the Water and Sewerage Services: AForecasted Failure in Cochabamba, Bolivia, World BankEmpowerment Retreat Powerpoint Presentation, 78 May,processed.

Litvack, J. and C. Bodart (1993) User Fees Plus Quality EqualsImproved Access to Health Care,Social Science and Medicine, 37 (3):36983.

Martimort, D. (1992) Multi-principaux avec Anti-selection,Annales dEconomie et de Statistique, 28: 137.

Martinez, J. and T. Martineau (1998) Rethinking Human Resources: AnAgenda for the Millennium,Health Policy Plan, 13 (4): 34558.

Miguel, E. and M. Kremer (2001) Worms: Education and HealthExternalities in Kenya, NBER Working Paper No. 8481, CambridgeMA: National Bureau of Economic Research.

Mills, A. (1997) Improving the Efficiency of Public Sector HealthServices in Developing Countries: Bureaucratic versus MarketApproaches, in C. Colclough (ed.),Marketizing Education and Health

in Developing Countries: Miracle or Mirage?Part 5, pp. 24574, Oxford:Oxford University Press.

Mowday, R. and R.M. Steers (1979) The Measurement ofOrganizational Commitment, Journal of Vocational Behavior, 14:22447.

Pauly, M.V. (1987) Nonprofit Firms in Medical Markets, AmericanEconomic Review, 77 (2): 25762.

Pritchett, L. and L.H. Summers (1996) Wealthier is Healthier,Journal ofHuman Resources, 31 (4): 84168.

Pritchett, L. and M. Woolcock (2002) Solutions whentheSolution is theProblem: Arraying the Disarray in Development, Kennedy Schoolof Government, Harvard University, processed.



25/25

Reddy, S. and J. Vandermoortele (1996) User Financing of Basic SocialServices, New York: United Nations Childrens Fund.

Reinikka, R. and J. Svensson (2001) Explaining Leakage in PublicFunds, Policy Research Paper No. 2709, Washington DC: WorldBank, available at http://econ.worldbank.org/search.php?term=2709&sp=&st=&scope=0

(2003) Working for God? Evaluating Service Delivery of ReligiousNot-for-Profit Health Care Providers in Uganda, Policy ResearchPaper, Washington DC: World Bank, forthcoming.

Republic of Uganda (2000) Tracking the Flow of and Accountability forUPE Funds, Ministry of Education and Sports report, Kampala:

International Development Consultants Ltd. (2001) Study to Track Use of and Accountability of UPE Capitation

Grants, Ministry of Education and Sports. Report by InternationalDevelopment Consultants Ltd (IDC), Kampala, Uganda, October.

Sahn, D. and D. Stifel (2002) Progress Towards the MillenniumDevelopment Goals in Africa,World Development, 31 (1): 2352.

Schultz, T.P. (1987) School Expenditures and Enrolments, 19601980:The Effects of Income, Prices and Population Growth, in D. Gale

Johnson and R. Lee (eds), Population Growth and EconomicDevelopment,Madison WI: University of Wisconsin Press.

Segall, M. (2000) Human Development Challenges for Health CareReform, in P. Ferrinho and W. Van Lerberghe (eds),Providing HealthCare Under Adverse Conditions: Health Personnel Performance andIndividual Coping Strategies, Antwerp: ITG Press.

Vaillancourt, D. (2002) Burkina Faso Health and Nutrition Project:Project Restructuring, Washington DC: World Bank, processed.

World Bank (1997)The State in a Changing World, World Development

Report, Washington DC: World Bank. (2001) India: Raising the Sights: Better Health Systems for IndiasPoor, Health, Nutrition, Population Sector Unit, South Asia RegionReport No. 22304, Washington DC: World Bank.

(2002) World Development Indicators 2002, Washington DC: WorldBank.


Documents

Devarajan & Reinikka (2004) Making Service Work for Poor People