Author: Shanta Devarajan, Director of the World Development Report 2004, and Chief Economist in the Human Development Network
The messages of the WDR 2004 on "Making Services Work for Poor People" are both positive and negative. The challenge is that too often key services such as health, education, water, sanitation, electricity, are failing poor people in access, in quantity and in affordability.
But we also have strong examples of where services do work for poor people. The common feature we find among these examples is that they all have to do with empowering poor people, and empowering them in three ways. One is to give them the ability to monitor. Closely linked to this is to enable people to encourage and, where necessary, to discipline service providers -- by rewarding or penalizing them -- to serve poor people. The third is to raise their voice in policymaking, for example through the ballot box.
Allow me to corroborate some of these messages.
First, what do we mean by services 'failing' poor people? One simple observation is that public spending and these services are fundamentally a public responsibility. Unfortunately, public spending in, for example, health and education is typically benefiting the non-poor or the rich rather than the poor.
In a poor country like Guinea, 48 percent of health spending benefits the richest fifth of the population, and less than eight percent benefits the poorest fifth of the population. Primary spending and primary health and primary education spending is slightly more pro-poor. Many countries reallocate spending towards primary schools and primary clinics as a way of making their spending more pro-poor. But this is where we face the second problem.
And the second problem is that money intended for primary education or primary health often fails to reach the front-line service provider. In a landmark study in Uganda, WDR04 co-director Ritva Reinikka showed that for every dollar of money that was due primary schools in Uganda, only 13 cents actually arrived in the schools. The distribution around this average means that there are poorer schools getting much less.
But even if a country can raise this, as the Ugandans have done by going from 13 all the way up to 80 cents on the dollar, there is a third problem.
And the third problem is that the quality of services that poor people receive is often very low. One symptom of this is the high degree of absenteeism observed in schools and in clinics. In primary health clinics in Bangladesh, the absenteeism rate for doctors was 74 percent, and there are examples of abusive treatment that poor people receive in some of these areas, and the disappearance of theft of pharmaceuticals, such as a 70 percent disappearance in Guinea, is another symptom of the poor quality of the resources that poor people receive.
Our view in the WDR 2004 is that these problems, the poor allocation, the fact that money fails to reach the front-line service provider , and the low quality are all symptoms of a fundamental failure of accountability. What do we mean by that?
Let us think in terms of a simple competitive market transaction. When I buy a sandwich in the market there is a direct relationship of accountability between me, the purchaser and the sandwich maker as the seller. I give resources directly to the sandwich maker. I can observe whether or not I got a sandwich. And, most importantly, if I do not like the sandwich, the sandwich seller knows that s/he lost a customer. S/he will not get repeat business from me.
Now turning to health, education, water and sanitation, and electricity, societies have decided for good reason that these services should not be provided through competitive market transactions or could not be provided through competitive market transactions. Rather they are provided by government because they are a fundamental public responsibility. They are provided through what we would call the indirect route of accountability. That is, by poor people or citizens influencing policymakers or politicians, and then the politicians influencing the providers.
There are at least two legs in this long route of accountability, which means that there are at least two places it can break down. Allow me to illustrate the first one. The first is the relationship between poor people as citizens and politicians. This is politics, the fundamental process of political decisionmaking.
What we find in this report is how profoundly political public services are; that in many societies, including in electoral democracies, politicians often use public services as the currency of political patronage.
To illustrate it, in the early 1990s, Mexico had a program called PRONASOL, a large poverty alleviation program that consumed about 1.2 percent of GDP per year to provide water, sanitation, education construction, and electricity to poor communities in Mexico. But even though it used 1.2 percent of GDP per year, over the 6-year period, the results showed that it reduced poverty only by 3 percent.
There are some studies that showed that had it been perfectly targeted, it would have reduced poverty by 64 percent. In fact, had they just given it out equally to the entire Mexican population, it would have reduced poverty by 13 percent. How can a program like this happen?
The per-capita PRONASOL expenditures to different municipalities, depending on which party the municipality voted for in the 1988 election, show that the municipality that voted for the party in power got a significantly higher share of the PRONASOL expenditures, as opposed to the other municipalities. It is not surprising that the effect of this on poverty was so minimal and we can conclude that this was basically not a poverty alleviation program but a political patronage program.
This does not mean that all electoral democracies have this. There are democracies and other types of regimes, including one-party states, that have managed to achieve pro-poor public services. Countries like Sri Lanka, Costa Rica, Cuba, China and others have managed to achieve a level of outcomes in health and education that are extremely favorable.
But even if the politician or the policymakers care about getting services to poor people, there is another problem, which is that the policymaker is not the person teaching in the classroom. The policymaker is not the person working in the clinic, so he or she has to get the provider to actually deliver the service. This is often challenging as the policymaker may not be in a position to monitor and may not be giving the provider the incentives to serve the poor.
This, however, does not mean that the problem is insurmountable. To share a good example, Cambodia was coming off a Civil War, and they needed to get rural health care to the clinics very quickly--rural health care to rural areas very quickly--and they introduced a system of contracts .
In fact, they had two types of contracts, called contracting out and contracting in , where NGOs would provide the services, and they would be compensated in return for independently observed quality of health and service delivery in those districts.
As Cambodians genuinely wanted to learn from this experience they introduced a control group, which was the government's own health clinic. The contractual arrangements were randomly assigned across 12 districts. The results are rather surprising or rather striking. The biggest increase or improvement, in just on this one indicator, was in the contracted-out districts, the second biggest in the contracted in, and almost no change in the government's, that is, the control group's. This program has been scaled up, not just in Cambodia, but around the world, thanks to the solid and rigorous impact evaluation that was introduced.
This does not mean that all services can be contracted out. Indeed, most services are very difficult to write contracts for because it is very difficult to monitor whether or not the service is being provided. You can monitor whether a child has been immunized, but you cannot truly monitor what goes on in a doctor's clinic or what goes on in a classroom, and this is what leads the some of the problems.
But I think here is where we sometimes forget the fact that there is somebody else who can monitor the service provider, and that is the client. Even if the policymaker in the capital city does not know whether the doctor is showing up for work, the patient knows, and the patient's family knows.
And this is one way in which we can make a lot of progress. Evidence shows that if we allow, and enable and empower clients to be active participants in the service provision process that guarantees progress. Documented cases share lessons and examples how having poor people participate actively in the service provision process improves service delivery. It is essential to reinforce the direct route of accountability between clients and providers when there are problems in the long route.
What does this mean in practice? It should be clear from this reasoning that there is no one size that fits all. It is not the case that public sector provision fulfills all of the requirements or will work in all circumstances. Nor is it the case that private-sector participation will work in all circumstances. In fact, the way we say it, somewhat tongue-in-cheek, "While no size fits all, maybe eight sizes do," and let me tell you why we say "eight sizes." Because we think that the important characteristics that give rise to a favorable service delivery arrangement are based on three questions:
- Is the service easy to monitor or difficult to monitor?
- Is the relevant client base homogeneous or heterogeneous?
- And, perhaps the most difficult, is the politics in the country pro poor or is it subject to clientelism?
Allow me to share three examples where the answers to these three questions actually determine service delivery arrangements.
1. Take the easiest example. The easiest one would be a service that is easy to monitor, where the population is homogeneous and the politics in the country is pro-poor. Think about a Scandinavian country and its service provision arrangement. It does not really matter what kind of arrangement you have. You can have central government provision of the service. You could have--because it is easy to monitor--central government provision with contracting.
2. In the second example there is one slight deviation from the previous one. That is, we are still in this pro-poor, homogeneous country but with service provision that is difficult to monitor, such as student learning. That is the case which requires central government provision.
3. Finally, let us take the polar extreme - a service that is hard to monitor, a population that is extremely heterogeneous, and a country in which the politics are not pro-poor, and decisions on services are subject to clientelism. This is the most difficult, but this is where the service delivery arrangement needs to emphasize the power of poor people. This is where demand side subsidies are extremely important, which essentially give power to the people to be able to choose. Because leaving it to the political system to deliver it on its own may not necessarily lead to poor people benefiting from the service. These are the cases where demand side subsidies, and indeed, co-payments and other forms of participation by households have a high premium. It does not mean it always is the case, but in this one case, in this extreme case we may need to think about it.
The core message that one needs to take away from WDR 2004 is that services can work when poor people stand at the center of service provision - when they can avoid poor providers, while rewarding good providers with their clientele, and when their voices are heard by politicians - that is, when service providers have incentives to serve the poor.