Teachers in Tanzania are absent 23 percent of the time; doctors in Senegal spend an average of 39 minutes a day seeing patients; in Chad, 99 percent of non-wage public spending in health disappears before reaching the clinics.
These and other service delivery failures have been widely documented since the 2004 World Development Report, Making Services Work for Poor People .
But why do these failures persist? Because they represent a political equilibrium where politicians and service providers (teachers, doctors, bureaucrats) benefit from the status quo and will therefore resist attempts at improving services. For instance, teachers are often the campaign managers for local politicians. They work to get the politician elected, in return for which they get a job from which they can be absent. Powerful medical unions ensure that their members can work in the private sector and neglect their salaried government jobs. The losers are the poor, whose children don't learn to read and write, or get sick and die because the public clinic is empty.
The plenary session of the AERC's biannual workshop , on Institutions and Service Delivery, corroborated this political equilibrium and suggested some ways of disrupting it. Tessa Bold and Jakob Svensson's paper  described a randomized control trial where a program of contract teachers in Kenya was implemented by both NGOs and government. The improvement in learning outcomes was only in schools where the program was introduced by NGOs. When the government ran the program, there was no discernible improvement--probably linked to the strong protest from the (permanent) teachers unions.
In another experiment, they provided teachers with more resources, but subjected some to monitoring by the local community. Only when the additional resources were accompanied by community monitoring was there a significant improvement in student learning.
These and other examples illustrate a fundamental point. We can think of several "gimmicks" to make service providers accountable, but unless the underlying politics is conducive to those innovations, they are unlikely to work.
A bonus to doctors for performance  seems to improve health outcomes in Rwanda, but a program to introduce time-stamp machines to monitor health workers' presence in another country failed: the night before the program was to start all the machines were vandalized.
How can we disrupt this equilibrium to improve services for poor people? One possibility is to provide them with information. We saw that community monitoring (poor people obtaining the information directly) had an effect on teacher performance. But more broadly, poor people do vote. If, with information, they vote along service delivery lines--rather than for politicians from their own ethnic or religious group, or who promise them private goods such as a job building roads--then it would be more difficult for politicians to ignore service delivery failures.
You might ask why we need to inform poor people. Don't they already know that the teacher or doctor is absent? Life is so difficult when you're poor that you may not associate absentee doctors or ill-equipped classrooms with public policy failures. As part of a village immersion program, I spent time with a poor woman in Gujarat, India. One day, her kid was sick. She took him to a private doctor whose qualifications were suspect. When I asked her why she didn't take him to the free, public clinic, she said (rolling her eyes), "because the doctor's not there." I then asked her why she thought the doctor wasn't in. "Because the rains didn't come," she replied.
If we want to inform poor people so they can vote for programs in their own interest, how do we do so? We know very little about what types of information campaigns work (link to my blog post about critiques of information campaigns). This is an area that is ripe for randomized control trials, such as this one in Benin .
Information may not be the only solution. The underlying problem is that politicians are behaving "clientelistically"--more interested in handing out private goods to their "clients" rather than public goods that benefit society, especially the poor--and getting away with it (they get elected). And if most politicians behave this way, it's in every politician's interest to follow suit. How can we move from clientelism being the norm to one where it is the exception?