#880 2004 · Citizen Report Card project, rural Uganda (Björkman & Svensson field experiment) · Public primary health care / community accountability

25 Ugandan clinics were graded by their own villages — child deaths fell by a third

问题

Rural clinics were failing but official supervision was broken — and each family only knew its own bad visit

背景

In the early 2000s, across sub-Saharan Africa — Uganda squarely included — roughly one child in five died before age five, much of it from diseases a functioning primary clinic could cheaply prevent or treat — if the clinic was open and the staff were in it. Unannounced surveys across seven developing countries of the period found health workers absent on average 35 percent of the time; drugs leaked, equipment sat unused, and care was casual when it happened at all. The standard remedy — more rules, more funding, more inspectors — ran through the same top-down hierarchy that had already failed: the institutions supposed to monitor the clinics were themselves weak and badly incentivised, so the official line of accountability from clinic to state was, in the authors' words, effectively dead.

Development orthodoxy was already turning toward the alternative — 'beneficiary control', letting the users themselves hold providers to account — but almost none of it had been tested rigorously, and the obstacles were obvious. A family knows only its own outcome, an anecdote a clinic can shrug off as partial or invented; monitoring is everyone's interest and nobody's job; and a village meeting is easy for local elites to capture. Nobody had yet shown that ordinary citizens, given the right information, could actually make a clinic work better.

换别人会怎么做

The standard responses to a failing clinic all route through the same broken hierarchy that let it fail: more money, more drugs, more training, more inspectors from the district. Every one of them assumes someone higher up can and will supervise the frontline — the exact capacity whose absence caused the problem — and none of them reaches the only party present at every failure: the patient.

他们看到了什么

Each family knew only its own bad visit — an anecdote a clinic could dismiss. 5,000 visits pooled and compared with peer clinics is a fact nobody can argue with. Given the aggregate, the customers become the inspectors.

那一手

In 2004 the researchers launched a randomized pilot in 50 rural dispensaries across nine districts. They surveyed roughly 5,000 households and pulled each clinic's own registers, then assembled the findings into citizen report cards — translated into six local languages, with artists' posters for non-readers — showing each clinic's utilisation, quality and informal charges next to its district peers and the government's own service standards. Half the clinics, chosen at random, then got two rounds of facilitated village meetings in which residents interpreted their own report card, confronted the staff with it at an 'interface meeting', and drafted their own monitoring plan. The other half stayed untouched as controls.

为什么管用

The report card did two things a complaint never can. It turned scattered anecdotes into one comparative fact — performance measured from the clinic's own registers and thousands of household interviews, laid against district peers and government standards — which the provider could no longer brush aside as partial, anecdotal or invented. And the meetings were load-bearing: piloting had shown that simply reporting facts changes nothing, so residents interpreted the data themselves, in their own languages, and wrote their own monitoring plan, which converted information into commitment and spread the watching past the local elite. Staff then behaved as if watched, because now they were: within a year, with no new money and no change in official supervision — nothing but effort, unlocked by visibility.

值了多少

One year later: visits up 16%, absenteeism down 10 points, waits cut by 14 minutes, infant weights up — and child deaths down by a third.

什么时候会失灵

The authors' own 2010 follow-up found the effect is uneven where the community itself is fractured: income inequality and, above all, ethnic fractionalization undercut the collective action the whole mechanism rests on — divided villages could not organise the monitoring that united ones did. It also extracts effort, not resources: monitoring cannot supply drugs, equipment or skill a clinic does not have. The design had to be defended against elite capture (meeting participants were deliberately drawn from across the village's groups), it depended on trained local facilitators who did not exist everywhere, and the evidence covers one year — whether the watching stays sharp after the meetings stop is not shown.

后来呢

Community scorecards existed before this — the paper itself cites Malawi's — but the Uganda trial supplied what none of them had: randomized evidence that beneficiary monitoring improves health outcomes. 'Power to the People', published in the Quarterly Journal of Economics in 2009, became one of the most cited field experiments in development economics, and report cards and community scorecards moved from advocacy tool to funded policy instrument across health and education systems in low-income countries. The authors' 2010 follow-up went looking for the boundary conditions and found them in the communities themselves (see whenItFails).

资料来源

[1]Power to the People: Evidence from a Randomized Field Experiment of a Community-Based Monitoring Project in Uganda (Policy Research Working Paper 4268)World Bank, 2007documents1.worldbank.org
[2]Power to the People: Evidence from a Randomized Field Experiment on Community-Based Monitoring in UgandaAbdul Latif Jameel Poverty Action Lab (J-PAL), MIT, 2009poverty-action.org
[3]When Is Community-Based Monitoring Effective? Evidence from a Randomized Experiment in Primary Health in UgandaJournal of the European Economic Association (authors' copy, Bocconi University), 2010didattica.unibocconi.it