n.V1-7.16 | COMMUNITY SCORECARDS
01| Every other solution in this book part adds something — money, training, an asset, a worker, a road. This one adds nothing. It tells people how their existing services are performing and lets them confront the providers about it.
02| Björkman and Svensson ran a randomised field experiment on community-based monitoring of primary health care in Uganda. Local NGOs convened village meetings and supplied report cards on staff performance, strengthening the community's capacity to hold providers to account. The results were large: increased utilisation, improved health outcomes, reduced child mortality, increased child weight. A follow-up found the improvements in delivery and outcomes persisted into the longer run despite minimal continued support.
03| Nothing was delivered. The clinics, the staff and the drugs were the same clinics, staff and drugs. What changed was that the people using them could see how their facility compared and were placed in a room with the staff to discuss it. Provider behaviour changed, and the change stuck after the convenor left — which is close to unique in this book part, where the standard finding is that everything reverts when the programme ends.
04| A caveat has to be carried with this one. The 2009 result has been the subject of a published replication study commissioned by 3ie, and anyone relying on the headline figure should read the replication rather than the abstract. This chapter treats the finding as strong and contested, not as settled.
05| Even taken at full strength, the result raises a question the enthusiasm around social accountability tends to skip. It worked in a setting where the providers were present, salaried and identifiable, and where the failure was effort rather than absence of resource. A scorecard tells you your nurse is often absent; it does not conjure a nurse where the post is unfunded. Accountability mechanisms work on the margin of effort, and are useless on the margin of supply. Which margin binds is a local fact, and applying the instrument to the wrong one produces a well-informed community and no change.
06| There is also a structural problem with the intermediary in this model that does not appear in the previous three chapters. The person or body convening a scorecard meeting is asking a community to confront a provider. The same person or body, in most designs, also depends on that provider for cooperation, referrals and access. Advocacy against and dependence upon are hard to hold in one role, and the literature does not say how the successful cases managed it — or whether the NGO convenors succeeded partly because they were outsiders who could afford the friction and leave.
07| The mechanism connects to a finding from a different corner of this book part. In the Indonesian block grant programme, the experiment on corruption found that raising the probability of audit reduced losses, and that reading the findings aloud in a village meeting immediately reduced them considerably more. Same instrument, different domain, same answer: immediate local publicity in front of the people affected outperforms formal enforcement, and costs almost nothing.
08| That is the most transferable finding in this book part and one of the few where the cost is trivial relative to the effect. It is also, notably, not a service. It is information delivered publicly and fast, which is why nobody has built an industry on it.
