Community co-financing of local public goods: Evidence from an experiment in Ghana

Blog State

Decentralisation is again growing in popularity among developing countries. There is a strong rationale for delegating greater responsibility for public decision-making and service delivery to local governments. Research suggests that giving local communities more say in policy decisions can enhance accountability, empower citizens and improve the quality of service delivery.

But as the responsibilities delegated to local governments increase, so must the funding. And revenue mobilisation is a particular challenge in countries where few people work in the formal economy and contribute to the income tax base. Finding supplementary sources of funding for community public goods is a pressing policy issue.

One option is to supplement government funding for local public goods with contributions from communities themselves. There is a long history of fundraising in religious organisations across Africa, for local and faraway causes. And given the significant amounts of money being sent home by migrants, a community’s diaspora could also contribute. But in trying to collect donations for public projects, we run into the classic ‘free-rider problem’. Even though citizens might all individually value a new road, public toilet or water pump, each person will contribute less than their fair share to its cost, expecting their neighbours to make up the difference.

Just as a wealth of experiments and impact evaluations have shown that the poor can and will save if given the right tools and incentives, the free-rider problem can be alleviated by changing the incentives to contribute. The fundraising industry has developed a range of techniques to encourage individuals to donate money. For example, charities have found that people give more if their donations are matched by a third party. Tax deductions for charitable contributions are a common way of doing this. Similarly, people may be concerned that their donation will be wasted if no-one else contributes and the project is underfunded. In this situation, offering a ‘money-back guarantee’ that the project will be fully funded has been shown to increase donations in laboratory experiments.

If these techniques can be used to encourage charitable donations, might they also work in raising funds for community projects? With the support of the IGC, we designed a field experiment to test this hypothesis in four rural communities in southern Ghana. Who would donate to a fundraiser for local public goods, and how much would they give? Would the added incentive of a matching offer or money-back guarantee encourage them to give more?

To kick off the experiment, our survey team called the residents of each community to a public meeting. We explained that we would be raising funds from the residents to pay for a small community construction project. Supposing the community could raise around 1,000 Ghana cedis, what would they most like to spend that money on?1 After some lively discussions and voting, three of the communities decided to raise funds for the construction or renovation of school buildings. The other community elected to build a maternity ward for an existing health clinic.

We conducted five rounds of the fundraising experiment, once every two months during 2009. In each of the four communities, we invited about 150 adults (randomly selected for a concurrent survey) to participate. In a closed room, we met with each participant in turn, and gave them a small amount of money as compensation for their time. They then drew a token to determine whether we would match their donation, and the proportion (between a half and 2) of the match. For example, if they drew a match ratio of 1, we would match their donation to the project one-to-one. If they drew a ratio of zero, there would be no match. Randomly varying the proportion of each participant’s match gave us a control group of individuals who did not receive a match offer, and also allowed us to test whether the size of the matching grant had an effect on donations. We explained the matching concept carefully to the participant, and asked them how much they wanted to donate. Donations were entirely voluntary and their decision was kept private. If the participant chose to donate, the contribution and the match were then put in the fundraising box.

In the last two rounds of fundraising, we added a money-back guarantee. We announced a funding target for the project, and explained how much had already been raised. If the target was not reached, we explained, every participant would have their donations for the last two rounds refunded. In the end, the fundraising targets were met in all four communities and the four projects were completed. Some community members contributed additional materials and labour to the construction.

The results from the study were very interesting. First, we found that almost all of the participants were willing to donate, and some contributed money in excess of what we had paid them for attendance. The matching grant offer raised donations by over 10% on average, and we found that a higher match ratio led to more generous donations. The money-back guarantee had an even greater effect on donations, almost doubling the amount contributed.

The next step would be to pilot these ideas on a larger scale. Local governments could employ these techniques to supplement funding for small infrastructure projects. The community might be required to raise a certain fraction – say 10% – of the funds for the project, the remainder being a ‘match’ from the government. The money-back guarantee could be incorporated by setting a condition that pledges would be refunded (and the project abandoned) if the target is not met. This would help avoid the government funding projects for which there is insufficient community interest. With contributions from remittances and the participation of religious organisations, more ambitious fundraising goals could be set.

Whether this scheme can work in practice will depend on the quality of implementation and the setting. It may work for some communities and projects, and not others. But our experiment in southern Ghana suggests that with the right incentives and leadership, people are prepared to contribute to the cost of small public goods for their communities.


Further reading

Walker, T (2011), ""Community Involvement in Public Goods Provision: Evidence From a Field Experiment in Ghana"", IGC Working Paper No. 11/0069.


[1] 1,000 Ghana cedis was roughly equal to US$700 at the time of the study.