Neal Bangerter and Brigham Daniels, BYU Law School

I worked with Professor Brigham Daniels on his ongoing research with Mark Buntaine at UC Santa Barbara. This is a summary of the results from the 2018 phase of the project.

Introduction

Transparency in government is widely considered to have a positive impact on governance, by equipping citizens with the ability to monitor governmental processes. Using a randomized control trial (RCT), we experimentally test the effect of community monitoring on the success of public programs by passing information from local citizens to the proper government officials tasked with oversight. This was done with the intention of promoting accountability for results and limiting corruption.

Our experiment takes place in the area surrounding the Bwindi National Park in Uganda. The villages surrounding the park have the opportunity to participate in a revenue-sharing program sponsored by the park; in this program, a percentage of park revenue from tourism is used to fund development projects proposed by committees in each village. The aim of this program is to spread the benefits of tourism and conservation with local villages, many of which suffer economic hardship (in the form of diminished access to natural resources or the destruction of crops by park wildlife, for example) due to their proximity to the park. This revenue-sharing program has been plagued by problems with corruption and mismanagement of funds; audits in previous years suggest that nearly 80% of revenue-sharing funds were not used for their intended purpose. This low rate of success makes for an ideal venue to test the impact of transparency and community monitoring.

Methodology

In order to test the impact of community monitoring and transparency on the revenue-sharing program, we designed an RCT wherein half of the villages were assigned to a control group and half were assigned to receive a treatment. Dr. Daniels and his colleagues had previously collected the mobile phone numbers of many people living in each of the 98 villages immediately surrounding the Bwindi National Park that were eligible for the revenue-sharing program. Individuals in treatment villages received timely and pertinent information regarding the approval of village projects and the transfer of funds through various levels of government, while individuals in control villages received messages with irrelevant information promoting public health. We also collected data on the status of village projects from individuals in treatment villages. This feedback was aggregated into reports that were sent from the Uganda Wildlife Authority (which oversees the national parks in Uganda) to the corresponding government Districts (the second-highest level of government, after the federal government) in the hope that Districts could use this information to better oversee the implementation of projects in these villages.

After the conclusion of the 2018 revenue-sharing program, each village was audited on its projects. This information, in conjunction with feedback from individuals in both treatment and control villages, was used to measure the rate of successful implementation of projects in each village. We used a regression analysis to control for relevant covariates and estimate the treatment effect.

Results

Surprisingly, we found no evidence that our treatment had any impact on the successful implementation of revenue-sharing projects in treatment villages. Treatment villages had success rates slightly lower than (though statistically comparable to) control villages, indicating that providing both the applicable information to individuals in villages and feedback to government officials tasked with oversight had no impact on whether projects were successfully implemented or not.

To better understand these results, we conducted interviews aimed at examining both the transparency and community monitoring sides of our initiative. Interviews with the Chief Administrative Officer (CAO) in each of the three Districts involved with the revenue-sharing program revealed that our attempts at providing information from the communities to government officials failed to have much of an impact; while the CAOs had all received our biweekly reports, they didn’t really use that information in any impactful way. Interviews that examined the impact of transparency were more positive; in three cases where money had been diverted from projects in treatment villages, villagers had used the information we provided them to demand accountability, resulting in the firing of one local official and the transfer of another away from overseeing revenue-sharing funds.

Discussion

These findings point to some important details. Our interviews with the CAOs revealed that simply making information from community oversight available to government officials is insufficient to drive action. It seems likely that having information available is a prerequisite to effective oversight, but that information doesn’t have any intrinsic power to drive change. Efforts at improving community input will need to also design a channel whereby the feedback can be effectively used to drive change.

The qualitative results surrounding transparency were more positive. While transparency didn’t have an impact on the results we measured (successfully delivered projects), it proved an effective way of allowing communities to hold their leaders responsible. Over time, such a shift could potentially drive a higher proportion of successfully implemented projects, but not in a timeframe as short as we measured.

Conclusion

Our treatment and measurement strategies failed to result in better outcomes in the Bwindi National Park’s revenue-sharing program. Treatment and control villages had comparable levels of unsuccessfully implemented projects, in spite of the information and community monitoring channels we implemented in treatment villages. Interviews conducted after the experiment indicated our efforts at providing feedback to relevant government officials were largely unsuccessful, but that the additional transparency in treatment villages had been used to punish corruption, albeit in a way not measured in the experiment.