How Cambodia’s program to strengthen local governance built roads and saved lives

A new report on the long-term impacts of locally-led development projects provides insights for those who work in post-conflict settings.

June 24, 2019
Christian Baehr, Sarina Patterson, John Custer
Commune council members meet to review the final draft list of poor households for each village.

Commune council members meet to review the final draft list of poor households for each village. Photo by GIZ, licensed under fair use.

In the 1970s and 1980s, millions of Cambodians were forcibly displaced to labor camps by the Khmer Rouge, where many were worked or starved to death, or executed. Local governments effectively ceased to function during this time, and by the mid-1990s, public trust in local institutions was severely eroded.

In order to rebuild local governance and improve economic welfare from the bottom up in post-conflict Cambodia, a $750 million-dollar program focused on local infrastructure and governance—the Commune/Sangkat Fund—was scaled up and institutionalized nationwide beginning in 2002. The program was initially supported by the Swedish International Development Cooperation Agency (Sida), the World Bank, and other donors in partnership with the Government of Cambodia, but has since transitioned to full country ownership. The Commune/Sangkat Fund therefore represents a rare instance of a development program that began as a donor-supported initiative with limited geographic scope, but evolved into a fully government-owned, nationwide program—one that has had profound socioeconomic impacts, as a new AidData evaluation finds.    

Cambodia made major strides during the period following the program’s expansion, with the national poverty rate falling from nearly 50% in 2006 to less than 15% by 2014. Yet, until recently it was unclear how much the Commune/Sangkat Fund contributed to these gains, as no rigorous impact evaluation (with counterfactual evidence of what would have happened in the program’s absence) had ever been done. In partnership with Sweden’s Expert Group for Aid Studies, a team of researchers from AidData and Open Development Cambodia has published the first geospatial impact evaluation of the Commune/Sangkat Fund’s long-term effects, filling this evidence gap.

Read the full report, Building on a Foundation Stone: The Long-Term Impacts of a Local Infrastructure and Governance Project in Cambodia, on the EBA’s website.  

The ‘communes' and 'sangkats' of the program’s name refer to the 1,621 jurisdictions where local governance takes place in Cambodia. In 1996, the Government of Cambodia, Sida, and other donors launched a pilot program—  called Seila—in just four communes and sangkats. ‘Seila’ means foundation stone in Khmer, and it introduced a suite of reforms and capacity-building efforts  to restore public confidence in Cambodia’s local institutions. In 2002, the Seila program was reconfigured and renamed the Commune/Sangkat Fund as it entered a nationwide scale-up.  

By 2010, the Government of Cambodia had assumed full financial and operational responsibility for the Fund, and nearly three percent of the central government’s budget was contributed to the program each year. These resources, in turn, were then allocated to local councils using a transparent formula, which facilitated an expansion of the program to all 1,621 of Cambodia's communes and sangkats. Ultimately, the program resulted in democratically-elected local councils managing the design and implementation of more than 40,000 local infrastructure projects across 14,073 villages. The vast majority of these projects supported the construction and rehabilitation  of rural roads, water supply systems, irrigation dams, and canals.

Development, illuminated

To determine the net, attributable effect of the Commune/Sangkat Fund on local development outcomes, the evaluation team measured changes in nighttime light intensity within villages that occurred over time as a result of the implementation of individual Fund projects. Likewise, they used village-level administrative data on key indicators of wellbeing—such as infant mortality rates—to compare outcomes in each village after treatment to counterfactual outcomes obtained from the same village’s preceding socioeconomic outcomes. This quasi-experimental method provided a rigorous way to estimate the effects of the scale-up and institutionalization of the program, as opposed to other, independent factors that could have brought about local development gains.

The evaluation finds that the completion of locally-managed infrastructure projects supported by the Fund—in particular, road improvements in densely populated rural areas—significantly improved local economic development outcomes, as measured by changes in the intensity of nighttime lights. On average, a village that successfully implemented a single project saw a roughly 20% increase in nighttime lights attributable to that project. Since the median (or typical) village received four projects over the course of the program, it experienced an average gain of nearly 80% over its baseline level of nighttime lights. Even more encouraging is the fact that these impacts appear to grow over time, suggesting that infrastructure improvements increased not only the level but also the trajectory of economic development.

The evaluation team also found that projects supported by the Fund reduced infant mortality in the villages where they were implemented. On average, in a village that implemented a single project, infant mortality (measured by the number of deaths of infants less than one month old) fell by 3.2% from 2008 levels, indicating that socioeconomic gains were broadly shared by village residents. In a typical village that received four projects between 2008 and 2016, this average effect is roughly equivalent to a 12.8% reduction in infant mortality.

A focus on local governance

The Commune/Sangkat Fund was a major decentralization effort that provided public resources to local governments (many newly created) for the first time in a generation. Consequently, the evaluation team also examined  how the socio-economic gains from Fund-supported infrastructure projects were influenced by community engagement, government responsiveness to citizen priorities, and the capacity of local village councils.

Previous research demonstrates that villages subjected to high levels of political violence under the Khmer Rouge have lower levels of social cohesion and community engagement today, so the evaluation team used data from Yale University’s Cambodian Genocide Program Interactive Geographic Database on the locations of prisons, mass graves, and memorial sites in Cambodia to proxy for pre-existing levels of community engagement. Government responsiveness was measured with village-level data on the percentage of community priorities that received funding during the first year of the Fund’s expansion (2002-03). Finally, the capacity of local councils was measured using several variables, including the total number of councilors per commune; the proportion of newly elected local council members and councils chief with previous experience; and the percentage of councilors that were female.

Counterintuitively, the evaluation finds that areas with lower levels of community engagement at the beginning of the program actually experienced larger gains from project implementation, and areas whose local councils were initially more responsive to citizen priorities experienced no greater impacts than other areas. Local government capacity, by contrast, was a key determinant of project success: communes with larger councils and more experienced councilors experienced larger economic gains, especially in areas with multiple projects. These results suggest that local councils’ ability to direct resources on the basis of citizen preferences may have been limited, and that while the Fund’s support of local institutions created some capacity for administering local projects, it did not substantially alter their responsiveness to citizen interests.

These findings reinforce a critical feature of the Commune/Sangkat Fund: unlike a typical community-driven development (CDD) program, in which a community exercises more direct control of planning decisions and more robust monitoring of project activities, the Fund prioritized the rebuilding of local government capacity over activities that would have required high levels of social trust and civic engagement. Recent studies show that CDD programs often fail to improve —and can even undermine—social cohesion and the quality of local governance. In settings where local government is severely capacity-constrained, CDD programs often underinvest in local government capacity, thereby setting in motion a vicious cycle of government inaction and citizen disengagement. But the Commune/Sangkat Fund seems to have avoided this “premature load-bearing” trap by first rebuilding local administrators’ credibility and capacity to perform basic functions that serve their constituents.  

This study also highlights the importance of early, sustained investments in host country data systems. Since Sida and its partners invested in village-level data collection and management at an early stage, and because those investments were sustained over time by the Cambodian government, our evaluation team was able to precisely measure the timing of the program’s rollout over geographic space, as well as changes over time in village-level development conditions as the program scaled up. With this granular intervention and outcome data, geospatial impact evaluation methods could be employed to establish a credible counterfactual of what would have happened in the absence of the program and rigorously estimate the program’s long-term impacts. This allowed the program to be evaluated retrospectively and at a modest financial cost—nearly a decade after donors made their last financial contributions to the Fund.

To read the full evaluation, Building on a Foundation Stone: The Long-Term Impacts of a Local Infrastructure and Governance Program in Cambodia, please visit the EBA’s website here.

Christian Baehr is a Junior Data Analyst at AidData.

Sarina Patterson is AidData's Communications Manager. 

John Custer is AidData's Deputy Director of Communications and Data Analytics.