Click here for search results

Brief on KDP's Anti-Corruption Work

 Monitoring Corruption Study
 Additional Anti-Corruption Evidence

Monitoring Corruption Study

Ben Olken’s paper, “Monitoring Corruption: Evidence from a Field Experiment in Indonesia” (225kb pdf) was recently highlighted in a March 18th 2006 article of The Economist. The paper stems from field research commissioned by the World Bank Kecamatan Development Program (KDP) task team to examine innovative approaches to reducing corruption. The study conducted a randomized, controlled analysis of corruption in 600 KDP road-building projects in East and Central Java during the period of 2002 to 2004. The study found that the intervention of announcing an increased probability of a government audit and reporting the audit results directly to a public village forum was more effective at reducing corruption as compared to increasing grassroots participation in the monitoring process.

The study further concludes that grassroots monitoring may be more effective in certain contexts than in others. For projects where villagers inherently have good information about potential corruption and a strong personal stake in minimizing theft of funds, such as subsidies for food, health care, education, or potentially micro-credit projects, grassroots monitoring is likely to be effective. For projects where information gathering is difficult and benefits from reducing corruption are more diffuse, such as infrastructure projects, grassroots monitoring alone is unlikely to be sufficient.

One of the controversial aspects of the study was the measurement of absolute corruption levels.  The study found that the increased probability of an audit reduced the percentage of “missing” funds – i.e., funds that could not be accounted for by an engineering inspection of the completed infrastructure project – by about 8 percentage points, from approximately 29 percent in control villages to 21 percent in villages receiving the audits. While much attention has been cast upon this absolute value of 29 percent, this figure should be interpreted with caution for several reasons:

Back to top

First, Olken makes it clear in his paper that the project’s estimate of the change in the percent missing due to the experiments, i.e., the 8 percentage point reduction, is substantially more reliable than the estimated level of the percent missing, i.e., the 29 percent in control villages.  In other words, the study team did its best to estimate levels as accurately as possible, but the numbers are much more useful for comparisons between villages then about absolute levels per se. What mattered most for this study was the accuracy of the relative changes between the treatments, not their absolute levels.

Secondly, as the author also explains in his paper, the absolute levels of corruption are subject to debate, and are heavily dependent upon calibration assumptions. Some of the missing amounts could be attributed to corruption but most of it could just as rationally be attributed to other factors such as standard engineering loss rates and sheer inefficiency.  Factors such as weather, erosion, compacting of materials, transport losses, difficulties of field measurement, and incompetence, are all difficult-to-measure ingredients in the missing amounts or “loss ratios”. Or as The Economist article put it, “…Not all of the gap can be put down to venality though, some of the gravel, for example, was probably worn away…” KDP’s engineering manuals, which are taken from the U.S. Army’s and Indonesian Public Works guidelines, estimate a standard 30 percent loss factor due to factors such as materials washing away in heavy rains, compaction, big rocks dropping off carts, etc. Simply using the standard engineer’s loss ratios would have lowered Ben’s loss figures by 10-20 percent. A very clear example is sand, which makes up a big percentage of village roads: the Olken study sets loss rates at 0 percent, the engineers calculate normal sand loss due to settling and rain at 34 percent. In sum, there is disagreement with technical experts over what constitutes normal construction losses, measurement errors, and assumptions about worker capacity, e.g., the healthy, skilled workers who run a pilot vs. the poor and women who normally participate in KDP infrastructure works.

Third, this study did not look at the effects of community participation versus top-down audits, but at the marginal effects of additional participation. All study quadrants were already experienced KDP sites with high community participation, even the control sites. The study was only testing whether interventions could increase KDP participation rates still further. Even more critically, the audits themselves were part of the participatory approach. The regressions showed clearly that the two single biggest determinants in the efficacy of the audits were whether the plan to conduct audits was announced and disseminated to villagers beforehand, and whether audit results were read out publicly in village meetings. In short, the study tells us how to make participation more effective, but it doesn’t say very much about whether participation in and of itself makes for greater effectiveness and it says nothing at all about whether top-down audits without participation would produce any impact.

Back to top

Additional Anti-Corruption Evidence

The Olken study is not the only study which the Bank and Government have commissioned to examine ways of combating corruption in KDP.  Evidence from other evaluations and audit sources reveals much lower levels of presumed corruption:

World Bank Audits -  The World Bank commissions annual external audits of all its projects.  Over the last several years, audits by Price Waterhouse and Moore Rowlands reported no significant qualifications. Fund irregularities were less than one percent.

Government Audits – (Badan Pengawas Keuangan dan Pembangunan) or BPKP is the official government auditor of KDP.  Each year, BPKP audits KDP activities on a five percent sample.  Over the last several years, there has been very little found by way of leakage. In 2004, BPKP visited KDP sites in 23 provinces, 196 sub-districts, and 626 villages. BPKP found no major abnormalities and gave KDP a good score of 81 points out of a possible 100 points. The weaknesses reported related to handling complaints and slowness in complaints resolution. 

Internal Financial Audit Team – In addition to government and Bank auditors, KDP maintains an internal, seven-person unit for financial supervision and training.  Over the last four years, the unit has audited an average of 18 percent of KDP kecamatan throughout the country. The Unit discovered 124 cases of funds misuse amounting to Rps 1.2 billion (US$135,800) between 2002 and 2004.  During 2005, the Unit found 28 cases of misused funds amounting to Rps 514 million or US$56,400, again less than one percent of total grant funds. About half of this amount was due to one specific case in North Maluku. 

Back to top

Evaluations of Cost-Effectiveness – There have been several studies showing that KDP is much more cost effective than other government programs. An independent evaluation carried out in 2004 of 113 villages in four provinces reconstructed unit costs for construction materials and labor, and found that village infrastructure built through KDP methods costs significantly less – on average 56 percent less – than equivalent works built through the Ministry of Public Works or through local government contracts. In a separate evaluation undertaken in 2005 of 288 villages and 355 infrastructure works, technical teams found that the calculation of unit costs yielded data consistent with these figures and also reconfirmed that KDP activities are substantially less expensive than other projects while still maintaining high quality of works. This would not be possible with high real corruption rates.

Preventive Measures – There is no doubt that KDP and all projects in Indonesia operate in a high-risk environment when it comes to issues of leakage and corruption. KDP is not immune to these threats. The project has long maintained an active anti-corruption strategy which includes: eliminating complexity in procedures and systems; information transparency, and responding to grievances. This strategy translates in practice into several concrete measures such as contracting some 28 provincial NGOs and journalists to monitor independently KDP, increasing technical oversight, follow-up and sanctions for culprits, increased auditing, information-sharing to the public, and studies such as the ones listed above to learn which methods are effective or not. While it will be difficult to reduce corruption in its entirety, the program has made significant strides. Olken’s innovative study was part of KDP’s own efforts to find further tools to improve the anti-corruption arsenal.

March 2006

 Back to top

More information:
blye arrow Indonesia Kecamatan Development Program

Permanent URL for this page:

KDP website (in Bahasa Indonesia)

Selected Information