Within the global anticorruption community, no topic has generated as much discussion as the measurement issue. Start with the most basic of questions. Is there an agreed upon definition of corruption? Get by the heated objections to claim there is none, and next consider: are there ways to measure something that by its nature is clandestine? Take for granted clever social scientists can, then ask if these measures are comparable. Across time? Different nations?
The methodological and epistemological debates over such questions have raged in the academy for millennia. But as corruption has gained ever more salience as a policy issue, the debate has ranged far beyond the academy. Just ask any political leader forced to explain to citizens why his or her country scored poorly on some corruption-rating scale.
The United Nations Office on Drugs and Crime has now brought needed clarity to the debate. At the request of the 189 state parties to the U.N. Convention Against Corruption, it has published the first draft of a comprehensive statistical framework to measure corruption (here) with a form for providing comments (here).
Bearing in mind my bias, I contributed (very slightly and with more comments promised), I think the draft is a first class piece of work. Two of many reasons why.
First, UNODC is very clear that the purpose of the effort is not to develop a method for comparing the level of corruption across nations (something not only impossible but certain to bring instant condemnation from many states). Rather, in the accompanying paper explaining its work UNODC writes:
“The main objective of the comprehensive statistical framework to measure corruption is to provide guidance to national governments to develop national information systems able to detect the presence, measure the magnitude and monitor trends of the different forms of corruption, as per the United Nations Convention against Corruption (UNCAC). The framework provides a list of indicators together with guidance on how to analyze the indicators together and on how to collect the needed data. The ultimate goal of the statistical framework is to contribute to the efforts of Member States to build the sort of scientific evidence that can underpin the design, implementation, monitoring, and assessment of anti-corruption policies,” para 10.
Hard to think of a more useful, salutary, and achievable goal.
Second, the authors strike a blow for realism about TI’s CPI and the World Governance Indicators:
“International organizations have made attempts to measure corruption by producing information based on perception-based indicators. Indices such as the World Bank’s Control of Corruption indicator or the yearly Transparency International’s Corruption Perceptions Index identify perceptions and drivers of corruption and raise awareness of the negative effects that corruption has on sustainable development but have important limitations as these do not provide sound information on the direct occurrence of corruption and on the different types of corruption, the sectors, procedures, and formalities that are more vulnerable to corruption,” para. 15.
UNCAC member states and the anticorruption community should welcome this solid, important, and authoritative contribution to the measurement question.
This looks like an excellent start to the extent that it gets away from the goal of providing yet another set of whole-country, one-number scores, and seems open to the idea of different kinds of corruption occurring in differing contexts, reflecting distinctive sets of incentives, opportunities, and (lack of) constraints — I’ll look forward to reading the actual document with much interest —
Thanks for sharing! Looks like it builds of the ICCS methodology. I guess the proof will be in the ability of some government agency in each country to run/coordinate the surveys with a representative enough sample, analyze the statistical data (I guess this could be done through automation?) and produce this sort of thing yearly? Or maybe take slices of the data? Will continue watch with interest
Thanks for sharing! Looks like it builds off the ICCS methodology. Would be interested to know more about proposed implementation as undertaking the surveys (with statistically relevant proportion of the population) and aggregating all the admin data could be a big lift for a lot of countries.
Thank you, Professor. It’s strange because I didn’t find any call for this consultation on the UNODC webpage.
Rick, thanks for sharing the draft work of the UNODC. A few observations:
1. Many of the indicators require ad how information collection in the form of surveys and in-depth analyses of enforcement actions. This means that the data is not likely to be collected. Instead, efforts should be devoted to getting more out of existing information systems (e.g., eprocurement and IFMIS’s) through automated collection. This also opens up the question about big data, which this UNODC paper has largely ignored.
2. Some of the indicators have not been defined clearly or are non-sensical, e.g., comparing disclosed assets with income to identify discrepancies (see 1.3.a).
3. Many indicators are based on averages and sums, failing to identify hot spots.
4. The indicators presented are not supported by any evidence to show why they are relevant and can be relied upon. Such evidence would also help respond to those that would sow doubts about the indicators’ robustness (and the anti-corruption agenda more generally).
5. What is presented is a laundry list of indicators related to economic crime. As a sourcebook I suppose that is helpful. The problem, however, is that one is left with many indicators searching for a problem. It would be more practical to first ask what decisions are we trying to make and then determine what indicators would be most helpful in making that decision. For example, only a small subset of indicators is relevant to any given area, whether HR, procurement, audit, corruption investigations, asset forfeiture, and so forth. And within each area, some indicators are likely to be more relevant than others, e.g., the share of firms reporting bribe demands is less actionable than the number of tenders with indicators of irregularity prior to contract award. My plea here is that we select indicators that are likely to trigger an action.
Thanks Richard, this is very timely. I will be giving my detailed feedback to this draft, as I carried out my PhD research on this topic in Kenya, and proposed an anticorruption scorecard to help countries detect, measure and monitor corruption. More later.