Do you remember the 2008 economic crisis? Greece was particularly badly hit. Not only did Greek debt prove to be unsustainable, but the crisis also revealed the systemic corruption afflicting the country. The problems had not been completely invisible prior to this: Greece had received critical evaluations from the Group of States against Corruption (GRECO), for example. But strangely, Greece had been performing well in the Corruption Perceptions Index (CPI) before the crisis.

Once the corruption of the past became known, and its impacts felt, experts rushed to downgrade Greece’s scores – just at the point when new governments and new leadership were actually working to fix the problems.

This proved that the major challenge in corruption studies was not only how to measure corruption in a specific and meaningful way, but also how to measure it across time, so as to capture change, not just reputation. We developed the Corruption Risk Forecast (CRF) for 120 countries, releasing it in 2019, as an answer to this problem, because if we cannot trace corruption across time no anti-corruption policy can be evaluated.

The difficulties of evaluating anti-corruption tools

For many years, I was among the few academics responding to the challenge launched by the even fewer policymakers who wanted evidence of what worked (and didn’t work) in the anti-corruption arsenal. By 2010 my team had evaluated existing top-level mechanisms and governance models for the Norwegian Agency for Development Cooperation (NORAD), including the ratification of the United Nations Convention Against Corruption (UNCAC). We found that countries that had ratified the Convention (about half the world) had not recorded more progress than those that had not.

Later, consulting for the Swedish International Development Cooperation Agency (SIDA), we looked at more anti-corruption tools: anti-corruption agencies, conflict of interest regulation, freedom of information acts, and so on. Two major issues hindered our efforts to evaluate them. First, we had to use as the ‘dependent’ variable over time a perception aggregate, either CPI or the Control of Corruption indicator from the World Bank.

These instruments were not suited for tracing corruption over time, nor were they suited to evaluation purposes. They lack specificity, meaning that even when something changes it is hard to know what changed – and the time lag inherent in these measures makes it harder still. Also, since these measures are statistical aggregates they display little variation across time, and the statistical aggregation method of the various individual sources on corruption means that any change they do signal often falls within the margin of error.

The second obstacle to effective evaluation was the lack of country context. No anti-corruption intervention should be assessed without an understanding of the broader factors that shape corruption, as corruption is the expression of a governance context. In my earlier assessments I used the Human Development Index as a measure of a country’s level of modernity, with modernity serving as a proxy for the governance context.

The problem with this approach was that modernisation is not external to corruption: they influence and reflect one another. For instance, a country should only be considered modern when it has a bureaucracy autonomous from private interest; and patrimonial countries may have skyscrapers, but they are not modern if the ruler uses the state budget as his own. Corruption and development influence one another, it’s not just a one-way street, meaning ‘modernity’ is not a sufficient measure of context when it comes to corruption.

The IPI: breaking down – and measuring – control of corruption

We needed a model of the main factors directly determining control of corruption. In 2016 I pulled these factors together as the Index for Public Integrity (IPI).18a9065ca75d The IPI conceived corruption as a balance between constraints (legal and normative) and resources (power discretion and material resources). The single composite score is based on an assessment of six national-level proxy indicators:

• Budget transparency
• Administrative transparency
• Online services
• Judicial independence
• E-citizenship (a measure of broadband subscriptions and internet users)
• Freedom of the press.

These apparently disparate components come together remarkably well in a principal component analysis, as they measure one latent variable – control of corruption – and thus prove highly

As the IPI score for each country is normalised on a 1–10 scale for ease of analysis (10 representing the highest integrity), the comparison across time presents the same issues as other aggregate indexes. We thus prefer to use aggregate IPI scores to compare across countries and not over time.

We decided it would be valuable to track the individual components of IPI over time, not just the aggregate. By doing so, one could know exactly which indicators progressed, which stagnated, and which regressed, bringing us closer to a full ‘diagnosis’ of country trends and turning the lack of a single score to an advantage.

Enter, the Corruption Risk Forecast (CRF).

The CRF: finding and explaining corruption trends

The CRF is an analytical rather than a measurement tool, designed for corruption risk analysis and evaluation purposes. In the first step to creating the forecast, we assess the IPI’s country trend for consistency, to ensure that we capture the whole and not just isolated parts. We consider a time interval of ten years; in that period, a country needs to change significantly in two or more IPI indicators in one direction, and none in the opposite one, for its change to be recorded.

The second, qualitative step is to solve inconsistent cases (say, three indicators in the positive and one in the negative) and update the data for all the cases. The ten-year time series has a time lag, and events might take place beyond the latest data point that subvert long-term trends. The unpredictable can always strike, for example Burkina Faso and Ukraine both showed improving trends when unforeseen coups or wars (respectively) set them back.

The methodologies for both IPI and CRF can be found on our website.

The Corruption Risk Forecast (CRF)

The result is a transparent trend diagnosis of a country. Where is this country and why? Will this or that anti-corruption tool give results next year? As an analytical tool, the CRF can be used for ex-ante as well as ex-post evaluation of anti-corruption policies and tools.

The components provide the context against which any individual anti-corruption tool can be assessed: if judicial independence has significantly declined the judicial measures will not work; if freedom of expression has shrunk significantly one can hardly expect any whistleblower law to perform.

Where these indexes are taking us

Anti-corruption tools operate in a broader context, not in a vacuum, and the evaluation of any given anti-corruption intervention should use the background controls offered by the CRF. We see in the data that the world has not progressed on anti-corruption, despite the technological advances on which we all pin our hopes (and which reduce opportunities for corruption), because these tech advances were offset by bad politics. Press freedom and judicial independence, the two key corruption ‘constraints’ components, have mostly declined or stagnated in those countries most in need of improvement.

To date, the CRF has been used mostly by governments and intergovernmental organisations, but the data (and our contact details) are available to all. Some US businesses also expressed interest in the project, including in acquiring it, but as academics we are reluctant to create a paywall. We hope for even more evidence-based anti-corruption in the future, and not just of the ‘let’s show them we do something’ kind that is still so frequent in anti-corruption policy.

The IPI and CRF complement one another: the IPI being used for comparisons between countries and the CRF for trend analysis over time. Freely available to all, any analyst or anti-corruption reformer, or any investor looking for a destination, can find both the trend of a country and its explanation. They can use these tools to answer their own research questions, whatever they may be, and analyse what works in their own context.