Beyond perceptions: Applying behavioural insights to the OECD Public Integrity Indicators

Imagine you are at a government office, applying for a business license. The paperwork is complete, the wait has been long, and just as you think you're nearing the finish line, the official leans in with a knowing look. A subtle hint – a suggestion that things could move faster with a ‘small contribution.’ Would you pay it? And, if you did, would the official accept it? More importantly, would anyone notice – or care?

This kind of everyday corruption happens worldwide, from business permits to border crossings. But here’s the problem: anti-corruption measures have paid little attention to the real-life decisions that drive corruption at the individual level.

Traditional anti-corruption approaches have focused on increasing transparency—through open data, freedom of information laws, and budget disclosures—and strengthening accountability via audits, oversight bodies, and anti-corruption agencies. While these measures are essential, their impact has often been modest or inconsistent, particularly when implementation is weak or enforcement is politically constrained (see OECD and World Bank).

To address this gap, the OECD Public Integrity Indicators (PII) offer a comprehensive framework for assessing institutional safeguards against corruption. The PII provide an objective, data-driven diagnostic of a country’s integrity system, helping policymakers identify strengths and areas for improvement.

So, how well do the PII assess the resilience against corruption risks? We put them to the test by comparing them to real-world behavioural data, as well as to well-known aggregate indices of corruption: Transparency International’s Corruption Perceptions Index (CPI) and Global Corruption Barometer (GCB).

Using behavioural data, we show that greater transparency in government decision-making processes corresponds to lower bribery incidence. When decisions are made publicly, it becomes more difficult for corrupt acts to go undetected.

In its 2018 report Behavioural Insights for Public Integrity, the OECD emphasised the importance of behavioural insights for identifying corruption risks and for creating evidence-based solutions. Unlike the CPI and the GCB, which primarily reflect perceptions and experiences of corruption, behavioural data captures the decisions individuals make when met with corrupt opportunities. This distinction is crucial for designing interventions that target the root causes of corruption. Behavioural data from incentivised bribery experiments from 2017, 2019 and 2023 have already demonstrated the drivers of corruption at both individual and societal levels.

Our behavioural data comes from bribery experiments conducted across 21 countries, drawing on both a recently published paper and an unpublished study on intra- and international bribery. In these experiments, participants play the roles of a citizen and a public official in a bribery game with real money at stake. As citizens, participants decide whether to pay a bribe to the public official to obtain a license. As public officials, they decide whether to accept bribes offered by citizens. A successful bribe (= a bribe is offered, accepted, and not detected) is profitable but negatively impacts the external climate-protection organisation “Atmosfair”.

Illustration of the bibery game. "Talers" are converted to a country's currency.
Source: Authors' elaboration.

We compared the PII results with CPI, GCB, and behavioural data across our sample. To ensure solid comparisons, we focused on the PII criteria for which there are data available for at least ten countries.  The figure below illustrates all PIIs considered and their relationship with behavioural bribery measures.

Heat-map correlation matrix indicating how the PII sub-indicators correlate with the CPI, GCB, and published (S1) as well as unpublished (S2) behavioural data from incentivised bribery experiments.
* indicates significant correlations. Blue indicates negative correlations, e.g. decreasing bribery with higher PII values. Red indicates positive correlations, e.g. increasing bribery with higher PII values.
Source: Authors' elaboration.

1. "Sunlight is the best disinfectant": Behavioural data provides a novel lens

Our experiments make one thing clear: the more open a government’s decision-making process is, the less bribery happens in our experiments. When decisions are made in the public eye, it’s harder for corrupt deals to slip through unnoticed.

For example, transparency of lobbying activities is associated with lower bribery incidence using behavioural measures. This finding supports the principle that "sunlight is the best disinfectant," emphasising the importance of transparency in curbing corruption, as seen above.

Interestingly, most PIIs are not significantly correlated with behavioural measures of bribery. This disconnect suggests that, while many countries have formal anti-corruption frameworks in place, these institutional safeguards may not effectively shape everyday decisions around corruption. The finding points to a gap between rules on paper and real-world behaviour—highlighting the importance of aligning institutional reforms with how individuals perceive and respond to corrupt opportunities.

2. There is a positive correlation between the anti-corruption and integrity strategic frameworks and day-to-day corrupt behaviour

Certain strategy-related indicators, such as coverage of strategic frameworks, show a strong positive connection with our real-world bribery data but not with the CPI or the GCB. This means that our behavioural approach captures aspects that commonly used perceptions and experience-based measures miss.

How so?

At first glance, it might seem counterintuitive that strategy-related components such as coverage of strategic frameworks are associated with more bribery behaviour (see OECD working paper). Yet this pattern may reflect reverse causality: countries with higher corruption risks are more likely to develop and implement such strategies. This explanation is plausible but requires further investigation because the present study only examined correlations and no causal effects.

The results emphasise the potential of the PIIs for identifying corruption risks and designing targeted interventions. However, several limitations must be acknowledged.

First, the findings draw on relatively small samples with missing data, which may limit the generalizability of the results. While the analyses provide valuable initial insights, they should be interpreted cautiously.

Second, the correlations observed in this study do not establish causation, and future research is needed to explore the mechanisms underlying these relationships.

Although novel and insightful, behavioural data need validation through larger, more diverse samples. Future research should expand studies across more countries and contexts and include longitudinal studies to assess the long-term effects of integrity measures.

Finally, while this study highlights key trends, it underscores the complexity of corruption dynamics. Effective anti-corruption strategies must account for these nuances, ensuring that interventions are evidence-based and adapted to specific contexts.

Behavioural insights can be used in combination with the Public Integrity Indicators to better understand what prompts ethical behaviour: Going beyond perception and combining real—world data sources can help to better understand corruption risks.

Transparency works: Focus on proactive measures such as openness, transparency, and public consultation to reduce opportunities for corruption.

This study demonstrates the potential of integrating behavioural insights with multidimensional governance indicators to understand and address corruption. However, given the study’s limitations, future research should validate these findings and expand our understanding of corruption dynamics. By addressing cross-national integrity gaps and leveraging actionable indicators like the PIIs, policymakers can develop more effective interventions to strengthen public integrity and reduce corruption in a globalised world.

This research was funded by the University of Cologne Center for Social and Economic Behavior.

OECD (2018), Behavioural Insights for Public Integrity: Harnessing the Human Factor to Counter Corruption, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/9789264297067-en.

OECD (2024), Anti-Corruption and Integrity Outlook 2024, OECD Publishing, Paris, https://doi.org/10.1787/968587cd-en.

This blog article is authored by the winners of the 2025 OECD Anti-Corruption Research Challenge, a competition showcasing innovative research on applying OECD Public Integrity Indicators to academic studies on anti-corruption. This blog article should not be reported as representing the official views of the OECD or of its Member countries. The opinions expressed and arguments employed are those of the authors.