Transparency International recently released its latest Corruption Perceptions Index (CPI). This index runs from 0 to 100, where higher scores indicate lower corruption and lower scores indicate higher corruption. The global average now stands at 42 out of 100 – the worst level in more than a decade. Of the 182 countries ranked, 122 score below 50. Only five countries score above 80. Even long-standing high performers like Luxembourg and Switzerland appear to be losing ground.
For researchers, policymakers, and journalists, the CPI is a primary reference point. These rankings quickly make headlines and spark political reactions. But beneath the headlines lies a simple question: What exactly are we measuring when we try to measure corruption?
Corruption is not directly observable. Official statistics do not record bribes, for instance. Favoritism or nepotism rarely leaves a paper trail. And politicians are unlikely to label their own behavior “corrupt.”
When looking at this year’s rankings, one important question is not who moved up or down the overall list. It’s equally important to understand what the index reflects (and what it does not). Corruption can be measured in different ways. Each approach highlights some aspects while ignoring others. Measuring corruption is not merely a technical exercise – this shapes how we diagnose problems and design solutions.
Why measure corruption?
People typically discuss corruption as a moral failure. This can be politically powerful but also analytically limiting. This view highlights individual wrongdoing. However, it does not help us diagnose institutional weaknesses or evaluate how to improve the situation..
For research and policy, moral language is not enough. Measuring corruption plays several essential roles:
1. It makes an invisible problem visible, as corrupt acts usually take place behind closed doors. Without systematic measurement, corruption is reduced to anecdotal stories about scandals, rumors, or individual cases. Measurement turns claims such as “this system is corrupt” into observable patterns – what we don’t measure becomes easier to ignore. And measuring corruption shifts the discussion to policy-relevant terms, rather than a moral accusation.
2. Comparisons across countries and regions become possible. Why does procurement corruption persist in some countries but not others? Why do some systems improve while others stagnate? Corruption indices enable researchers to identify broader trends.
3. New policy insights and explanations can result from this type of corruption analysis. Scholars ask whether higher public sector pay reduces corruption, whether independent courts matter, or whether international organizations constrain abuse of power. Without measurable variation, these questions remain unanswered.
Measurement is also essential for policy evaluation. Governments and international organizations invest heavily in anti-corruption efforts as they adopt new laws and create new agencies. Without a closer look at the indicators of corruption, we cannot assess whether those attempts alter incentives or merely change formal rules. Measurement, in this sense, is a form of accountability.
4. Measurement shapes real-world decisions. Corruption indicators influence aid allocation, loan conditionality, investment decisions, and risk assessments by firms and donors. In short, even imperfect measures of corruption help guide billions of dollars.
Why is corruption so difficult to measure?
Measuring corruption is difficult for several reasons. Corruption is inherently hidden. Bribes, kickbacks, and favoritism are designed not to be observed. Even when corruption is exposed, it may be selective and politically driven.
Another problem is the lack of a single, universally accepted definition of corruption. As I discuss in an earlier Good Authority piece, narrow definitions focus on bribery. Broader definitions include other forms of misconduct such as nepotism, cronyism, favoritism, and revolving-door abuses. What counts as corruption in one context may appear as normal operations in another. Context-dependent definitions of corruption complicate comparison.
A final point to consider is that corruption data are politically sensitive. Governments may resist data collection or attempt to shape narratives. Therefore, data availability and quality are not politically neutral. Efforts to measure corruption, thus, can become politicized.
The various ways to measure corruption
Given these challenges, researchers have developed several approaches to measure corruption. Each captures a different dimension of corruption, and each approach comes with trade-offs.
Perception-based measures
Perception-based measures ask how corrupt a system appears to informed observers. These measures aggregate expert surveys, business assessments, and risk evaluations. The focus of this approach is not on observed behavior. Instead, these measures capture expectations and reputations. Transparency International’s CPI, the World Bank’s Control of Corruption, and V-Dem’s Political Corruption Index fall into this category.
These measures are useful for capturing “grand corruption” – the systemic abuse of power that may be invisible to the average citizen. These findings allow for cross-national comparisons and time-series analysis. Scholars use these datasets to explore the causes and effects of corruption.
But perceptions may be slow to change. A country that implements genuine policies may not see immediate improvements in its score. Media coverage and past scandals can shape assessments. Perception-based measures capture how experts see corruption, not necessarily how people experience it.
Experience-based measures
Experience-based measures offer an alternative. A survey, for instance, can ask individuals or firms about direct encounters with corruption. Were they asked to pay a bribe? Did they make informal payments to access services? Examples include Transparency International’s Global Corruption Barometer, the World Bank Enterprise Surveys, and the Quality of Government Institute’s European Quality of Government Index (which focuses on both perceptions and experiences with public sector corruption).
These datasets capture corruption as it is lived. They are useful for analyzing “petty corruption” – the abuse of public office by officials during day-to-day interactions with the public. Researchers use experienced-based measures to examine the impact of everyday corruption and evaluate attempts to reduce it.
Experience-based measures also have limitations. They capture little about those instances of corruption that rarely involve direct citizen contact, such as elite corruption or procurement fraud. These indices are also vulnerable to underreporting due to fear, especially in contexts of low trust or authoritarianism. Low levels of reported bribery do not necessarily mean clean governance.
Structural measures
Structural measures take a different approach, shifting the focus from observed corruption to institutional vulnerability. This approach doesn’t ask how corrupt a system appears. Nor does it ask what corrupt acts people experienced. Instead, these measures examine the conditions that make corruption more or less likely.
These measures rely on (mostly) objective indicators, such as administrative discretion, budget transparency, digital access, judicial independence, media freedom, and trade openness. The Index of Public Integrity, the Transparency Index, and the Corruption Risk Forecast are some leading examples.
Structural indices are especially useful for diagnosing long-term corruption risks. They help explain why corruption persists even when a country has formal rules in place. Researchers use these indicators to trace trajectories and identify institutional bottlenecks. These indices are also helpful in distinguishing countries that receive similar perception-based scores but differ substantially in institutional capacity.
Structural measures also have limits. They are not fully objective. Some components, such as assessments of judicial independence, rely on expert evaluations. These indices also depend on imperfect proxies. Indicators that can approximate citizens’ capacity for collective action include digital access or social media use, for instance. But such measures are only rough substitutes for how accountability works in practice.
What measurement can (and cannot) do
Of course, no single indicator captures corruption in full. Perception-based measures capture reputations. Experience-based measures capture interaction. Structural measures capture areas of vulnerability. Research suggests these indicators often strongly correlate with each other. But they are not interchangeable.
The key question here is not which measure is “best.” After all, no measurement protocol could possibly reveal every illicit transaction. But without systematic measurement efforts like the Corruption Perceptions Index, corruption remains anecdotal and thus easier to dismiss or politicize.
Mert Kartal is a 2025-2026 Good Authority fellow.
