Stuck at 43: why the world's corruption score won't move

For more than a decade, one of the most-watched numbers in global governance has barely moved. Transparency International's global Corruption Perceptions Index — a 0-to-100 score aggregating 13 expert and business surveys across 180 countries — has hovered near 43 since the current methodology began in 2012. In 2024 it held at roughly 43, and the dataset records a recent slip to about 42.5 for 2025, a record low under this methodology. More than two-thirds of countries still score below 50, the index's de-facto integrity threshold.

Why a flat line is itself a finding

The striking part is not the number but its stillness. Since 2012 the world built more anti-corruption machinery than at any point in history. The UN Convention against Corruption (UNCAC), adopted by the UN General Assembly on 31 October 2003, remains the only legally binding universal anti-corruption instrument, with roughly 190 states parties obliged to criminalize, prevent and cooperate against corruption. National reforms multiplied: Ukraine's bureaus and transparent procurement, Romania's prosecutorial drive, Georgia's service overhaul. Yet the global needle did not climb.

The tug-of-war model

Factrail reads this stasis as the visible balance of two opposing forces. On one side, anti-corruption enforcement capacity — independent bureaus, transparent procurement, merit-based hiring and the political will to prosecute. The Factrail model links UNCAC to a structural, worldwide strengthening of this driver: a binding treaty obliging every signatory to build enforcement infrastructure. On the other side, state-capture pressure: the continuing effort by concentrated private or partisan interests to bend regulators, courts, procurement and oversight bodies to their benefit.

In the dataset's driver-to-indicator links, enforcement capacity pushes the global CPI up with medium strength and a long lag of roughly three years, while capture pressure pulls it down with comparable force on a shorter lag. When the two forces are near balance — gains in some states offset by backsliding and capture in others — the aggregate stays flat. That is precisely what the 2012-2025 series shows.

What changed relative to the normal line

The normal line for this indicator is 50, Transparency International's integrity threshold. The global average has sat persistently below it, near 43, never closing the gap. The recent drift downward to about 42.5 is small in absolute terms but notable because it breaks even the appearance of stability, suggesting capture pressure may be edging ahead of enforcement gains in the aggregate.

What is verified versus uncertain

The CPI anchor values are real reported annual averages and are well sourced; UNCAC's adoption and legal status are verified. What remains interpretive is the attribution: CPI measures perceptions, not corrupt acts directly, and the two-driver tug-of-war is the Factrail model's reading of why the line is flat, not a measured causal proof. Driver intensities are analytical estimates, and the driver-to-indicator links carry medium confidence.

Alternative explanations

The flat line could also reflect methodological design: CPI is built to be stable and comparable year-on-year, so genuine but uneven national change may simply wash out in a 180-country average. Survey-based perceptions can lag real reform, and large-country dynamics can dominate the mean. Each of these would dampen movement without invoking a capture-versus-enforcement balance at all. The dossier therefore treats the tug-of-war as a plausible framing rather than a settled conclusion.

What may happen next

If enforcement capacity continues its post-2017 erosion seen in the driver series, and capture pressure keeps rising, the global average could slip further below 43 over the next several years rather than recover toward 50 — a baseline expectation the Factrail forecast layer treats as needs_review. The sources supporting this analysis are Transparency International's 2024 CPI release and the compiled annual-average series, together with the UNCAC record cited in the dossier.