It’s an ordinary day. You’re driving through your neighborhood and you get to that same intersection you pass every day — bam! — you hit the same giant pothole as always. Your shocks screech while you remember that apparently there’s no budget to fix the asphalt. At home, you turn on the TV and see the mayor cruising around in an official car that costs more than your house, or you find out the cleaning contract went to his brother-in-law’s company at a golden price.
Your blood boils, right? And then, at Sunday lunch, someone drops the classic bar-stool bombshell: “It’s because we’re a country of crooks, it’s our culture, that doesn’t happen in Northern Europe.”
Well, pay attention, because we’re about to dismantle one of the biggest bar-talk myths of our time. Political corruption isn’t a defective gene in the DNA of certain cultures, nor a historical curse, nor a lack-of-ethics problem you can fix with a few morality classes.
The principal-agent problem
To understand embezzlement without getting sentimental about it, we need to strip away the romanticism. In social science, institutional corruption is explained through what’s called the principal-agent problem [1]. You are the principal (the taxpayer) and the politician is your agent (the employee you delegate the management of your money to). The issue is that you can’t watch your employee twenty-four hours a day.
It’s like when you take your car to the shop because it’s making a weird noise. You have no idea about mechanics, you don’t know what’s wrong, but the mechanic tells you the head gasket blew and the repair costs a thousand euros. You have no choice but to trust him blindly. It’s the same in public administration, just at an industrial scale. When the mayor says paving your street costs two million, you can’t go count the bags of cement yourself. And when a human being — doesn’t matter if they were born in Copenhagen or Bogotá — has access to a box full of cash and near-total certainty that nobody has the technical knowledge to watch them, the odds that they’ll dip their hand in are high.
The corruption index you read about in the news is a mirage
Every year, the media runs front pages on Transparency International’s famous Corruption Perceptions Index (CPI). That world map painted in colors, with Nordic countries glowing fluorescent green and half of the Global South in depressing red. We rely on that index to say who’s corrupt and who isn’t, and the Global South tends to come out looking bad.
But there’s a methodological detail almost everyone ignores. The CPI doesn’t measure actual corruption. It measures exclusively subjective perception, and it doesn’t ask the average citizen — it’s built by aggregating surveys of international experts, risk analysts, and multinational executives [2]. In other words: cognitive bias, maxed out.
Benjamin Olken, probably the greatest living genius in the empirical measurement of state fraud, demonstrated in 2009 that relying on these perceptions is a first-class academic trap. He cross-referenced perception surveys with real data on stolen money that he himself had objectively measured on the ground in Indonesia, and found that perceptions are deeply contaminated, and people often have no idea how much is actually being stolen from them [3]. Corrupt politicians are specialists at hiding fraud in the murkiest budget line items, where no surveyor can ever see it.
Think about the mental shortcut at play: when a country is rich and the streets are spotless, experts assume “zero corruption” through a simple halo effect. But they forget about tailor-made bank bailouts, legalized corporate tax havens, or consulting contracts handed out by hand-picking in wealthy-capital ministries. These are forms of institutionalized corruption that are far more sophisticated and invisible on a quick questionnaire. Using this index to claim one culture is more thieving than another is a glaring error. Corruption isn’t a matter of opinion — it’s measured by digging into the dirt.
The science of measuring real theft
How do you measure embezzlement without relying on what people think? In the early 2000s, Olken ran one of the most spectacular field experiments in the history of political economy. The Indonesian government was distributing millions to hundreds of rural villages to build roads. Olken designed a Randomized Controlled Trial — the same rigorous methodology used in pharmacology to test medicines — and split more than six hundred villages into intervention groups [4].
In some villages they applied the “democratic” strategy: mass invitations to residents to attend public meetings and monitor the accounting. In others, the institutional-fear strategy: they were told they had a hundred percent chance of receiving a surprise technical audit from the central government.
And here’s the part of the methodology worth framing. To find out how much had been stolen without relying on falsified invoices, Olken hired independent road engineers, sent them out with heavy machinery, had them extract cylindrical samples of the asphalt, and calculated in a lab exactly how much sand, gravel, and cement was actually in the ground. They then compared that physical cost against the invoices sent to the government. The exact gap between the invoice and the stone was the money that had been stolen.
The result was that citizen-assembly monitoring was an almost total failure. It barely reduced the missing money. Why? Because of the relentless free-rider problem: monitoring public works invoices is tedious, requires technical knowledge nobody has, is incredibly boring, and can earn you the mayor’s enmity. Everyone thinks “someone else will raise their hand.” However, in villages with an announced government audit, material theft dropped by eight percentage points [4]. Knowing a relentless inspector was going to review the paperwork terrified local thieves.
The Brazilian lottery drum
But for fear to work, it has to be tied to the politician’s survival. And here science hands us another gem from South America.
In 2003, Brazil’s federal government launched an anti-corruption program that used the national lottery drums, broadcast live on television, to randomly select which municipalities would be investigated that month. Once the number came up, an army of federal auditors would show up unannounced at the town hall. Claudio Ferraz and Frederico Finan took advantage of this giant natural experiment and cross-referenced the findings with subsequent election results [5].
They found a chilling pattern: first-term mayors, desperate for reelection, stole on average 27% less than “lame duck” second-term mayors who could no longer run again. Without the imminent threat of being fired by citizens, restraint disappeared and they plundered as if there were no tomorrow.
But the second part of the finding is the critical one: auditing a thieving politician before an election only led voters to punish him if there was an independent local radio station translating the technical data and telling residents about it [5]. Without that channel, the corrupt mayor got reelected exactly as if he’d stolen nothing at all. The equation was proven: you need objective technical information, a mass distribution channel, and a real possibility of punishment at the ballot box.
Saying “corruption is a problem” can make it worse
Peiffer and Alvarez (2016), and later Cheeseman and Peiffer (2022), ran experiments in Nigeria where they exposed citizens to typical anti-corruption messages: “corruption is destroying our country,” “we must all fight against it.” The result: these messages, in certain contexts, increased people’s willingness to pay bribes [6]. By emphasizing how widespread corruption is, you normalize it. The implicit message is “everyone does it, so you can too.” It’s the equivalent of putting up a “no stealing” sign in a shop that just got robbed: you’re reminding people that stealing is an available option.
A fingerprint scanner doesn’t have cousins to hand contracts to
What happens if we cut the middleman out of the equation? Let’s travel to India, to the state of Andhra Pradesh. The government spends fortunes on social assistance programs, but the money had to pass through an endless chain of bureaucrats who demanded their cut at every step. Or worse: officials invented hundreds of “ghost” beneficiaries — dead or nonexistent people — and collected their pensions month after month.
In 2016, Muralidharan, Niehaus, and Sukhtankar published the results of a massive trial: they randomly assigned millions of citizens to a new system where they collected benefits using smart cards with biometric fingerprint verification [7]. Signatures didn’t work, stamps didn’t work. For the money to come out, a living person had to put their finger on a scanner.
The results demolished any argument against government digitization. Leakage of stolen money fell by about forty million dollars a year [7]. And the most striking finding was a human one: more than ninety percent of the poorest overwhelmingly preferred the new system, because payment was faster and no local thug extorted them for a cut just to hand over their own money. A silicon fingerprint scanner doesn’t understand political loyalties, doesn’t have cousins to hire, and doesn’t need electoral favors.
Fisman and Miguel (2007) analyzed more than 150,000 parking violations by diplomats in New York, who have immunity and don’t pay fines. Diplomats from countries with high corruption on the CPI racked up dozens of violations each; those from Sweden or Norway, practically zero [8]. Does that mean corruption is “cultural”? In one sense, yes: internalized norms travel with people and operate even without oversight. A Norwegian diplomat doesn’t park badly because his social environment has drilled into him that you just don’t do that, and he carries that norm in his suitcase to Manhattan. But Swedes weren’t born honest — they have a past of bribery and patronage too. Any country can change its institutional design, and with it, behavior.
Catching the kingpin
And what about the corrupt ones? If we catch them and they go to prison, why is there still corruption — aren’t they afraid? The reality is, yes, it’s true: according to research, the greater a politician’s certainty of getting caught, the less likely they are to break the law.
This is confirmed by Fisman and Miguel’s (2007) study: in New York, while politicians were immune, diplomats from the most corrupt countries racked up 18 million dollars in fines. Starting in 2002, the rules changed — politicians could lose their international license plates, and violations were deducted from foreign aid. The result was that violations dropped by 98%.
But unfortunately, the problem is that if a country is corrupt, its justice system tends to be corrupt too. If an attorney general is a position appointed by a politician or political party, and judges depend on that person for their promotions, prosecuting corrupt members of that same party becomes a much riskier move. China has been fairly ruthless about it: since 2012 it’s run a massive anti-corruption campaign among officials, including death sentences, which has reduced judicial favoritism toward connected parties.
Singapore and Hong Kong have had success with dedicated agencies that pursue corruption in both the public and private sectors. Though there’s one detail worth noting: both raised civil servant salaries to match the private sector, which is a factor that tends to favor low corruption levels.
Marked digital money as a solution?
Central Bank Digital Currencies (CBDCs) are currently in active development in more than a hundred countries. What would happen if public money weren’t just digital, but mathematically “programmed” through smart contracts? Imagine a hospital budget where every euro carries embedded code preventing it from being transferred to tax havens or spent on anything that doesn’t match the tender. If a mayor tried to divert half a million to his brother’s company, the money itself would block the transaction because it doesn’t meet the cryptographic rules.
There are several projects currently running this way around the world. In Kazakhstan, the “Coulouring Money” project uses its digital tenge to pay public contracts, with the money digitally traceable so it can’t end up with blacklisted or unsuitable entities. Brazil and Australia have pilot projects using the same philosophy for public aid and payment management.
We’d save on auditors and endless trials. Embezzlement would become technically impossible by design. But — there’s always a but — if the State can track and program every single cent of public money, it can also apply that same pinpoint control to citizens’ private money.
Fix the incentives and the potholes get filled
The global evidence converges with a clarity that leaves no room for evasion. Corruption in your city isn’t an immutable cultural trait. Nor is it cured with ethics sermons or appeals to goodwill during election campaigns. It’s been thoroughly diagnosed: it’s the predictable byproduct of a system that offers too many opportunities to steal, too much opacity to hide the money, and too little chance of political ruin for doing it.
What works — proven through randomized experiments in Indonesia, Brazil, India, and Uganda [9] — is combining three pieces: independent technical audits that measure the stone rather than the invoice, free press channels so voters can punish the thief at the ballot box, and automating the flow of money through biometric infrastructure that exterminates middlemen. If even one piece is missing, the whole system limps.
If city hall still hasn’t fixed the pothole on your street, it’s not because the politicians in your country were born with a defective evil gene. It’s because the administrative system around them is still designed in a way that makes stealing the asphalt money scandalously cheap. Fix the incentives, apply relentless monitoring, slap an unbreakable digital fingerprint on the money, and watch how fast those potholes get filled.
REFERENCES
[1] Klitgaard, R. (1988). Controlling Corruption. University of California Press. — Theoretical framework
[2] Transparency International. (2023). Corruption Perceptions Index: Methodology and Framework. Transparency International. — Methodological framework
[3] Olken, B. A. (2009). Corruption perceptions vs. corruption reality. Journal of Public Economics, 93(7–8), 950–964. — Reliable
[4] Olken, B. A. (2007). Monitoring corruption: evidence from a field experiment in Indonesia. Journal of Political Economy, 115(2), 200–249. — Reliable
[5] Ferraz, C., & Finan, F. (2008). Electoral accountability and corruption: evidence from the audits of local governments. American Economic Review, 98(4), 1274–1311. — Reliable
[6] Cheeseman, N., & Peiffer, C. (2022). The paradox of anti-corruption messaging. Nature Human Behaviour, 6, 1349–1357. — Reliable
[7] Muralidharan, K., Niehaus, P., & Sukhtankar, S. (2016). Building state capacity: evidence from biometric smartcards in India. American Economic Review, 106(10), 2895–2929. — Reliable
[8] Fisman, R., & Miguel, E. (2007). Corruption, norms, and legal enforcement: evidence from diplomatic parking tickets. Journal of Political Economy, 115(6), 1020–1048. — Reliable
[9] Reinikka, R., & Svensson, J. (2004). Local capture: evidence from a central government transfer program in Uganda. The Quarterly Journal of Economics, 119(2), 679–705. — Reliable