An uncle wires the money. It is an investment, a friend of a friend, returns that are good but not impossibly good, statements that arrive on time and always in the black. For years there is nothing to see. Then one day the statements stop, or the man stops answering, or a letter comes from a lawyer, and the savings are gone, and the worst part is not even the loss. It is what happens next, which is usually nothing. No one is arrested. If someone is, the sentence lands years later and feels small against the wreckage, and the money does not come back. The family is left with a double injury, the theft and then the impunity, and a question that sounds like a cry but is really an analytic one: how does this keep happening, and why does no one seem to pay for it? This chapter answers both halves. The persistence and the light punishment are not two mysteries. They are the same fact seen twice, and the fact is that fraud, looked at coldly, is a business with a favorable balance sheet.
The scam is a business
Strip the outrage away for a moment, which is hard to do and necessary, and look at a scheme the way the person running it does. It has revenues, the money brought in from marks. It has costs, the things that can go wrong: being detected, being charged, being convicted, being made to pay. A rational operator weighs the take against those costs, and the cost that matters is the penalty discounted by the chance of ever facing it, not the penalty on its own. If the odds of being caught are low, then even a severe punishment, multiplied by a small probability, comes out small. The scheme runs whenever the expected take exceeds that expected cost, and for a great many frauds it does, by a wide margin.
Once the scam is seen as a business, both halves of the chapter’s question fall out of the same picture. Scams persist because the balance sheet is good: the take is collected up front, the cost is a small penalty made smaller by long odds of detection. And the punishment looks light because it is the penalty term alone, the part a victim sees at a sentencing, ripped out of the multiplication that actually governed the decision. The operator never expected to face the penalty in full, and mostly does not, so a system that catches about a third, on the one population anyone has measured, and sentences a fraction of those to a couple of years is not being lenient by accident. It is producing exactly the low expected cost that made the scheme worth running.
Fraud pools in the lag, and the boom widens it
If fraud is a business, it is one that thrives on a particular condition: a gap between a new way of moving money and the world’s ability to understand and check it. Every genuine innovation opens such a gap. A new security, a new market, a new technology arrives faster than the auditors, regulators, and ordinary counterparties who would police it can learn to read it, and into that lag fraud pools, because deception is easiest where verification is hardest. This is why each era’s frauds wear the era’s newest clothes, and why the people running them can look, for a while, like its cleverest innovators.
The lag is not constant. It widens in booms and closes in busts, and the classic statement of why belongs to John Kenneth Galbraith, writing about 1929. He gave the hidden stock of fraud a name. At any moment, he wrote, there is an inventory of undiscovered embezzlement in the country’s businesses and banks, and this inventory, which he proposed calling “the bezzle,” runs to many millions and varies with the business cycle. In good times, he observed, people are relaxed, trusting, and flush with money, audits are lax and few questions are asked, and the bezzle grows; in depression all of this reverses, audits turn penetrating and meticulous, and the bezzle shrinks. The crash, as he put it, “enormously advanced the rate of discovery.” The insight is that fraud is not revealed at random. It is revealed when the money that was papering over it stops flowing, when the marks ask for their capital back and it is not there, when the tide, in the older metaphor, goes out and shows who was swimming without a suit.
Madoff, at the record rather than the legend
The largest of these is worth telling carefully, because the way it is usually told has already drifted from what the documents say, and the drift is instructive. Bernard Madoff ran the biggest Ponzi scheme on record, paying old investors with new investors’ money for decades behind a wall of respectability, having chaired a stock market and advised regulators. The popular version has him confessing to his sons, who then turned him in. The Securities and Exchange Commission’s complaint says something more precise: on or about 10 December 2008, Madoff informed two senior employees that the advisory business was a fraud. In the words the complaint quotes, he told them “it’s all just one big lie” and that it was “basically, a giant Ponzi scheme,” with the losses running, by his own estimate, to at least 50 billion dollars. He was arrested the next day. The gap between “his sons” and “two senior employees” is small, and it is the subject’s standing hazard in miniature: even a fraud reported as heavily as this one carries a legend that the primary record does not support.
The rest is court record. Madoff pleaded guilty on 12 March 2009 to eleven counts, and on 29 June 2009 he was sentenced to 150 years in prison, the stacked maximum. The size of the scheme is quoted many ways because it is genuinely several different quantities, and keeping them apart is the discipline the subject demands. The 50 billion dollars is Madoff’s own estimate as the complaint alleges it. The court-appointed trustee later put the principal actually lost by investors at close to 20 billion dollars, and the claims process that trustee runs, backstopped by the securities industry’s protection fund, has allowed about 17.5 billion in principal for the customers who filed. Each number answers a different question, and a figure that adds up the fictional profits on the fake statements would be larger still. What surfaced Madoff was the same crash that runs through the figure, not a diligent auditor: when the market fell in 2008, investors asked for their money, the redemptions could not be met from new deposits that had dried up, and the scheme collapsed under its own arithmetic. The bezzle was shrinking, and Madoff was inside it.
Why the punishment looks light, and what that is not
Return to the uncle, and to the fury that no one paid. The impunity is real; the enforcement ledger confirms it. Detection is the exception: the estimate that exists, for large United States public companies, is that only about a third of frauds are ever caught in normal times, and the same body of work estimates that in any given year roughly a tenth of large public firms are engaged in securities fraud, destroying on the order of 1.6 percent of equity value a year, about 830 billion dollars in 2021. Those numbers are for large corporations; no one measures the private swindle that empties a retirement account, and there is little reason to expect it fares better, being smaller, quieter, and often gone before anyone with subpoena power hears of it. When a case is made, the sentence is real but bounded: across federal convictions under the guideline that covers theft, property destruction, and fraud, the average is about 22 months. Where fraud can be measured at all, it mostly goes unpunished, which is the picture the victim already had.
What is weaker is the reason usually given for it, that no one cares about the little guy. Indifference cannot be ruled out, since examiner budgets and disclosure rules are chosen by someone, but it is not needed. The impunity is the output of a detection problem and a penalty structure. Fraud is expensive to find, because the good ones are built to pass the checks that exist; expensive to prove, because intent must be shown beyond reasonable doubt; and expensive to punish proportionately, because the losses can dwarf anything a prison term or a forfeiture can answer. A system could care enormously about victims and still catch a minority, because caring does not by itself make a hidden fraud visible. The ledger locates the shortfall: once a fraud is caught and convicted, prison is the norm, and it is being caught that is the exception. The remedies that address that term are the unglamorous ones, better disclosure, funded examiners, faster clawbacks, each aimed at the odds of detection rather than at a penalty that mostly never comes due. This is private fraud, the swindle run for private gain; when the looting is done through the state itself, by officials diverting public money, the mechanism and the cure are different, and that case has its own home in the volume on countries. It is also worth saying what a credit score, the three-digit number asked to vouch for so much, can and cannot screen: it can price the risk that a borrower will not repay, and it was never built to detect a lie told by someone whose credit is excellent, which nothing about running a long fraud prevents, a limit the volume on history takes up where it treats what scores measure.
Scammers get away with it because no one cares about the little guy.
Oversimplified Moderate confidence
The outcome the claim describes is real, which is why the ruling is not that its opposite holds. Scammers do get away with it, often. Only about a third of frauds at large public companies, the one population anyone has measured, are ever detected in normal times; no one measures the private swindle at all; and even a conviction ends in a sentence that a victim reasonably feels is small against the loss. Anyone who answers that fraud is reliably caught and punished is contradicted by the enforcement record, so the reverse of this claim is the less reliable reading, and that is what keeps the ruling off backwards. What the claim gets wrong is the reason, the word “because.” The impunity does not need indifference to be produced. It follows from the economics of detection and penalty: fraud is expensive to find, because good schemes are built to pass the checks; expensive to prove, because intent must be shown; and its expected cost, the odds of being caught multiplied by the penalty, sits below its expected take, which is what makes the scheme worth running in the first place. A system that cared deeply about victims would still catch a minority, because caring does not make a hidden fraud visible, though how hard a society looks is a funded choice, so concern is not nothing; the levers that look harder are duller ones aimed at the odds of detection, disclosure, examiners, and clawbacks. The claim takes a true outcome, real impunity, and pins it all on a cause the ledger cannot confirm and does not need, a failure of concern, where the structure of incentives and evidence would produce the impunity either way. It is oversimplified rather than false or reversed.
Sources
- The impunity is real: Dyck, Morse and Zingales, “How Pervasive Is Corporate Fraud?” Review of Accounting Studies (2023) — in normal times only about one in three frauds at large U.S. public firms is detected, with roughly a tenth of such firms committing securities fraud in a given year and about 1.6 percent of equity value destroyed annually, some 830 billion dollars in 2021. These are large-public-company estimates; the private swindle has no measured detection rate at all.
- The penalty term: U.S. Sentencing Commission, Quick Facts, Theft, Property Destruction, and Fraud (§2B1.1), fiscal year 2024 — median loss about $210,410, average sentence about 22 months, 74.2 percent imprisoned. This is a population of people already convicted, not a measure of how often fraud is punished.
- The mechanism, not indifference: expected cost equals the probability of detection times the penalty, and both terms are low, so the scheme’s expected cost sits below its expected take. Fraud pools in the lag between innovation and comprehension, and the bezzle, the stock of undiscovered embezzlement, grows in booms and shrinks in busts: J. K. Galbraith, The Great Crash 1929. The cases surface when the tide goes out, as Enron, WorldCom, Madoff and Stanford did on or around downturns, though Wirecard and FTX show the tendency is not a law.
- Boundaries: this is private fraud; graft run through the state has its own home in the volume on countries, and what a credit score can and cannot screen is treated in the volume on history. The wider credit cycle that drives the boom and bust is the subject of a later chapter in this volume.
- Confidence is moderate under the rubric, which scores the weaker of evidence directness and construct match. Construct match carries a gap of its own: the claim pairs an outcome with a cause, and the ledger measures the outcome, impunity, while the cause the claim names, a failure of concern, is convicted only indirectly, by showing a mechanism sufficient to produce the impunity without it. The binding, weaker leg is still evidence directness. The figures that establish the impunity are drawn from populations that are not the uncle’s scammer, the detection estimate from large public companies and the sentencing figures from people already convicted in federal court, and no source publishes a direct measure of how often a private swindle is caught and punished. The direction is not in doubt, impunity is real and widespread, which keeps the ruling firm and off low; the distance between the measured populations and the retail scam the claim has in mind is what keeps it off high.
What the ledger explains
The two halves of the opening question turn out to be one. Scams persist because they are businesses with a favorable expected value, and the punishment looks light because the light penalty, discounted by long odds of detection, is precisely the low cost that made the business viable. The bezzle grows quietly through good years and is uncovered, all at once and too late, when the boom that fed it ends. None of this requires anyone to be indifferent to the victims. It requires only that deception be cheaper to commit than to catch, which, in the lag between every new way of making money and the world’s slow learning to check it, it usually is.
That lag, and the boom that widens it, are not confined to fraud. The same trusting, leveraged, audit-light optimism that lets the bezzle swell is the condition of the wider cycle itself, the long calm that ends in a crash, and the swindles the crash uncovers are often just the most criminal edge of a much larger and mostly legal build-up of risk. A later chapter in this volume turns from the scam that hid inside the boom to the boom and bust themselves, and asks why an economy that has survived a dozen crises keeps arranging the next one.