A financial adviser from Chicago stands in front of a cartel accountant and does something almost nobody in his position does: he explains, with numbers, why he is more valuable alive. He proposes a jurisdiction, a set of cash-intensive businesses, a volume, and a timeline. It works, and it keeps working for four seasons, because he is genuinely good at it. The show is frequently described as being about a man who becomes a criminal. Read through the theory that explains him best, it is about a man who did not have to change his method at all.
// The theoryAn equation, and the parking ticket behind it
Gary Becker's “Crime and Punishment: An Economic Approach” (1968) is one of the most consequential papers ever written about crime, and its origin story is the best summary of it. Becker was late for a doctoral examination and had to decide whether to park illegally: he weighed the chance of a ticket against the cost of the fine and the value of his time, parked illegally, and realized while walking that he had just modeled the thing criminology claimed was pathological. The paper formalizes it. An individual offends when the expected utility of the offense exceeds the utility of the alternative use of their time and resources, where expected utility depends on the gain, the probability of apprehension, the probability of conviction given apprehension, and the severity of the sanction. Three consequences follow immediately and all three remain influential. Offenders are not distinctive — they face different prices, not different natures. Certainty and severity are substitutes in the equation, which is why raising one can offset lowering the other. And the optimal amount of crime is not zero, because enforcement is costly and there is a point past which preventing the next offense costs more than the offense does.
Laundering is the purest test case
Money laundering is the offense the economic model handles best, and the reason is structural: it has no victim in the room, no emotion, and an explicit success metric. The practitioner's problem is to move a known quantity of cash into the legitimate economy at a cost—losses, fees, exposure—lower than the value of the cash, and every decision in it is a genuine optimization. Which businesses absorb cash without implausible margins? A strip club, a funeral home, a marina, a casino — Marty's portfolio is selected exactly as an investment portfolio is, for absorption capacity and covariance. What ratio of declared to actual revenue survives inspection? How much reporting threshold can be avoided without the avoidance itself becoming the signal, which is structuring, an offense in its own right precisely because the regulator anticipated the optimizer? Becker's framework does not merely describe this behavior; it is the framework the regulators use, which is why anti-money-laundering regimes are built around thresholds, reporting duties and the deliberate raising of transaction costs. Marty and his opponents are running the same model from opposite sides.
// The subjectThe variable he keeps getting wrong
What makes him a case rather than an illustration is where the model breaks, and the series is disciplined about it. Marty's estimates are excellent for everything that behaves like a market and catastrophic for everything that behaves like a person. He can price the probability of an audit and cannot price Darlene Snell, whose decisions are driven by honor, territory and insult rather than by return. He can model cartel logistics and not his own wife's ambition. Each of the series' major reversals arrives through an actor whose utility function is not the one he assumed — which is, precisely, the standard criticism of the economic approach, dramatized over four seasons. The model is not wrong about him. It is wrong about the people he is embedded with, and he is the last to notice because his instrument has no place to record it.
There is a second observation worth making about his trajectory, because it cuts against the show's own mythology. Marty does not descend. He is, from the first episode to the last, doing the same activity with the same tools for progressively larger clients, and the moral distance he travels is almost entirely a matter of what he is now willing to have happen at the far end of a transaction he does not witness. Becker's model is silent on that, and the silence is the point of the next file in this hub.
Marty Byrde
Get the map of theories for free
Subscribe to the criminal archive —one case a week, no spam— and I'll send you The map of theories: the great currents of criminology on a single sheet.
// The crackThe most useful wrong theory in the field
The economic model has four well-established problems. First, the numbers do not support the policy it inspired. If offenders respond to expected cost, raising severity should deter; the evidence, reviewed repeatedly by Daniel Nagin, is that certainty effects are real and severity effects are weak to nonexistent, which is an asymmetry the basic model does not predict and which mandatory-minimum policy ignored for thirty years. Second, offenders are poor estimators: the perceptual deterrence literature finds that people's beliefs about detection probability are badly calibrated, updated erratically from personal experience rather than from actual rates, and frequently formed after the decision. A model built on expected probabilities is fragile when nobody knows the probabilities. Third, discount rates and impulsivity do more work than the framework allows: a great deal of offending is committed by people who heavily discount the future or are intoxicated, and once you build that in, the model starts to look like the low-self-control theories it was meant to replace. Fourth, the framework is silent on preferences — it takes the desire as given, which is fine for a parking space and useless for explaining why one person wants what another does not. What survives is substantial and worth being honest about: for planned, acquisitive, organisationally embedded offending — fraud, laundering, corporate and tax offenses — the model performs well, because those really are decisions made by people who estimate. It performs badly for exactly the offenses that dominate the public conversation. And its greatest practical legacy is not deterrence at all but design: the entire anti-money-laundering apparatus is Becker applied, raising transaction costs until the optimizer optimizes elsewhere.
Nothing to diagnose
Marty launders hundreds of millions of dollars for an organization that kills people, and the fact that he never touches a weapon is a feature of the division of labor rather than a mitigation. The model explains him better than any psychological account would, and the uncomfortable part is what that implies: there is nothing to diagnose. He is not compensating, not damaged, not seeking thrill. He found an arbitrage and he is good at it, and the only thing that ever stops someone like him is a change in the arithmetic. That is not a comforting conclusion, and it is the reason enforcement against this kind of offending looks like accounting rather than like policing.
Three key ideas
- Becker (1968): a person offends when expected utility exceeds the alternative, given gain, probability of apprehension, probability of conviction and severity. Offenders are not distinctive, certainty and severity are substitutes, and the optimal amount of crime is not zero.
- Laundering is the model's best case: no victim in the room, an explicit metric, and portfolio decisions selected for absorption capacity. Regulators run the same model from the other side, which is why thresholds and reporting duties exist — and why structuring is itself an offense.
- Its failures: severity does not deter as predicted, offenders estimate probabilities badly, impulsivity and discounting do unmodelled work, and preferences are taken as given. It works for planned acquisitive offending and badly for everything else.
Sources · to keep reading
- Becker, G. S. (1968). “Crime and Punishment: An Economic Approach”, Journal of Political Economy, 76(2).
- Ehrlich, I. (1973). “Participation in Illegitimate Activities”, Journal of Political Economy, 81(3).
- Nagin, D. S. (2013). “Deterrence in the Twenty-First Century”, Crime and Justice, 42(1).
- Levitt, S. D. (2004). “Understanding Why Crime Fell in the 1990s”, Journal of Economic Perspectives, 18(1).
- Paternoster, R. (2010). “How Much Do We Really Know about Criminal Deterrence?”, Journal of Criminal Law and Criminology, 100(3).
- Levi, M. y Reuter, P. (2006). “Money Laundering”, Crime and Justice, 34(1).
Case-file questionsFrequently asked questions
Who is Marty Byrde?
Marty Byrde is a villain from Ozark, analysed on KRIMINIS through the lens of criminology. He does not ask whether it is wrong.
Why does Marty Byrde become a villain?
Marty Byrde and the economic model of crime: the financial adviser who launders with a spreadsheet. The file on Gary Becker's equation — and on why its author said it came out of a parking violation.
Which criminological theory explains Marty Byrde?
Marty Byrde's case is analysed through Economic model of crime. In 1968 Gary Becker published the equation that won him a Nobel and still organizes criminal policy across much of the world: a person commits an offense if the expected utility exceeds that of devoting the same time and resources to something else. No pathology, no subculture — preferences, prices and probabilities. Marty Byrde is that model with a body, and the interesting part is not that he launders money but that he does it with the same instruments he used to advise on retirement funds: estimate risk, diversify, work out the threshold past which an operation stops being worth it. This file asks how far that equation reaches and where it starts to fail.
