Here’s a personal tidbit about me – I’m a bit of a true crime buff. I often listen to true crime podcasts and audio essays while unwinding at the end of the day, or during my time in the gym. (My wife finds the idea of listening to podcasts talking about horrific crimes to be an incomprehensible method of winding down, nor does she understand how I can possibly enjoy horror films.) One thing I’ve noticed is that in many of the cases I’ve heard, the criminal drastically overestimates their odds of getting away with the crime.
One of my favorite shows in the genre is called The Casual Criminalist. In this show the host, Simon Whistler, is provided with a script from one of his writers that he’s never seen or read before recording. He then reads it off for the first time on the show, frequently breaking from the script to add in some of his own side commentary. Very often, this takes the form of him absolutely roasting the criminal featured in the episode for their ineptitude both in committing their crime and their attempts to cover up what they did, such as this case involving a young man. More famously (or infamously), there is the case of Leopold and Leob, two young men convinced they were capable of masterminding the perfect crime, but whose actual performance was didn’t exactly rise to the level of Professor Moriarty.
In a recent post I once again touched on Gary Becker‘s model of criminal behavior. To briefly recap, Becker modeled choices about committing crimes as a form of rational behavior. Criminals consider the expected payoff of their crime, and compare it the expected cost of committing the crime. The expected cost is a function of the probability and severity of punishment. A severe punishment might provide little deterrence if the odds of being punished are minuscule. And near-certain punishment might also provide little deterrence if the punishment itself is trivial. To put more specific numbers to it, a 0.1% chance of a $1,000 penalty provides as much deterrence as a guaranteed $1 penalty – not much. In the case of Leopold and Leob, even though their crime carried the possibility of a death penalty or a life sentence, their (false) certainty that they could evade detection nullified any deterrence.
This highlights an important point. When speaking of criminals behaving rationally accounting for the probability of punishment, what’s relevant is the criminal’s own subjective estimation of the probability of being caught and punished. When economists talk about people making “rational” decisions, that does not mean their decisions can’t be mistaken or ill-informed. If a criminal drastically underestimates their likelihood of being caught, their estimation of the expected costs of the crime will be lower than the actual costs. As a result, they might end up carrying out a crime that is “not worth it” by their own lights.
This does not show that the criminal was behaving “irrationally” as economists use the term. But if criminals systematically underestimate their likelihood of being caught and punished for a crime, that would lead to criminals overproducing their crimes – criminal acts that might have been deterred if they had a better understanding of the likely outcome will fail to be deterred. This provides one argument in favor of stringent punishment for crimes. If criminals underestimate the expected costs of their behavior because they systematically underestimate the odds of punishment, the only way to raise the expected cost is to increase the severity of punishment.
But sometimes, criminals are able to more accurately assess their odds of getting caught. Consider the case of Joseph DeAngelo, known by many monikers, most famously as the Golden State Killer. His criminal activities carried on from 1974 to 1986. He was finally apprehended in 2018, over 30 years after his spree ended. DeAngelo, it turned out, was a police officer. As a result, he knew far better than most how to avoid detection and how investigations would be carried out. He was very careful, for example, to never leave fingerprints behind. But there’s something particularly unusual about DeAngelo – the 30+ year gap between when his crime spree ended and him being caught. In those years, he didn’t strike again. This is noteworthy – serial killers almost never stop. A serial killer who stops killing is so unusual that it’s usually taken as a sign that the killer has themselves died or perhaps been arrested for some other crime.
So why did DeAngelo stop? Well, shortly after his final act as the Golden State Killer, something happened that made waves in the law enforcement world. The first court case where DNA evidence was used to secure a murder conviction occurred. This put DNA evidence on the radar for law enforcement officers everywhere – including DeAngelo. And he realized what this meant. While he had taken great care to avoid leaving evidence that police investigators could link to him up until that point, he also knew his DNA would be present at previous crime scenes and very likely new ones as well. Suddenly, the “probability of punishment” variable for him shot up drastically, and combined with the quite high severity of punishment that would await him for his acts, carrying out further crimes was suddenly “not worth it” anymore.
And thus a very brutal and pathological serial killer whose criminal acts had been intensifying and escalating for over a decade suddenly stopped. This makes perfect sense under Becker’s theory of crime. And while rational choice theory isn’t a perfect model to describe all human behavior in all circumstances, I think this shows that it explains a much larger scope than most give it credit for.
READER COMMENTS
David Henderson
Jan 8 2025 at 1:31pm
Great story.
One correction. He was apprehended 32 years, not 40 years, after his crime spree ended.
Craig
Jan 8 2025 at 11:19pm
They caught him because they ran his DNA through a genealogical database and they found two relatively distant relatives. It was something like third cousins even where nobody was even aware of the relation, but they got two hits and then researched the family of both looking for overlaps and that left 5 people and then just used police work to eliminate before deciding DeAngelo was probably the killer after which they grabbed something he tossed away like a soda can and confirmed it.
Kevin Corcoran
Jan 9 2025 at 12:56pm
Aaaaagh, thank you for catching that. I have no idea how I let that happen. I’m going to blame gremlins. (You don’t hear them denying it, do you?)
David Henderson
Jan 9 2025 at 1:22pm
You’re welcome, Kevin.
Monte
Jan 8 2025 at 9:39pm
RCT certainly has predictive value in modeling many types of criminal behavior where the payoff is money, power, or revenge. Even a serial killer, as in the unique case of Joseph DeAngelo, can exhibit some level of rational behavior (see Spatial Patterns of Serial Murder). But RCT struggles in its ability to predict criminal behavior when applied to cases of mass murder and terrorism.
Mass murderers and terrorists operate in a vacuum void of rationality. They are utility maximizers only in the sense that they attempt to kill as many as they can for the sake of killing. A rational choice framework involving material gain or self preservation obviously doesn’t apply to actors whose revealed preference is death or infamy.
Jon Murphy
Jan 9 2025 at 5:54am
Why doesn’t rational choice theory apply? You present a good case that it does, but then say it doesn’t. You say it struggles to provide an explanation, but then provide that explanation. I don’t understand your point.
Monte
Jan 9 2025 at 11:19am
Because death and infamy are objectives or preferences that defy rational self interest in the traditional sense. RCT assumes that actors are rational in that they seek to maximize rewards and minimize risks. Self-destructive behavior is antithetical to what RCT would predict, that a rational actor would choose the utility-maximizing options of self preservation or material gain. Notoriety and self sacrifice , unlike wealth, status, or power, are non-rational, intangible benefits that RCT doesn’t fully account for.
Jon Murphy
Jan 9 2025 at 12:33pm
Nope. They’re perfectly consistent with rational self-interest.
Monte
Jan 9 2025 at 4:59pm
You and Kevin both are expanding the boundaries to include non-rational motivations, which is a conceptual shift away from traditional RCT. Death and infamy are non-rational goals that don’t fit into the framework of self-interest or utility maximization as defined by traditional RCT. This is a standard criticism of RCT by many economists. That’s why they attempt to fill in the gaps with things like emotional choice theory or theoretical pathology.
Jon Murphy
Jan 9 2025 at 6:09pm
I don’t think we are. There’s a large literature on how suicide, crime, etc can be rational.
Monte
Jan 9 2025 at 7:07pm
Ok, so you’re redefining traditional RCT to include non-material rewards like honor, revenge, and martyrdom. I can accept that, but the initial premise of my argument was that RCT was not traditionally applied to these types of motivations.
Kevin Corcoran
Jan 15 2025 at 3:59pm
Well that’s just mistaken – traditional RCT has always included non-material considerations. Nowhere in the RCT literature will you ever find it stipulated that material considerations are the only things people seek to maximize. Pretty much the only place you’ll find people saying that is straw-man summaries of what economists say, written by people who have never actually read any of the literature.
Kevin Corcoran
Jan 9 2025 at 1:08pm
Who says the “rational choice framework” must involve “material gain or self preservation”? The whole point here is that what is being valued within the “rational choice framework” depends on the specific preferences and views of the subject making the decision. If someone highly values material gain and self preservation, they will make very different choices than someone who places little or negative value on material gain and self preservation. That doesn’t mean one person is more “rational” than the other in the context of rational choice theory – it means that the “rational” choice for each will be very different.
No, that’s simply false. RCT assumes that people’s choices are rational in that their choices are means-ends oriented. That’s all. RCT carries no assumptions about what specifically people value or what they seek to maximize – there is no assumption that people seek to “maximize” their “material gain” or to “minimize” anything risking their “self-preservation.”
Again, no. RCT tells us that if a person highly values self preservation and material gain, then in that specific context we would not predict (intentional) self-destructive behavior. (They might make choices that do have those effects because they were ill-informed about the facts, but that’s also why I made the point that “rational” does not mean “well-informed,” nor does it mean “virtuous.”) If someone does not value those things, or places negative value on them, then RCT absolutely would predict such behavior.
Monte
Jan 9 2025 at 7:17pm
Based on this universally accepted, formal definition of RCT:
Your argument totally ignores this side of the RCT equation. What losses do lone wolf mass murderers and suicide bombers try to minimize?
Kevin Corcoran
Jan 15 2025 at 3:37pm
You’re still missing the point, methinks. Yes, rational choice theory says people try to maximize their gains and minimize their losses, but RCT by itself does not say anything about what constitutes gains or losses. If someone places negative value on life (or positive value on death, because of a nihilistic outlook or a belief in an afterlife paradise awaiting them or whatever) then RCT would easily account for “lone wolf mass murderers and suicide bomber.” And, in fact, many people do place a negative value on life, even their own, and place a positive value on death.
I think you’re making some tacit assumptions that certain things can only and for absolutely everyone be put into the gain or loss columns, but the whole point here is that anything can go into either column – based on the beliefs and preferences of any given subject.
Monte
Jan 9 2025 at 7:58pm
This, from Dr. Jeff Victoroff (USC Dept of Psychiatry):
And some terrorists (e.g., “lone wolf” terrorist Theodore Kaczinski) commit violence due to unequivocally irrational motives (in his case, paranoid schizophrenia). Thus, the rare and idiosyncratic decision to become a terrorist [and I think we can include Shamsud-Din Jabbar here, as well as mass murders Stephen Paddock, Nicolas Cruz, and Brandon Scott Hole] cannot be explained by rational choice theory.*
* The Mind of the Terrorist: A Review and Critique of Psychological Approaches
steve
Jan 10 2025 at 11:28am
RCT probably applies to most planned crime. Even when a schizophrenic commits a terror act it’s something that makes sense from their POV. However, a lot fo crime, especially violent crime is impulsive done out of anger often affected by alcohol or drugs.
AS an aside there is pretty good data looking at IQ of those convicted of crimes and they run below average. Most criminals arent that bright.
Steve
Kevin Corcoran
Jan 15 2025 at 3:39pm
Yes, you have successfully demonstrated that what psychiatrists mean by “rational” is not the same as how “rationality” is defined within economics and how it’s applied to RCT. But the fact that psychiatrists and economists might use the same term to mean different things was already common knowledge, so you’re simply trying to kick down a wide open door here.
Thomas L Hutcheson
Jan 8 2025 at 10:09pm
This suggests police departments improving and touting their success record woud be a good deterrent.
Craig
Jan 8 2025 at 11:16pm
“In those 40 years, he didn’t strike again.”
Are they confident about this. I had actually seen this before and I also speculated that his awareness of DNA testing led him to stop, but I am curious if that was confirmed or simply inferred from the circumstances? I mean, another inference is that he changed his tactic so that the victim/his DNA would never be found at all?
Billy Kaubashine
Jan 9 2025 at 3:04pm
I agree with Thomas Hutcheson’s comment.
Publicity is helpful. Potential criminals need to be aware of the likelihood of being caught and punished.
Toby Keith had the solution in “Beer for My Horses”
Take all the rope in Texas find a tall oak treeRound up all them bad boys, hang them high in the streetFor all the people to see.
Thomas L Hutcheson
Jan 9 2025 at 6:11pm
If you rounded then ALL up there woud be no one left to deter and so no need to hang them. 🙂 The more successful the roundup, the lesser the needed punishment to deter the un-rounded up.
David Seltzer
Jan 9 2025 at 7:26pm
Kevin: Great stuff. So why did DeAngelo stop? The first court case where DNA evidence was used to secure a murder conviction occurred. DeAngelo stopped because his prior probability of being caught changed with new (updated) information. To wit. DNA evidence. Maybe he was a closet Bayesian wherein his hypotheses was; the likelihood of being apprehended increased significantly, yielding a posterior probability.