Mercy has long been a hallmark of a just legal system. Judges are often given considerable leeway in determining punishment for a crime (except in cases like mandatory minimums or “three-strike” legislation) for exactly this reason. Extenuating circumstances can result in a lower punishment for some criminals than others who commit certain crimes, and so on.
Some claim that mercy is a weakness, however. They want no mercy, at least for certain crimes. So called “tough on crime” politicians, for example. The list of crimes where mercy is a weakness is long and variable: illegal immigration, pedophilia, murder, rape, treason, drug dealing, prostitution, etc. The idea is if you greatly increase the penalties (up to and including the death penalty), you get less crime. The empirical evidence of deterrence is mixed and probably not going to be resolved any time soon.
But while total effects are important, what about the mix of crimes? Could tough-on-crime legislation create a mix toward more violent crime even if the total crime rate is falling? The economic way of thinking gives us reason to think so.
Let us assume that a criminal is economically rational. That is, a criminal will only conduct a crime if the marginal benefit of the crime exceeds the marginal cost in their estimation. The marginal cost of the crime will subsequently be the expected punishment and the marginal benefit is whatever benefit the criminal gets by committing the crime. The expected punishment is the probability of getting caught times the punishment if caught. By this assumption, we can see that a tough-on-crime stance could deter petty crimes. For example, if the punishment for stealing $100 is a $10,000 fine, then even a 1.1% chance of getting caught would deter the rational criminal: marginal benefit = $100. Marginal cost = 0.011*$10,000 = $110. $110 exceeds $100, so the rational criminal would not commit the crime. At 1%, the criminal is indifferent between committing the crime and not.
If the criminal does act and is caught, they face a choice: surrender and pay the fine, or resist and get a heavier sentence. For the rational criminal, the relative cost for surrender is lower than that of arrest. He will thus surrender.
But let’s change the scenario and have an extremely tough-on-crime policy. Let’s say that the legislature, to combat crime, orders that all crimes are punishable by death. One might think such a policy would deter crime. After all, the marginal cost has dramatically increased. But I argue not necessarily; it would change the mix of crime toward violent crime, since it reduces the cost of violent crime relative to that of lesser crimes.
Let us again look at the criminal who aims to steal $100. He attempts to commit the robbery but gets caught by a police officer. The criminal now faces a choice: he can resist arrest (say, by shooting the police officer) or he can submit to arrest. If he resists, let us say there is a 10% chance he successfully escapes. Under the tough-on-crime policy currently in place in this hypothetical, it is rational for the robber to resist arrest. Let us see why:
Option 1: Submit to arrest
Marginal benefit: none
Marginal cost: 100% chance of death
Result: 100% chance of death
Option 2: Resist arrest
Marginal benefit: 10% chance of escape
Marginal cost: none
Result: 90% chance of death
Option 2 is the better option here for our criminal. In the first option, he will die. No ifs, ands, or buts. In the second option, he has at least some chance of survival. The cost of resisting relative to surrender has fallen when compared to the pre-tough-on-crime policy. There is no marginal cost to the criminal as he faces certain death if he surrenders. So, paradoxically, the tough-on-crime policy could encourage violent crime by reducing its relative cost.
So, from an economic perspective, there is a case to be made for mercy. Mercy lowers the cost of surrender relative to resistance, encouraging more criminals to peacefully surrender. Conversely, a tough-on-crime policy regime increases the cost of surrender relative to resistance. Those poor people who are detected and caught are doomed; fighting their way out is the cheaper option now.
A tough-on-crime policy could reduce the total number of crimes committed. On the margin, committing petty crimes is cheaper when compared to committing no crimes. But once a crime is committed, the choice calculus changes to encourage more violent behavior. A merciful policy could result in more crimes in total, but the mix would be less violent as the options to resist or commit more violent crimes are more expensive. From an economic perspective, mercy is a good thing. Thus, we are left with the question: is it better to have a (relatively) small number of violent crimes or a (relatively) large number of petty crimes?
PS, there is a Japanese anime that deals with these issues called Psycho-Pass. The basic plot is that a government system judges people’s “crime coefficients,” or how likely they are to commit crimes. They are arrested, or if their crime coefficient is sufficiently high, executed even without committing a crime and without a trial. In the first episode, a man with no criminal record is determined to have a high crime coefficient, so his arrest is ordered. The man realizes this and decides to kidnap and attempt to rape a woman because he is going to jail either way. In this case, the system designed to reduce criminality ended up increasing it.
READER COMMENTS
nobody.really
Apr 11 2025 at 4:43pm
All fine game theory.
But does this theory depend on the idea that people make rational decisions about whether to engage in crime? In the case of the professional thief (or the perrenial tax-evader), perhaps they do. But I suspect much crime is driven by people with low impulse control. Admittedly, perhaps such people DO make rational decisions, but merely weigh short- and long-term consequences differently than most people. In that case, perhaps it still makes sense to structure punishments to reflect rationality. But perhaps not.
One theory about the decline of violent crime in the US is that a small share of the population commits a disproportionate share of the crimes–and you can capture that small share if you’re willing to lock up ENOUGH people for ENOUGH time. Even if the threat of strict punishment prompts criminals to engage in desparate measures, the strategy may prove optimal if it results in incapacitating these criminals when they are caught.
That said, here’s a theory for mercy that does not depend upon the rationality of the criminal: I understand that, in the US, it costs less on average to incarcerate a prisoner for life than to execute one. This reflects that cost of litigating cases through mulitple rounds of appeals. Under such circumstances, a rational public policy would not provide for executing a prisoner unless the prisoner proved to be abnormally expensive to incarcerate (say, abnormally violent or abnormally sick).
As a bonus, research shows that most violent crime is committed by young-ish men. Thus, incarcerating a man until he ages out of his violent impulses (35? 45? 55?) might suffice. Releasing prisoners thereafter would spare the public both the cost of the execution and the cost of running geriatric prisons. Ultimately, the policy is designed to show mercy to the taxpayer.
Jon Murphy
Apr 12 2025 at 8:06am
Sure. As do pretty much all theories based on crime. And supported by evidence. Criminals are just like regular people. Will there be those who act irrationally? Sure. But the theories of rational choice, that people respond to incentives, is sufficiently general.
MarkW
Apr 13 2025 at 9:59am
There are good reasons to think even the most depraved criminals are rational about their crimes. Decades old serial killers are being caught now due to DNA testing. It appears to be no longer possible to get away with a ‘career’ of being a long-term serial killer. And when is the last time you heard of *new* ‘Night Stalker’ or ‘BTK killer’? Also, up until reliable forensic pathology existed, poisoning was believed to be a much more common form of murder. But again, for rational reasons, it declined.
Bill Conerly
Apr 11 2025 at 6:34pm
Good commentary, Jon.
We economists rely on the concept of diminishing marginal utility. The prison time equivalent is diminishing marginal disutility. That is, your second year in prison is not as bad as your first; your third not as bad as your second. (I’m sorry that I can only hypothesize; no personal experience in the clink.)
Expected value is much harder to apply when we don’t know an individual’s or group’s disutility function. Maybe it’s logarithmic, sort of. Hard to nail down the parameters.
Jon Murphy
Apr 12 2025 at 8:08am
Thank you Bill.
Interesting comment. I hadn’t thought about the diminishing marginal did utility angle.
Student
Apr 11 2025 at 7:45pm
Interesting way to make these points.
Mactoul
Apr 11 2025 at 9:51pm
You are confusing mercy with justice. The proper punishment of a particular crime comes under preview of justice. Justice is what a person deserves. And criminal justice treats a criminal as belonging to a particular class ( class of persons who have committed a particular crime)
Mercy is to particular individuals based upon specific circumstances. Strictly speaking, judicial system has nothing to do with mercy which is an executive prerogative.
But loosely, judges can and do take note of extenuating circumstances, which does take character of mercy.
In one sense, to mete out justice, one needs to know the person as a person, not as a member of a particular class ( which is what the criminal justice involves).
In that case, which is a God-like view, mercy merges with justice. But the criminal justice and the judges etc aren’t God.
Jon Murphy
Apr 11 2025 at 10:18pm
No sir. As I state in the very first sentence, I see mercy as an aspect of justice.
Mactoul
Apr 12 2025 at 11:33pm
Mercy is particular but you are arguing for a general leniency in sentencing.
How would you distinguish mercy from leniency?
Jon Murphy
Apr 13 2025 at 8:22am
No I am not. Please reread more carefully.
Warren Platts
Apr 12 2025 at 6:05am
The recent example in El Salvador proves that getting tough on crime reduces crime..
steve
Apr 13 2025 at 9:21am
China under Mao and Cuba under Castro also had little crime. Low crime rates under authoritarian rule arent that unusual, if you are willing to pay the price.
Steve
Jon Murphy
Apr 13 2025 at 9:49am
Please reread my last paragraph (before the post script) more carefully
Jose Pablo
Apr 12 2025 at 7:50am
Interesting post, Jon.
But are “tough on crime” policies truly designed to be tough on crime, or merely to appear tough on crime?
There’s a much more effective way to combat crime: by increasing clearance rates. In the U.S., these rates are shockingly low, around 50% for homicides, one-third for rapes, and a staggering 10% for property crimes. Essentially, if you’re an above-average intelligent thief, you can be pretty sure of going unpunished.
Improving clearance rates can deter crime just as effectively, without the harmful side effects that often accompany performative “tough on crime” approaches. The problem, however, is that this requires the government to actually do things, to become more competent at solving crimes, rather than simply passing punitive legislation.
Anyone can draft harsh laws. It takes real institutional capability to solve crimes effectively.
Unfortunately, the incentives in government often favor the very easy and easy to market to voters, passing stricter laws over the harder, less visible work of improving investigative and prosecutorial abilities.
Jon Murphy
Apr 12 2025 at 8:03am
A legitimate question. I’m assuming here that they are not just theatre.
Bill Conerly
Apr 12 2025 at 4:31pm
Jose makes a good point. But implementation is very complicated.
Arrests depend on city and county police and sheriffs. Dialing up their activity to increase clearance rates and eventually decrease incidence of crime requires an up-front expenditure in hopes of for long-term gains.
Not only that, but the long-term budget gains accrue to the state government, which is responsible for prisons; while the enforcement costs are borne by local governments. No easy solution. Maybe the state provides a rebate to cities/counties that reduce need for prison space? Maybe policing becomes state-financed rather than local-financed? Or maybe the state charges cities/counties for the prisoners they send up? But it’s clearly nonsensical to have one level of government send people to prison and another level foot the bill.
This would be a good research project for a young economist.