Second, [Lee Kuan Yew] learned that fear of brutal punishment can deter crime:

The Japanese Military Administration governed by spreading fear. It put up no pretence of civilised behaviour. Punishment was so severe that crime was very rare. In the midst of deprivation after the second half of 1944, when the people half-starved, it was amazing how low the crime rate remained. . . . As a result I have never believed those who advocate a soft approach to crime and punishment, claiming that punishment does not reduce crime. That was not my experience in Singapore before the war, during the Japanese occupation or subsequently.

On the other hand, there are plenty of sources that claim that harsh punishment does not deter crime, e.g.:

And starting in the early 1990’s, crime began a two decade long decline that the public seems by and large not to have noticed. Yet there’s little evidence to suggest that the threat of punishment—even the threat of very harsh punishment, such as the death penalty—is responsible for the drop in crime. A massive 2014 study undertaken by the National Research Council announced that one of its “most important conclusions is that the incremental deterrent effect of increases in lengthy prison sentences is modest at best.” Put in less academic-ese: Threatening people with increasingly harsh punishments doesn’t discourage crime.

What explains the differences in the two viewpoints? I can think of many possible explanations, but I can't seem to find any sources to back them up - e.g. it's commonly asserted that it's the perception of how likely one is to be caught that deters crime, but I can't find any sources that say the Japanese Military Administration are very good at catching criminals. Similarly, Lee Kuan Yew's description of Japanese brutality implies that terror is effective at deterring crime, yet terror seems at least closely related to harsh punishments, and so is explicitly contradicted by the second source.

  • 19
    Is there any reason to believe that a population living under a brutal foreign occupation would behave the same way as a population living under a democratic government that raises prison sentences a bit to sell a tough on crime story to their electorate? The two situations seem so different that I see no conflict to explain in the paragraphs cited, even if I assume everything they say is 100% objectively correct.
    – Ben
    Commented Jul 19, 2023 at 3:48
  • 7
    In particular the second one says "the incremental deterrent effect of increases in lengthy prison sentences is modest at best" (emph mine). That fits exactly what I would expect; nobody committing serious crimes is going "X years is a price I'm willing to pay, but if it were Y then I wouldn't do it", so raising X to Y has little deterrent effect (when X is already high enough; more than a few years, say). But it's a totally different situation from suddenly changing the entire regime to impose extremely harsh penalties for even very minor crimes.
    – Ben
    Commented Jul 19, 2023 at 3:54
  • @Ben that makes it sound like the harsh punishment not deterring crime (as in the second source) is because the punishment isn't harsh enough, or alternatively that the punishment (length of prison sentence) isn't actually harsh. If that's the case, it would also answer the question.
    – Allure
    Commented Jul 19, 2023 at 7:51
  • 3
    Correlation is not causation and Yew was probably wrong about the causation. The Japanese were incredibly law abiding in the face of hardship long before WWII.
    – ohwilleke
    Commented Jul 19, 2023 at 18:25

7 Answers 7


This question is comparing apples to oranges. To quote a highly upvoted comment, which should really have gone as an answer:

Is there any reason to believe that a population living under a brutal foreign occupation would behave the same way as a population living under a democratic government that raises prison sentences a bit to sell a tough on crime story to their electorate? The two situations seem so different that I see no conflict to explain in the paragraphs cited, even if I assume everything they say is 100% objectively correct.

Let me cite from memory a book I read about counter-insurgency warfare a long time ago. In it, the author said something along the line of:

Despite what is claimed, extreme repression can work to counter insurgencies. Nazi Germany did that with some success in its anti-partisan warfare on the Eastern front.

The book then goes on to discuss how, in the context of liberal and rule-of-law countries, anti-insurgent warfare is most effectively carried out with restraint.

Few objective people would claim that extreme punishment, doled out indiscriminately, without judicial procedures (and often against merely bystanders *) would have no dissuasive effects. That's what Nazi Germany and Imperial Japan were up to.

Likewise, it is possible that extreme punishments like amputations deter car thieves.

Pretty sure castration would help limit sex crimes. Certainly would lower recidivism rates.

Singapore may very well have a low drug trafficking rate, given that it has the death penalty on the books for it. And is not averse to using it (11 executions in 2022, from 5.5M pop. compared to 18 in 2022 for the US).

But are those societies whose response we should evaluate when assessing how we run our criminal systems?

* In the aftermath of the Doolittle Raid, Japan's army may have killed up to 250k Chinese in the region where the B25s landed. That's what Lee is talking about.

  • 3
    +1 And I would add that both Nazi and Japan occupations were as brief as they were brutal, it is not clear if the terror regime would have worked persistently if it was permanent. This is an article commenting how, despite the harsh penalties, drug consumption in Singapore is (or was) still on the rise: thecabinsingapore.com.sg/blog/…
    – Rekesoft
    Commented Jul 20, 2023 at 9:10

Your second quote doesn't actually say that harsh punishment does not deter crime.

From the conclusion of the article:

[...]even if severely punishing people in order to deter crime would be ethically justified, to appeal to deterrence doesn’t seem sensible if increasing the punishments doesn’t decrease the crime.[...] The 18th century philosopher Cesare Beccaria hypothesized that whether punishment deters crime depends on its severity, certainty, and swiftness of imposition. If I am right, then perhaps our criminal justice system would be more effective if it concentrated on making punishment more certain and more swift rather than more harsh.

This introduces two additional criteria for our punishment. It has to be:

  • Harsh, so that it denies the perpetrator the profit he would be getting by committing the crime;
  • Swift, so that there's no time to use gained profit to offset the punishment;
  • And finally - it has to be certain, so that the criminal doesn't rely on the chance that he will just evade punishment, devaluing harshness completely.

So the point isn't that harsh punishment is useless, but that, on one hand, established punishment is already harsh enough, as demonstarted by diminishing returns gained from simply increasing the punishments, and on another - that harshness is undermined by uncertainty and/or ponderousness of the system responsible for its delivery.

After all, if a car thief knows that there's 99% chance he will be caught and put in prison - that's deterrence. But if there's 99% chance he will be uncaught and 1% he will get a death sentence - that's a gamble someone very desperate (or very arrogant) may well take.

Comparing that to the anti-insurgency tactics used by Nazi Germany and Imperial Japan - they achieved extreme harshness (death) and extreme swiftness (as soon as the troops can be directed to the site), and totally sacrificed certainty (attacks on basicaly random people) for fear factor - if the whole village is exterminated as a reprisal for guerilla attacks in the area, then it is more likely that at least some of the guerillas lose someone in the purge; and even if they don't - as other villages will know that guerillas' actions present a threat to their lives, they would be more active in helping find and/or trap them. As certainty is actually important within context of rule of law (you can't just punish random people hoping to hurt actual criminals by association), this scenario isn't really applicable.

  • "guerillas lose someone in the purge" - that purging was as much amongst the cause of guerrilla activity.
    – Steve
    Commented Jul 21, 2023 at 0:58
  • @Steve well, that was the reasoning for the attacks, whether actual effect matched the expectations of the planners is up to discussion. Note, however, that guerillas in areas where allied support was relatively available (France, USSR, China) fared much better than similar movements in South-East Asia (of which Singapore would be a prime example), where guerillas' powerbase was local and more vulnerable to reprisal attacks. Commented Jul 21, 2023 at 4:44
  • 1
    I think you need to be careful about conflating the deterrence of guerillas from their simple and complete destruction. The Nazis destroyed a lot of people, but overall they clearly were never able to bring deterrence to bear in excess of the undeterred opposition they provoked (with further opposition often provoked by the very measures which were supposed to deter further opposition). Harshness, swiftness, and certainty, were mentioned in the context of discussing the deterrence of criminals, not their destruction.
    – Steve
    Commented Jul 21, 2023 at 9:30

I would think of several factors right away

  1. Context and Culture: The context within which a punishment system operates is crucial. During the Japanese occupation, Singapore was under a military regime where civil liberties were suspended, and the fear of harsh punishments was pervasive. This is dramatically different from a peacetime democratic society where rights are respected and due process is observed.

  2. Certainty vs Severity: Research shows that the certainty of punishment can be a more effective deterrent than the severity of the punishment. If people believe they are likely to be caught and punished, they are less likely to commit a crime, regardless of the severity of the potential punishment. During the Japanese occupation, the certainty of punishment might have been perceived as very high, contributing to the low crime rate.

  3. Immediate vs Long-term Effects: Harsh punishments may deter crime in the short term, especially in highly controlled environments, but they are less effective in the long term, particularly when normalcy is restored. Long-term crime prevention is more successfully achieved through addressing underlying social issues, such as poverty, education, and unemployment.

  4. Type of Crimes: Harsh punishments might deter certain types of crimes more than others. For instance, they might be more effective in deterring crimes of impulse or opportunity rather than premeditated crimes or those driven by desperation or necessity.

  5. Fear vs Rehabilitation: Lee Kuan Yew's observations highlight the role of fear in deterring crime. However, a justice system built solely on fear is unlikely to be sustainable or desirable in the long term. Modern justice systems aim to balance punishment with rehabilitation, aiming to deter crime while also helping offenders reintegrate into society and reduce recidivism.


The answer is that sources claiming that policing doesn't deter crime are simply biased and try to follow the mainstream narrative in academia. Serious, unbiased studies show plenty of correlation between harsh punishment and the reduction of crime. For one recent example, see the case of Salvador:

The Salvadoran gang crackdown began in March 2022 in response to a crime spike between 25 and 27 March 2022, when 87 people were killed in El Salvador. The Salvadoran government blamed the spike in murders on criminal gangs in the country, resulting in the country's legislature approving a state of emergency that suspended the rights of association and legal counsel, and increased the time spent in detention without charge, among other measures that expanded the powers of law enforcement in the country.

As of 12 July 2023, 71,479 people accused of having gang affiliations have been arrested, which has overcrowded El Salvador's prisons. As of 16 May 2023, 5,000 people who were arrested have been released.


In January 2023, Minister of Defense René Merino Monroy announced that the government registered 496 homicides in 2022, a 56.8% decrease from 1,147 homicides in 2021. He attributed the decrease in homicides to the gang crackdown. That same month, the government opened the Terrorism Confinement Center (CECOT), a prison with a capacity for 40,000 prisoners

Harsh punishment works, because the objective reality is that someone behind bars can no longer commit crimes and if they spend a long enough time in prison they'd be too old to effectively commit crime upon release. The only issue with harsh punishment is that the false positive rate of prosecution can be quite high, so you pay for extra security with the lives of innocent men. This is why people in academia dislike this approach but moral questions aside there's no doubt that it works well.

  • I think the important fact to note is that perception of crime rates doesn't match reality. For example talk of crime is news and gets peoples attention but no talk of crime isn't news and doesn't get peoples attention. Even in the US crime is dropping but people think it is increasing. pewresearch.org/short-reads/2016/11/16/…
    – Joe W
    Commented Jul 21, 2023 at 15:08
  • 1
    @JoeW yes and crime did increase in 2020 due to less efficient policing in the US. Commented Jul 21, 2023 at 15:11
  • And there was no other factors that occurred in 2020 that could account for changes in crime?
    – Joe W
    Commented Jul 21, 2023 at 15:34
  • 1
    @JoeW yes but if policing stayed the same we'd likely see fuller jails but not as much extra crime overall Commented Jul 21, 2023 at 18:55
  • 1
    @ItalianPhilosophers4Monica it’s been done in NYC during the 1990s (broken window theory), likewise with good success. A complete suspension of due process would include rounding up every aggressive looking young man hanging out on the street - and I suspect it would likewise result in a huge reduction of crime. Commented Jul 25, 2023 at 16:10

There are a few related issues with punishment

  • How quickly after committing a crime is a person punished.
  • How likely is a person to be punished, and do people believe they will be punished. Eg if there are no ticket inspectors there is little risk of being punished for not having a ticket.
  • How brutal the punishments are.

It tends to be that counties with brutal punishments also have a high risk of being punished and a short time from crime to punishment. So the low crime may not be due to the brutal punishments.


First of all is there a sound basis for that statistic in the first place? Like obviously a crime statistic only includes the crimes that were recorder and reported and there are various factors for why that would not happen:

  • criminals might be more cautious about getting caught.
  • reduction of petty crimes, which are most numerous but least noteworthy.
  • the government might be lying about statistics to defend their terror as effective.
  • The population might cover up the criminals because the punishment is seen as excessive.
  • The media is not allowed to cover crimes
  • the government is becoming the biggest criminal organization, so people inclined to crime might join the government and perform crimes legally. You want to murder someone? Just accuse them of a crime and let the government do it for you.
  • Or if you can't trust the government to sort out victim and perpetrator you might see an increase in extrajudicial violence that also doesn't end up on the records.

And that list is probably still incomplete. Another set of reasons might be economic conditions. Like at least some part of the crimes committed stem from excess, inequality, status and necessity. So given that such extreme measures are often the result of fucked up situations you might enter a short period of a sweet spot where the excess is eroded, society is more egalitarian and thus status differences become negligible and which is just above necessity problems (yet).

Another set of reasons might be associated with totalitarian system that take away the privacy to both commit crimes but also to publicly enjoy the result of crimes, so that privacy itself becomes a crime. So that even if crimes still happen you have less of an awareness of them, because people just in general try to hide things and what is happening publicly is "clean". That again can for a short period of time decrease the public perception of crime, until people snap because that puts a lot of pressure on the individual that has to find a valve.

Not to mention that, as others have already mentioned, this isn't just a tuning of the system it can end up being an entirely different system with entirely different problems and you might just have the focus on the wrong parameters. Like if you exert all energy to solve one problem, chances are a lot of other problems pop up elsewhere. So is it really a good idea to reduce theft at the price of killing the economy by ramping up the police and thinning out the workforce thus reducing the economic output and the profit margin of thieves? Or is it better to accept some level of theft but also enjoy a higher standard of living in general and cover the biggest problems with insurances.


Your first case was a foreign occupation scenario. That is different from the normal scenario. Even the scenario will be different if a country has a repressive regime.

I personally believe that the degree of harshness of punishment doesn't have any effect on how people will behave.

In my observation, crime depends on the following factors:

  1. The promptness of justice - The more prompt the justice will be served the lower the crime will be. For example, in the USA, law enforcement is prompt but justice takes a long time. The same is in India and similar 3rd world countries where the law enforcement and judiciary are either corrupt, inefficient, or simply lackluster.

  2. Anthropology and Culture - The more culturally and ethnically diverse people will be, the more crime there will be. For example, immigrant-heavy countries have more crime in general.

  3. Politico-economic Situation - If we are done with the above two factors, the final thing boils down to the ongoing political and economic situation of the country. For example, Iran has a sufficiently harsh punishment regime. However, crime in Iran is high because of economic reasons.

  • All other things being equal, add the functioning of the mental health system to the list. The absence of mental healthcare brings some people into a situation which is more crime-prone than otherwise.
    – o.m.
    Commented Jul 19, 2023 at 20:15
  • @o.m., add the functioning of the mental health system to the list. --- No, I won't. This is essentially part of #3.
    – user366312
    Commented Jul 19, 2023 at 20:30
  • "in the USA, law enforcement is prompt but justice takes a long time" - the reason justice takes a long time is primarily because the system struggles in the first place to cope with the quantity of crime relative to the criminal justice resources it previously planned for. And shortcuts or arbitrariness simply generate riots or other swingeing costs for the system to cope with.
    – Steve
    Commented Jul 21, 2023 at 1:13
  • (1) sounds plausible, but is it true?
    – gerrit
    Commented Jul 21, 2023 at 14:01
  • @gerrit, Yes it is. In my country it is true. I am from South Asia.
    – user366312
    Commented Jul 21, 2023 at 17:58

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .