Tuesday, April 24, 2018

Outrage vs Deterrence

Yesterday the Delhi High Court asked the Government of India a difficult question. The context is the Criminal Law (Amendment) Ordinance 2018, recently approved by the Union Cabinet and the President of India, which increases the severity of punishment for rape of minor girls. Coming soon after some ghastly recent reports of this crime, it may seem perfectly legitimate to amend the law in this way. Public outrage is at an all-time high on this issue, and with good reason.

So what could go wrong if rapists, particularly those who prey on minors, are put away for eternity or even executed? Here is what the honourable Court asked the Centre: "Did you carry out any study, any scientific assessment that death penalty is a deterrent to rape?". The Court had in mind that when any crime is elevated to a capital crime, the probability of murder of the victim (even if that was not initially intended) actually goes up. To my knowledge this was first observed by the Italian jurist and penologist Cesare Beccaria, to whom I'll return below.

The present post is not specifically about the death penalty, a controversial topic that I would like to leave for another occasion. I would prefer to focus on a more general scientific question: when we increase the severity of a punishment, does this have a greater deterrent effect on the associated crime? Put simply, if the penalty for burglary goes from one year in jail to ten years in jail, will there be less housebreakers? If so, will their number fall by a factor of ten? And are there other factors that are more effective in deterring crime, than merely increasing its severity? In several years of online discussions, articles and rants about crimes in India, I've rarely - if ever - seen this question even posed. This is not for lack of studies on the subject, but simply represents the sad state of contemporary society: everyone has strong opinions but not many want to read and reflect.

In 1764 Beccaria published a lengthy treatise, "Dei delitti e delle pene" (On Crimes and Punishments) of which a nicely formatted English translation is available here. In Chapter 27 he informs us that "Crimes are more effectually prevented by the certainty, than the severity of punishment." He observes that "If punishments be very severe, men are naturally led to the perpetration of other crimes, to avoid the punishment due to the first." This is essentially the point raised by the Delhi High Court in its query described above.

In support of his thesis, Beccaria observes that "it is the nature of mankind to be terrified at the approach of the smallest inevitable evil, whilst hope, the best gift of Heaven, hath the power of dispelling the apprehension of a greater". In other words, if punishment is guaranteed then criminals will be more afraid, but where there is a chance of evading punishment they may decide that the crime is worth the risk. This means the certainty of being punished is the major deterrent.

Much research has been carried out since 1764. Not being an expert in the field I cannot summarise it all, but will only convey some impressions. In modern times the discussion is framed under the heading of "Deterrence Theory". The parameters typically discussed are: (i) severity of punishment, (ii) certainty of punishment, (iii) celerity of punishment (i.e. the speed at which punishment is implemented). Assuming these three to be independent parameters, one can do case studies to see how much their variation actually deters crimes. (Even without case studies, I will guess that celerity is the biggest casualty in the Indian justice system, followed closely by certainty).

Such studies have revealed curious paradoxes. For example, some authors found in the year 2000 that "a fine for a previously unfined behavior may increase, rather than decrease, the unwanted behaviour. This happens as the fine replaces a previous set of moral or ethical norms." Other studies appear to contradict these findings. The deterrence impact of a given punishment clearly depends on the type of crime under discussion, since different crimes are committed by different types of criminals (compare the type of person involved in armed robbery, bank fraud or environmental crimes). A risk-averse person, or someone with a respectable public image, would respond strongly to certain but mild punishment, while a hardened criminal may respond more to the severity.

There is yet another complication. We assumed severity, certainty and celerity to be independent variables, but they might in fact be interdependent. This fascinating 1972 article published in the Journal of Criminal Law, Criminology and Police Sciences, questions a standard assumption of deterrence theory: that severity and certainty are "additive factors". In other words, is it even possible to increase both severity and certainty together? And if we try to do so, will we achieve the combined effect of increasing either one separately? The authors point out that in some cases, "severity and certainty of punishment are inversely related". Their own study, while not completely conclusive, offers "a fairly consistent inverse trend in the relationship between ... severity and certainty". It is not hard to guess why such a relationship might exist. Suppose there is the possibility of convicting a suspect on somewhat scanty evidence, then a judge would be more hesitant to do so if the punishment is severe, since there could be a major adverse effect on a possibly innocent person. In recent years, DNA tests have been carried out in the US on people who were executed for murder at a time when such tests did not exist. In many cases the new tests have completely exonerated the convicts. In the backdrop of these results, any decent modern judge would have greater reason to hesitate when the sentence is harsh.

The short conclusion is that it's complicated. But I do want to stress that howling in outrage about crimes, and rationally studying which parameters in our criminal justice system need to be modified to best prevent future crimes, are two distinct activities. The first one appears to make (some) people feel better while the second one will actually prevent future rapes and murders. We need to decide which one is the goal.


Meta Dynamic Systems said...

Though after the narrative, this sounds obvious, and the realization that the issues and rationale indeed were going round and round in the back of the mind, the blog helped me put them together cohesively, particularly useful when involved in a discussion. Another blog that gives some scathing case studies, specially effective ones would make things better and clearer.

S said...

The ideas that the efficiency, and not the severity, of legal punishments is what matters, and that more severe punishments can induce more crime, are both much older and occur, e.g., in the Arthashastra (as cited and analyzed here: http://acorn.nationalinterest.in/2008/10/12/reading-the-arthashastra-the-rule-of-law/, English translation here: https://en.wikisource.org/wiki/Arthashastra/Book_I)


"No," says Kautilya; for whoever imposes severe punishment becomes repulsive to the people; while he who awards mild punishment becomes contemptible. But whoever imposes punishment as deserved becomes respectable. For punishment (danda) when awarded with due consideration, makes the people devoted to righteousness and to works productive of wealth and enjoyment; while punishment, when ill-awarded under the influence of greed and anger or owing to ignorance, excites fury even among hermits and ascetics dwelling in forests, not to speak of householders.


A rather interesting feature for me in that excerpt is the explicit taking down of the notion of punishment as an outlet of anger.

A somewhat similar but more narrowly applied idea also appears just a few decades after Beccaria, in the Eight Amendment to the constitution of the USA