Tuesday, June 22

The expected costs of releasing sex offenders

An interesting post on a very contentious topic over at The Commons caught my eye. Author Jonathan McLeod weighs into the debate over whether we should change how we pardon sex offenders in Canada. He argues we should use hard numbers rather than emotions to make these decisions. Although like many Canadians I have strong emotions on this topic, the economist in me tends to agree with McLeod.

McLeod then describes some data he found, which suggests that sex offenders are less likely to re-offend than the average criminal. Thus, he concludes that "statistics demand that we offer no special punishment to sex offenders seeking pardons."

But after looking into this a little deeper, I think the data shows the opposite: releasing sex offenders poses more of a cost to society than the average criminal.

The key flaw in McLeod's analysis is that by looking only at the chance of re-offending, he is assuming that all crimes are created equal. This, I believe, is a very unreasonable assumption.

To do a proper analysis, one would need to obtain three pieces of information. First, we would need to know how likely it is that a sex offender re-offends compared with the average criminal. McLeod obtains this number from a Canadian Department of Justice report: 12.4% of sex offenders re-offend, compared with 30.1% of general criminals.

Second, we need to know whether sex offenders who re-offend commit different types of offences than general criminals who re-offend. Intuitively, one might expect sex offenders to be more likely to commit another sex crime than a general criminal. It turns out this is true. A study by the U.S. Department of Justice found that 5.3% of sex offenders were re-arrested for a sex crime within three years of being released from prison, compared with 1.3% of general criminals.

Third, we need to know whether all crimes are in fact created equal, as McLeod assumes. As economists, this means we probably want to attach a monetary cost to different types of crime. But coming up with the cost of a crime can be difficult. If someone steals my watch, there is the obvious cost of replacing my watch. But when word of the crime gets around, others will experience a psychological cost whereby they start looking over their shoulders in case the watch thief sneaks up on them. How do we calculate the cost this? And what about the security guards that watch stores must now employ to ensure no more watches get stolen?

Calculating the costs of crimes is not my area of expertise. But poking around online, it is clear that some people have attempted to do this. I found a 2000 study by the U.K. government that calculates the costs of various types of crime. They find that the cost of the average crime is £2,000. Not surprisingly, their cost estimate of the average sex offence is much higher: £19,000. Their estimates are based mostly on direct effects of crime. They do not take into account costs to the victim's quality of life, prevention costs, or costs of people fearing potential crimes. My guess is that this would cause the cost of the average sex offence to be underestimated relative to the average crime.

Putting the three pieces of data together (the chance of re-offending, the chance of re-offending with a sex offence, and the cost of sex offences relative to other crimes), we can then compare how releasing a sex offender compares with releasing an ordinary offender.

Let's assume a sex offender has a 12.4% chance of re-offending, and that they have a 5.3% chance of re-offending with a sex crime (obviously, one number comes from Canada and one from the U.S. Also, the first number presumably has no time frame on it, whereas the second number has a three-year time frame. But it's better than nothing). Putting these numbers together, we can assume that a sex offender will have a 5.3% chance of committing another sex crime, and a 7.1% chance (12.4 - 5.3 = 7.1) of committing a general crime.

We can multiply these probabilities by the cost of the two types of crime to see what the expected cost of releasing a sex offender is. The calculation is (0.053 x £19,000) + (0.071 x £2,000) = £1,149.

Now we can repeat for general criminals. They have a 1.3% chance of re-offending with a sex crime, and a 28.8% chance of re-offending with a general crime. That works out to an expected release cost of £823.

My method is by no means flawless. But these back-of-the-envelope calculations suggest that it is, on average, 40% costlier to society to release a sex offender than it is to release a general criminal. It appears the push to make it tougher for sex offenders to obtain a pardon is in fact supported by data.

3 comments:

  1. David,

    A few quick thoughts:

    1. Thanks for the further inquiry into this. You're quite right that I could have gone deeper in my research. (And, though I didn't phrase it as such, my post was meant to be an invitation to further discussion.)

    2. I disagree with this sentence:

    "Putting these numbers together, we can assume that a sex offender will have a 5.3% chance of committing another sex crime, and a 7.1% chance (12.4 - 5.3 = 7.1) of committing a general crime."

    This isn't backed up by the numbers. The 12.4% rate and the 5.3% rate come from different studies in different countries. You can't use them to accurately gauge the rate of rex-offenders re-offending with a general crime. The statistics in the reports I quoted in the original post clearly state (though I didn't quote the numbers) that sex offenders are quite unlikely to re-offend in any manner other than sex crimes.

    3. One Canadian study also broke down the recidivism rate for sex offenders with a previous sex crime conviction and re-offense (ie they'd been convicted, released, committed another crime, released, and then committed another crime). This might be useful to plug in to the numbers.

    4. We have gotten far off track on this (and that's because of me... and I don't really mind getting off track), but do your calculations prove that we should make it tougher for sex offenders to get pardons? Is there any evidence that a pardon increases their chances of re-offending (as, these people who are free - not on parole, not in jail)?

    The statistics also imply that about half of all re-offending sex offenders do so within five years - exactly the span of time that they aren't currently eligible for a pardon. Wouldn't we have to crunch your numbers dealing specifically with people re-offending after five years... or, better still, after getting a pardon?

    I think if we were to adjust your math accordingly, this would be quite a good way to - at least partially - evaluate our system of pardons.

    ReplyDelete
  2. 1. Awesome, glad you're happy to debate!

    2. Hmm. It wasn't clear to me from the wording of the studies on your blog that sex offenders are unlikely to re-offend in any manner other than a sex crime. At any rate, if that's true, it means that I've underestimated how much costlier it is to release a sex criminal than a general criminal.

    3. I did see a few studies that have numbers showing that the more crimes a person has committed, the more likely they are to re-offend. But trying to do something with those numbers would make the math a lot more complex I think.

    4. You are quite right. My calculations suggest that the expected cost of releasing a sex offender is more than releasing a general criminal, which doesn't directly say anything about pardons. I drew the connection because your original post was using data on the risk of re-offending to make an argument about pardon policy, so I took what I thought was your argument and ran with it. Your tweaks to the math would make for a better analysis -- but that would require digging for more data, which I'm too lazy to do.

    ReplyDelete
  3. David,

    Regarding point #2, there was just a throw away line in the middle of my post about it. I certainly didn't draw attention to it (my post was long enough as it was).

    Regarding #4, it's true, I jumped from the pardons to recidivism (because much of the debate was framed as a public safety issue), so it's fair game for you to continue on that path. But if we're going to have a deeper and reasoned investigation, we should probably steer back on course a bit... but I'm in agreement, I don't have the time or energy to harvest and analyse all the numbers.

    ReplyDelete