Excellent post. I have a strong prior that academic literature on criminology is biased, so I am more inclined than you to guess that consensus estimates for criminal justice reform not having net negative effects on crime are too optimistic. So my guess for second-order effects is that they make criminal justice reform even less valuable relative to other global health/wellness causes.
Putting that aside, I think one reason Open Phil might have been so favorably inclined to criminal justice reform was the bipartisan consensus that pursuing it was a good idea. The 2010′s were a uniquely good time to purse criminal justice reform (until ~ 2020, when increasing crime rates made criminal justice reform less bipartisan).
Perhaps you could call this the “The Hinge Hypothesis”—during the years that Open Phil made large donations to criminal justice reform efforts, it was a uniquely good time to do so. I think this was a reasonable guess, though I don’t think it brings the QALYs/$ to parity with other global health/wellness goals.
A lot of people, myself included, had relatively weak priors on the effects of marginal imprisonments on crime, and were subsequently convinced by the Roodman report. It might be valuable for people interested in this or adjacent cause areas to commission a redteaming of the Roodman report, perhaps by the CityJournal folks?
That’s an interesting idea. It seems like an effort that would require a lot of subject-matter expertise, so your idea to commision the CJ folks makes sense.
I do wonder if cause areas that rely on academic fields which we have reason to believe may be ideologically biased would generally benefit from some red-teaming process.
Putting that aside, I think one reason Open Phil might have been so favorably inclined to criminal justice reform was the bipartisan consensus that pursuing it was a good idea. The 2010′s were a uniquely good time to purse criminal justice reform (until ~ 2020, when increasing crime rates made criminal justice reform less bipartisan).
Perhaps you could call this the “The Hinge Hypothesis”
Thanks. Yeah, having a negative tail really reduces expected values. E.g., playing with a toy models, having a 25% that the impact is of the same magnitude but negative ~halves the expected impact:
qualysPerDollarAllPositive = 1/lognormal(9.877,1.649) // taking from systemic change model
qualysPerDollarWithNegativeTail = mx(qualysPerDollarAllPositive, -qualysPerDollarAllPositive, [0.8, 0.25])
[mean(qualysPerDollarAllPositive), mean(qualysPerDollarWithNegativeTail)] // 2e-4 vs 1.13e-4
I wonder if this paper, which appears to show that incarceration reduces prisoner mortality relative to non-incarcerted but criminal-justice-involved people, should change your estimates of CJ reform benefits. Given that, it seems plausible that reducing prison stays actually increases mortality for prisoners.
Another interesting thing about this paper is the implication that the previous work on this topic (which used the general population as the control group) was flawed in an obvious way. That should generally lower our opinion of the academic literature on this topic.
I imagine that the benefits of marginally increased mortality wouldn’t be the most important facto here: the vast majority of prisoners would prefer to be outside prison, even if this leads to an (I presume small) increase in mortality.
So I imagine this would have an effect, but for it to not be too large.
I think it is an impressive effect, though I agree people not wanting to be in prison is more important.
Using a panel of all defendants over the seven years after sentencing, we fnd that incarcerated defendants have a more than 60% lower mortality rate during the time of incarceration than similar defendants who were not incarcerated (an average of 230 deaths per hundred thousand annually as compared to 587 deaths per hundred thousand annually). The main sources of these differences are dramatically lower risks of mortality from homicide, overdose, or suicide during the period of incarceration. Defendants also have a lower risk of mortality from natural causes of death such as heart disease while incarcerated, potentially due to increased access to medical care
Excellent post. I have a strong prior that academic literature on criminology is biased, so I am more inclined than you to guess that consensus estimates for criminal justice reform not having net negative effects on crime are too optimistic. So my guess for second-order effects is that they make criminal justice reform even less valuable relative to other global health/wellness causes.
Putting that aside, I think one reason Open Phil might have been so favorably inclined to criminal justice reform was the bipartisan consensus that pursuing it was a good idea. The 2010′s were a uniquely good time to purse criminal justice reform (until ~ 2020, when increasing crime rates made criminal justice reform less bipartisan).
Perhaps you could call this the “The Hinge Hypothesis”—during the years that Open Phil made large donations to criminal justice reform efforts, it was a uniquely good time to do so. I think this was a reasonable guess, though I don’t think it brings the QALYs/$ to parity with other global health/wellness goals.
A lot of people, myself included, had relatively weak priors on the effects of marginal imprisonments on crime, and were subsequently convinced by the Roodman report. It might be valuable for people interested in this or adjacent cause areas to commission a redteaming of the Roodman report, perhaps by the CityJournal folks?
That’s an interesting idea. It seems like an effort that would require a lot of subject-matter expertise, so your idea to commision the CJ folks makes sense.
I do wonder if cause areas that rely on academic fields which we have reason to believe may be ideologically biased would generally benefit from some red-teaming process.
Cheers.
Thanks. Yeah, having a negative tail really reduces expected values. E.g., playing with a toy models, having a 25% that the impact is of the same magnitude but negative ~halves the expected impact:
I wonder if this paper, which appears to show that incarceration reduces prisoner mortality relative to non-incarcerted but criminal-justice-involved people, should change your estimates of CJ reform benefits. Given that, it seems plausible that reducing prison stays actually increases mortality for prisoners.
Another interesting thing about this paper is the implication that the previous work on this topic (which used the general population as the control group) was flawed in an obvious way. That should generally lower our opinion of the academic literature on this topic.
I imagine that the benefits of marginally increased mortality wouldn’t be the most important facto here: the vast majority of prisoners would prefer to be outside prison, even if this leads to an (I presume small) increase in mortality.
So I imagine this would have an effect, but for it to not be too large.
I think it is an impressive effect, though I agree people not wanting to be in prison is more important.