These are great and well-reasoned Jack! The only question where I think I disagree with you by more than a few percentage points is the Supreme Court composition question. I was lower than the crowd last year on Metaculus, but, this year, as you pointed out, either Sotomayor or Kagan might want to / feel pressured to resign. I don't have high confidence in my assessment, but Manifold feels closer to me on this one. It's an important priority for Democrats and this could be the last time in a long while that they have a chance to install a new justice.
Thanks! Interesting, I'm curious what probability you put on one of them retiring? I made a market on stepping down specifically: https://manifold.markets/jack/will-a-us-supreme-court-justice-ret - note that this also includes retiring due to health problems, so the total probability should be this OR death. I figure that they aren't nearly as old as Breyer was, and also Breyer was probably the exception rather than the rule (see RBG)
For chances of a vacancy, Manifold is now at 13% and Metaculus at 15%, so that lines up reasonably with my prediction.
After consideration, I think you're probably right. I think the chance that Sotomayor will retire is probably around 5 - 6%, and my estimate of a probabilistically independent retirement from any of the remaining justices is around 1% (with about 85% of that 1% being on Kagan). Meanwhile, barring any planned retirements, I was pretty close to your 10% death / incapacitation estimate for 2024, but, after reflection, adjusted down on that slightly. I think if a conservative justice experiences an event that would be otherwise incapacitating, they'll probably do whatever they can to try to hang on for as long as possible if not for a Republican president than at least maybe for a Republican controlled Senate. I made some assumption adjustments and ran them through a simulation program that I've used for previous estimates like this and came back with about 9% for death / incapacitation for the year as a whole and about 8% for the remaining year. Putting the two together gives me an estimate of about 14% (1 - 0.935 * 0.92). I'll probably write up my estimate with all the assumptions and code and post it in the next few days when I have time, but this was a really interesting exercise!
Will be interesting to read more about your detailed analysis!
I was thinking one thing that should be possible to improve the estimate is to get some better data on risk of mortality and severe illness, especially broken down by socioeconomic status and other relevant factors. Crunching the data on the severe illness part doesn't look easy, but I found some data on the effect of SES on mortality at https://www.soa.org/resources/research-reports/2020/us-mort-rate-socioeconomic/ which shows mortality rate of 9.2% for all US females 60-69 vs 6.2% for the top 10% SES. That might mean that my off-the-cuff estimate of 10% for death + incapacity is too low?
I think your point about Republican appointees hanging on even if they are severely ill is very true, and vice versa that if one of the older Democratic appointees has an illness that has a good prognosis but makes continuing to work tough, they may be much more inclined to retire this year.
As I think about it more, it occurs to me that if a justice were clearly incapacitated, but refused to resign for obvious political reasons, Biden might nominate a new one anyway and there could temporarily be 10 justices. I'm probably getting into the realm of rapidly diminishing returns on forecast improvement, but it's interesting to think about.
yes, you are correct that legislation would be required to change the number of justices. I think this branch of reality can be safely rounded down to zero for forecasting purposes.
These are great and well-reasoned Jack! The only question where I think I disagree with you by more than a few percentage points is the Supreme Court composition question. I was lower than the crowd last year on Metaculus, but, this year, as you pointed out, either Sotomayor or Kagan might want to / feel pressured to resign. I don't have high confidence in my assessment, but Manifold feels closer to me on this one. It's an important priority for Democrats and this could be the last time in a long while that they have a chance to install a new justice.
Thanks! Interesting, I'm curious what probability you put on one of them retiring? I made a market on stepping down specifically: https://manifold.markets/jack/will-a-us-supreme-court-justice-ret - note that this also includes retiring due to health problems, so the total probability should be this OR death. I figure that they aren't nearly as old as Breyer was, and also Breyer was probably the exception rather than the rule (see RBG)
For chances of a vacancy, Manifold is now at 13% and Metaculus at 15%, so that lines up reasonably with my prediction.
After consideration, I think you're probably right. I think the chance that Sotomayor will retire is probably around 5 - 6%, and my estimate of a probabilistically independent retirement from any of the remaining justices is around 1% (with about 85% of that 1% being on Kagan). Meanwhile, barring any planned retirements, I was pretty close to your 10% death / incapacitation estimate for 2024, but, after reflection, adjusted down on that slightly. I think if a conservative justice experiences an event that would be otherwise incapacitating, they'll probably do whatever they can to try to hang on for as long as possible if not for a Republican president than at least maybe for a Republican controlled Senate. I made some assumption adjustments and ran them through a simulation program that I've used for previous estimates like this and came back with about 9% for death / incapacitation for the year as a whole and about 8% for the remaining year. Putting the two together gives me an estimate of about 14% (1 - 0.935 * 0.92). I'll probably write up my estimate with all the assumptions and code and post it in the next few days when I have time, but this was a really interesting exercise!
Will be interesting to read more about your detailed analysis!
I was thinking one thing that should be possible to improve the estimate is to get some better data on risk of mortality and severe illness, especially broken down by socioeconomic status and other relevant factors. Crunching the data on the severe illness part doesn't look easy, but I found some data on the effect of SES on mortality at https://www.soa.org/resources/research-reports/2020/us-mort-rate-socioeconomic/ which shows mortality rate of 9.2% for all US females 60-69 vs 6.2% for the top 10% SES. That might mean that my off-the-cuff estimate of 10% for death + incapacity is too low?
I think your point about Republican appointees hanging on even if they are severely ill is very true, and vice versa that if one of the older Democratic appointees has an illness that has a good prognosis but makes continuing to work tough, they may be much more inclined to retire this year.
As I think about it more, it occurs to me that if a justice were clearly incapacitated, but refused to resign for obvious political reasons, Biden might nominate a new one anyway and there could temporarily be 10 justices. I'm probably getting into the realm of rapidly diminishing returns on forecast improvement, but it's interesting to think about.
It seems that if a justice is incapacitated, the only way to remove them is via impeachment (https://www.politico.com/magazine/story/2019/01/16/ruth-bader-ginsburg-supreme-court-health-224014/), and I'm pretty sure a new justice cannot be nominated without first removing them - I believe any change to the number of justices requires Congress to pass legislation.
yes, you are correct that legislation would be required to change the number of justices. I think this branch of reality can be safely rounded down to zero for forecasting purposes.