I believe there have been studies into how social life impacts longevity, and probably cognitive decline as well. For some people, like my great-grandmother who kept working well into her 80s by choice, jobs can be a big social outlet. For others a job can be very socially isolating. Those factors probably matter a lot.

Side note: I'm sure we'll see research into these areas used to propose delaying retirement age more in the near future.

It's 48 pages and I haven't read it fully, but it seems almost childlike that the paper doesn't address the obvious confounding variable:

"Does Unemployment Make It More Likely for Late Middle-Aged People, Particularly Men, To Drink Alcohol? Evidence From We Obviously Should Have Considered This In The Paper, Perhaps We Are Too Sheltered"

To be clear I am not being pedantic. The paper explicitly endorses the policy of pushing back the retirement age specifically because doing so likely reduces cognitive decline. I agree with this, in the same sense that shooting car thieves in the street without a trial reduces automotive theft. "Reducing cognitive decline in people near retirement age" might be better met with psychiatric intervention, so that unemployed people also get some of the benefits. Ignoring this confounding variable and prattling about "causal explanation" - while endorsing the policy of snatching away people's pensions until they work a few more years - is evil born from ignorance.

I finally found the reason why my cognitive function feels like that of a 7years old child.