Wednesday, September 2, 2009

Did Social Security lower suicide rates?

In a generally silly article claiming – no kidding – that Republicans want to kill people, Slate's Jacob Weisberg claims that Social Security reduced the suicide rate for retirees. Unlike some of this other claims (which NRO's Ramesh Ponnuru deconstructs here), this one isn't totally made up: Weisberg cites an NBER paper by David Cutler and Ellen Meara on the overall decline in U.S. mortality through the 20th century. Cutler and Meara say:

Since 1930, suicide rates among the elderly have fallen by 56 percent, while rates among teens have tripled. The decline in suicide among the elderly is coincident with the large absolute increase in income for the elderly stemming and pre-Social Security social programs and the phase-in of the Social Security system. Following the introduction of formal Social Security benefits, benefits rose rapidly over the period from 1950 until 1970, with a particularly rapid rise in the late 1960s. During these same time periods, suicide rates among the elderly fell most rapidly.

Sounds reasonable – I wouldn't dispute the possibility that giving early retirees massive transfers of money would improve their mood. But does it hold up? Cutler and Meara don't run any statistical tests themselves; they draw their conclusions entirely from their Figure 14.

Suicide rates for people aged 65 and over do decline significantly from 1935, when Social Security began, to today. But note a couple things:

  • From 1930 through 1940, a period when Social Security had not begun paying benefits yet, seniors' suicide rates declined at a rate faster than they did afterwards. Moreover, this was a period in which seniors' incomes were depressed, which formed part of the justification for the Social Security program. If so, shouldn't suicide rates have increased during the Great Depression?
  • Suicide rates for Americans aged 65 and over closely track those for individuals aged 55-64 and 45-54. These folks weren't receiving Social Security benefits, so pretty clearly something else was going on. These three age groups move closely together throughout the period, while those of individuals aged 15-24 and 25-34, which were generally rising, also track each other.
  • Beginning in 1980, seniors' suicide rates jump up. One explanation is that this was the beginning of a period of retrenchment for Social Security benefits, although even retirees in the 1980s got a pretty good deal from the system. However, this was also a period when Medicare benefits ramped up. Shouldn't we have seen a similar effect there? Or may all those seniors just hated Reagan and couldn't go on.

Cutler and Meara's data doesn't appear to be available online so it's not possible to do any formal analysis. (If anyone has the data, please send it!) I would personally be pretty surprised, though, if the Social Security/suicide conclusion held up under stress.


Bruce Webb said...

"improve their mood"? Yeah I suspect a minimal ability to pay rent and feed yourself might reduce your impulse to blow out the pilot and stick your head in the oven but the whole suggestion is kind of dismissive.

Plus the argument entirely discounts benefits paid between 1936 and 1940 under Title 1 of the 1935 Act. Maybe the drop in suicide rates had something to do with the fact that there was some relief from absolute destitution before Title 2 kicked in?

Andrew G. Biggs said...

Sorry if I was a bit snarky. But still, the tight fit between the three older age groups over the period tells me there was something else going on. Maybe the 45 year olds decided not to kill themselves because they had Social Security to look forward to, but I doubt it. Sorry, that was me being snarky again.

Michael said...

I suspect that the decline in the suicide rate has more to do with improvements in medical care. Diseases that were once considered fatal became treatable. My limited personal observation is that more people kill themselves to avoid a slow wasting away that to avoid financial ruin.

Andrew G. Biggs said...

Good point. I bet if you did a regression and put in dummy variables for widespread availability of certain medical treatments that you'd find something there. Granted, higher incomes for seniors through Social Security would allow them to buy more health care, which by this theory would lower suicide rates. But the evidence I've seen on the introduction of Medicare (which, admittedly, came somewhat later) is that seniors didn't actually consume more health care but just tended to pay less for what they did consume. In any case, I suspect you've got a good point here.