I hear a lot about deaths rising with recessions. What does the data indicate about the robustness of such a relationship?
Below is a plot of the deaths/population ratio, in logs.
Figure 1: Log ratio of deaths to population. NBER defined recession dates shaded gray. Source: CDC, Census via FRED, NBER, and author’s calculations.
Estimating a log-log regression of deaths on population, 1999-2018, one finds that the semi-elasticity of deaths with respect to recession is -1.5%, although it is not statistically significantly different from zero.
Interestingly, augmenting the regression with a “Trump dummy” leads to a Trump effect of 6.4% (statistically significantly different from zero at the .01%).
Well we thought Trump was toxic, but now the science is in. Who would have guessed!
Funny, but not meaningful.
Bruce Hall not meaningful
Well, you’re right. But not for the reason you think. The regression discounts the claim that recessions cause suicides. In other words, the claim that you and sammy have been making about recessions causing suicides is “not meaningful.”
Bruce,
Why is this “funny”? You seem to have a sick sense of humor.
I would like to see some other variables in that regression, with I suspect a seasonal component being quite important. But it is not clear that including that would alter the finding all that much. There is much about this we do not understand.
In the current context, I am drawing on anecdotal personal sources, but despite a massive increase in depression, there so far does not seem to be a noticeable increase in suicides, with, again, the main group appearing to have that problem being frontline medical people who are overemployed and overstressed, not unemployed. You think this is funny, you sicko?
There is a serious possible irony here. Again, despite all the depression, a reason we might not have seen suicides up has been this wartime mentality, with in fact many who have not been well feeling sort of justified and supported as so many others have been affected. But now we have all these reopenings and claims of “normalcy,” with our prez out playing golf without a mask at one of his clubs two days in a row over Memorial Day weekend (rather than going to church as he urged others to do). Everything is supposed to be peachy keen. But the recovery will not be a V, it will be slow, whichever of the other shapes it takes. So those who continue to be unemployed will no longer have this wartime mentality and may be in danger of becoming even more depressed as their circumstances fail to improve. The economy may be growing and thus technically not in a recession, but we may well see an increase in suicides.
All of this is just knee-slappingly hilarious, of course.
I would like to see some other variables in that regression, with I suspect a seasonal component being quite important.
The eleven dummy variables ending with “DM” along with the constant represent the twelve seasonal variables. Visually there doesn’t seem to be any evidence of a unit root in the seasonality.
You are right, 2slug. Good point.
There is a widely reported relationship between suicide and recession, for instance:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4473496/
It may be that, in a display of a common mental error, “suicide” was generalized to “death”. Death by certain other causes, auto accident and work accident for instance, fall during recession. Together, auto and work accidents account for more deaths each year than suicide, so would tend to offset the increase in suicides during recession.
Bruce, who has soiled himself pretty badly with his magic malaria elixir rants, seems to have adopted a new hit-and-run tactic in comments here. He declares some idea to be “not meaningful” without any explanation and then goes back into hiding, This is part of the paid-troll toolbox. Regulars here are not swayed by anything Bruce writes, but we are not his intended audience. Aside from his masters, his audience is the casual reader of this blog. Anyone who shows up just to see what’s being said may be led to doubt what Menzie writes by a a flippant dismissal. Job done from a paid troll.
Anyone familiar with the techniques off “junk science” propaganda knows the drive-by dismissal technique Bruce is using. It’s a sin against intellectual honesty, which makes it perfect for paid trolls.
md,
In 2018 while traffic deaths were about 38,000, suicides were about 48,000, substantially more. I have not counted work deaths, but obviously if you start adding a bunch of sources of deaths they will eventually outnumber suicides. Why were you doing something so silly while largely taking on Bruce more or less reasonably?
Ah, there’s Barkley’s grudge-based thinking again. He has a hard time thinking things through after a apparently joined team-Moses. So, Barkley, here is what you would have realized for yourself if your thinking hadn’t been clouded by emotion:
The issue is whether deaths increase as a result of recession. My point was that suicide deaths often do rise in recession, but deaths by some other causes decline in recession. I did not offer any old cause of death in my example. Rather, I offered two examples of causes of death which result in fewer deaths during recession, to show why overall deaths should not be expected to rise in recession simply because suicides may rise.
Having reread what I wrote, I can see why Barkley might have missed what is obvious to others. I failed to tug my forelock in Barkley’s direction, and that clearly masked my intended meaning.
Barkley, you need to get a grip on that attack reflex. It leads you into silly behavior such as displayed here. That behavior runs contrary to your intellectual pretenses. Nose, meet face.
Macroduck,
This seems a bit overdone, macroduck. Did you notice that Mark was claiming that we have 1.25 million deaths by autos per year in the US. He was also quite proud of not having actually looked at any data, but he was quite sure that declines in auto deaths must certainly far outweigh any change in suicides, given the vast difference in their numbers that he inaccurately listed. Nobody bothered to comment on my correction of him, certainly not you, md.
As it is, you just happen to be wrong about the total of auto and work-related deaths in the US. Their sum is less than the number of suicides. Again, 38,000 auto deaths, 48,000 suicides, and work-related deaths in the US in 2019 were about 5,100. I shall presume you can add and thus realize that you have gone on a completely foolish rampage sort of resembling our old friend Moses, ranting and raving while wrong.
As a technical point, I looked more closely at the linked article, which seems reasonably well done. It is possible that it in fact says that declines in auto and work deaths do offset increases in suicides during receissions, although it is clear that numbers vary across recessions, so there may be no definite answer here, with different outcomes in different recessions. I note that even if there are or were more auto and work deaths than suicides at times, which is probably true, that does not mean that changes in them would more than offset changes in suicides. These are separate matters, and I suspect we do not know the answer.
That is very much part of why there is so much vigorous discussion about this going on here, with few definite answers, if any, so far. This is a completely new situation with many cross-cutting effects, and I am not at all out to declare some particular answer. I do not know, although if you want to point out death-lowering effects of a lockdown-induced recession, please note fewer due to reduced pollution. Heck, I posted on this on Econospeak, and made refeence to some of the related arguments in a comment here nobody noticed, but according to a recent report, carbon emssions globally declined 17% to hit bottom around April 7 and have been rising since, which suggests the world economy may have been rising since then, although we know the US is lagging on both the virus and the economy.
I am really not pushing any particular answer here and do not know what the bottom line is on suicides, although I have now noted a couple of times without any comment from anybody that the most reported change in suicides in the US has not been by people upset about being unemployed or even those bored to death as 2slug has suggested might be going on. It is frontline medical personnel, guilt-ridden and overstressed, although their numbers are not huge. I happen to have two daughters who are in that situation, and my information on this has come from them, so in fact I do care about this quite personally, but I do not know what the answers are.
But I do find it annoying to see people like you and Mark dismissing suicide because of supposed changes in auto deaths or work deaths, when neither of you seems able to get your basic facts straight. This is actually pretty nauseating, Bruce Hall kind of stuff.
Barkley Rosser or even those bored to death as 2slug has suggested
Just to be clear, I don’t believe people are committing suicide because they are bored. I was restating explicitly what sammy was implying in some of his comments. I don’t believe people are likely to “off themselves” because they can’t go see the latest action movie or get drunk down at the local watering hole. I completely agree with you about suicides among healthcare workers. The PBS show Frontline had a special report on the toll the disease was taking on medical workers in northern Italy. It was hard to watch. I have a niece (my goddaughter) who is a Navy doctor based out of the Bethesda/Walter Reed complex but was detailed to hotspots in support of COVID patients. She was worn out by the experience. And the PPE situation was unforgiveable. It’s ridiculous when her aunts and uncles had to scrounge around to find and make masks for her because she was issued one N95 mask and told to make it last.
Got it, 2slug. Sorry for the misunderstanding. Point taken.
BTW, I have figured out where Mark got his 1.25 million auto deaths from. That is cloase to the global number, which is actually 1.35 million. Globally there are more auto deaths than suicides, with only 800,000 of the latter recorded, although as has long been noted in many countries suuicides get underreported for various reasons.
In any case, we have more suicided in the US than auto deaths, with our deviation from the world pattern probably mostly due to all the guns we have, whose numbers in the US across states are highly correlated with suicide rates, and we have way nore guns per capita than any other nation.
Barky,
Ah, yes, now comes the faux-reasonable tone, with many words piled on – part of your intellectual pretense.
Yep, it appears I made a data error. You, on the other hand, made a analytical one. I admit to my error. You have not admitted to yours, Instead, you trot out your usual bile (for those following the exchange, see Barky’s last paragraph.)
Even if the sum of deaths by two causes which are sensitive to recession is less than total suicides, the cyclical swings in those two serve to offset cyclical swings in suicides. That was the point I made. That was the point you called silly. I chose the wrong data source. You have chosen an anti-intellectual attitude masked by intellectual pretense.
So, you ad hominem tossing homunculus, let’s go. You have been entirely dishonest in the your claim of “annoyance”, as well, implying that I have dismissed suicides as a concern. You stoop to dishonest to puff yourself up. That, you sad, sad little man, is truly disgusting.
“Macroduck,”
“Barky”? Ah, you out yourself as Moses Herzog. Well, do please obey Menzie and stay away from arguing that Nancy Pelosi is demented.
As for your “analytical” point, it is crap. There is no evidence that I see in the linked study, which does show that most of the time suicides seem to rise with unemployment, in contrast to Menzie’s apparent finding, that consistently or even at all that changed in auto and work deaths offset changes in suicides. Indeed, you are so out of it you do not get that this is not an analytical point at all but an empirical one. I completely agree and never disputed that during recessions we are likely to see a reduction of deaths from autos and work accidents. The question is the empirical one about the relative strengths of these effects, and that is unclear.
I thought you were intelligent, Macroduck, but going all gonzo on your inner Moses is leading you to spout just rank nonsense.
A Trump dummy.
how appropriate
Actually we are having an argument downunder about the costs of our lockdown against no lockdown. The counterfactual is nonsense however our government response should mean the costs are greatly lessened.