1. You asked for cost-benefit analysis? Here is Luigi Zingales.
we are talking about roughly $9 million per lost life, which, multiplied by the 7.2 million extra deaths, yields $65 trillion.
Thus, even the simplest cost-benefit analysis suggests that the US government should be willing to spend up to $65 trillion to avoid extra deaths. Since $65 trillion is 3 times the US GDP, the United States should be willing to stop production for up to 3 years in order to eliminate the extra deaths.
Satisfied? I’m not. Forget the wild assumptions (7.2 million extra deaths). We need to be careful about thinking on the margin.
Consider three scenarios:
a) zero social distancing
b) social distancing decisions made entirely by the private sector
c) social distancing imposed by the government
Scenario (a) is irrelevant, but people tend to forget that. But relative to (a), (b) is going to both reduce deaths and impose economic costs.
The relevant policy choice is between (b) and (c). The economic case for (c) is that when I go to work I calculate the risk to myself, but I do not include the risk I impose on the people who ride mass transit with me. But describing the externality is not enough. We need to guess how much social benefit comes from moving from (b) to (c) and compare it to the cost.
My own view is that, to a first approximation, moving from (b) to (c) imposes no economic cost, because I think that if the death rate soars under (b) the economic cost will be enormous, also. That’s just my intuition, based on what I think is the importance of social capital, which in turn I think depends on people feeling that their fellow citizens care about them. So it’s a very soft concept, not something I can put a hard number to.
What I would like to better estimate is how many deaths (c) saves relative to (b). But you know how frustrated I am with the inability to do reliable analytics on something like this. Meanwhile, I am willing to assume that the marginal life savings are high.
Under this plan, a person whose earnings fall to zero this year keeps all of the social insurance payments and does not pay the surtax. A person whose earnings fall by half keeps half of the payments and returns half. A person whose earnings remain the same (or increase) returns everything: They will have just gotten a short-term loan.
The specifics are at the link. This is clever. Maybe too clever. Any scheme like this invites gaming. For example, suppose my earnings go up, so I should not get to keep any of the money. In that case, I just sell some stock, take the capital losses, and as far as the IRS is concerned my income did not increase. It is always somewhat advantageous to me to take losses, but this would produce a proportionately larger payoff from doing that.
Still, Mankiw’s idea is far better than what I’m guessing what Congress will come up with.
Honestly, I don’t know what everyone is worried about with all this stimulus stuff. In 1945, we had just been through four years of war, and the economy bounced back way more quickly than the Keynesians thought possible (please click on the link). I’d give it a chance to bounce back on its own after a war against a virus that has fewer casualties and shorter duration.
If we don’t do (c) deaths will increase to the point where (c) is imposed 2-4 weeks later. So we will:
1) Get more deaths
2) Likely pay an even greater economic costs because (c) will need to last longer and be harsher to get deaths down
We already saw this happen in Italy. People don’t respond to collapsing healthcare systems and bodies being shoveled to the side by rejecting C.
The best case scenario is to keep C as small and short as possible. The earlier you act, the better.
You’re forgetting to take into account life-years. Most of the excess deaths are of the elderly with comorbidities. Average foregone life expectancy will tend to be only a few years
Yes, this. +1
Slightly more reasonable: $2.5m per life, 750k deaths = almost $2T
> Giving every tax-filing household $5k (2500/mo for 2 months to start) is roughly $725B.
> Give a similar amount to struggling businesses as a back stop for those same 2 months.
> Total bill: $1.5T
> Re-evaluate in 6 weeks.
Suppose we give up about $9 trillion in the next two-three years in lost GDP due to Covid-19 government-mandated restrictions.
Suppose we save 90,000 lives.
That is $100 million per life, and that life generally an older person with co-morbidities.
You can ramp up the lives saved, put it at 900,000 lives. That is still $10 million dollars per life.
I hope the economy does bounce back. The experience after 2008-9 suggests a long drawn-out and slow recovery. Some countries, such as Greece, never did bounce back. In the US, we shifted to lower productivity increases and never bounced back on that particular level.
Since 1980, after each recession we see lower interest rates and inflation. Perhaps we will be at zero bound and no inflation going forward.
“Forget the wild assumptions (7.2 million extra deaths).”
I don’t think the number is that wild. It would correspond to a total mortality rate of about ten million people this year, instead of the normally expected ~2.8 million, i.e. an increase of total mortality by a factor of about 3.5.
This would be consistent with the current data from some of the worst-affected places in Italy, where since the beginning of the epidemic the total death rate has soared. (Which means it’s not even necessary to look at testing, hospitalization, etc. to see that something terrible is happening — just a simple look at the total number of deaths compared to any other period of the same duration immediately shows that mortality is going through the roof.)
For example:
https://www.straitstimes.com/world/italy-coronavirus
“Last year, 120 people died in Nembro, 10 a month; now 70 have died in the space of just 12 days.”
Now, it’s likely that the population of these worst-affected towns is very old on average, and presumably these extreme spikes in mortality would last for much less than a year before the epidemic would burn itself out. But nevertheless, they show that an uncontrolled epidemic could quite plausibly lead to the total death rate for the year spiking by a factor of three or four.
That’s not remotely consistent with current data, even in Italy. First of all, the observed mortality rate is probably 5-10 times the actual mortality rate. The consensus is there are 5-20x as many people infected as reported. 2nd of all, Italy probably greatly over reports coronavirus deaths: if you die and you have coronavirus, they attribute it to the virus. This one of the big reasons they’re reporting such a high mortality rate.
Realistically, if the governments did nothing at all, the absolute max imo is maybe a million, and probably well fewer. Most of the relevant action that can reduce deaths is fairly costless to the state, e.g. imposed social distancing.
Lastly, cynical as it sounds, the average life lost is not worth anywhere near 9 million. The average fatality is in their 70s, has no work years ahead of them, has a few years of life left, and will cost someone (society or their loved ones) a six or seven figure expenditure in medical bills during the remainder of their life. Not saying we should t care, but if we accept that a 75 year old dying is less bad than a 20 year old, and we insist on assigning value to life, the average death cost from this virus is way less than the average.
I think Zingales’s overall estimate off by nearly 2 orders of magnitude.
You charactize the I only people at risk as very old people who are about to die anyway. They are certainly more at risk, but the risk profile for average adults isn’t exactly thrilling. Not just mortality, but also permanent lung damage, a trip to the ICU.
Would you continue to conduct yourself as normal given these risks? What about the risks to your family if you transmit to them.
My father would almost certainly die if he got this. I’m in my 30s but with asthma and other health problems I might go too. My daughter has asthma. What’s the mortality rate for young children with complications? What is the odds of her lungs being damaged? Would I like to see her hospitalized even if she turned out fine. Under these circumstances, how can we expect people to carry on as normal.
Mark Z,
What’s remarkable and shocking about these Italian total mortality statistics is that they make it unnecessary to guess any of the uncertain variables to prove that the situation is quite dire.
The stark truth that speaks for itself is that we can see towns where in the last three or four weeks, five or more times more people have died than they do in a normal month, which is completely unprecedented in modern times. It’s already visible from the obituary sections of local newspapers, which have been taking up several times more space than usual.
There’s simply no way to explain this away, no matter how optimistic assumptions one makes about underestimated infection rates, confounding of COVID-19 with other deaths, etc., etc. Something out there is killing people at a rate several times higher than normal, so unless there’s another yet unidentified cause of mass death, this must be it.
And extrapolating these rates to the U.S., with its regular mortality of about 2.8 million people a year, clearly shows that 7-8 million deaths would be within plausible range if the epidemic were to be left uncontrolled.
It’s certainly true that these would be largely old people, and how that should be accounted for in utilitarian trade-offs is a question I’m not addressing here. I’m only pointing out that we already have evidence that this disease would be capable, at least under some circumstances, of increasing the total yearly mortality in a country to several times the usual number.
“I’d give it a chance to bounce back on its own after a war against a virus that has fewer casualties and shorter duration.”
Ezra Klein has an article out on Vox where Mark Zandi speaks about four waves of economic pain. Here, Zandi describes the third wave (I believe that’s what causes at least a part of the fourth wave):
The third wave, according to Zandi, will be “all these folks who’ve seen their nest egg wiped out. They thought they were set for retirement and they’re not. They’ll go into panic mode.” The shattered stock market will be a disaster for those in or near retirement. They’re watching wealth they worked their whole lives to build crumble in the space of weeks. They won’t purchase that new car, buy that new house, plan that vacation — and unlike some of the direct economic stoppages, which will lift when the virus ease, their reticence to spend will slow economic growth long after the direct crisis ends.
Link to full article: https://www.vox.com/2020/3/23/21188900/coronavirus-stock-market-recession-depression-trump-jobs-unemployment
This is certainly evidence that if an economist wants to justify giving a blank check to the state, he won’t let reasonable assumptions or a sound understanding of data get in the way him dressing up an outrageous assertion as a scientific analysis. 65 trillion is plainly ridiculous. It doesn’t pass a basic sanity check.
I am super disappointed in Zingales who is normally good. Not only is 7 million deaths super unlikely but $9 million per life is the wrong measure. A better start is to consider QALYs which are typically valued at 50-150K each. A plausible estimate range is that policy measures have headroom to save 10-50M QALYs; at 100K per QALY that’s 1-5 trillion dollars worth of value. Still very large but very far from 65 trillion.
The economic downturn is caused by government edict and hospitals overwhelmed. Under these circumstances there is no stimulating, there is merely a short term lending problem. This is not about getting people to work it is about getting the lending system to respond to a Big Black Swan.
Work it out, in place. There are not secrets here, we are paying Black Swan costs, let Coase apply and firms can make a standard deal, get through the mess.
My hat’s off to Mr. Zandi for attempting to put a number on the cost of “option a)”.
The next step, is to refine this and come up with a better estimate, specifically tailored to whatever country you live in. My guess is that most of the assumptions used in his analysis result in significant overestimates for the U.S. For example, his estimate uses willingness-to-pay to come up with his numbers for the value of a life lost. This method has been widely discredited. Then his estimate refers to EPA numbers to refine this based on the age of those likely to die, reducing it by 37%. Why not use EPA numbers for both? The EPA number for life lost is about $9.4M. I am guessing that a close look at the actual age profile of those who have died will result in a reduction for age much higher than 34%. Some commenters here have attempted more reasonable estimates. Cudos to them. It would be more helpful if a respected economist took a shot at this.
Some other factors to consider would be that the US has more hospital beds, ventilators, and overall capacity in the medical system than Italy. The US is younger than Italy. The US is less densely populated than Milan, the source for his estimate. I’m sure many other refinements are possible. The way Italy calculates the number of deaths has been criticized.
It would be more accurate, although not very PC, to also account for future expenditures avoided due to cornavirus (h/t Mark Z). This would be appropriate for a purely economic analysis. My guess is that this is would result in a significant reduction, especially given the age profile for deaths.
Adding in costs for those who become ill but don’t die would also be an improvement (h/t Asdf). Ideally it would include things like lost wages, pain and suffering, and some cost for those with permanent effects.
The next step would be to run this for Option b). Taking the example of our esteemed blog host, while riding the bus I might not “include the risk I impose on the people who ride mass transit with me”, but the other people on the bus are considering the risk you are imposing on them. Solving for the equilibrium, you might get a situation where fewer ride the bus, and everyone on the bus not wearing a mask suffers social or legal sanctions.
With a more accurate cost estimate, better decisions could be made, responses could be tailored to those most at risk or most likely to transmit the virus, and costs could be significantly reduced.
I am also wondering if Mr. Zingales accidentally proved that the current approach in the US is ineffective with his Covid-19 Lag Tracker chart. The source for the chart was not accessible, but it would be interesting to identify when each country implemented business restrictions, curfews, and other responses, to see what affect, if any, they have had on the curves.
“It would be more accurate, although not very PC, to also account for future expenditures avoided due to cornavirus (h/t Mark Z). This would be appropriate for a purely economic analysis. My guess is that this is would result in a significant reduction, especially given the age profile for deaths.“
Excellent, but controversial for sure.