Richard Epstein is a highly respected free market economist. He recently caught a lot of grief for predicting that coronavirus deaths in the US would peak at about 500:
While Epstein does not have a medical or epidemiological background, he is a fellow at the Center for Medical Ethics at the University of Chicago and had worked on research about the spread of AIDS during the 1980s and 1990s. Amid what he called a “full state of panic,” his March 16 article boldly declared that the total number of deaths from COVID-19 would top out under 50,000 and that in the United States we would see “about 500 deaths at the end.”
Five days ago, I came across this article in Reason magazine:
Richard Epstein Cops to a ‘Stupid Gaffe’ in Controversial Coronavirus Essay That Caught Trump Administration Attention
But he stands by his reasoning and predicts that global deaths will peak under 50,000.
In that article, the reporter indicates that Epstein revised his estimate of US deaths from 500 to 5000.
Today (March 29), it seems like the 5000 estimate for the US and the 50,000 global estimate are both too low, perhaps far too low.
There’s no shame in making false predictions, I’ve certainly made plenty. But I see a disturbing tendency for predictions to be correlated with ideology. The world is what it is. An epidemic may or may not be a reason for more government involvement in the economy. (I’m not convinced either way.) But that implication has no bearing on the severity of the coronavirus crisis. So why do predictions on this issue so often seem correlated with ideology?
Just to head off any partisan comments, I am not suggesting that liberals have done a better job of addressing the crisis than conservatives, as we can see when looking at New York City. Governments all over the Western world obviously underestimated this problem (as did I.) Rather I’m worried about science becoming politicized by wishful thinking. For instance, I have no idea whether chloroquine is a useful treatment for coronavirus, but I do know that its utility is utterly unrelated to the fact that President Trump rather than Barack Obama happened to mention it in a tweet. The world is what it is.
Before even beginning to think about appropriate policies we need an accurate evaluation of the facts of the matter, as best as they can be ascertained. I don’t know about you, but I have increased respect for pundits who a few weeks ago were saying that America would be an unrecognizable place in the very near future. They may or may not have been right about the appropriate policies, but at least they had a clear-eyed view of what was about to happen. And without that unbiased perspective, there is no hope.
This post is not endorsing any current set of predictions. Indeed I rather doubt some of the most pessimistic predictions will come true. So Epstein may be right that we are too pessimistic. Rather, I’m focused here on process. Why are predictions so often (but not always) correlated with ideology?
PS. Razib Khan and David Siegel are two pundits that have been extremely worried about the coronavirus, and might be viewed as somewhat right-of-center.
PPS. The title of this post was taken from a line in “A Bend in the River”, the first V.S. Naipaul novel I ever read.
READER COMMENTS
Russ Abbott
Mar 29 2020 at 7:14pm
I find this a useful presentation of the data.
This Minute Physics video explains how it works.
Scott Sumner
Mar 29 2020 at 7:17pm
Thanks. Very interesting. You can see how Spain might be in even more trouble than Italy.
AMT
Mar 29 2020 at 10:23pm
Can anyone give a better explanation why idea 3 (don’t plot against time) is good? The explanation there seemed just terrible. E.g. the graph shown at 2:45 looks to show the trend pretty clearly to me…
Brian
Apr 3 2020 at 8:10pm
Some smart people put that graph together but I agree the explanation was lacking. The problem of plotting time on the horiz axis is that if you use absolute time (March versus April) then different countries sort of don’t belong on the same graph because each country’s patient zero (first case) came into existence at a different time. So the reason for not plotting against time is to put all counties on the same graph.
If you don’t need to put all countries on the same graph then you just put each country on its own graph and then there’s no reason why you cannot use time on the horiz axis. On the vertical axis you would still have new cases (or in their case, a running average of new cases) and it will be obvious when there has been a deceleration (meaning less growth but still positive growth) in cumulative cases.
Another way to do it is to use the horiz axis for relative time. That is t=0 means the day that country hit 100 cases. I just say 100 cases instead of 1 case because may be a small number of cases is not indicative of what will happen later, so let the number grow to 100 before starting the horiz axis. If the horiz axis measures time relative to a certain event (like the first day cumulative cases was 100) then it seems you can put all countries on the same graph. I haven’t tried this so I’m just using my intuition to suggest this will look like a useful graph.
In the end, it seems the graph is clever but really not that enlightening because you can just go to the worldometer website and see individual countries graphed and you can learn which ones have a decreasing number of new daily cases. You would learn which countries are experiencing some success in dealing with the problem but you would have to click around the website because each country is graphed separately.
Jacob Egner
Mar 29 2020 at 7:43pm
Scott, you might be interested in Robin Hanson’s book The Elephant In The Brain. If you’ve already read it, could you take a sentence or two to say what you thought of it? Thanks.
Scott Sumner
Mar 29 2020 at 9:50pm
Yet, excellent book.
Daniel Hill
Mar 29 2020 at 8:45pm
You mean to say facts matter? Unfortunately the past three years prove that to a very large proportion of the population, that is not true. And sadly, many of those people won’t realize the cost of ignoring the truth until it literally kills them.
Weir
Mar 30 2020 at 1:21am
Bill de Blasio is nobody’s idea of a libertarian, and what was he saying three weeks ago?
“We have 25 cases as of this morning. We care deeply about each of those individuals, but against the backdrop of 8.6 million people, and for the vast majority of New Yorkers, life is going on pretty normally right now. We want to encourage that. We have to look out for people here, especially those who are over 50 years old and have pre-existing conditions, like lung disease, heart disease, cancer, diabetes. These are the folks we really, really need to be careful for. But if you’re under 50 and you’re healthy, which is most New Yorkers, there’s very little threat here. This disease, even if you were to get it, basically acts like a common cold or flu. And transmission is not that easy.”
Bill de Blasio isn’t an economist or highly respected, but for a moment there he was trying to think about trade-offs, trying to take account of the differences between old folks and the rest of us, trying to avoid a panic and trying not to lump everyone together as equally at risk from this disease regardless of the pre-existing conditions that genuinely put people in danger.
And Scott even referred to New York City up above, so when you talk about people ignoring the truth, there’s a really good example you’re not seeing.
blink
Mar 29 2020 at 9:45pm
Epstein’s “revised” 5,000/50,000 prediction may yet prove silly and naive, but I am willing to give him the benefit of the doubt about not intending 500 as a US cap. After all, it belies statistics to think that a country with ~5% of the world population would account for just 1% of the deaths.
As for ideology, what this correlation may show is that liberalism is also associated with pessimism and risk-aversion, conservativism (at least the free-market variety) more with optimism and risk-seeking (or at least higher risk tolerance. Haidt’s shows that several “big five” personality traits correlate with the liberal-conservative spectrum, giving plausibility that other traits may be as well. I would love to see some empirical work testing these differences.
Scott Sumner
Mar 29 2020 at 9:55pm
blink, I do believe the optimistic view is often more correct, as with the famous Simon-Erhlich bet.
But in this case, the optimistic view seems clearly to have led to excessive complacency in numerous governments. The more pessimistic mortality estimates may not come true, but if they do not it will probably be because we took an enormous economic hit, and one that I would argue was unnecessary if we had moved more quickly. (Especially more quickly with testing.)
Scott Sumner
Mar 29 2020 at 9:58pm
BTW, I like Epstein, so I’d also give him the benefit of the doubt. But I think he made a serious error in the second prediction. This will prevent people from taking seriously his other claims, which are worth considering.
Thomas Hutcheson
Apr 2 2020 at 8:02am
This seems exactly backwards. Who is afraid of foreign imports destroying jobs? Immigrants polluting our culture? Low income people (if allowed to vote) would vote them selves “stuff”? Who is afraid that a tax on net CO2 emissions will snuff out the flickering flame of capitalism?
West
Mar 29 2020 at 10:20pm
<i>Why are predictions so often (but not always) correlated with ideology?</i>
Because the primary purpose of predictions is to move policy, not to predict any particular outcome.
People inherently understand this, and criticize prediction on the basis of the policy it’s intended to achieve.
Sure, there are still some who still use the old fact -> policy paradigm. They are, as someone memorably told me, ready to betray their side at the drop of a fact. But given their innate untrustworthiness in the red-blue battle that flesh-and-blood Americans actually care about, it’s unsurprising that they are mostly ignored.
And as for assuming that lives matter more to most people than ideological inclination, I’m not certain that assumption would survive examination in historical context.
But then this crisis has not been good for lowering my cynicism.
Ahmed Fares
Mar 29 2020 at 10:39pm
Epidemic Calculator
http://gabgoh.github.io/COVID/index.html
Ahmed Fares
Mar 29 2020 at 10:51pm
Further to my comment…
Note that besides adjusting the sliders on the calculator, you can also moved the vertical dashed line around which changes the date of intervention, and also input a value into the top slider to give a new R number based on an estimation of how that intervention reduces the R number.
Also, mousing over the graph shows the data for that date on the left-hand side.
Mark
Mar 29 2020 at 11:24pm
Most people seem to have mood affiliation towards the President. During the Ebola epidemic, it seems that the mainstream mood on among Democrats was “no big deal” while the mainstream mood among Republicans was hysterical, which is the reverse of the coronavirus situation. I think this is primarily because Obama was in office then and Trump is in office now (part of it might also be that Ebola was always a primarily foreign virus whereas coronavirus is now primarily a domestic one).
Also, some people are always optimistic and other people are always pessimistic. I’ve read quite a few articles and watched quite a few YouTube videos that seemed to accurately predict that coronavirus was going to be bad, but then when you looked at those authors/channels’ histories, they have been predicting everything was going to be bad for years or longer. It’s hard to assign respect based on a single good prediction; the most respect should be for the people who correctly predicted that coronavirus was going to be worse than mainstream opinion, but who also correctly predicted that Zika, Ebola, and swine flu were going to be not nearly as bad as mainstream opinion at the time believed.
Weir
Mar 30 2020 at 12:49am
“Hysterical” was the word Biden used back in January when he thought Trump was over-reacting and doing more than the situation warranted.
Robert EV
Apr 1 2020 at 9:25pm
According to this January 27 op-ed by Biden, he thought Trump was doing too much of the wrong things, and not enough of the right things.
https://www.usatoday.com/story/opinion/2020/01/27/coronavirus-donald-trump-made-us-less-prepared-joe-biden-column/4581710002/
Mark Z
Mar 30 2020 at 5:45am
I think this is definitely true in general (though I think it depends on how the president reacts specifically; e.g., if a president reacts strongly to a crisis, supporters will follow while opponents will dismiss it as misdirection; I’m amazed more presidents don’t do this; presidents during an external crisis are almost untouchable); I don’t whether if it’s true for Epstein. Some libertarians are just habitual optimists.
Michael
Mar 30 2020 at 5:55pm
There are a number of differences between the Ebola situation in the US in 2014 and the current global coronavirus pandemic.
I think the most salient difference, by far, is that there were a total of 11 cases of Ebola in the US in 2014, only 2 of which were contracted in the US, and only 2 of which (a different 2) were fatal.
So, at least one reasonable basis for thinking that our current situation is a much bigget deal than Ebola in 2014 is that the number of cases and deaths were insignificant. It’s kind of insulting to say that politicization is the only reason why Democrats would have been comfortable with Obama’s handling of one crisis and critical of Trump’s handing of this different crisis.
Trump, himself, and some in the conservative media were no optimist regarding Ebola in 2014.
Daniel Klein
Mar 30 2020 at 3:11am
Your title makes me think of The Deer Hunter, when Mike tells Stanley, “This is this.”
nobody.really
Mar 30 2020 at 4:27pm
I thought of Ayn Rand’s “A is A” from <I>Atlas Shrugged</i>.
Or perhaps <I>Fihi Ma Fihi</i> [It is what it is] by the 13 century writer Rumi.
Mark Z
Mar 30 2020 at 6:00am
Scott, I think one may need to look back at what pundits predicted during similar, previous crises to really know if they’re good at predicting. My concern is this: if one pundit consistently predicts a recession this year, and another consistently predicts there won’t be one, then even if 12/13 years the latter pundit is right, and the former is only right the 13th year when the recession hits, I think the tendency of the public and journalists is to portray the 1 for 13 pessimist as a prophet and the 12 for 13 optimist as a deluded Pollyanna. Robert Lucas gets pilloried for his remark before the Great Recession about the business cycle having been conquered; but then some of the prominent economists or pundits who are heralded as successfully predicting the recession had been predicting an immanent financial crisis since as far back as the mid-90s.
I suspect that someone predicting COVID19 would be a huge problem before late January was probably just prone to crisis predictions in general. Up to then at the very least, I think, the ‘it’ll be no big deal’ prediction was probably rational. Of course, if you lived in Wuhan, it might’ve been rational to predict a catastrophic global pandemic a few weeks before that even, but then it was made sure that we didn’t hear much from those people.
Scott Sumner
Mar 30 2020 at 2:55pm
I’m not so sure about that. The experts not only predicted a bad outcome, they told us WHY that bad outcome would occur. The pointed to the high R0, and the unacceptable death rate (albeit lower than SARS). And those two well understood facts have turned out to be exactly why the crisis got so bad. They didn’t just get lucky, they applied rational analysis
Countries that took them seriously are being hit much less hard.
Michael
Mar 30 2020 at 8:46pm
That’s the key point. This is a pretty solid (though harsh) critique of Epstein’s analysis:
https://rexdouglass.github.io/TIGR/Douglass_2020_How_To_Be_Curious_Instead_of_Contrarian_About_Covid19.nb.html?fbclid=IwAR0O2ofveESjceunkvG1M7NxlVWB7BTy80Ws1YGgV3i56RyKVyGkWHUF3II#fnref15
zeke5123
Apr 1 2020 at 9:05am
To channel Taleb, it doesn’t matter how often you are right or wrong, but the payoff / survival. If you are betting 100 to win 5, then you need to be right more than 12/13 to survive. Likewise, if you are betting 5 to win 100 you only need to be right 1/13 times.
Being right is not about frequency but about payoff.
Thaomas
Apr 2 2020 at 8:13am
Lucas was fairly pilloried because he did not make explicit his assumption of reasonable monetary policy. Ditto Krugman’s prediction of a recession in respect to the 2013 tax increase. He implicitly assumed that the Fed would not accommodate.
Thaomas
Mar 30 2020 at 7:42am
I think the process is quite clear. One expects that the policy reaction will be proportional to, but with costs far greater/lesser (including costs of empowering the “wrong” people to impose other future unrelated costs) than the harm being estimated. This is very clear in ACC “deniers” and “alarmists.”
Michael
Mar 30 2020 at 9:13am
Recent interview with Epstein.
https://www.newyorker.com/news/q-and-a/the-contrarian-coronavirus-theory-that-informed-the-trump-administration
To me, he’s pretty clearly way out over his skis (even disregarding the initial 500 deaths prediction). His lack of specific subject area expertise is on full display.
Roman Lombardi
Mar 30 2020 at 9:51am
I am reminded of Taleb’s observation that boils down to, “We often ask the wrong expert.” Unless Epstein has some hidden sheepskins laying around attesting to his work and training in epidemiology, statistics and Risk Management we would all do well to put him in the same category as state sponsored lotteries. “For Entertainment purposes only.”
Robert Schadler
Mar 30 2020 at 3:21pm
Prediction.
The word might be misused.
Think it worthwhile to go back a century to Frank Knight’s distinction between “risk” and “uncertainty” Risk is insurable because there is sufficient data to predict the likelihood of a specific incidence. Uncertainty is where there is not enough data/experience.
With a NOVEL virus, there is no data accumulated over time. One can make guesses, maybe even “projections” based on assumptions. Whatever the assumptions, it is then a matter of math.
There is, then, no basis for a prediction about something that is novel, or unique to our experience.
Extreme projections or “predictions” are likely to get more attention. That may be part of why the MSM likes it, as well as some academics. In today’s world, it may have an implied message for more government action — the worse something is the more we need the government to do something.
Mark Brady
Mar 30 2020 at 4:33pm
“Richard Epstein is a highly respected free market economist.”
This may seem a small point in the big scheme of things, but summarizing Richard Epstein’s qualifications as an “economist” doesn’t really capture his competencies. He has written a great deal on law, not least on economics and law, and he is a distinguished defender of the classical liberal order. But why call him “a highly respected free market economist” to the exclusion of anything else, when addressing a readership, some of whom may be unfamiliar with his work?
And, yes, it is unfortunate that all too many commentators and readers are apt to write off serious arguments that are offered by people whose ideology does not agree with that of the observer. That, of course, is true for all debate, not just on Covid-19.
David Henderson
Mar 30 2020 at 6:35pm
Scott,
Although Richard Epstein is economically quite literate, I would not call him an economist. He’s a brilliant legal scholar.
Scott Sumner
Mar 31 2020 at 1:27pm
Thanks for pointing that out. I guess I’m most familiar with his work on economic issues.
Mark Bahner
Mar 31 2020 at 8:44am
I’ll probably have many more comments on this post, but my first is that it’s an excellent post…as usual. Great work, Scott.
Theodore Lopez
Mar 31 2020 at 12:00pm
I think Epstein’s “gaffe” was he originally meant to say “5000” but said “500”. So it’s not a case of revision as much as a simple case of misspeaking.
Scott Sumner
Mar 31 2020 at 1:26pm
I accept that, although he never really explains why he predicted 5000/50000 for the US and the world. The US has 4% of the world’s population, and the severe part of the crisis hit us later than Asia or Europe. We had more time to prepare. The US usually has lower death totals from natural disasters than most countries, even developed places like Japan and Europe. Why so pessimistic about the US?
Epstein says that given our population it’s obvious he meant 5000. Actually it’s not obvious at all. Nonetheless, I like Epstein and take him at his word. One can be a brilliant legal scholar and not necessarily have a good intuitive sense of statistical data. But he never should have made the second set of predictions on March 24. By that time it was obvious that he had badly goofed.
Michael
Mar 31 2020 at 8:14pm
The problem has little to do with his numbers being off. The problem was that he very clearly didn’t understand either the models he was critiquing or the underlying biology of the system he purported to model. He engaged in cherry picking of data and unsupported assumptions about how the virulence of COVID-19 would evolve during the course of the pandemic.
Worse, in his intial paper, he offered a justification for how he came up with the number 500, and then just tacked a zero on when the death toll exceeded 500.
“it seems more probable than not that the total number of cases world-wide will peak out at well under 1 million, with the total number of deaths at under 50,000 (up about eightfold). In the United States, if the total death toll increases at about the same rate, the current 67 deaths should translate into about 500 deaths at the end.”
67*8=536. That appears to be how he got to his number.
In his revised prediction, he said:
“it seems more probable than not that the total number of cases world-wide will peak out at well under 1 million, with the total number of deaths at under 50,000 (up about eightfold). In the United States, if the total death toll increases at about the same rate, the current 67 deaths should reach about 5000).”
As Rex Douglass noted in the article I linked to above: “This is not great. This alters the prediction but does not bother to alter the logic which led to the calculation. Multiplying the then global death count by 8 would lead to a global prediction of 50,000 and so multiplying the then U.S. death count of 67 by 8 would be 536, hence the forecast of 500. Likewise the 50,000 global total is left unchanged. Simply adding zeros to the prediction every time it is proven wrong doesn’t alter the underlying model given in the same sentence.”
My guess, and I admit it is only a guess, is this: Epstein looked at the huge numbers of predicted cases and deaths, decided that that could not possibly be right, and set out to develop a model with more reasonable (to him) numbers. And that is what he did, but when he did so he left biology, epidemiology, and the scientific method behind. And when he got called out on it, he obfuscated.
I cannot speak to his work as a highly-regarded conservative legal scholar, but this particular work of his is irredeemably awful. And not because the number he got to at the end was wrong (sometimes it’s possible to learn from a flawed model, and all models -even good ones – must neceesarily be imperfect), but because of the approach he used to get there.
Mark Bahner
Apr 1 2020 at 1:32pm
Excellent comments. Especially these:
Here’s an article in the New Yorker in which Richard Epstein comes off very, very badly:
When bad thinking informs public policy
Mark Bahner
Apr 1 2020 at 1:41pm
Hi Michael,
Oh, I see you already linked to the New Yorker article. Sorry about that. He was indeed “Way out over his skis.” In fact, I think an even better analogy to skiing–and I know this dates me–would be that he was experiencing the “agony of defeat” in his New Yorker interview.
The agony of defeat
Michael
Apr 2 2020 at 9:03pm
The agony of defeat is unfortunately appropriate.
Comments are closed.