Category: Uncategorized
An econometrician on the SEIRD epidemiological model for Covid-19
There is a new paper by Ivan Korolev:
This paper studies the SEIRD epidemic model for COVID-19. First, I show that the model is poorly identified from the observed number of deaths and confirmed cases. There are many sets of parameters that are observationally equivalent in the short run but lead to markedly different long run forecasts. Second, I demonstrate using the data from Iceland that auxiliary information from random tests can be used to calibrate the initial parameters of the model and reduce the range of possible forecasts about the future number of deaths. Finally, I show that the basic reproduction number R0 can be identified from the data, conditional on the clinical parameters. I then estimate it for the US and several other countries, allowing for possible underreporting of the number of cases. The resulting estimates of R0 are heterogeneous across countries: they are 2-3 times higher for Western countries than for Asian countries. I demonstrate that if one fails to take underreporting into account and estimates R0 from the cases data, the resulting estimate of R0 will be biased downward and the model will fail to fit the observed data.
Here is the full paper. And here is Ivan’s brief supplemental note on CFR. (By the way, here is a new and related Anthony Atkeson paper on estimating the fatality rate.)
And here is a further paper on the IMHE model, by statisticians from CTDS, Northwestern University and the University of Texas, excerpt from the opener:
- In excess of 70% of US states had actual death rates falling outside the 95% prediction interval for that state, (see Figure 1)
- The ability of the model to make accurate predictions decreases with increasing amount of data. (figure 2)
Again, I am very happy to present counter evidence to these arguments. I readily admit this is outside my area of expertise, but I have read through the paper and it is not much more than a few pages of recording numbers and comparing them to the actual outcomes (you will note the model predicts New York fairly well, and thus the predictions are of a “train wreck” nature).
Let me just repeat the two central findings again:
- In excess of 70% of US states had actual death rates falling outside the 95% prediction interval for that state, (see Figure 1)
- The ability of the model to make accurate predictions decreases with increasing amount of data. (figure 2)
So now really is the time to be asking tough questions about epidemiology, and yes, epidemiologists. I would very gladly publish and “signal boost” the best positive response possible.
And just to be clear (again), I fully support current lockdown efforts (best choice until we have more data and also a better theory), I don’t want Fauci to be fired, and I don’t think economists are necessarily better forecasters. I do feel I am not getting straight answers.
Monday assorted links
1. Why isn’t the public more supportive of free trade?
2. How easy the spread on college campuses?
3. A bunch of medical claims, interesting, neither endorsing nor damning.
7. Are children’s face-to-face social skills declining?
8. “Church of England moves valuables to Tower of London amid fears of lockdown looting.”
9. Robin Hanson poll-based risk indicator.
10. An epidemiological account of how risk-spreader heterogeneity matters.
11. Harvard to sell $1.1 billion bonds. And pending cuts to higher ed.
12. Stephon Marbury, prophet for the NBA (WSJ). And Scott Gottlieb on employer testing (WSJ).
Sunday assorted links
1. Excellent thread, showing an actual understanding of public choice.
2. The mask as fashion statement.
3. Scott Sumner on the lower mortality estimates.
4. AMC movie theatres likely to go bankrupt soon.
6. The Belgium heterogeneity? And the Austrian heterogeneity?
Happy Easter
Credit: Saint Simeon Stylites the elder. Credit: Wellcome Collection. Attribution 4.0 International (CC BY 4.0)
What does this economist think of epidemiologists?
I have had fringe contact with more epidemiology than usual as of late, for obvious reasons, and I do understand this is only one corner of the discipline. I don’t mean this as a complaint dump, because most of economics suffers from similar problems, but here are a few limitations I see in the mainline epidemiological models put before us:
1. They do not sufficiently grasp that long-run elasticities of adjustment are more powerful than short-run elasticites. In the short run you socially distance, but in the long run you learn which methods of social distance protect you the most. Or you move from doing “half home delivery of food” to “full home delivery of food” once you get that extra credit card or learn the best sites. In this regard the epidemiological models end up being too pessimistic, and it seems that “the natural disaster economist complaints about the epidemiologists” (yes there is such a thing) are largely correct on this count. On this question economic models really do better, though not the models of everybody.
2. They do not sufficiently incorporate public choice considerations. An epidemic path, for instance, may be politically infeasible, which leads to adjustments along the way, and very often those adjustments are stupid policy moves from impatient politicians. This is not built into the models I am seeing, nor are such factors built into most economic macro models, even though there is a large independent branch of public choice research. It is hard to integrate. Still, it means that epidemiological models will be too optimistic, rather than too pessimistic as in #1. Epidemiologists might protest that it is not the purpose of their science or models to incorporate politics, but these factors are relevant for prediction, and if you try to wash your hands of them (no pun intended) you will be wrong a lot.
3. The Lucas critique, namely that agents within a model, knowing the model, will change how the model itself operates. Epidemiologists seem super-aware of this, much more than Keynesian macroeconomists are these days, though it seems to be more of a “I told you that you should listen to us” embodiment than trying to find an actual closed-loop solution for the model as a whole. That is really hard, either in macroeconomics or epidemiology. Still, on the predictive front without a good instantiation of the Lucas critique again a lot will go askew, as indeed it does in economics.
The epidemiological models also do not seem to incorporate Sam Peltzman-like risk offset effects. If you tell everyone to wear a mask, great! But people will feel safer as a result, and end up going out more. Some of the initial safety gains are given back through the subsequent behavioral adjustment. Epidemiologists might claim these factors already are incorporated in the variables they are measuring, but they are not constant across all possible methods of safety improvement. Ideally you may wish to make people safer in a not entirely transparent manner, so that they do not respond with greater recklessness. I have not yet seen a Straussian dimension in the models, though you might argue many epidemiologists are “naive Straussian” in their public rhetoric, saying what is good for us rather than telling the whole truth. The Straussian economists are slightly subtler.
4. Selection bias from the failures coming first. The early models were calibrated from Wuhan data, because what else could they do? Then came northern Italy, which was also a mess. It is the messes which are visible first, at least on average. So some of the models may have been too pessimistic at first. These days we have Germany, Australia, and a bunch of southern states that haven’t quite “blown up” as quickly as they should have. If the early models had access to all of that data, presumably they would be more predictive of the entire situation today. But it is no accident that the failures will be more visible early on.
And note that right now some of the very worst countries (Mexico, Brazil, possibly India?) are not far enough along on the data side to yield useful inputs into the models. So currently those models might be picking up too many semi-positive data points and not enough from the “train wrecks,” and thus they are too optimistic.
On this list, I think my #1 comes closest to being an actual criticism, the other points are more like observations about doing science in a messy, imperfect world. In any case, when epidemiological models are brandished, keep these limitations in mind. But the more important point may be for when critics of epidemiological models raise the limitations of those models. Very often the cited criticisms are chosen selectively, to support some particular agenda, when in fact the biases in the epidemiological models could run in either an optimistic or pessimistic direction.
Which is how it should be.
Now, to close, I have a few rude questions that nobody else seems willing to ask, and I genuinely do not know the answers to these:
a. As a class of scientists, how much are epidemiologists paid? Is good or bad news better for their salaries?
b. How smart are they? What are their average GRE scores?
c. Are they hired into thick, liquid academic and institutional markets? And how meritocratic are those markets?
d. What is their overall track record on predictions, whether before or during this crisis?
e. On average, what is the political orientation of epidemiologists? And compared to other academics? Which social welfare function do they use when they make non-trivial recommendations?
f. We know, from economics, that if you are a French economist, being a Frenchman predicts your political views better than does being an economist (there is an old MR post on this somewhere). Is there a comparable phenomenon in epidemiology?
g. How well do they understand how to model uncertainty of forecasts, relative to say what a top econometrician would know?
h. Are there “zombie epidemiologists” in the manner that Paul Krugman charges there are “zombie economists”? If so, what do you have to do to earn that designation? And are the zombies sometimes right, or right on some issues? How meta-rational are those who allege zombie-ism?
i. How many of them have studied Philip Tetlock’s work on forecasting?
Just to be clear, as MR readers will know, I have not been criticizing the mainstream epidemiological recommendations of lockdowns. But still those seem to be questions worth asking.
Saturday assorted links
1. MIE: “This Man Owns The World’s Most Advanced Private Air Force After Buying 46 F/A-18 Hornets.”
2. Romer tweet storm states his plan.
4. Is American innovation speeding up? (WSJ)
6. Non-exemplary lives (ouch). And what do the humanities do in a crisis?
7. Instagram strippers (NYT). And Bret Stephens: our regulatory state is failing us (NYT).
8. “Believe women,” selectively.
9. BloombergQuint on Alex and Shruti.
10. A proposal for releasing British young people (ever listen to early Clash?).
11. Arnold Kling annotates (and likes) my Princeton talk.
12. A Swede explains Sweden to an Israeli: “Some maintain that the Swedish policy can succeed only in Sweden, because of its distinctive characteristics – a country where population density is low, where a high percentage of the citizenry live in one-person households and very few households include people over 70 cohabiting with young people and children. Those are mitigating circumstances which the Swedes hope will work to their advantage.”
What should I ask Adam Tooze?
I will be doing a Conversation with him, no associated public event. He has been tweeting about the risks of a financial crisis during Covid-19, but more generally he is one of the most influential historians, currently being a Professor at Columbia University. His previous books cover German economic history, German statistical history, the financial crisis of 2008, and most generally early to mid-20th century European history. Here is his home page, here is his bio, here is his Wikipedia page.
So what should I ask him?
Friday assorted links
1. Balaji on heterogeneities and data integration.
2. Citizen’s handbook for nuclear attack and natural disasters. Do we need a new version of this?
3. The Amazon: “We show that, starting at around 10,850 cal. yr BP, inhabitants of this region began to create a landscape that ultimately comprised approximately 4,700 artificial forest islands within a treeless, seasonally flooded savannah.”
4. How much distance do you need when exercising? And against crowded spaces.
5. Dan Wang letter from Beijing in New York magazine.
6. Trump pushing to reopen by May 1.
7. Lots of new testing results from Germany, consider these as hypotheses but still a form of evidence.
8. Good and subtle piece on Tiger King (NYT). And betting markets in everything.
9. The Vietnamese response seems pretty good so far.
10. Joe Stiglitz discusses his love of fiction (NYT)
12. Ronald Inglehart on the shift to tribalism.
13. Explaining the Fed lending programs.
14. MIT Press preprint of new Joshua Gans book on Covid-19, open for public comment.
Does working from home work?
Better than you might think. Here is a paper from a few years back, by Nicholas Bloom, James Liang, John Roberts, and Zhichun Jenny Ying:
A rising share of employees now regularly engage in working from home (WFH), but there are concerns this can lead to ‘‘shirking from home.’’ We report the results of a WFH experiment at Ctrip, a 16,000-employee, NASDAQ-listed Chinese travel agency. Call center employees who volunteered to WFH were randomly assigned either to work from home or in the office for nine months. Home working led to a 13% performance increase, of which 9% was from working more minutes per shift (fewer breaks and sick days) and 4% from more calls per minute (attributed to a quieter and more convenient working environment). Home workers also reported improved work satisfaction, and their attrition rate halved, but their promotion rate conditional on performance fell. Due to the success of the experiment, Ctrip rolled out the option to WFH to the whole firm and allowed the experimental employees to reselect between the home and office. Interestingly, over half of them switched, which led to the gains from WFH almost doubling to 22%. This highlights the benefits of learning and selection effects when adopting modern management practices like WFH.
Via Matt Notowidigdo. Of course in that paper, the schools were not all closed…
The economics of supply cut-offs
As a number of people have pointed out, cable TV, cable internet connections, and cable streaming have been remarkably robust throughout this crisis. Why might that be? Let’s think through a few basic points about the economics of supply cut-offs. This will not be a complete model, but it will focus attention on perhaps one possible factor of interest.
Imagine a seller with market power who comes close to perfect price discrimination. That supplier will take great care to avoid supply cut-offs (imagine an electric utility investing in emergency capacity, for instance). If a cut-off were to happen, the profits of the supplier would be much lower. As a first-order approximation, such suppliers will invest a near-optimal amount of resources to prevent such supply interruptions.
Alternatively, imagine a nearly perfectly competitive situation where all of the surplus goes to consumers and producers earn the going rate of return. Fixed costs are not significant. A market collapse or supply cut-off doesn’t cut much into profits, and in essence the suppliers do not care about the losses of the inframarginal consumers, were a supply interruption to occur.
As a simple theorem, if the market is good for the producers in the first place, supply interruptions are less likely. If the market is good for consumers in the first place, supply interruptions are more likely.
Might this also apply to health care systems? The U.S. hospital system, in normal times, spends way too much. Still, it has the “cultural mentality” for making capital expenditures, far more than say Britain’s NHS does. And so the United States has far more ICU units per capita than does Britain. Whether justly or not, the U.S. health care system might come out of this crisis looking not entirely bad.
Fiction and classics to read under lockdown
A number of you asked me for a list of books to read during lockdown, mostly novels and fiction (like Plato, right?). Here is a list I drew up maybe fifteen (?) years ago, with only slight revisions since. I feel a current list might be quite different, but actually the early list is perhaps closer to most of your tastes? Here it is. It starts with classics and then goes through more recent novels maybe up through 2000 or so.
Thursday assorted links
1. “We are at a critical juncture for the market.”
2. Pandemic insurance for Wimbledon cancellation.
3. Borjas on who is undertested, from NYC data.
4. Japanese cook draws every meal he eats.
5. How to close a bag of chips with no clip.
7. How is the Swedish approach working out?
8. Re-entry stickers for the Florida Keys — get the picture?
9. Stapp and Watney, masks for all.
10. Hong Kong quarantine diary.
11. How the Faroe Islands aced it (so far).
12. “Many brands are using keyword blocklists to stop their adverts appearing next to stories about Covid-19, meaning that even though news websites are getting record traffic from readers they are barely earning any money from the clicks.” Link here.
13. The Pandemic Challenge, from Singularity University.
14. Will Covid-19 induce a decline in religiosity?
Where we stand
I thought it useful to sum up my current views in a single paragraph, here goes:
I don’t view “optimal length of shutdown” arguments compelling, rather it is about how much pain the political process can stand. I expect partial reopenings by mid-May, sometimes driven by governors in the healthier states, even if that is sub-optimal for the nation as a whole. Besides you can’t have all the banks insolvent because of missed mortgage payments. But R0 won’t stay below 1 for long, even if it gets there at all. We will then have to shut down again within two months, but will then reopen again a bit after that. At each step along the way, we will self-deceive rather than confront the level of pain involved with our choices. We may lose a coherent national policy on the shutdown issue altogether, not that we have one now. The pandemic yo-yo will hold. At some point antivirals or antibodies will kick in (read Scott Gottlieb), or here: “There are perhaps 4-6 drugs that could be available by Fall and have robust enough treatment effect to impact risk of another epidemic or large outbreaks after current wave passes. We should be placing policy bets on these likeliest opportunities.” We will then continue the rinse and repeat of the yo-yo, but with the new drugs and treatments on-line with a death rate at maybe half current levels and typical hospital stays at three days rather than ten. It will seem more manageable, but how eager will consumers be to resume their old habits? Eventually a vaccine will be found, but getting it to everyone will be slower than expected. The lingering uncertainty and “value of waiting,” due to the risk of second and third waves, will badly damage economies along the way.
So there you have it.
My Conversation with Emily St. John Mandel
I am a fan of her two latest novels Station Eleven (about a post-pandemic world) and The Glass Hotel, and many other smart people like them too. Here is the audio and transcript. Here is the CWTeam summary:
She joined Tyler to discuss The Glass Hotel, including why more white-collar criminals don’t flee before arrest, the Post Secret postcard that haunts her most, the best places to hide from the Russian mob, the Canadian equivalent of the “Florida Man”, whether trophy wives are happy, how to slow down time, why she disagrees with Kafka on reading, the safest place to be during a global pandemic, how to get away with faking your own death, how A Canticle for Leibowitz influenced her writing, the permeability of moral borders, what surprised her about experiencing a real pandemic, how her background in contemporary dance makes her a better writer, adapting The Glass Hotel for a miniseries, her contrarian take on Frozen II, and more.
By the way, I would fake my own death by going on a cargo vessel and bribing them to claim I fell overboard. Here is one bit about the pandemic:
COWEN: Have people been more or less cooperative than you had thought?
ST. JOHN MANDEL: My impression — and the problem is, we don’t see people anymore — but my overall impression is they’ve been more cooperative.
Definitely in the literary community, I’ve seen a lot of people really trying to support their independent bookstores, which has always been a thing. But I think there’s been a greater awareness that if you don’t buy your books from your independent bookstore — and by the way, they do all sell online mostly — then that store might not be there when all of this ends. So I see people pulling together like that, to try to support the businesses they love. That’s been a major one.
I wish I could see people and bring back a report from actual humanity, [laughs] but that is my impression. There’s been more cooperation.
And:
COWEN: In so many postapocalyptic novels, it seems that people wander a lot. Do they wander too much? Should they just stay put?
ST. JOHN MANDEL: I had this conversation with another postapocalyptic novelist. Would everybody stop walking? Why is everybody wandering endlessly in a postapocalypse?
And:
COWEN: How good is Frozen II, if I may ask?
ST. JOHN MANDEL: It’s pretty good.
COWEN: Pretty good?
ST. JOHN MANDEL: Yeah. This is a controversial statement. I know a lot of parents who hate it, but I find it more interesting than Frozen I.
Perhaps I like The Glass Hotel a wee bit better than Station Eleven, but maybe Station Eleven is better to read first?
Happiness and the quality of government
From John F. Helliwell, Haifang Huang, and Shun Wang:
This chapter uses happiness data to assess the quality of government. Our happiness data are drawn from the Gallup World Poll, starting in 2005 and extending to 2017 or 2018. In our analysis of the panel of more than 150 countries and generally over 1,500 national-level observations, we show that government delivery quality is significantly correlated with national happiness, but democratic quality is not. We also analyze other quality of government indicators. Confidence in government is correlated with happiness, however forms of democracy and government spending seem not. We further discuss three channels (including peace and conflict, trust, and inequality) whereby quality of government and happiness are linked. We finally summarize what has been learned about how government policies could be formed to improve citizens’ happiness.
Having read through the paper, I thought the main interesting result was that quality of service provision (effectiveness, rule of law, regulatory quality, and absence of corruption) is correlated with happiness whereas kind of democracy is not, with the latter democracy variable being an index related to voice, accountability, stability, and freedom from violence.
Of course it would be very interesting to rerun such a test during pandemic times.