European fatality rates and lawfulness

Correlation ain’t causation, but nonetheless it is worth looking at correlation:

Via Daniel Wilson.  And here is a story about defiant Iranians.


The Germans are expecting their first true wave of cases to arrive in the next week to 14 days. And they have already changed the recommendations concerning how funerals are performed - no funerals in enclosed spaces, only outdoors, no throwing dirt on the casket as customary, and no open caskets.

And this is from the Spiegel international edition.

"We will be looking at the behavior of the population this weekend,” Chancellery head Helge Braun told DER SPIEGEL. "Saturday is a crucial day. We have our eye on that day in particular."

Braun referred to the fact that in the past week, despite repeated requests from politicians to avoid social contacts to the extent possible because of the risk of infection, many people were continuing to meet in groups - in public parks, for example. In her address to the nation on Wednesday night, Chancellor Merkel made an urgent appeal for people to adhere to the restrictions on public life imposed by the government. .... Braun stressed that the German government wants to avoid imposing a mandatory shelter-in-place policy.

"For now, we are counting on people understanding the measures and their readiness to restrict their social lives,” he said. "And when we look to our neighboring countries, which have already imposed curfews, it becomes clear: It would be an enormous additional burden.”

"After all, very few people, especially in cities, have a backyard or a large piece of property. That’s why we are calling on everyone to take to heart and implement the measures decided so far. And that means refraining from all social contacts except the nuclear family if possible." He also noted that government expects "that the public trusts us and will not allow itself to be driven crazy by fake news and that it will follow our recommendations rigorously.”

The Germans had their first cases at the same time as Italy. Actually the first Italian cases in Lombardy came from Bavaria.
They have have 2/5 the number of cases but < 1/50 the deaths. They also have 5 times the number of ICU beds as Italy. It’s unlikely it will get that bad there.

At a doubling time of about 6 days, five time the beds means a couple of weeks more before crisis.

That is pretty much what the German health system is expecting, and why the last week has been spent attempting to create as much new capacity as possible - while absolutely knowing that it will not be enough.

There is other evidence that social discipline plays a major role in health. Empirical work on antibiotic resistance that controls for other factors shows there remains a strong relationship between a lack of social discipline and a higher rate of antibiotic resistance. The relationship is arguably driven by factors such poorly regulated use and over use of antibiotics, but also breakdowns in other area such as poor management of sanitation and water quality, ineffective management of food storage and distribution, etc. that permit the drug resistant microbes to spread unchecked. For example, Collignon et al. report results over 70 countries that uses a Governance Index creates from four published measures of corruption, and finds Governance along with quality of infrastructure as the most factors explaining country level antibiotic resistance. .

@Tyler during the Econtalk crossover epp you said you're spending more time cooking at home. Given your expertise at dining out, I'd be very interested to hear a bit about what cuisines you cook at home and any rules of thumbs you might have for picking good recipes.

I'm not Tyler, obviously, but this is what I look for when evaluating a source of recipes:

1. For independent blogs, the less prominent the writer's life the better. Food blogs which spend the first half of an article writing about their family indicate that it's someone using food as a vehicle to perform a mind-dump rather than someone who takes the practice seriously.

2. Articles which reference their sources.
The more the writer mentions who they're inspiration is, the more likely they've put some proper thought into the dish. The more sources, the better. A sentence like "Traditionally parsnips are used but some local recipes call for swede or turnip." is a sign of someone who knows what they're talking about.

3. Sites which mention the history of the dish.
An extension of #2, but again, detailing the origin of the recipe shows that the author has done their research and takes food seriously, instead of a weekend hobby.

4. Avoid sites with a trendy theme.
"Paleo for One", "The Low-carb Vegan" are rarely quality sources, for hopefully obvious reasons.

5. Telling you why, not just how.
For me this is key. I don't put together a house full of IKEA furniture and call myself a carpenter, likewise you can't really call yourself a cook simply by following instructions. Any recipe which explains why you are performing a step is invaluable (and if you've adhered to the prior tips, the author isn't likely to be throwing out some "seal in the juices" nonsense). The most important part of cooking is understanding why, so you can then create your own quality dishes, or tinker with existing recipes to fit them to your tastes/larder correctly.

An excellent example is Hunter Gardener Angler Cook. Very well written by someone who has been in the game for a long time, used to be a professional writer and knows the dishes he presents inside out.

The culture that is Dutch: “if your survival chances are low, you’re probably not dying on the IC.” (Note that this by patient choice!)

Yes, the Dutch are incredibly fatalistic. They also don't believe much in flu shots or medicine short of a few doses of Paracetamol. That extends to the doctors, who often won't treat things that are readily treatable. Read the last two sentences of the article (translated): "If the worst case scenario models come out, we will have 500 to 1,000 Corona patients at the ICs next week. I hope it won't work out, but we better be ready for that. ”

What are they doing to get ready for that? There's no mention of it from the Dutch media or health ministry. Cases are increasing at about 20%/day. What it being called the "worst case" here is not that, it's pretty much a lock to happen. Factor in that about a quarter of all documented cases in the Netherlands are health care professionals (they're being tested more often - civilians aren't tested unless they have significant symptoms), and the system is going to break down. I give it 10 days max until the Dutch, who are proud that they don't panic, do just that.

These numbers do not mean anything right now because deaths lag infections.

However, if I wanted to spin a story out of this I would say that high rule of law index correlates with more widespread testing, and thus larger denominator. But it would be really stretching it, to clickbaiting news site level.

"Rule of law" indexes usually measure whether you can get a fair trial, whether officials are corrupt, whether police will protect your property against gangs, etc. What is needed here is a much more specific "obedience to law" index (Germans high, Italians low).

Agreed. FWIW, from the US Rule of Law Index is 0.72 to the left of Spain and the right of Italy on the x axis.

True for Norway, but isn't a get out clause for the German results.

If you predict Italy's fatality rate on a polynomial fit from either days since first recorded case or total cases (which is correlated with time), Germany's death rate is still low by two orders of magnitude.

51 days since first infection, so roughly 4 "death cycles" on.

Nope, "deaths lag infections" ain't gonna cut it here.

(The get out clauses there are "False positives", "Germany isn't counting cadavers!").

I mean, let's take literally the idea that deaths lag cases by about two weeks and this explains lower "Germanosphere" death rates (working definition here excluding UK+NLD).

Italy records on 18/03/2020, 2505 deaths , and two weeks earlier (14 days) on 04/03/2020, 2502 cases. This would imply a 100% fatality rate (which is of course completely absurd and above ebola, etc.).

Germany on the same day, 13 deaths and two weeks earlier 262 cases. 5% fatality, taking it literally that the case count is close to accurate (and not off by an order of magnitude) and it does take a full 14 days to progress to death.

(Note, if you instead extended case progression and death lag even longer, perhaps by a week, you would be coming up with death rates that are beyond 10x actual recorded cases, and that in itself would indict the idea that the case counts are even close to accurate on the same order of magnitude.)

There is a factor of 20 off here which is not explained by time lag.
Germany is of course the only "Germanosphere" nation with a long enough timespan of cases to compare here, so time will tell if it overlaps the same way elsewhere.

Perhaps the real correlate with mortality rates is not "lawfulness" per se, but rather "inclination towards semantic precision." If you ask an Italian to fill out a cause of death form, he/she will immediately try to figure out how to work the system in the most favorable way. A Norwegian, on the other hand, will spend ten silent, neurotic, melancholy minutes making sure he/she isn't making a mistake.

The skies cleared and I got to go out and do some yard work, which was very nice. I have a feeling I missed some things.

But here's the deal. The flaws in our leadership, and the flaws in our national health system (such as it is) are very much on topic.

They are both exactly about how we approach these problems going forward.

And sure, so is social responsibility. The idea that virtue might not just be signaling.

"Harvard is laying off nearly all dining workers. While the univ. has agreed to provide 30 days' pay for the directly hired dining workers who work at the College, they are refusing to provide this pay for the subcontracted dining workers at @Harvard_Law"

The system we have. While you're working you may have insurance. When you're laid off you may be able to acquire insurance. Assuming you have that safety fund set aside. And assuming for-profit insurance companies think it's a good time to keep the portals open.

And every single uninsured person who skips a doctor visit (because they aren't sure if they're sick or how they'll pay) drives infection rates.

Your desire to politicize a crisis is Krugmanesque. The leadership issues are especially a matter of captains of industry, our elites arranged it so that even our antibiotics are made in China. This predated President Trump.

Think about it again.

As we approach this election, isn't it exactly about how to deal with this crisis?

The entire theory of democracy is that we choose our leadership, including and especially to solve the problems of our age.

Portugal and Austria have low death rates as well. What's with the cherries?

Austria went into lockdown a week before Bavaria did. The rest of Germany still hasn't.

US rank appear to be 0.71 on this Rule of Law index
and yes, correlation is not causation

Now plot death rate versus indictments against the chief executive.

I'm thinking that the effectiveness of the Federal government will be about the proportion of the propriety of FISA applications. And the care and attention on issues that matter will be about as good as those judges.

I wonder if the low crime countries in this index do actually have as high a hurdle for investigations.

There was an earlier post on here about Japan's mysterious success in mitigating the virus's effects. I assume Japan also scores high on Rule of Law?

Regarding correlation and causation. I assume lawfulness was measured before the pandemic. How would expectations about future mortality during an at-the-time unexpected pandemic cause changes in lawfulness? Can we at least agree that, if there is causation, it runs from lawfulness to mortality and not vice versa? In other words, people aren't breaking the law to avoid death nor are they asking why bother following the law if they're going to die anyways.

Japan is 0.78 and the US is 0.71. As you'd expect, Australia is about the same as Japan at 0.80.

The reason why the spread of the Coronavirus hasn't been nearly as bad as it could have been is because people took some precautions there.

There are two countries where Coronavirus has been turned around. China and South Korea. Both of these had high rates of face mask wearing and this is also high in Japan.

China engaged in lockdown measures now being followed in many other countries while South Korea did extensive testing and intrusive surveillance and contact tracing.

Japan is an interesting case and I suspect we still have some things to learn about why they have been successful. Mask-wearing, no hand-shaking, and a pre-existing culture of hygiene and rule-following probably all helped. Also, maybe even the high levels of introversion, loneliness and low levels of Christianity or Islam may have helped slow the spread.

Tyler your puritanical side shows itself again

The problem with this is that these sorts of indexes simply are not precise enough to measure differences between Norway and Belgium. Between Norway and Mozambique? Sure, a difference in the rule of law index between the two countries probably reflects some real, on the ground, difference in what it purports to measure. But not differences among Western liberal democracies.

That’s quite an amazing chart, causation or not. It highlights cultural differences. If the government in Austria or Norway issues a health recommendation, people will treat it like a rule and comply. Italians will treat it like an appeal not to engage in jaywalking - it’s something to be ignored. Unfortunately, it may turn out that places like New York are more like Italy.

Utah should do well. People are conscientious, and they have a basement full of food.

Maybe you might interpret low level of Rule of Law as healthy skepticism toward pompous asses in uniform and their politician bosses.

If so, I am afraid to feel like a collectivist for a while, but I can’t help but being proud of the piece of dirt I happen to be born on, however stupid the feeling is.

Agree with M that this looks like cherry picking. Brazil has a very low death rate yet "rule of law" is far lower than Italy. Russia's death rate is microscopic, I have not noticed that they excel in rule of law.

Eleven data points. I'm not convinced that this is a meaningful correlation (never mind causation).

Please understand that if you have no idea what the following phrase means:

" The 'cause in fact' or the 'but for' test with which all law students who have to sit through 'first year' tort law is well reflected in the difference between Italian coroner reports and German coroner reports regarding the coronavirus named after last year."

if you have no understanding of what that sentence means.... then you have no right to opine as if you know what you are talking about on what the chart in question might possibly mean, and you might in fact be one of those unthinking bigots who think they understand the world but do not.

so if the actual deaths, in any given jurisdiction, that were actually caused by the 2019 coronavirus (passing the 'cause in fact' and 'but for' tests) ---

then this graph, with roughly a two to one slope, establishes something between approximately zero percent to approximately six hundred percent of what it purports to show.

to be fair, I translated it into German and then back into English, but still you should get it.

Or maybe the hidden variable is state capability rather than size (ala Italy and UK)


The x axis is distance from the equator

Yes, but north vs. south Italy???

Once again, Southern Europeans have shown cultural and possibly genetic inferiority to Northern Europeans.

Data is cherry picked. Put Eastern EU on the Graph and there will be no correlation

Cultures in eastern EU, on a global scale, are very rule-based. Not as rule-based as Germans or East Asians, but on par with Scandinavians. It's just that the rule of law is not the only source of rules they're following. Non-conformist mavericks are almost non-existent.

Central Europe is very ordnung, yes. "Papiere bitte". Southeast Europe less so. This shows up very clearly on things like uncertainty avoidance or tightness-looseness scales.

On the other hand, "rule of law" is not so much about how much people abide by the law, but law as an alternative to charismatic and populist leadership, and as a check on the powerful personalities within government who come up with all sorts of constitutional schemes to extend their power. Rule of law is about how much powerful people are constrained by the law. How much law really rules, and despots and technocrats don't. Not how much commoners follow rules rigidly. On this measure East-Central Europe is absolutely certainly not Scandinavian (though not as far from them in the grand global scheme as we could imagine).

MR specializes in cherry picked presentations, they are a staple on the menu.

This seems to be a cherry pick slight of hand assertion with only a handful of data points. COVID.19 is just being spread for a couple of months if not weeks, the mortality rate cannot be confidently established yet, beside that is a reflection of the health care system not rule of law. Using log(NConfirm per MPop) as the dependent variable with 111 data points shows a completely different picture. It is statistically very very significant but the correlation is negative, i.e. increasing Rule of Law increases the per capita COVID.19 infection rate.

log(NConfirmMPop) = +11.8*ROLndx -4.71; #n=111; Rsq=0.498; p=0 *** (VVSig)

Sorry. Positive correlation

Let's also do a graph of European fatality rate vs. desirability as a mid-winter holiday destination. I wonder if it wouldn't look the same.

Of maybe European fatality rate vs. desirability of native food culture.

Oh, yeah, the article on religious nut jobs in Iran. Two coworkers from different eastern European countries tell me that in their home countries the Orthodox Christian Church is taking a similar view - that the church is a healing place, that the government is giving out misinformation, etc.

And of course we're seeing anecdotal reports of freedom loving Americans with remarkably similar beliefs and defiance. But hey, let's focus on those other people.

The Christmas Market effect vs Southern Europe being not actually that
warm in winter*, will mess with your prediction ;)

*People in Europe often go to the Canary Islands and the like instead these days, though I guess they are nominally Spanish.

My point was that the graph Tyler reposted was nonsensical. The fatality rate is clearly tied to the duration of high case levels (whoever gets a lot of cases first is going to be first to a lot of fatalities) and to the stress load on the medical and public health systems (again, whoever gets a lot of cases first . . .). My suggestion is that winter cross-boarder travel destinations were in fact a real reason that the pandemic hit Italy and Spain and France hard and earlier.

It's well known that the ski resorts in northern Italy and in France were the draw for a lot of infected travelers. Spain is also an highly desirable winter travel destination. Norway and Denmark don't attract so many faraway travelers in January/February. Basically, that graph tracks farily well where people like to travel to in the post Christmas season, for different reasons. The disease wasn't spreading much before Christmas, so I don't understand the mention of Christmas Markets. And the graph doesn't include warm weather winter destinations like the Caribbean, Canary Islands, etc., so not sure why you mention that.

And just in case you're confused, my suggestion of graphing fatalities with good cuisine was to show how easy it is to find spurious correlations, as the original graph also demonstrates.

I'm doing a riffing on your joke with another, guy.

Wait, you were serious that travel to Southern Europe was likely to explain early infection rates? Lmao.

Besides the time lag included in the naive version of the mortality rate, there is also the question on how deaths are attributed to various causes. If the direct cause (say, heart failure) happened because of covid, it might not get reported as covid-related death (and yes, counting indirect causes is hard and subjective).

Does lawfulness and social cohesion correlate with mask wearing?

It seems that citizens of most eastern countries are wearing face masks in this crisis. But here in the west we are told that the standard face masks are not useful for healthy people, and that the useful masks (N95 etc) are, currently at least, best reserved for medical staff where they are desperately needed.

But the citizens of those countries are not stupid, so why are they wearing these masks?
Is it virtue signalling? Does wearing masks encourage a wider compliance with hygiene and social distancing? Or what?

Also, given that the true rate of infection will likely be higher than that revealed by testing so far, might it be that the wearing of these masks is in fact helping reduce R0, by preventing asymptomatic wearers from spreading droplets?

This is a very bad chart. I can't comment on the Lawfulness measure (though I hate these subjective measures in general), but the mortality rate is garbage. Every country has a different testing rate and different criteria for reporting COVID-19 deaths. Shame on Tyler for sharing this and not including caveats.

Don't forget the difference between population density and average density of the square mile the median person lives in. Germany has a lot of people, but they live in far less dense urban areas than Spaniards and Italians: It's not a matter of looking at entire countries, but places dense enough to have a really high R0. On top of that, Spain had plenty of large marches during the first cases' incubation period, leading to horrible early spreads.

This is why I hold far more hope for the US' Midwest: The contact networks are so much sparser.

This is why I suspect

What about mortality rate and sunny days or mortality rate and olive oil consumption?

Interesting though this is about what one would expect if one guessed, no? -- that chaotic and poorly governed countries will fare worse than better governed ones. And the US will probably be somewhere in the middle -- with most of Africa and the middle east on one end and parts of Asia and Europe on the other. And no, correlation does not show causation, but often it does.

One other angle on this; consider the your plot is almost (not quite, almost) the inverse of European life expectancy.

In general, Italians and Spaniards live longer than Germans and the Nordic countries. Together with Italy having higher population aging, their oldest old people at the right tail are literally older.

Since this thing increases chance of death pretty linearly through the lifespan, that may be worth some consideration.

I clearly think most of this is just testing frequency, but this is worth consideration.

Related, it's kind of adorable to see the Nords about two weeks ago imagining that their low death rate is due to their healthy practices and the fact that their old people are just so darn young and vigorous -

The Nordics have "a population that is generally healthy and vaccinated against illnesses, universal health care, not many smokers or much industrial pollution," said Oystein Olsvik, a professor of medical microbiology at Tromso University in northern Norway.

"The Scandinavian population is generally less susceptible to this kind of illness that can develop in densely-populated regions of China," he added.

In addition, an elderly person in Scandinavia is generally in good health.

"When you're 80 in China, you're really old, whereas a Norwegian in their 80s can complete the Birken (a famed cross-country ski race) or cycle from Oslo to Trondheim," noted professor Olsvik.

And yet they generally have lower life expectencies and generally die younger than the Italians with high apparent death rates.... Industrial pollution and vaccination isn't a differentiating factor with Italy.... etc.

Defiant Iranians is not helpful in supporting the point Tyler is making above

If you you look at timeseries of cases/deaths you see consistent exponential growth at the same rate for all the countries in Tyler's plot, whilst Iran achieved the most dramatic levelling off outside of Asia.

I keep being surprised it seems that Tyler and Alex are not following the numbers that carefully, just sharing mood affiliations.

Download this stuff, plot it, and make sure that your narratives match what you see!

Comments for this post are closed