Hempton: Do Sweden or Melbourne, just not America

Advertisement

Below is an excerpt from Bronte Capital’s John Hempton and his most recent client note:

Viruses in nature

Epidemiologists have modelled viruses in nature – say – to use an Australian example – myxomatosis amongst rabbits, or phages in bacteria. They have a curve – familiar in all the text-books – which is exponential growth on the way up until something interferes with viral growth (mostly herd immunity) and sharp decline on the way down. Observed reality closely matches these mathematical models, indicating the models are sound.

Incidentally these models work regardless of the severity of the disease. If the disease causes mild sickness it will spread through the population of animals roughly the same way as it would if the disease were mostly fatal. Speed at which the disease spreads is a more notable parameter in these models than severity of the disease.

The virus in humans

The early stages of this virus in the West matched these models quite closely. This was particularly true in February where the spread was below the threshold of public perception. What was spreading in February was a silent disease that would wind up killing
a large number of people.

Since the latter part of March, no country has had a virus infection/fatality curve that matched these mathematical models. And the new curves are being interpreted according to one’s preferred ideology. A libertarian interpretation is that the virus isn’t that bad and you should eliminate lockdown. Sweden is the favorite example for an anti-lockdown
stance. Alternatively, those with opposing points of view observe the lower death rates as indicative of the success of public policy and maintain a pro-lockdown stance. Sweden is characterized instead as a failure, with a higher death rate than its nearest neighbors (Norway, Finland).

We argue that the new curves do not indicate anything much about the virus. They are just what you would expect to happen (although to be fair we did not expect it at the time).

The idea is simple. For any individual there is a risk to social activity. Getting on a bus entails risk. Singing in a choir much more risk. Working from home and minimizing exposure entails very little risk.

And there is a benefit from taking risk. You might be able (or need to) go out to work to earn a living and feed your family. You might have a good time at a party (or choir practice). Any individual will make a decision (partly in knowledge, partly in ignorance) that equilibrates the perceived risk of doing something with the perceived benefit of doing something.

This is true across society.

As the virus becomes more common the real and perceived risk of going out becomes higher. People will choose to stay at home or take less risk. As the virus becomes less common the risk of going out and doing things is lower so risk taking will increase.

The end result is a rough equilibrium where the virus goes sideways. To use the jargon, R0 tends to one. This seems to happen everywhere. (The idea has been expounded elsewhere – see this article by Joshua Gans.)

Incidentally again R0 tends to one irrespective of the severity of the virus. It will tend to one with Ebola where you bleed from the eyes and there is a 60-90 percent mortality. It will tend to one with SARS (which had about ten percent mortality) and it will tend to one with COVID-19 which is far less nasty than either of those diseases. (Obviously this generalization does not work in extrema. If the virus is too mild to bother avoiding like the common cold R0 will not tend to zero. Likewise, if the virus were airborne and transmissible at a huge range then it might be too contagious for behavior to slow it, and R0 will not tend to zero.)

That the virus is going sideways is neither proof that the virus is much less dangerous than the public are led to believe (a belief common amongst our libertarian friends). Nor is it proof that the public health measures (such as lockdown) are working and the solution is stronger public health measures.

It took us a while to realize this. Almost all of Johns personal (and very smart and very opinionated) friends are wrong (regardless of political persuasion and policy recommendation) and for the most part we were wrong too.

Having accepted that R0 tends to one, the question is how long does it tend to one for. And the answer seems to us to be until it is interrupted, either by herd immunity or by artificially generated herd immunity (i.e. a vaccine) or by enforced lockdown. If there is no vaccine it will tend to one for several years and the bulk of the population will eventually get it.

If that were Ebola (60-90 percent mortality) that would be an unbelievably bad result. If this had a mortality of 10bps (a wild underestimate of the current virus) then it would be tragic but would not threaten civilization as we know it.

If this were Ebola we would really want to eliminate the virus. Eventual outcomes where half the population dies are not acceptable. Fortunately fear alone will drive Ebola to manageable case numbers, simply because people would be justifiably very scared of the virus. But fear alone will not eliminate the virus because once case numbers are low people will take more risks.

To solve the behavioral dilemma when people are no longer scared of the virus you need them to behave as if they are still scared of the virus. You might be able to do this by brainwashing (i.e. convince people risks are higher than they perceive) or by massive social pressure, but the easiest way is probably just to make them scared of the police. That is how China controlled the virus and is how Melbourne, Australia is currently controlling the virus.

In other words enforced lockdown works.

That doesn’t mean that enforced lockdown is the right policy. It would be with Ebola (huge mortality). It almost certainly was the right policy with SARS (it worked, and mortality was high). But it is not the right policy with say a seasonal influenza. We will leave open where this virus sits (although we have our personal views).

You can however control this virus if you are prepared to trample civil liberties enough. China did. Suffice to note authoritarianism is good for the virus and the virus is good for authoritarians. The virus has cemented Xi’s power in China for instance.

Again, note that all of this is true for a wide range of virus severity.

The behavioral model explains some cross-country comparisons

The behavioral model is extremely useful for cross-country comparisons of the virus. Sweden famously never enforced any lockdown. It had an explosion in virus cases which rapidly ameliorated, and case numbers went sideways at fairly low levels. America did have a wide range of variable lockdown policies (but relative to say Melbourne weak policies with weak enforcement). It also had an explosion of cases, but the virus is now going roughly sideways at levels much higher than Sweden.

Early in the outbreak we thought that Sweden’s policy would be a spectacular disaster. We were wrong. Libertarians however have argued that the relative success of the Swedish policy (versus say the American policy) means that if you had adopted the Swedish policy in America you would have got better results in America. They are likely wrong too.

The reason is obvious. Sweden is one of the strongest welfare states in the world. America is the weakest welfare state amongst rich countries. The cost of not going out (i.e. not taking risk) in Sweden is largely that you collect (generous) welfare. The cost of not going out (and hence avoiding risk) for a low-income American is that they and their family starve.

Americans – especially poor Americans – will take more risk because the lack of welfare dictates that they take more risk. The net effect is that the virus will stabilize at a higher level in America than Sweden at any level of non-effective lockdown policy.

Incidentally this is true for a wide range of virus severity.

Also, you should note that the cost of not going out and taking risks in America differs quite a deal depending on profession and wealth. If you have a nice home to stay in and you do not starve by staying home then you are much more likely to stay home. If you are poor and do not have savings and will starve if you do not go out then you will go out.

Again, this is true for a wide range of virus severity.

It should be no surprise the virus skews very poor in the US. There may be little difference in mortality by say wealth status (we do not know). But there is likely to be far higher mortality amongst people with lower financial security.

The behavioral model is also predictive

The behavioral model is predictive too and in quite useful ways. We think (for good scientific reasons) that the virus is much harder to contain when the weather is 4 degrees centigrade than when it is 30 degrees centigrade. (This is for instance consistent with the only serious Australian outbreak, in Melbourne in winter. It is also consistent with data showing meat processing facilities, which are really large cold-rooms, have been major clusters.)

If that is the case, the risk of socializing in the northern hemisphere increases in Winter.

Much of the US goes to 4 degrees centigrade as Australia enters Summer.

This means that virus numbers will just rise and rise as we are going into an American election. But they will not rise indefinitely because behavior adjusts to the new riskier reality and R0 will still tend to one, albeit at a higher underlying level of virus cases and deaths. If we are right about this there will be fear as the virus numbers go up. But that fear will
ameliorate.

Note again all of this is independent of how severe the virus actually is. You do not need to estimate case fatality rates for this model to be explanatory.

What would happen if you removed American lockdowns?

We are writing this up as American because most of the friends we have debated this with are American. But this could apply to lockdown measures in other countries as well. America has a partial (and variable) policy which does not appear to be very effective. In some instances public schools are closed but private schools are open. There are high risk behaviors but there are also old people who have barely been outside since February because of fear.

We have friends who think that sharply reducing American lockdowns would result in unconstrained spread like Milan or New York in their epidemic stages. They think the outcomes would be instantly terrible. We also have friends who point to Sweden and think that you would get better outcomes and less economic damage.

It is likely both positions are wrong. If you reduce restrictions on some people it is likely that virus spread will go up. But that will make everyone’s activity riskier. Other people will thus constrain their behavior more. Mortality will rise, but not very much.

Incidentally if it makes it riskier for people (say middle aged people) who have high economic value it is entirely possible that you reduce lockdowns and the high-value people constrain their behavior and GDP goes down.

A quick policy comment

All of the above analysis works for a wide range of virus severity. R0 tends to one whether this is Ebola or COVID-19.

But we can say something about rational policy.

As we have explained, partial lockdown does much less than you would think. It doesn’t change mortality very much. And it doesn’t even have predictable effects on economic activity (that could go either way). But it is expensive.

Full lockdown (with police pressure and/or massive social pressure) however is effective. It has been effective in China and will almost certainly be effective in Melbourne.

You either thus go the “drive to elimination” route (Melbourne) or the “welfare plus no lockdown” route (Sweden). Which you choose depends on the mortality of the virus and how you weight that mortality versus the considerable costs of harsh lockdown.

The middle ground here looks pretty stupid actually. Strangely we agree with both our libertarian friends here (who hate all lockdowns) and our public health orientated friends here (that think the lockdown is not harsh enough). Both are rational positions. It is hard for an instinctive centrist (like John) to say this – but in this case the middle ground is plain
stupid. America alas is in that middle ground and the current American policies are plain stupid.

Give us the very strong lockdown policies of the Premier in Melbourne. He is colloquially known as “Dictator Dan”. That is harsh enough to defeat the behavioral trap. Or give us Sweden. Just do not give us the middle.

About the author
David Llewellyn-Smith is Chief Strategist at the MB Fund and MB Super. David is the founding publisher and editor of MacroBusiness and was the founding publisher and global economy editor of The Diplomat, the Asia Pacific’s leading geo-politics and economics portal. He is also a former gold trader and economic commentator at The Sydney Morning Herald, The Age, the ABC and Business Spectator. He is the co-author of The Great Crash of 2008 with Ross Garnaut and was the editor of the second Garnaut Climate Change Review.