≡ Menu

Weekend reading: Corona-crisis, round two

Weekend reading logo

What caught my eye this week.

The US may be about to see another surge in Covid-19 deaths. This is clearly tragic from a human perspective. We might also wonder what it will do to the market.

I’ve been pretty relaxed about the UK’s de-lockdown. Daily cases have steadily fallen, contrary to – soon forgotten – warnings at every stage that we were moving too fast.

Indeed it was probably only around 4 July that we went much past where I’d have set the limits for the actual lockdown! Albeit partly with the benefit of hindsight.

My hunch is 90% of the heavy lifting of preventing a rapid escalation of Covid-19, at least in summer, comes from basic social distancing and hygiene, and perhaps wearing masks inside.

Throw in today’s testing capacity – spectacularly absent back when many of us thought we had the virus – and it looks to me like enough to at least keep the situation contained.

More or less

Shutting down the economy, paying extra billions in furlough, not leaving the house more than once a day, keeping lovers apart for three months – I doubt it made much more difference. We should have focused instead on ring-of-steeling care homes. From day one it was the elderly who were clearly most at risk.

Still, reasonable people can disagree.

Early, when total cases are very low, maybe extermination from a heavy lockdown is an appropriate goal. Maybe if you’re going to get a 15% hit to GDP you might as well take a 20% hit and try to wallop the thing on the head. Fresh flare-ups in Australia and Hong Kong reinforce my doubts on that score, but even experts debate this – and I’m far from that!

But I still believe the case for uber-lockdown ignores myriad unforeseen costs and consequences.

Something I learned from Tim Harford this week is called the identifiable victim effect.

Nuclear cul-de-sac

For example, after the Fukushima nuclear disaster in Japan, hundreds of thousands of people were made to relocate from their homes for years on end.

Obviously, right? Nobody wants citizens getting sick from radiation.

Right, except it’s now thought relatively few lives were saved by the relocation. In fact it’s possible more evacuees killed themselves from the emotional wrench of separation and upheaval than would have died from radioactive fallout.

At least 50 suicides have been recorded. And there will have been tens of thousands of smaller tragedies, from job losses and broken relationships to farms falling into disrepair.

It’s all incredibly sad – but try making that argument when the Geiger counters are going haywire. I’m sure I’d have relocated them, too. Most politicians wouldn’t hesitate.

Some more emotive arguments about Covid-19, especially in the early days, saw this identifiable victim effect loom large.

America dreaming?

Still, I know that one thing all1 readers of this website will agree on is that Covid-19 is not a hoax…

…that being asked to wear a mask isn’t a plot to turn us into communists.

…and that saying a less full-on lockdown might have been more proportionate doesn’t mean I’m saying we should do nothing, or pretend the virus isn’t happening.

Yet, incredibly, swathes of Americans appear to have reached exactly those conclusions. A lunatic fringe apparently even believes ex-President Obama created the virus and is spreading it.

(So this is how democracy dies. Not with a bang but a “WTF? Really?”)

You can understand why the saner end of this happened. The US is a huge country. Much of the US was in lockdown when it probably didn’t need to be. Local lockdowns and exclusion zones might have been more pragmatic, and caused less frustration. Although this is with hindsight – I imagine the US health officials didn’t expect things would eventually go so far off the deep end.

Anyway, the end result is the virus seems to be taking off rampantly again.

I have tried to stay open-minded about this. In particular the US is now doing over 600,000 Covid-19 tests a day. It’s a truism (not a Trumpism) that if you increase testing in a pandemic you’ll find more carriers. So it was plausible that part of the surge in US cases was down to the huge ramp in testing.

(Imagine if we’d been able to test everyone in London over a few days in early April. We might have found half a million people with the virus!)

As you test more people though, what you don’t want to see is the percentage of positive results going up. That means you’re not just finding more Covid-19 carriers through more tests – it also implies there’s a growing number of them out there.

And what you really don’t want to see is daily deaths go up.

Unfortunately the US is now seeing dramatically more positive test results:

(Click to enlarge the misery)

This might not have been definitely terrible. I suspect young people have had enough of putting their lives on hold for something that will only badly affect a tiny minority. When respected US medical figures say as much as 50% of the US population will probably have had Covid-19 by the end of the year, you might well ask why not get it over with? Sure enough, the average age associated with a positive result has been getting younger in the US.

However, the counter-argument to the ‘let rip’ thesis is that eventually the pandemic ebbs up to and kills a swathe of the vulnerable.

And in my daily chart checks, that’s what I’ve noticed may now be happening:

(Click to enlarge: Notice uptick at far right-hand side)

Two questions for the market.

Firstly, is this uptick just an artifact of the natural history of the virus?

I’ve noted before it’s still not clear why thousands die in some places and other places get off fairly lightly. Maybe Covid-19 has taken a while to get embedded in some larger and more clement US environs? This is grim news for people living in such places. But it could mean a nationwide resurgence isn’t inevitable.

Secondly, is the US effectively going for a Swedish approach? Not openly stated, but de facto.

Again, this always seemed a plausible endgame to me with a virus this transmissible, and yet apparently ultimately harmless to most. The cost of repeated shutdowns is just too great.

It’s worth noting that for all the wailing and gnashing of teeth, the Swedish approach has sort of worked.

Yes, before *you* say it, plenty of people have died. As I always reply, we’ll only see in the end how premature much of this death really was.

But more importantly for this discussion, many were arguing in April and May that only stringent lockdown could curb the spread of the virus.

The Swedish experience, painful as it was, seems to show that’s not true:

(Click to enlarge the counterfactual)

Through one lens Sweden’s approach looks like a failure. It comes fifth ranked on deaths per capita from Covid-19. (Though that’s still better than lockdown-happy Britain at third).

But this isn’t the argument I’m making here. I’m simply saying a looser policy that accepts the virus will spread and run its course isn’t a catastrophic policy from the set of bad choices on offer.

The US may be about to find out, one way or another. Frustratingly, I could see the market taking this news either way.

Bluntly and economically-speaking, it probably fears another fully-loaded economic shutdown more than a higher death tempo among the elderly.

On the other hand, a return to New York-like scenes on the news is hard to square with a recovering economy let alone buoyant share prices – even if the very worst fears and tail risks of this pandemic have now been lopped off the distribution graph.

Hold on to your hats (/allocation to bonds and cash). Things could be about to get bumpy.

From Monevator

The UK’s worst stock market crash: 1972-1974 – Monevator

From the archive-ator: The investor sentiment cycle – Monevator


Note: Some links are Google search results – in PC/desktop view you can click to read the piece without being a paid subscriber. Try privacy/incognito mode to avoid cookies. Consider subscribing if you read them a lot!2

Sunak’s Summer Statement: Key points at-a-glance – BBC

No help for ‘excluded’ three million self-employed in Sunak’s mini-Budget – ThisIsMoney

Stamp duty cut up to £500,000: how much will you save? – Which?

The stamp duty cut is a boost for buy-to-let landlords [Search result]FT

House prices to fall 5% in 2020 and 10% in 2021 despite the cut, says CEBR – ThisIsMoney

VAT cut for restaurants, pub meals and hotels: what will it mean for you? – Guardian

High-rise flat owners are effectively being told they cannot sell or remortgage for years – Guardian

Comparing different rebalancing tempos with long-term buy-and-hold [Note the tight axes!]Morningstar

Products and services

How much will people get in the ‘Eat Out to Help Out’ scheme and where can you use it? – Metro

Hargreaves Lansdown revives ‘best buys’ after Woodford scandal – Which?

Sign-up to Freetrade via my link and we can both get a free share worth between £3 and £200 – Freetrade

Will your local John Lewis be closed down? Eight are going, putting 1,300 jobs at risk – ThisIsMoney

Homes off the beaten track [Gallery]Guardian

Comment and opinion

Is it crazy to own gold? – Value Stock Geek

Is it possible to save too much money? – Money Maven

Using an all-in-one fund in a downturn [Ignore the US-relevant comment about taxes]Oblivious Investor

Dead people don’t get to retire – A Teachable Moment

Busting the pension Lifetime Allowance is music to my ears [Search result]FT

Safe bonds are the best diversifier for stocks… – Morningstar

…but on the other hand  [US but same difference]Alpha Architect

The danger of learning from your mistakes – Validea

The bear market in happiness – A Wealth of Common Sense

Fancy fintech’s fishy fun: a rant – Simple Living in Somerset

Institutions banking on private assets mini-special

The nerdy, contentious, and existential debate raging inside one of Britain’s biggest funds – Institutional Investor

Alternative assets [and the pseudo-Yale model] are a “a loser’s game”Investment Magazine

Naughty corner: Active antics

Charlie Songhurst: Lessons from investing in 483 companies [Podcast]Invest Like The Best

Why Ted Baker has been a disaster and a gift in equal measure – UK Value Investor

Buffett’s $90 billion Apple stake is now 43% of Berkshire Hathaway’s entire stock portfolio – Business Insider

GMO, long bearish, loves US banks – Institutional Investor

First-half review: Running up that hill – IT Investor

More coronavirus chatter

Why hasn’t the UK seen a second wave? – New Scientist

Herd immunity questioned after Spanish antibody study – CNBC

“I’ve spent years taming the OCD monster. Coronavirus has ruined everything”Guardian

Data show panic and disorganization dominate the study of Covid-19 drugs – Stat

Space jam – The Reformed Broker

Can big countries realistically eliminate COVID-19 without a vaccine? Four experts discuss – The Conversation

Patricia Lockwood: Insane after coronavirus [Beautiful and poetic]The London Review of Books

Kindle book bargains

You’re Not Broke, You’re Pre-Rich by Emilie Bellet – £0.99 on Kindle

The Economics Book: Big Ideas Simply Explained by Niall Kishtainy- £1.99 on Kindle

Alchemy: The Surprising Power of Ideas That Don’t Make Sense by Rory Sutherland – £0.99 on Kindle

When Genius Failed: The Rise and Fall of Long Term Capital Management by Roger Lowenstein – £0.99 on Kindle

Off our beat

Instinct – Indeedably

An open letter on justice and debate – Harpers (and an on-point tweet from Malcolm Gladwell)

Three women fight for two spots in the US surf team at the Tokyo Olympics – Marie Claire [via Abnormal Returns]

“How I became a poker champion in one year”The Atlantic

Elon Musk and other spaceflight pioneers rally around a failure-struck rival – Business Insider

Our top story – Seth Godin

Regrets – Of Dollars and Data

And finally…

“The lesson is that no amount of sophisticated statistical analysis is a match for the historical experience that ‘stuff happens’.”
– Mervyn King, The End Of Alchemy

Like these links? Subscribe to get them every Friday!

  1. Okay, 99% []
  2. Note some articles can only be accessed through the search results if you’re using PC/desktop view (from mobile/tablet view they bring up the firewall/subscription page). To circumvent, switch your mobile browser to use the desktop view. On Chrome for Android: press the menu button followed by “Request Desktop Site”. []

Receive my articles for free in your inbox. Type your email and press submit:

{ 72 comments… add one }
  • 51 Ruby July 14, 2020, 3:29 pm

    @ ZX Spectrum – I don’t think I quite understood what Spiegelhalter was saying and it seems he’s had a spat with Michael Levitt so I doubt my reference was terribly useful. Quantifying or rationalising ones own risk is difficult I find – I’m in the 55-59 bracket, and on a child’s dose ICS, but fit and routinely cycling 150 to 200 miles a week so I don’t really feel at much risk. However, as a non helmet wearing red light jumping London cyclists I wouldn’t argue if you said I needed a risk assessment course. Those two words again ‘risk assessment’ which one sees everywhere now and have to be the two most depressing words in the English language.

  • 52 Grumpy Old Paul July 14, 2020, 5:20 pm

    @xxd09 – The median age of UK prisoners is around 35 which is much younger than the population as a whole.

    @ZXSpectrum48k – I agree that the IFR is likely to fall. Two additional and linked factors are improved protection of care home residents (PPE, testing, limitations on visitors, fewer discharges from hospital) and a fall in the average age of newly infected people. I haven’t seen the stats for the UK but the southern US states are showing a massive drop in the average age of new cases, in Arizona for example from 65 to 35.

  • 53 Brendan Jackson July 15, 2020, 12:30 pm

    @TheInvestor: there have been plenty of anecdotes, and now even occasional clinical reports of reinfections. Medicine if full of exceptions; I’d fully expect those with a compromised immune system to be reinfected for instance. But what matters for the epidemiology (and the actuaries…) is the typical behaviour, and it doesn’t seem that it’s typical for reinfection within six months or so.

    Beyond that, we don’t know – we can only speculate based on similar viruses and typical immune response, which usually results in a milder disease, and perhaps less infectious. Waning antibodies level shouldn’t trouble you too much, as this is normal (T cells then preserve a ‘memory’ of the infection to produce antibodies faster in the next infection).

    To those writing about Ferguson and the IFR (especially @ZXSpectrum48k): Ferguson got his values from early estimates by Verity et. al (though he was of course last author on that paper, so he was hardly uninvolved!). But the credit should probably go to the first authors on that paper; they did surprisingly well given the limited early information.

    The IFR is going to be a moving target; it will depend on demographics and evolve as our treatments do. There will also be substantial excess mortality from the collapse of healthcare systems if the pandemic exceeds our capacity to handle it. Assuming we avoid such scenarios, IFR is not much bigger than 1%; it’s not much less than 0.5%.

    It sort of doesn’t matter, though, other than to calculate worst case scenarios. Letting R get too high for too long will be a public health disaster of sufficient scale – no matter the exact number of deaths – that there will need to be a counter response. The government then has to minimise the damage that causes, ideally preemptively (by keeping R low) rather than reactively (by instituting a national lockdown when cases have already reached high levels).

    I’d be surprised if we aren’t having many local lockdowns throughout the winter. Hopefully we don’t have a national one again.

  • 54 The Investor July 15, 2020, 1:38 pm

    @Brendan — Interesting comment, especially about waning antibody levels and the relationship with T cells. Need to brush up my immunology!

    Do you have a source for clinical reports for reinfection? As far as I am aware the ones that made headlines were all false alarms. Would be interesting to read, even if as you say they are unfortunate edge cases (e.g. immunocompromised and already in hospital).

  • 55 Brendan Jackson July 15, 2020, 2:42 pm

    A clinical report (as in…from the clinic) but a clickbait headline:


    There’s of course plenty of explanations that might be false negative/positive tests, even beyond the fact these might be outliers. But there are clinical, even if they’re ultimately incorrect or edge cases.

  • 56 Simon T July 15, 2020, 6:54 pm

    Anybody else using the Zoe app to report each day, interesting update today https://covid.joinzoe.com/post/skin-rash-covid

  • 57 The Investor July 15, 2020, 11:18 pm

    @Brendan — Thanks for that link. Interesting piece, although as you say potentially an outlier. But if the patient featured is just an unfortunate outlier at the front a vanguard of reinfections, it would be especially worrying.

    I just realized an error in my thinking on this; I’ve assumed we’ll hear about proven reinfection first from China because they got the virus first. But of course there’s apparently very little Covid-19 in circulation now in China. So perhaps if reinfection does become commonplace, it’d make sense if we learn about it in say the US first.

    Play the worst case scenario outlined by the physician in that piece forward though, and what we’d need really is not a vaccine but infinitely better treatments (/a ‘cure’) as the worst case would see even vaccine protection rapidly waning, presumably. 🙁

  • 58 ZXSpectrum48k July 16, 2020, 10:14 am

    I hadn’t spotted this (thanks to Ruby who made me check Spiegelhalter’s Twitter feed) but the following paper on risk factors associated with COVID has now come out in Nature (https://www.nature.com/articles/s41586-020-2521-4). This paper uses the one clear advantage the NHS has, it’s scale. It’s one of the most comprehensive analyses of risk-factors out there that I’ve seen (I’ve been looking quite hard).

    When paper first came out as a preprint it scared me. It had some fairly high hazard rates for those who were obese and asthma. I’m not actually obese but overweight and when added to be male and asthmatic, the compound risk factor to the topside was pretty unpleasant.

    When I looked at the numbers, however, something seemed odd. The hazard rates for obesity, in particular made, no sense at all but even those for asthma and gender seemed too high. Either errors in the input data or the regressions/spline used to fit the data. It didn’t match some other papers using UK data. I even wrote to them.

    Well in the revised version in Nature those anomalies disappeared. The fully adjusted hazard rate for being male has dropped from 1.99 to 1.59, a massive difference. The hazard rate for something like OCS asthma (normally mild) has dropped to 1.11 to 0.99 i.e. basically no additional risk. For modest obesity (BMI 30-34.9) from 1.27 to 1.05 i.e. only a 5% increase in marginal risk. These numbers look consistent and the CLs have narrowed.

    Using the above paper, this preprint (https://www.medrxiv.org/content/10.1101/2020.05.21.20108969v1) converts the hazard rates into a “male COVID age” risk score i.e. a white man aged 45, BMI 36, with controlled diabetes has a Covid-age of (45+5+4) = 54 years or a healthy Asian woman, aged 40, with has a Covid-age of (40+5-8) = 37 years. This is useful since we have the most data on age/gender related death rates. Personally, I prefer looking at “orthonormal” hazard rates but others may see this as useful in a risk assessment.

    I do find it frustrating that the media continues to trump headlines like “overweight at more risk of dying” when actually the more recent/larger/accurate studies are showing there is little or no evidence that overweight people are dying at a higher rate. In fact another UK study using ICNARC data found that overweight (BMI 25-29.9) people were less likely to require ICU than those with normal BMI (18.5 – 24.9). Albeit I tend to think they might be some data issues aswell! Even modest levels of obesity seem to be showing only low orthonormal risk factors once other correlated factors are removed (such as diabetes). No journalists are reporting these corrected papers. It clearly doesn’t fit the narrative. We desperately need journos (and politicians/policymakers) who have at least a basic understand of probabilities. Rant over.

  • 59 Brendan Jackson July 16, 2020, 11:19 am

    I’m not really sure what to make of China’s data – reporting seems even worse than America. But they have taken much more draconian measures, so it would probably make sense there are now fewer cases.

    This is a good article on T cells to read (by an industry expert that is worth following anyway):


  • 60 The Investor July 16, 2020, 2:55 pm

    @Brendan — Really interesting article/link. (Could also back up what that statistician was saying when he was talking about unknown ‘dark matter’ that apparently gave the German population some greater resistance to Covid-19 in his assessment?) Will include in the links.

    @ZXSpectrum48k — Cheers for that extensive data-driven rant. Apologies that your comment sat in moderation for a while (happens automatically with multiple links) as I was engaged for a couple of hours elsewhere and didn’t get a chance to moderate.

    Glad to hear your odds of being with us and contributing for many years to come have risen! 😉

  • 61 Vanguardfan July 16, 2020, 3:25 pm

    Worth just clarifying @zx that the paper is talking about mortality risk in those infected with covid (I wasn’t clear from your post whether it was referring to risk of contracting covid or risk of serious illness rather than mortality).

  • 62 Vanguardfan July 16, 2020, 3:28 pm

    I also find it irritating and illogical that they express risk in relation to a ‘healthy white male’. Why not pick the group with the lowest risk, the ‘healthy white female’? It is usually the convention to pick the lowest risk group as the baseline….

  • 63 Boltt July 16, 2020, 4:38 pm

    @ Vanguard

    The modelling software I used in Insurance automatically set the base level to be the factor level with the most data, although you could change the base level manually.

    Most policyholders were male too, so they were the base level too – obviously this was back in the good old days when we could use gender as rating factor!


  • 64 The Investor July 17, 2020, 2:58 pm

    Readers may remember I warned months ago that it seemed that all deaths *with* Covid-19 were being counted as ‘Covid-19 deaths’, specifically suggesting someone who tested positive could be hit by a bus and go into the figures.

    Some readers pushed back, but it seems Hancock may be coming around to the view.

    Today he’s ordered an ‘urgent review’.

    According to Oxford researchers who sounded the alarm:

    “A patient who has tested positive, but successfully treated and discharged from hospital, will still be counted as a COVID death even if they had a heart attack or were run over by a bus three months later.”



  • 65 Vanguardfan July 17, 2020, 3:07 pm

    @ti. Two quick points.
    1. Excess deaths remains unbiased method and we have clocked over 60,000 of them (not all directly due to covid of course)
    2. I don’t know how PHE are recording deaths, but I know that cause of death on death certificates (data collated by ONS from death registrations, not PHE) is recorded differently.
    No method is straightforward or free from bias. The more you know about it the more you know it’s not simple. There are likely to be far more Covid deaths that were never tested (most of our deaths occurred in April) than incidental deaths incorrectly classified.

  • 66 Vanguardfan July 17, 2020, 3:22 pm


    Is this the source paper? If so, you’ll see that the total PHE estimate is less than the ONS death registration estimate. So rather than over estimating Covid deaths it seems they underestimate.

  • 67 The Investor July 17, 2020, 3:31 pm

    @Vanguardfan — Cheers for the follow up. Yes, I was more struck by someone using the same ‘bus’ hypothetical that I did than the consequences for more/less deaths per se, immediately fired that memory circuit.

    Agree with your read on first glance.

    Also agree excess deaths are the gold standard. Of course we’ll need to see how quickly they drop out of the figures (i.e. will we get a three month run of below average deaths?) If it happens quickly then that will be an interesting counterpoint.

  • 68 Al Cam July 17, 2020, 4:17 pm

    @ZX &TI
    Assuming I am reading Extended Data Table 2 correctly in the Nature paper, this suggests to me that high blood pressure or diagnosed hypertension actually reduces your risk. This also seems to somewhat fly in the face of the narrative to date?

  • 69 Ruby July 17, 2020, 4:21 pm

    @ The Investor – there’s an interview with Carl Heneghan on Unherd on this which is worth a listen. https://unherd.com/thepost/prof-carl-heneghan-can-we-trust-the-covid-19-death-numbers/
    I believe a longer interview on covid more generally is coming up shortly also. Dr John Lee has been writing about the problems surrounding death certification in The Spectator for a few months now so you’re in good company.

  • 70 ZXSpectrum48k July 17, 2020, 5:31 pm

    @AlCam. You are correct in that there are some non-intuitive results.

    The technique they use is a very standard approach called a multivariate Cox (proportional hazards) regression. It’s a method for investigating the effect of several variables upon the time a specified event takes to happen (in this case mortality). In a proportional hazards model, the unique effect of a unit increase in a covariate is multiplicative with respect to the hazard rate. The method does not assume any particular “survival model” but it is not truly nonparametric because it does assume that the effects of the predictor variables upon survival are constant over time and can be additive in one scale.

    It’s that assumption that can cause some non-intuitive results. If you look at say High Blood Pressure you find the age-sex adjusted hazard rate is 1.09, the fully-adjusted is 0.89. The fact that the fully-adjusted is lower than than the age-sex adjusted is because they are trying to isolate the impact of high blood pressure vs. all other factors that might have driven the fatality. Remember though that those other factors (often more dominant factors) are also covariates in the regression and could be correlated with it.

    One such factor is age. This is correlated with high blood pressure. Moroever, age is the dominant risk factor, much than High Blood Pressure, in the regression. Because so many of the deaths associated with COVID are those above 70y of age, this large age risk factor causes the blood pressure risk factor to actually end up below 1.0. It gets “pushed down” by the dominant age factor. Hence you see sub 1.0 numbers for the full sample but numbers above 1.0 once adjusted for age-sex.

    I always want to see both age-sex adjusted and fully adjusted because age, in particular, is such as dominant risk factor and such as large part of the sample of deaths is 70y+. Fully adjusted is still important since if you use age-sex adjusted only you can overestimate the total risk by “adding up” too many risk factors. Some will overlap.

    It also doesn’t change the the reality is that some factors like blood pressure just don’t seem significant drivers even once the age factor is taken out. Think about it. I’ve aged about six months since COVID was first talked about. In that time, my probability of death due to age risk factor has increased by about 5-10%. So blood pressure isn’t the thing you want to be worried about.

  • 71 ZXSpectrum48k July 17, 2020, 6:08 pm

    Addendum. Just to be clear when i said at the end “my probability of death due to age risk factor has increased by about 5-10%”, I don’t mean I’m now 5-10% more likely to die. I mean if my probability of death was say 0.5%, then 6 months later it might be 0.525%-0.55% i.e. 5-10% greater. So the impact of high blood pressure is perhaps equivalent to six to 12 month in age terms.

  • 72 Al Cam July 18, 2020, 8:43 am

    As you say, somewhat non-intuitive.
    But, to slightly recast a couple of well-worn clichés: we all live (hopefully) & learn; and every day is (should be?) a school day.
    Thanks very much for taking the time to explain.

Leave a Comment