Monday, July 27, 2020

What is Gigi actually saying about COVID?

* Read Gigi's piece at The Conversation

Twitter is a mob, and the mob has a new enemy. Prof. Gigi Foster has been on a few television shows lately trying to make the case that economic lockdowns cost lives, just as COVID does, and therefore we want to make sure that we aren’t inadvertently killing more people over the long-term by crushing economic activity and livelihoods.

After all, a functioning economic system is what delivers high-quality healthcare, safe roads and workplaces, investment in public works, the ability to devote resources to research medical treatments and new drugs, and more. It delivers livelihoods and careers, leisure and happiness. It delivers quality of life as much as it delivers quantity.

However, I don’t think gotcha television and infotainment current-affairs shows are the best venues for this discussion. From what I saw, these outlets were actively avoiding putting a number on the trade-off, or even acknowledging it. A cabal of social media economists avoided acknowledging this trade-off a few months back. They then back-peddled, said there was a trade-off, and botched their attempts to quantify it.

The puzzle to me is that everyone seems to want to say Gigi is trading off lives for the economy. Her point is that a functioning economy delivers health and welfare outcomes, and hence the trade-off is about lives for lives.

For some reason, saying this is a new taboo.

Policy decisions that explicitly make this trade-off occur all the time. Should we fund more medical research? Should we install traffic lights? Should we make people wear seat belts? Should we ban alcohol and cigarettes? Should we legalise recreational drugs?

Policy analysts, particularly economists, spend careers looking at these welfare and livelihood trade-offs in all sorts of policy domains.

When she says "man up" she means that we need to face up to the fact that we cannot create a pre-COVID world. There are going to be losses of life quality and quantity, either from the virus or our response to it.

I want to go through some of the strange things I see when talking about our COVID policy response, and some of the things people say to avoid facing the reality of this trade-off. My personal view is that the reasonable thing to do is to make sure our policy response does not shorten lives and reduce their quality more than COVID would. I hope that this helps people to understand where Gigi is coming from.

1. The exponential growth and tail risk story
One of the big claims early on was that those talking down the risk didn’t understand exponential growth. Strangely, exponential growth doesn’t usually apply to virus propagation. The pattern is well-understood to be logistic growth, which is going to saturate the population at some point. The unknown was merely where that point would be. 

2. Virus prevalence estimates
How much of the population has been exposed to the virus? This is another area where the worst-case scenarios got all the airplay, and where more sensible estimates were ignored. The more prevalent the virus was, the lower the overall mortality. You can see the media incentive for publicising the high mortality estimates, even though it was known quite early on what the realistic estimates were.

3. Infection and case fatality rates
Initially, the highest estimates were promoted, but the reality is that the range of 0.25-0.65% for infection fatality is the current view. If two-thirds of the population is infected overall, that is a worst-case scenario of about 0.16-0.4% of the population dying from COVID, or a few months of normal deaths brought forward in time. This is a generous worst case. The "Swedish disaster" has seen a crude COVID death rate of just 0.05% of the population, under a third of my lowest estimate. 

4. Getting the orders of magnitude right
I asked my Mum when she was panicking about the COVID outbreak how many people she thought had died. She said 5.

When people tell me about the shocking number of COVID deaths I like to ask them how many people die each day in normal times. No one seems to know, or care.

Nearly 8,000 people die every day in the US. Fear does not care for statistics.

The 6,000 coronavirus deaths likely to come from the virus in Sweden are equal to around 24 days of normal expected deaths.

A good rule of thumb is that 8 in 1,000 people die every year (0.8%), or about 60 million globally.

For perspective, the seasonal flu in Australia kills 1,500 to 3,000 people, with about 18,000 hospitalisations.

In Queensland last year 285 people ended up in ICU due to flu. 

5. Getting the cost of life right
Everyone dies, so dying of one thing today simply stops you dying from something else later. Deaths are life-shortening. In the trade-off I described above, economic recessions and lower future output are also life-shortening. There is no point talking about “prevented deaths”, only shorter lives (that whole quality/quantity thing). If people die a year to two younger than otherwise from COVID, then that’s not too bad. If they die 30 years younger than otherwise, the loss is fifteen times worse per death.

6. But what about transmission!
Another argument is that coronavirus can have lasting effects on some people. Yes. And? Others say that you might feel bad transmitting the disease to others. Yes. And? These issues are true of all viruses. They were true of last year’s record 1,300 flu deaths. There are hundreds of people out there who transmitted the flu virus to someone last year and it killed them. Where was the outrage then?

7. Or do recessions save lives?
I’ve heard the argument that recessions decrease traffic and workplace fatalities, reducing crude death rates. I can’t make a judgement about whether this is true, but it makes perfect sense and could be. But it only raises another question—if recessions save lives as a general matter, why aren’t we trying to orchestrate recessions all the time? If it is logical to do it for coronavirus, then it is logical to do it for traffic fatalities, workplace deaths, or whatever other indirect mechanisms of death are at play during economic expansions. 

8. The poorest countries suffer the most
There are global costs to lives from large scale lockdowns. Global vaccination programs for preventable diseases are being delayed, costing lives right now. Construction of health facilities is being delayed, costing lives in the future. Their general development and progress are hampered. Worse still, with very young populations, most poor countries have relatively few people at risk of COVID. 

9. The endgame. What endgame?
After a month of “flatten the curve” rhetoric, which basically had the right intention, there was a silent shift towards “crush and eliminate”. How does this make sense in a globalised world where the virus is going to saturate the rest of the world population? What endgame does that entail?

Being a national bubble with no international travel for years until a vaccine adds to the human cost of our policy response. If (or when?) the bubble is breached we get outbreaks anyway. New Zealand is hailed as a success on this front. But until when? The first person who arrives with COVID will simply take NZ back to square one. 

10. The counterfactual
This is tricky to consider. Is there a “no panic” counterfactual where the media doesn’t whip up society into a frenzy? I think there is, and this means that the “people would voluntarily lockdown” argument doesn’t fly. Why would they voluntarily lockdown?

The 2017 flu season was nearly 4x worse than the 2016 flu season, with 1,255 deaths compared to 464 the prior year.

In relative terms, the 2017 flu season was huge. It also had the risk of being multiple times bigger given the state of knowledge in the early stages. Yet no one panicked and shut down society.

If people don’t notice a 4x jump in flu deaths, would they notice a 10x jump from coronavirus deaths and voluntarily lockdown? I would argue they wouldn’t.

The “don’t panic, don’t lock down, invest in health resources” counterfactual is a plausible one. 

11. The un-science cancel culture
The Twitter mob has decided it can decide what is science and what is not, while at the same time attributing all variation in the COVID outbreaks across different countries or states to the policy response, leaving no space for randomness or luck.

This is “un-science”.

There is also nothing the un-science mob loves more than cancel culture. If my Twitter searches are anything to go by, plenty of people now want Gigi to be fired from her job. Yes, for raising the point that we should try and save the most lives possible by accounting for the cost to lives from our response to COVID, she is being “cancelled”. 

12. A final thought
One thing I have learned to do to help maintain perspective is to turn a problem around and ask the reverse question. How many early deaths would we tolerate to avoid a large recession? Is your answer really zero? Even a global one?

How many early deaths do we tolerate by not spending more on the public health system? Where is the outcry?



  2. great writing Cameron, thanks for your insight.

  3. also, I did find Gigi confronting the first time I heard her on TV, then last night on Q+A , I didnt think she sounded so unreasonable and also thought she made some very good points, Bohdan

  4. Hi Cameron,
    Succinctly put. Usually I agree with a lot of what you publish, but there seems to be a few things that you haven't factored in.
    1) Hospitalisation rate versus fatality rate. The proportion that need to be hospitalized is higher than the fatality rate, and typically those that need to be hospitalized can be for 2-3 weeks. Choking up hospitals with a surge of Covid patients means that the health system is unable to treat other ailments, which causes more deaths.
    2) Infection Rate: What explains the assumption that two-thirds will only be infected? I'm familiar with the models stating the necessity of this level for herd immunity, however that's a very different modelling assumption than what we're talking about. Herd immunity assumes that we're releasing a pathogen into a population that 60-70% won't catch the disease. This is not a valid assumption to make when modelling our current scenario.
    3) Unequal distribution of burden - The figures used above take an aggregate view of the population. This viewpoint is valid and needs to be considered, but it also needs to be supported by more nuanced statistics of different population groups. We know that the virus does not affect everyone equally, and it's increasingly looking like disadvantaged groups in society will be disproportionately affected by the Covid.
    4) Fatality Rates Used: The fatality rate is still being quantified and your figures will be extremely sensitive to changes. Multiplying a fraction of a percentage by a large number (i.e. the population) will be extremely sensitive to slight changes. I think it is premature to abandon the precautionary principle when what're discussing could be impacted extremely by a small change.
    5) Assuming no innovation: Point 9 presumes that the endgame is to isolate until Covid is no longer a threat. This ignores innovation in both treatment and vaccines. There is an enormous amount of research being done on both these fronts. There is talk of a vaccine being discovered/proven by October/November (still needs a long time to mass produce). There is also ongoing research into which drugs work best to treat those who are infected. Both of these innovative fronts present a potential endgame with a significantly reduced fatality rate.

  5. It would be interesting to hear your analysis/view on the Covid and Spanish Flu correlation in terms of response and economic outcomes. I often hear quotes about Spanish Flu, generally but also comparisons between US regions, that areas where early and stronger lockdowns occurred, had better economic outcomes over the longer term - shorter duration of the downturn and stronger sharper rebound back.

  6. "For perspective, the seasonal flu in Australia kills 1,500 to 3,000 people, with about 18,000 hospitalisations."

    Where do these figures come from?

    Only 141 people died from influenza in Australia in 2018 (ABS, Causes of Death, Australia, 2018, 3303.0, Table 1.1).

    It is true that 2961 people died from pneumonia. Perhaps you mean pneumonia instead of flu.