Phil Magness is an economic historian and Senior Research Fellow at the American Institute for Economic Research. He’s also a classical liberal and a lockdown sceptic. During the pandemic, he’s written articles about masks, lockdowns, pandemic modelling and the Great Barrington Declaration. I interviewed him via email.
On 28th January, you gave a talk at Hillsdale College titled ‘The Failures of Pandemic Central Planning’. You’ve since written a full-length paper with the same title. Could you briefly summarise your argument?
I argue that the political response to the Covid pandemic is best understood as an exercise in failed central planning. In a sense, it closely parallels the mindset behind mid-20th century economic planning. It’s the mindset that says complex human interactions may be tweaked, corrected, and managed by expert scientists with sophisticated models of the same society-wide systems. If a problem emerges, simply follow the model’s directions and pull the correct policy levers and all will be fixed – or so they claim.
With Covid, most of the world’s governments adopted an aggressive policy response built upon then-untested modelling that advised when and where to impose the ‘non-pharmaceutical interventions’ (NPIs) we’ve all come to know – things like social distancing requirements, school closures, event cancellations, and lockdowns. If an outbreak crosses a threshold, then lock everything down and the outbreak can be managed.
The problem, as we’ve seen time and time again, is that the models guiding the NPI approach were wrong – often catastrophically so. I focus on the Imperial College-London (ICL) model of Neil Ferguson, which had an outsized influence on the adoption of lockdowns and other NPIs. I show that, as of its one year anniversary, ICL’s main model overstated mortality projections in 189 out of 189 countries. It also severely exaggerated the effectiveness of NPIs, and even failed to account for the acute vulnerability of nursing home and old age care facilities.
Combined together, Imperial gave us a roadmap for centralized NPI planning that turned out to be fundamentally unsuited for the Covid pandemic. And yet once we were locked into that policy trajectory, politics intervened and made it nearly impossible to change course, despite mounting evidence that the NPIs were failing to deliver as promised.
You work for the American Institute for Economic Research, which hosted the conference that led to the Great Barrington Declaration – a public statement advocating focused protection. Could you tell us what happened at that conference?
In early October 2020, AIER hosted a small academic conference for the purpose of calling scientific attention to the costs of lockdowns. Up until that point, the media and political figures such as Anthony Fauci had been working to create a false impression of strong scientific consensus behind the lockdown measures – even as they were failing to perform as promised (recall “two weeks to flatten the curve”). This new consensus was an outright falsehood. As recently as 2019, the WHO, leading epidemiology research institutions such as Johns-Hopkins University, and even Fauci himself had gone on record stating that lockdowns would not work in a respiratory pandemic, and should be ruled out as a policy response.
The conference would call attention to the largely ignored harms of lockdowns, while proposing alternative approaches that were in keeping with the pre-2020 public health science. We hosted three eminently qualified scientists from top research institutions, who presented the case against lockdowns in a filmed discussion panel. This was followed by interviews with journalists who specialize in pandemic coverage. On the last day of the conference, the three scientists then drafted a general statement of principles that (1) summarized the case against lockdowns and (2) called for an alternative “focused protection” strategy. They dubbed this the Great Barrington Declaration (GBD), and released it publicly the next morning.
Much to everyone’s surprise, the Declaration went viral. The scientists’ statement had tapped into growing scholarly dissent from the lockdown approach, which had thus far dominated the Covid-19 response, and quickly amassed tens of thousands of signatures from other scientists and medical practitioners.
While we expected some pushback from the pro-lockdown side, we weren’t anticipating the vilification campaign that followed. Instead of engaging the scientists’ arguments as laid out at the conference, the pro-lockdown side went on the political offensive. They made ad hominem attacks, spun together wild conspiracy theories about the GBD’s supposed funding, and falsely claimed that the GBD scientists were guiding U.S. and U.K. policy responses.
For a few weeks after its October 5th publication, some pro-lockdown scientists even claimed the GBD was “arguing with the past” – that the lockdowns were behind us, and that bringing them up again was just a “strawman.” Of course, we all know how that turned out. Within a month, many of those very same scientists endorsed another round of lockdowns. So not only did they refuse to engage in scholarly debate, they engaged in outright duplicity about their own motives – first denying the prospect of more lockdowns, and then embracing a second round as soon as the opportunity presented itself.
At the same time, however, the GBD provided something that opponents of lockdowns had thus far lacked – a succinct statement of scientifically grounded principles that challenged the dominant political paradigm. It opened the door for more scientists to speak out against lockdowns, while shattering the media-cultivated myth that lockdowns were backed by an overwhelming scientific consensus.
Some people, such as the U.K. Chancellor Rishi Sunak, have claimed there’s “no trade off” between health and the economy. What do you make of this claim?
The understanding of trade offs is an essential tool of economics itself, so to assert that there is “no trade off” associated with lockdowns is to deny economic reality. I suspect that Sunak was peddling what he believed to be a political talking point, with the aim of rationalising the policy decisions of his government, which were both extreme and unprecedented at that point in history.
We’ve seen very clear evidence that lockdowns and other NPIs impose severe economic harms on society, including its least well-off members, who often do not have the luxury of telecommuting from home. This became apparent once some countries and U.S. states began to reopen in the summer of 2020 after the initial lockdown. Employment typically rebounded in those locations, while remaining high in places that still had lockdowns. At the same time, we’ve seen no conclusive evidence that regions under lockdown performed any better on their covid metrics than regions that reopened, and quite a few examples where they performed worse.
In your paper ‘The Failures of Pandemic Central Planning’, you criticise some of your fellow liberals for supporting lockdown. Likewise, the journalist Freddie Sayers recently asked, “why have the most nominally liberal governments consistently reached for the most illiberal interventions?” How would you answer that question?
There’s no single answer to that question, but much of the illiberalism comes from an unwarranted faith in collective action solutions to the pandemic, particularly technocratic ones. I was surprised early on at how many otherwise sensible people fell captive to the ‘externality’ argument for aggressive NPI regimes. All we heard for months was how the spread of disease created an externality, and that the very existence of this externality somehow necessitated an aggressive policy response. They completely forgot Ronald Coase’s warning about the political difficulties of effective externality correction:
The fact that governmental intervention also has its costs makes it very likely that most “externalities” should be allowed to continue if the value of production is to be maximized. This conclusion is strengthened if we assume that the government is not like Pigou’s ideal but is more like his normal public authority–ignorant, subject to pressure, and corrupt.
Unfortunately, the ICL-Ferguson model presented an extremely appealing set of policy interventions – hit a threshold of X number of cases or Y number of hospitalizations, and all you have to do is pull an NPI lever and cases are supposed to go down. Except it did not work as promised, and it turns out that the model wasn’t even suitable for the characteristics of this disease.
Some of the liberal/libertarian supporters of lockdowns were nonetheless unambiguous in their enthusiasm for what ICL was offering. Tyler Cowen, for example, praised Ferguson’s approach as a model of “good policy design.” However, some of these commentators updated their priors and moved away from lockdowns as evidence amassed that they were not delivering what they promised. But others dug in.
In the paper, I’m very critical of self-described neoliberals like Sam Bowman and the U.K. CovidFAQ website. They started from the same externality position at the beginning of the pandemic, but rather than adjusting to account for evidence that lockdowns were not working as claimed, they doubled down with highly unpersuasive rationalizations. For example, they circulated the heavily criticised pro-lockdown paper by Flaxman et al, and they tried to infer causality by simply eyeballing a time series in post hoc ergo propter hoc fashion.
The result, unfortunately, is that many liberal/libertarian voices have ended up defending some of the most aggressive and far-reaching government intrusions on individual liberty in our lifetimes. People who once argued for open borders worldwide now rationalize multi-year travel bans and quarantine encampments, or they end up praising the alleged lockdown ‘successes’ of monstrously illiberal regimes like China, credulously repeating Covid data that shows clear signs of political manipulation.
According to a tweet sent by Imperial College London’s official Twitter account, “Professor Ferguson and the Imperial COVID-19 response team never estimated 40,000 or 100,000 Swedish deaths”. That isn’t quite true though, is it?
It’s not true at all. First the context.
Back in the Spring of 2020, a separate team of researchers from Uppsala University directly adapted the Ferguson-ICL model (which originally only projected numbers for the U.S. and U.K.) to Sweden. They ran the numbers for Sweden and got catastrophic results – 96,000 dead if Sweden failed to act, and around 40,000 dead if they eschewed lockdowns and went with a lighter touch approach. Well Sweden did not follow the lockdown/NPI strategy that we saw in the rest of Europe, and by the summer of 2020, Sweden had only had a few thousand deaths.
I was one of the first people to notice this aspect of the model’s dismal performance, and called attention to it on April 30, 2020.
In the early summer of 2020, Matt Ridley directly questioned Ferguson about the failure of his model in Sweden during a House of Lords hearing. Ferguson responded by denying that he had ever modelled Sweden, and attempted to blame the wildly inaccurate projections on errors in the Uppsala team’s adaptation of his model. Shortly thereafter, ICL’s media team picked up this talking point, and ever since they’ve been denying any connection to a model for Sweden.
Here’s the problem with Imperial’s PR messaging though. Shortly before the Uppsala team ran its own adaptation of Ferguson’s U.S. and U.K. model, Ferguson’s team at ICL also produced a second report containing a trimmed down version of their model for every country on earth. The data file for that model – released March 26, 2020 – is still downloadable from the ICL website. Imperial College projected up to 90,000 deaths in Sweden without mitigation and up to 42,000 deaths under a social distancing approach – almost the exact same numbers that the Uppsala team came up with.
In short, Sweden presented an embarrassing complication for Ferguson and the ICL team’s model because it showed a real world natural experiment for a country that did not lock down. Rather than address that shortcoming in their model though, Ferguson & ICL decided to mislead the public.
You’re an American. Given what we know now, what should Donald Trump have done in March of 2020?
For starters, he should not have listened to anything Anthony Fauci was feeding him. Nor should any president. I base this judgment on Fauci’s horrific track record during the AIDS crisis. In 1983, Fauci helped to unleash a nationwide panic by making the wildly unfounded speculation that AIDS could transmit through regular household contact. His 40 year career from that time until the present has been a succession of similar missteps, almost always arising from his attempts to build his own political influence.
For specific policy advice in March 2020, I would have urged Trump (and any other leader) to heed the cautions against lockdowns that were openly stated in the respiratory pandemic guidelines the WHO adopted in late 2019. These guidelines specifically warned that the evidence behind lockdowns was shaky, untested, and over-reliant on models such as Ferguson’s ICL team. Similar guidelines from Johns-Hopkins warned that lockdowns were likely to be ineffective and carry extreme social costs. We’d be in a much better place today if policymakers had simply followed their own plans from just a few months before the pandemic.
I also would have advised Trump (or any other leader) to focus his measures on nursing homes and similar facilities with acute vulnerabilities. The first major U.S. outbreak was in a nursing home in Washington state, so we knew about this vulnerability early on. Due to the ICL model and similar missteps though, almost all of our early response efforts were focused away from nursing homes and on hospital capacity. In fact, they were so focused on hospital capacity that some states ended up turning nursing homes into de facto overflow facilities. This is how we got the situation in New York state where Gov. Cuomo ordered nursing homes to take in covid-positive patients, with catastrophic death tolls and an ensuing political coverup.
One idea I first floated back at that time was to subsidize nursing home staffers to reside on site as a way of limiting their contact with the outside world, and thus the chance of carrying the virus into vulnerable facilities. A few private nursing homes did this, with high rates of success – including one that rented RVs on site for their staff. The cost of subsidizing this and even paying staffers a premium to isolate would have been a tiny fraction of the cost of lockdowns. But the ICL model, Fauci and Birx in the U.S., Hancock in the U.K., had already settled on lockdowns, and pursuit of that end became a recurring pattern of sunk cost fallacies overlaid with technocratic hubris.