A look at the evidence.
Lockdowns are not lockdowns. A Melbourne lockdown is different from a Sydney lockdown, be it only because the two cities are not the same. On top of this, the rules are different. As is the enforcement of the rules.
Last but not least in light of the current “Fortress Australia” rhetoric about lockouts being preferable to lockdowns — disappointingly also pushed by Labor politicians — , here is my attempt to understand what lockdowns really do and, importantly, at what costs they come, citizen and civil rights being one sacrifice of many.
The basic idea of lockdowns is to slow down interaction between humans and hence the transmission of the SARS-C0V-2 virus. Several Aussie states have done this repeatedly and the preferred strategy they use are stay-at-home orders with exceptions of different stringency. Typical are the stay-at-home orders of the NSW government that allow for people to go to work and school (if working or schooling from home is not possible), to do shopping for essential goods and services, and to exercise (presumably in your local government area, in small groups). Medical care or compassionate reasons are also on the list of exemptions, the latter implying partners not living together can visit each other under certain restrictions. Mask wearing on public transport and in shops etc is also mandatory, for good reason.
The more than two dozen lockdowns of various degrees that Australia has experienced all seem to have had the desired effects of slowing the number of daily cases and to reduce them (“bending the curve”) although it took longer in some cases (e.g., Melbourne) than others (e.g., the Crossroads and Avalon/Berala/Croydon outbreaks in NSW). The exact timing of lockdowns remains a perennial and contentious issue because, well, there are trade-offs between lives (and misery) today and lives (and misery) later. Fact of life lest you are a no-trade-off economist who prioritizes public health today lexicographically, a stance that I consider intellectual malpractice. Of course, “bending the curve” has its own problems as many countries have had to experience painfully, with more aggressive variants of the virus (such as the delta version) making it less likely that an outbreak can be contained. Trade-offs, trade-offs, trade-offs.
Containment, sometimes also called mitigation(if not in cases than at least in the damage that the cases can do), will ultimately require vaccination. And short of even more vicious strains emerging, vaccination seems to work surprisingly well.
Does SARS-CoV-2 elimination, not mitigation, create best outcomes for health, the economy, civil liberties, etc? In light of the Australian situation (about which more below) this claim in a comment published in The Lancet is hard to believe. Nevermind the illustrious bunch of authors, the comment is deeply flawed from the lack of references to research that comes to opposite conclusions or that provides evidence of significant voluntary changes in behavior, via the definition of elimination and identification of countries that tried it, to the lack of control for important covariates, or fixed effects, and so on. Frankly it is a pitiful piece and another good example of people claiming authority where they clearly do not have it.
For a number of reasons, elimination, or eradication is simply not an option. And so is herd immunity, at least in the sense of preventing everyone from being infected. Aschwanden argued earlier this year that there are five reasons why COVID herd immunity in this sense is impossible. First, it is currently unclear whether vaccines prevent transmission; second, the vaccine roll-out is uneven; third, new variants might change transmission scope and speed; fourth, immunity might not last forever (and possibly only a few months); and fifth, vaccines might change human behavior in that it invites more recklessness, a phenomenon known as prevention paradox. Like many others she concluded that in light of these five reasons, long-term prospects for the pandemic probably include COVID-19 becoming an endemic disease, much like influenza. Which is why independent data scientist Youyang Gu has changed the name of his forecasting model from ‘the path to herd immunity’ to ‘the path to normality’ @ https://covid19-projections.com/path-to-herd-immunity/
So if herd immunity, or elimination, seems out of the question, what do we know about the effects of lockdowns? There is at least a handful of papers that speak to the issue. (I appreciate you pointing me to other papers that should be included. I will update this piece accordingly.)
Berry et al. (2021) find, somewhat counterintuitively, that implementation of SIP [Shelter-in-place] policies did not lead to reductions in new COVID cases or deaths. They also find small but measurable effects on mobility that dissipate over time. Berry et al. stress that their study is not meant to deny social distancing; instead they argue that SIP policies are overwhelmingly ineffective because many (most?) people have changed their behavior before the introduction of SIP orders. Their results are based on the US data (including smartphone data) from Feb through May, 2020, i.e., the first pandemic wave in the USA. One problem might be the deaths data that they use. Arguably, “COVID-related deaths” are a poor outcome measure.
Goolsbee & Syverson (2021) find, based on cellular phone records data on customer visits to more than 2 million individual businesses across more than 100 industries, that the COVID-19 induced contraction of economic activity in 2020 was overwhelmingly driven by massive (pre-cautionary) changes in consumer behavior rather the government imposed restrictions: “While overall consumer traffic fell by 60 percentage points, legal restrictions explain only 7 percentage points of this. … Traffic started dropping before the legal orders were in place; was highly influenced by the number of COVID deaths reported in the county; and showed a clear shift by consumers away from busier, more crowded stores toward smaller, less busy stores in the same industry. States that repealed their shutdown orders saw symmetric, modest recoveries in consumer visits, further supporting the small estimated effect of policy. Although the shutdown orders had little aggregate impact, they did have a significant effect in reallocating consumer visits away from ‘nonessential’ to ‘essential’ businesses and from restaurants and bars toward groceries and other food sellers.” This finding is broadly in line with what we know from the experimental literature on guessing games and also what we know from models that account for such precautionary behavior changes (e.g., Cochrane’s early ruminations on SIR models with behavior or the excellent paper by Krueger et al. that uses more refined substitution arguments to evaluate the Swedish way.)
Haug et al. (2020) find, somewhat intuitively, that it takes a suitable combination of NPIs (“non-pharmaceutical interventions”) to curb the spread of the virus and that the optimal mix of NPIs is very contextual. Their exercise is impressive, quantifying and ranking more than 6,000 NPIs used during March and April 2020 and studying for 79 territories their effects on the effective reproduction number. They also validate their findings with two external datasets (with more than 40,000 additional NPIs from more than 200 countries). They do not have a measure of behavioral changes though and they also do not look at the long-term consequences of various combinations of NPIs, as they should since the standard reproduction number does not tell us anything about them.
Agrawal et al (2021) use a better outcome measure — excess mortality — to study the impact of the COVID-19 pandemic and policy respones for 43 countries that implemented SIP measures and all U.S. states. They find — like much of the relevant literature it seems — that SIP policies have a modest effect on mobility and identify people’s changed behavior in response to COVID-19 risk as the likely major reason for that result. They discuss potential confounds that might affect the measurement of their outcome variable (such as drug overdoses, homicides, and unintentional injuries, or reduction in non-COVID-19 health care).
Goldstein et al. (2021; for the full paper see their references) offer the results of an empirical analysis of the impact of virus NPIs on the virus’ transmission and death toll, for a panel of 152 countries, from the start of the pandemic through December 31, 2020. They find that lockdowns tend to significantly reduce the spread of the virus and the number of related deaths but this impact declines significantly over time. It is noteworthy that their outcome measures are the effective reproduction number and the number of COVID-19 deaths per million. That is, like all the other papers discussed above, they do not look at the long-term consequences of NPIs, as they should since the standard reproduction number does not tell us anything about them.
Spiliopoulos (2021), in an ambitious and very technical paper (a behavioral four-equation structural model that accounts for the endogeneity of many variables and correlated unobservable country characteristics, as well as several important controls), uses data for 132 countries between 15 February 2020 and 14 April 2021. He finds that at least 90% of the maximum effectiveness of NPIs is attainable with interventions associated with a Stringency Index in the range of 31–40, which is significantly less than the average current stringency level of implemented policies across the world. In other words, there are strongly decreasing returns to the stringency of NPIs. Spiliopoulos also, unfortunately, uses as his main outcomes of interest (growth rates) of confirmed cases and deaths which ignores the massive downstream costs that interventions bring about.
Scott Morrison is reported as having said it would be “unwise to surrender up” Australia’s “advantage” compared with other countries’ Covid situations, as he reaffirmed Australia’s suppression strategy over opening the country up. Presumably he refers to the fact that Australia has managed to use lockdowns more than two dozen times to bend the curve and contain outbreaks.
But one has to ask: Has he looked at the hits education and tourism industry have taken and continue to take? How about the tens of thousands of businesses (and livelihoods) lost? How about the two million un(der)employed and hundreds of thousands of “contractors” trying to make ends meet as delivery boys (and increasingly girls)? How about the massive withdrawals from superannuation accounts (summing up to about 40% of the costs of the JobKeeper fiscal stimulus)? How about the national (mental) health crisis that by all accounts is developing? How about the consequences of delayed cancer and colon screenings and similar interventions? How about the massive losses in human-capital formation that a year of online learning (and counting) have brought about? How about the increased homelessness (ironically as a consequence of a booming real-estate market in houses from which people like him undoubtedly benefit)? And how about the increased number of people and kids in particular that have fallen into poverty? How about the massive increase in debt? How about the massive infringements on citizenship and civil rights … All of these are very real costs.
Lockdowns. They work, there should be little doubt about it. But they come at potentially huge costs that need to be factored in properly. On balance, especially if implemented with high stringency, lockdowns are a less than persuasive containment, or elimination, strategy.