Sweden eschewed lockdowns. It’s (still) too early to be certain it was wrong.
In fact, …
Note: This is a substantially updated version (Version 2.0) of a heavily edited piece in The Conversation a couple of months ago. Version 3.0 is planned for early 2021.
The case of Sweden has attracted considerable attention, and opinionated — but often poorly informed — commentary (e.g., here, here, here, and here). We have been told that Sweden has become a cautionary tale and does not qualify as a model that Australia ought to pay attention to, nevermind follow.
The circumstances of the two countries (e.g., Australia’s insular status against the more integrated Scandinavian situation Sweden finds itself in, different population densities, different willingness of the people here and there to follow public-health advice voluntarily, etc.), and the strategies they have chosen to battle the pandemic caused by SARS-COV-2 (orders versus recommendations, heavy fiscal stimuli vs moderate ones), are quite dissimilar. Still … it is useful to understand the facts as they pertain to the Swedish patient. When everything is said and done, a couple of years down the road at the minimum, it is likely to be a useful reference point for an overall assessment of the consequences of various health, economic, and other interventions and lessons to be learned for future pandemics.
Commentators typically make much of the relatively high death rate per capita that Sweden has run up. And indeed, as of the publication of the earlier version of this article on August 1, worldometer suggested that Sweden had done poorly on this count and trailed only Belgium, the UK, Spain, Italy, and Peru, ahead of then-attention grabbers such as the USA, Brazil, Chile, and Mexico, and way ahead of countries such as Germany (then 110 deaths per million of population), Austria (then 80), and Czechia (then 36), which at that point were considered examplars of sorts.
What a difference a couple of months make!
I predicted in the Conversation piece that Sweden would, by the end of this year, have dropped out of the Top Twenty of that particular league table (deaths per millon of population). As of the writing of the present version, two months before the end of the year, Sweden has dropped out of the worldometer Top Fifteen (trailing the USA, the UK, France, Belgium, Spain, Italy, and several Middle — and South American countries, not counting Andorra, San Marino, and Vatican City) and is well on its way to verify my prediction although it reports as covid-19 deaths those dying with the virus, not those that die of the virus.
The same commentators that routinely dismissed the Swedish approach, often with moralizing arguments clearly designed to derail the public discourse, argue(d) also that these high losses in human life cannot be justified by better economic performance. A key piece of evidence, also used by Duckett & Mackey, is a study by researchers from the University of Copenhagen that suggests that aggregate spending dropped by around 25 percent in Sweden and by an additional 4 percent points in Denmark only. Many commentators suggest that these economic gains, if they exist, are not worth the pain that current lost lives impose. One of my colleagues has made a similar argument. As have other Aussie economists that in essence have denied that intertemporal trade-offs exist.
There are obvious responses to these arguments:
First, we are most likely still in the early stages of the pandemic, even though in Victoria, most US states, and many countries (prominently, in Europe see also here) second or third waves stronger than the first waves are well under way — at least in terms of diagnosed cases (but maybe not in terms of deaths). It therefore simply does not make sense to compare at this stage death rates without controlling for the stage in which a particular state or country is in, its (socio-)demographic idiosycracies, and its policy responses. Arguments that rely on a snapshot of a moment in the past such as the second quarter or even earlier windows, strike me as silly, misleading, and in fact outright irresponsible. Sweden for example has currently very few deaths and the evidence suggests that there haven’t been excess deaths for months. It is possible that the number of deaths for the year will be in the norm (see here and here, see excess in z-scores for the latter) even with the dramatic increase in cases it has seen since late September.
While cases have increased in Sweden, the number of “covid-19 deaths” has stayed, for now, at very low levels. While they are bound to go up — it typically takes 3–5 weeks for diagnosed cases to turn to death and ICU numbers have crept up — there is currently no indication the numbers will take the dramatic turn they have taken in the Melbourne outbreak, or a large number of US states. That’s because infections in Sweden currently seem to afflict mostly those in the 20–60 years range for which probability of fatality has been known for a long time to be substantially reduced. (I understand that Sweden has currently 10 new ICU cases daily. This will result in up to 3 deaths daily with a lag of a couple of weeks. When it peaked in the April/May it was around 50 ICU cases and later 15 deaths from ICUs per day. The big unknown currently is what happens in the old-folks homes.) Clearly though, the Swedish authorities have not taken the second wave lightly since they have imposed localized restrictions that are often tougher (albeit more confusing and hence inducing less compliance) than those during the first wave.
As mentioned, as of the writing of this version (November 5), Sweden has dropped out of the Top Fifteen deaths-per-million-of-population league table.
Around the same time I published the original version of this article, Anders Tegnell, the chief epidemiologist at Sweden’s public health agency, was quoted as saying,
“In many ways the voluntary measures we put in place in Sweden have been just as effective as complete lockdowns in other countries.” He added:
“We are now seeing rapidly falling cases, we have continuously had healthcare that has been working, there have been free beds at any given time, never any crowding in the hospitals. The failure has of course been the death toll that has been very much related to the long-term care facilities in Sweden. Now that has improved, we see a lot less cases in those facilities”.
Tegnell’s summary is broadly in line with this history of the Swedish covid-19 response, dated October 31, by a doctor working in the emergency room of one of the big hospitals in Stockholm. Rushworth’s earlier blog entry, dated August 4, about how bad the virus really is remains a useful — albeit outdated — read.
Even now, in early November, with cases having increased substantially, Tegnell’s approval is high (and actually keeps increasing), currently being above 70 percent.
To re-iterate, what we see currently is a snapshot at a still early stage of the pandemic, as the currently exploding numbers in the USA and parts of Europe and several other countries (namely Middle and Latin America but also Russia) suggest. To compare death rates at the present stage is hence misleading and unhelpful to the extent that it is meant to inform policy responses. The game is not over til it’s over. Which by most accounts is years down the road.
Second, and along a similar line of reasoning, I submit that it is too early to conclude that the economic gains were not worth the pain. For starters I note that the Andersen et al. study often used to argue against the Swedish model is not only from 12 May 2020 but uses data from 11 March to 5 April 2020. It is in my view disingenuous at best to parade data that old in an article published at the end of July 2020, nevermind today. Likewise to build synthetic doppelgangers of Sweden for the second quarter and draw conclusions from it is at best an academic, and in my view severely misleading, exercise.
In September 2020, for example, the retail trade sales volume increased by 3.9 percent in September 2020 compared with the same month a year ago. Retail sales in durables increased by 4.3 percent, while retail sales in consumables (excluding Systembolaget, the state-owned chain of liquor stores) increased by 1.8 percent.
Sweden’s economy is set to make a stronger recovery than expected in the middle of the year, with Sweden’s GNP at market prices currently expected to shrink by between 3–4 % for 2020, according to the National Institute for Economic Research. That’ s up from earlier predictions, a couple of months back of about 5% shrinkage but does not take into account the strong localized restrictions that have been imposed over the last couple of weeks.
Even a 5% shrinkage would be a much better economic outcome than the European Commission predicted in its summer (July) interim forecast, ie. a forecast that — save the model-based scenario analysis in Box 1.1. — did not yet factor in the second (and third) waves that forced almost all European countries into lockdowns of various degrees. The model-based scenario analysis predicted for a second wave an output loss three times that size (slightly above 10% and slightly below 12% relative to pre-pandemic path). Make no mistake about it: any such dramatic output loss will have consequences down the road, last but not least because good public-health outcomes require reasonably well-functioning economies.
Third, Sweden, at this stage, has the considerable advantage of having reached some form of behavioral equilibrium and compliance — Sweden’s Civil Contingencies Agency estimated it in July to be well above 80 percent. That was around the same time when Australian media reported that 1 out of 4 in Melbourne that tested positive did not obey stay-at-home-orders. And while there have been vocal people in Sweden opposing Tegnell’s strategy (see here for some relevant background), there have not been the kind of regular protests against (lockdown) restrictions that other countries such as Germany have seen. This surely reflects the very European willingness to accept that there are trade-offs between health, educational, economic, and other outcomes (anxiety, loneliness, mental health, suicides) and that the avoidance of covid-19 deaths should not be prioritized over the avoidance of other deaths.
There are legitimate questions to be asked about the Swedish strategy (e.g., which conflicted role its former state epidemiologist Johan Giesecke played in defining Sweden’s public health agency’s strategy, how and why the old-age care sector under-performed in Sweden, and why the death rate for ethnic Swedes seems about half of the country’s overall death rate).
In the end, it will take years to arrive at better grounded assessments of the different strategies that various countries have adopted. Any such assessment — if creditable — will have to compare health, educational, economic, and other outcomes and will require to weigh the lives lost today and in the future, properly weighted by their age distribution, the damage done to those that do survive encounters with the virus, and the damage that has been done to the millions on JobKeeper and JobSeeker, out of work (e.g., thousands of casuals let go by universities and not eligible for JobKeeper payments), or underemployed, the tens of thousands of livelihoods lost, the millions of superannuation accounts raided, and the educational and professional prospects of the young ones further dimmed. We have not even started to try to understand, for example, what the costs are of the disruptions to education we currently experience in Australia although preliminary indications are shocking. And we are far from quantifying the mental-health complications that inevitably will eventuate as the pandemic marches on, movement restrictions are maintained to various degrees, and the consequences of opportunities lost during this pandemic become more salient.
Update 19 December 2020:
And here we go again. Number of deaths as the sole criterion for the success of a pandemic strategy. The usual gutter journalism that I have come to expect. Of course with no-trade-off public-health-uber-alles economists working their privileged megaphones it’s hardly a surprise.
“”We are doing what we have done throughout this pandemic, the right measures at the right time. And this is what this is,” Lofven said.
With the number of deaths close to 8000, Sweden’s death rate per capita is several times higher than that of its Nordic neighbours but lower than several European countries that opted for lockdowns.”
Consider following me on twitter @aortmannphd