Sweden has attracted global attention for not imposing a full lockdown, as seen in most of Europe, to contain the coronavirus pandemic.
Nonetheless, data released from the country’s central bank and a leading Swedish think tank show that the economy will be just as badly hit as its European neighbors, if not worse.
This is from Holly Ellyatt, “Sweden had no lockdown but its economy is expected to suffer just as badly as its European neighbors,” CNBC.com, April 30.
The article’s title is accurate. The second paragraph is not. The data don’t show that at all. How could they?
The next two paragraphs clarify:
Sweden’s central bank, the Riksbank, gave two possible scenarios for the economic outlook in 2020, which it said “depend on how long the spread of infection continues and on how long the restrictions implemented to slow it down are in place.” Both possible economic outcomes are bleak.
In the first scenario (scenario A in the chart below), gross domestic product contracts by 6.9% in 2020 before rebounding to grow 4.6% in 2021. In a more negative prediction (scenario B), GDP could contract by 9.7% and a recovery could be slower with the economy growing 1.7% in 2021.
So the Swedish central bank made a prediction. The prediction could be too high or too low. The odds that it’s just right are miniscule. But the Swedish central bank did not show that Sweden’s “economy will be just as badly hit as its European neighbors, if not worse.”
Everyone understands this in other contexts. I predicted in early November 2016 that Hillary Clinton would win the presidential election. I did not show it.
READER COMMENTS
Charles Rosett
May 15 2020 at 2:10pm
It’s the same kind of thinking that takes predictions about global warming as factual truth, however faulty the models may be in practice.
Alan Goldhammer
May 15 2020 at 3:25pm
“It’s the same kind of thinking that takes predictions about global warming as factual truth, however faulty the models may be in practice.”
The problem with your statement is that there is accumulating data that points to global warming taking place. Surface temperatures in many geographical areas have been rising and ice caps have been shrinking. If this is not warming, do you have an alternative hypothesis?
Philo
May 15 2020 at 3:39pm
Warming has, indeed, been taking place. The issue concerns predictions of future warming.
Mark Bahner
May 15 2020 at 4:00pm
Your question isn’t really relevant to his comments. He’s commenting about what climate models predict for the future. Climate models don’t “show” the global surface or lower tropospheric temperature in the future will be warmer, just like these economic models don’t “show” what will happen to the Swedish economy in the rest of 2020 and in 2021.
Alan Goldhammer
May 16 2020 at 3:48pm
At some point in time, someone has to use information from a model to make a prediction about the future irrespective of the given area. I use models all the time running my family investment office and many others do to. Of course the easy alternative is just to ignore all the past data and not even attempt to predict anything at all. I see this all the time, particularly with climate science.
Mark Z
May 16 2020 at 10:27pm
I don’t think that’s what’s being suggested. As you say, you make predictions. Saying “showing” =/= predicting isn’t a rejection of predictions. For example, ‘the data from the past x years shows that inflation was lower than previously expected’ and neither ‘from the data we’d predict that inflation will be lower than currently expected in the next x years’ aren’t equivalent to ‘data from the past x years shows that inflation will be lower than expected for the next x years,’ the latter assuming something much stronger about the consistency of an observed pattern with future events.
At some point of course an observed pattern is clearly sufficiently predictive of the future that saying it ‘shows’ what will happen is perfectly valid, e.g., observing a planets orbit about a star may indeed show us when it will be at a certain point in its orbit next year.
AMT
May 15 2020 at 4:18pm
Do you really need to bemoan about using the term “data” instead of “projections” [obviously based on data collected to date] when the title, not to mention the remainder of the sentence (“WILL”), makes it perfectly clear these are projections? Do you contend that data collected so far contradict those projections? You’d have to be actively trying to misinterpret the article to find it problematic enough to think it would actually mislead anyone. I think most people understand the future hasn’t happened yet…
David Henderson
May 15 2020 at 5:22pm
I don’t think I can make it clearer. The data do not show that the Swedish economy will be just as badly hit as the other European economies.
I wouldn’t have written this if I hadn’t seen a number of people on Facebook writing as if they think the data do show that.
AMT
May 15 2020 at 6:30pm
That, along with the rest of your post, seems to be saying the problem was using the word “data” instead of projection. Of course, projections are not guarantees, or “showing facts.” I think the article makes it quite clear these are projections.
That sounds more like you think the projection is likely to be far from reality. What are the reasons you think these projections are erroneous?
Otherwise, you are just stating the obvious, that projections are not facts (projections do not “show” exactly what will happen). No one is disagreeing with that. I’m not sure who your facebook friends are that cannot differentiate between the present and the future. But given how you grotesquely mischaracterize the overall thrust of this article based on what is obviously a single poorly chosen word, I’m projecting it’s quite likely that your friends actually understand that the projections for Sweden are something close to what is likely to happen (e.g. “projections”). And it doesn’t matter if there is a “miniscule” chance of it being exactly correct. When you hear the weather forecast is calling for a hurricane, do you complain “The odds are miniscule whether it is going to actually rain exactly 4.25 inches and have wind gusts of 72mph over the next 24 hours!”? The precision and accuracy of forecasts matter, but obviously getting it exactly right is a ridiculous demand. Saying there is a “miniscule” chance of a projection being “just right” is a baseless complaint.
So, why do you think these projections are so grossly inaccurate that Sweden’s economy will perform far differently than they predict?
David Henderson
May 15 2020 at 6:41pm
I’m not going to bother. I was going to, but then I got to the “minuscule” part of your comment. On that, I was simply explaining why it’s minuscule and wasn’t trying to criticize that part of the article in any way.
I think you’re trying to pick a fight and I’m not going to fight.
Jon Murphy
May 15 2020 at 10:08pm
Here’s the thing, AMT:
You may be intelligent and careful enough to understand that the model is making a prediction, but I promise you a lot of people are not, especially political decision-makers (see, for example, this earlier post by Prof. Henderson). I can’t tell you how many people I’ve seen on Facebook and elsewhere pointing to this very Swedish article and claiming that it proves the Swedish efforts didn’t work (notice the change from the future to present).
Words have meanings. We must be very careful with them.
One final point: in my experience, it is very important to constantly state the obvious. I’ve noticed people so very rarely observe the obvious because it is so obvious. It is obvious my apartment has a ceiling fan (one can see that simply by observing). But after 2 years in the apartment, my roommate just discovered it today.
What is obvious to some people is not obvious to others. It’s obvious that politicians are just like us, and yet it took James Buchanan and Gordon Tullock many years and getting chased out of the University of Virginia to get the econ profession to start thinking about that fact (and indeed many economists still do not see it). The obvious should be repeated over and over again lest the very mistakes Prof. Henderson highlights become commonplace.
Mark Bahner
May 15 2020 at 9:55pm
The data collected can’t contradict projections of the future. In a similar situation, Robert Lucas, Jr. projected that global economic growth would slow down from approximately 3% per year in the late 20th century to about 2% per year in the year 2100. In contrast, I projected that global economic growth would increase from 3% per year in the late 20th century to more than 10 percent per year in the year 2100.
The data collected at the time of our respective projections could not possibly contradict our projections. We both started with the same data. He simply thought that global economic growth would slow due to poor countries catching up to rich countries, and thus everyone growing at the late 20th century rich-country growth rate of about 2% per year. In contrast, I said (and still say) that the source of economic growth is human brains. And computers will provide a staggering number of human brain equivalents (think quintillions of human brain equivalents) even by the middle of the 21st century.
Global economic growth in the 21st century
Again, data collected to date can’t contradict (or confirm) projections for the remainder of 2020 and the year 2021.
It’s not obvious. People take the output of models as being facts all the time. In fact, this statement is exactly the situation of a person reporting the output of a model as a fact:
Sean Kennedy
May 16 2020 at 6:40am
Great post. It is so frustrating when I see people using this as evidence that the Swedish model doesn’t work. It’s a very week argument based on what you’ve mentioned. Also it’s not comparing forecasts which are created by the same person so the methodology could be completely different. There’s a lot of judgment involved forecasting and small adjustmebts to assumptions can have big impacts. Using the Riksbank forecasting model could show those other countries much more heavily impacted than Sweden so read anything into this as a basis for action is foolish.
David Bauman
May 16 2020 at 10:40am
Just viewed your on-line debate. Without a denominator of how many people have contracted Covid-19 and were asymptomatic and/or recovered from it, how can a death be determined and compared to the flu ? Why did neither of you bring up anti-body testing data that exists to compare with covid death rates ? Also ignored was evidence that incentives exist (hospital reimbursement per death when they are losing $) for overcounting of Covid-19 deaths.
See Dr. Jay Bhattacharya on two samplings of anti-body testing that suggests the virus is just beginning and it’s it’s death rate (assuming anti-body protection against reinfection) is similar to the flu. Many politicians relish the opportunity for a socialist/communist economic revolution that never ending covid-19 provides.
https://www.hoover.org/research/dr-jay-bhattacharya-his-new-mlb-covid-19-study-and-dilemma-lockdown
https://www.hoover.org/research/fight-against-covid-19-update-dr-jay-bhattacharya
Thomas Hutcheson
May 17 2020 at 8:07am
Wow! Could the US, please, have the “disastrous” results of Sweden’s worst case scenario?
Thomas Hutcheson
May 17 2020 at 1:18pm
I’m so used to ignoring predictions that do not lay out their conditionals that I did not and (do not) pay attention to “show.” Is the Swedish central bank targeting NGDP, the PL, the inflation rate? the exchange rate?, something else? and did it make the same assumptions about foreign central banks? And what does the model say about a no NPI scenario?
Mark Z
May 17 2020 at 4:04pm
I don’t know what Sweden officially targets, but it seems to have been stabilizing NGDP at least until a few years ago (no idea about since then). David Beckworth argued this was why Sweden was recovering faster from the great recession in 2011.
Thomas Hutcheson
May 18 2020 at 10:54am
Actually, I meant to say which assumptions about central bank behavior in Sweden and abroad are built into the model?
David Seltzer
May 18 2020 at 6:22pm
I was one of Fischer Black’s research assistants at The U of C. He was quite circumspect about econometric models or their projections because the coefficients were not policy or regime invariant. Fama’s take on historical stock price behavior yielded his now famous Efficient Market Hypothesis.
Comments are closed.