This is a follow-up
to my June paper on the numbers relating to the COVID epidemic world-wide. That
paper is at https://wattsupwiththat.com/2020/06/20/covid-19-understanding-the-numbers-coronavirus/.
This time, the main focus will be
on the question: how well have lockdowns worked in different countries? I will
look first at those countries in Western Europe, which show evidence (or not)
of the various lockdowns having had a significant effect on the daily new case
counts. Then, I will visit some of the more “interesting” countries (from a
COVID statistics point of view at the present time) in other parts of the
world.
As for the first
paper, my data comes from Our World in Data. It can now be downloaded from https://ourworldindata.org/coronavirus-source-data. They are still keeping their spreadsheet (now
34,600 rows and counting) updated each day. They have also added a “stringency
index” column to the daily record, which indicates how tightly the particular
country was locked down on the particular day. I’ll discuss that soon. All the
data I use in this paper came from their spreadsheet dated August 4th,
2020.
But before I begin,
an explanation for non-Brits of the word “cock-up” in the title. The Cambridge
Dictionary defines it as “something that is done wrong or badly.” A North
American equivalent might be “snafu,” although it usually takes many cock-ups
to make a full snafu. Cock-up is to snafu, roughly, as dime to dollar.
The question I aim
to answer is: Is there evidence that the “lockdown” reactions to the epidemic
of governments around the world have significantly helped to alleviate the
effects of the virus, compared with what would have happened without those lockdowns?
Or has it, perhaps, been a lot of pain for no, or little, real gain? You shall
judge.
The Stringency Index
The “stringency
index” is a summary number (expressed as a percentage) of how far a country is
locked down. 0% is business as usual, while 100% represents maximum
restrictions for all the criteria considered. The stringency data supplied to
Our World in Data, and the supporting .csv file of daily data, both come from
the Blavatnik School of Government, which like Our World in Data is a part of
Oxford University. The file I used was the version from August 5th.
You can download the
.csv file from https://www.bsg.ox.ac.uk/research/research-projects/coronavirus-government-response-tracker. (They now have an additional secondary
dataset on US states’ responses to COVID-19.) The guide which explains the
individual data fields is at https://github.com/OxCGRT/covid-policy-tracker/blob/master/documentation/codebook.md. The full codebook (working paper) and the
documentation on the new US data are available as PDFs from https://www.bsg.ox.ac.uk/research/publications/variation-us-states-responses-covid-19.
There are nine
factors included in the stringency index. They are: school closing, workplace
closing, cancel public events, restrictions on gathering size, close public
transport, stay at home requirements, restrictions on internal movement,
restrictions on international travel, and public information campaign. In
calculating the stringency index, they weight all nine factors equally; an
approach which, for me at least, seems as good as any.
I did, however,
identify what seem to me to be a couple of things missing from their
nine-factor system. Firstly, there is no mention of mask-wearing requirements
at all. And secondly, they have only a single “workplace closing” factor, whereas
I would have preferred two. One for workplaces open to the public, such as
shops, restaurants and bars; and one for other workplaces, such as offices and
factories. This is because some countries, including the UK, have locked down
considerably harder on the first than the second; and have also loosened that
lockdown in several phases. It’s unfortunate that, because of the conflation of
the two types of workplaces, none of these relaxations appear as changes in the
stringency index for the UK at all.
All that being said,
it’s the best data I’ve got, so I’m going to explore it!
The underlying mathematics
My first question on
how to assess the lockdowns was: What should I plot the stringency index
against, to try to get an idea of how changes in it might (or might not) have
affected new cases? To make that decision, I needed to learn a (very) little
about the epidemiological mathematics.
Since I was only
interested in new cases, I only needed to think about one transition: the one
from Susceptible to Infected. Obviously, the number of infections that are
detected as new cases will depend on the level of testing being done, as well
as the number of new infections. But it’s rare for the level of testing to
change significantly and very suddenly. (Sweden at the beginning of June is the
only case I know of). So, if I keep my time horizon short enough – say, a week
– it should be a reasonable assumption that the number of new cases
(eventually) detected will be roughly in proportion to the number of new infections.
Now, the most
appropriate of the epidemiological models to COVID is probably the SEIR model. But if I make the (over-) simplifying assumption that the incubation
period is roughly constant, then I can just use a basic SIR model. Such models have
a new infection rate of the following form:
Here, N is the number of new infections per day; I the number currently infected; S the number of susceptible; P the total population; C the contacts per person per day, weighted by
time spent; and τ
is the transmissibility of the disease (the chance
of disease transmission in a contact of a given length between a susceptible
and an infectious). The units for C and τ
must be chosen so that their product τC is a dimensionless number (it is often given
the symbol β).
Since, of the parameters in this
equation, only C is amenable to change by individual actions or
even by government edict, the intent of a lockdown must be to reduce N by reducing C. And this would still be so, even
if I was using a far more complex model, with all the bells and whistles that
“experts” like to hang on such things.
Thus, I would expect any lockdown which
is at all effective to result in a reduction of N, and thus of the number of new
cases. This reduction might occur anywhere from all but immediately, or with a
delay of up to the incubation period of the virus plus however long it takes
people to start complying with the lockdown. So, say, one to two weeks. On the
other side, a partial or full release of a lockdown should produce a
corresponding increase of
, and so of the number of new cases
– unless, of course, herd immunity has been nearly, or actually, reached.
An extra complication is that the number
of new cases in an epidemic has its own dynamics. To mis-quote Larry Wall’s
Harvard Law: “Under controlled conditions of light, temperature, humidity, social
interaction and other factors, the organism will do as it damn well pleases.” So,
I thought it wouldn’t be useful to try to look for the effects of lockdown or
release directly in the new case numbers. Instead, I thought, I’ll look at the rate
of change of new cases. The effects of a lockdown or release ought to be
visible as a “knee-bend” (maybe not instantly, but certainly within a couple of
weeks) in this rate of change.
How I graphed the lockdowns for the UK
In my earlier paper, I used a graph of daily
growth of weekly averaged new cases, to enable me to pick out early outbreaks
of the virus which were too small to be easily visible on the headline graph of
daily cases and deaths. So, I adapted this into a graph of growth of new cases
on a weekly basis. I started out by trying to plot the ratio of the raw new
case count to the count from 7 days previously; but the data proved to be too
noisy for that to be useful. So, I had to settle for the week-on-week change in
new case counts which were already weekly averaged. The purist in me gags at
such a kludge; but climate scientists can get away with doing such things, so
why can’t I?
So, the weekly growth percentage shown
for a given day is the ratio of the (already weekly averaged) new cases 4 days
after that day, to the new cases 3 days before that day. The date underneath a
point on the graph is as close as I could make it to the centre of the period
from which its data was taken.
I also decided, in order to keep the
height of the graph within reasonable bounds, to cap the weekly growth
percentages at a maximum of 200%. Anything above 200% is simply shown as if it
was 200%. The resulting graph for the UK, my starting point for this journey, comes
out like this:
The blue line tells us that the week-on-week
growth has been negative for most of the time since late April; superficially
at least, good. Though it has crept up again a bit in July. It also suggests
that the week-to-week new case growth had already started to come down from its
peak, before any significant lockdown was put in place.
To confirm that last statement, I made a
version of the graph that didn’t constrain the growth percentage to 200%. To be
doubly certain, I also switched back to using the raw data instead of the
weekly averaged data. The result was remarkable:
Say what?
The week-on-week growth in cases was already well on its way down, before any
kind of legislative lockdown was even contemplated. Now I’ll grant you, divide
a small number from a noisy and uncertain distribution by another small number
from the same distribution, and the result is uncertain in spades. But there’s
a decline there, and it isn’t a hidden one.
Now, what might
have caused this decline? My best guess is that when expanding into new, fresh
fields, the virus (which takes time, 2 to 12 days I’m told, to incubate) can’t
keep up with its old rate of progress. It’s always going to be the first
cluster that expands fastest.
The April to July changes, on the other
hand, show just as clearly on the basic daily cases and deaths graph:
The suspicious among you may be thinking:
“But that isn’t a bit like the graph you showed us last time!” No, indeed.
Here’s the one from six weeks ago:
The reason for this revisionist tendency
in the UK’s data is that, since June, they have moved from showing new cases against
the day they were reported, to showing them against the day the test was done;
just as the Swedes were in the process of doing at the time of my earlier
paper. The effect is to separate the main peak into two, and to move the first
peak earlier by a few days. (Notice, also, that the second peak comes shortly
after the Easter week-end.)
Another effect of this means of reporting
– not visible on this graph, but you can see it in several other countries’
data – is that new case counts for the very latest few days will usually be
understated. They will be added to as delayed test results come in. So, if you
see an apparent downturn in new cases in the last few days before the end of
the data, it’s very probably spurious, and should not be taken into account in
any conclusions. Similarly, perhaps, for an apparent upturn; maybe reports from
the cities come in quicker than those from the boondocks?
To return to the plot of weekly case
growth against lockdown stringency. I used the raw daily data provided by the
Blavatnik people to produce a history of the changes in UK regulations over the
epidemic so far. (The texts in the “Measure” column are my own summaries, in
text form, of the numeric codings which are used for each of the nine factors
in the Blavatnik stringency data). Here’s the result:
|
Date
|
Measure
|
Scope
|
Stringency %
|
|
February 2nd
|
Co-ordinated
public information campaign
|
National
|
11
|
|
March 16th
|
Recommended
workplace closing
|
National
|
15
|
|
March 17th
|
Recommended public event cancellations
|
National
|
20
|
|
March 21st
|
Required some workplaces to close
|
National
|
30
|
|
|
Mandatory public event cancellations
|
National
|
|
|
March 22nd
|
Recommended not to travel
|
National
|
35
|
|
March 23rd
|
Mandatory all schools to close
|
National
|
70
|
|
|
Restrictions on small gatherings (<=10 people)
|
National
|
|
|
|
Required stay at home with exceptions
|
National
|
|
|
|
Mandatory restrictions on internal travel
|
National
|
|
|
March 26th
|
Recommended public transport closures
|
National
|
76
|
|
May 13th
|
Recommended stay at home
|
National
|
69
|
|
|
Mandatory restrictions on internal travel
|
Regional
|
|
|
June 1st
|
Mandatory all schools to close
|
Regional
|
68
|
|
June 8th
|
Quarantine international arrivals from high-risk regions
|
|
73
|
|
June 15th
|
Recommended stay at home
|
Regional
|
71
|
|
July 4th
|
Restrictions on small gatherings (<=10 people)
|
Regional
|
70
|
|
July 6th
|
Recommended not to travel
|
Regional
|
64
|
|
August 1st
|
Mandatory restrictions on internal travel
|
Regional
|
68
|
|
No stay-at-home measures
|
|
|
I was tempted to insert in there: “April
12th (Easter Sunday): Prime minister Boris Johnson comes back from
the dead!”
Now, look back at that graph of weekly
case growth versus stringency limit. The hypothesis that lockdowns and relaxations
ought to trigger drops and increases in weekly case growth would suggest that
the blue line ought to move in the opposite direction to the brown, soon – or,
at most, in a week or two – after it. Since the big lockdown in March, does
that graph show any evidence of such a movement? Not to my eyes. Remember,
also, that shops and pubs re-opened in three phases on June 1st,
June 15th and July 4th, none of which show up in the
Blavatnik data. If I squint hard, I can maybe see a tiny upward knee-bend in
the blue line in the first week of June; but it’s not conclusive.
A tour of Western Europe
Next, I thought, I’ll re-visit some of
the Western European countries I looked at last time round. First, Iceland. Six
weeks ago, it was the land of the perfect bell-curve. Now:
Hmmm… Ma Bell has spawned Baby Bell, and
there’s a third bell just starting. Here are the growth rates and lockdown
stringencies:
The Icelanders managed to snuff out the
first phase of the epidemic completely in just two months from the first case,
or six weeks from lockdown. Pretty impressive. What was it down to – low
population density, good track and trace, strong Viking constitutions? Or could
it have been due to the government locking down? Let’s have a look at the Icelandic
timeline:
|
Date
|
Measure
|
Scope
|
Stringency %
|
|
January 23rd
|
Co-ordinated public information campaign
|
Regional
|
8
|
|
January 28th
|
Co-ordinated public information campaign
|
National
|
11
|
|
January 29th
|
Quarantine international arrivals from high-risk
regions
|
|
17
|
|
March 15th
|
Restrictions on medium size gatherings (11-100
people)
|
National
|
25
|
|
March 16th
|
Required some schools to close
|
National
|
51
|
|
|
Required some workplaces to close
|
National
|
|
|
|
Mandatory public event cancellations
|
National
|
|
|
March 20th
|
Ban on international arrivals from some regions
|
|
54
|
|
May 4th
|
Schools open
|
|
46
|
|
May 25th
|
Workplaces open
|
|
36
|
|
|
Restrictions on large gatherings (101-1000 people)
|
National
|
|
|
June 15th
|
Quarantine international arrivals from high-risk
regions
|
|
33
|
|
July 31st
|
Recommended workplace closing
|
National
|
40
|
|
|
Restrictions on medium size gatherings (11-100
people)
|
National
|
|
Well, there you have it. As in the UK,
the weekly case growth was declining before any effects on Icelanders of even the
first lockdown edict could kick in. Moreover, the first cases for more than a
month appeared right after they went from banning high-risk international
arrivals back to quarantining them. They did close bars and close-proximity
businesses like hairdressers from late March to May. But otherwise, I think the
Icelanders have got the job done; maximum protection at minimal cost. It will
be interesting to see how quickly the latest outbreak of new cases dies down.
Next, to two central Western European
countries I looked at in detail last time. First, Italy. Here’s the weekly case
growth versus stringency:
The maximum lockdown level in Italy was
94%, and they were at over 85% all the way from March 11th to May 4th.
Ouch.
Now, there are a couple of instances here
of what I would have expected to see when a lockdown, which has had a real
effect, is lifted. There’s a definite knee-bend in weekly case growth around
May 4th, and another more debatable one about June 4th.
According to the Blavatnik data, on May 4th there was a package of
five measures:
1.
“Recommended workplace closing”
(had previously been “Mandatory all but essential workplaces to close”),
2.
“Public transport open” (had
previously been “Mandatory public transport closures”),
3.
“Recommended stay at home” (had
previously been at a mixture of “Required stay at home with exceptions” and “Mandatory
stay at home with minimal exceptions”),
4.
“Recommended not to travel”
(had previously been “Mandatory restrictions on internal travel”),
5.
“Quarantine international arrivals
from high-risk regions” (had previously been “Ban on international arrivals
from some regions”).
And on June 2nd and 3rd
all restrictions on travel were ended, both internal and international.
Here’s the daily cases and deaths graph:
There has been a small increase in new
daily cases since early June. But it’s piffling compared with the horrors back
in March. Barring some unexpected development, the Italians do seem now to be
through the worst.
To Switzerland:
The measures since late April have been:
|
Date
|
Measure
|
Scope
|
Stringency %
|
|
April 27th
|
Required some workplaces to close (previously Mandatory
all but essential workplaces to close)
|
National
|
69
|
|
May 11th
|
Required some schools to close (previously Mandatory
all schools to close)
|
National
|
66
|
|
May 30th
|
Restrictions on medium size gatherings (11-100
people) (previously Restrictions on small gatherings (<=10 people))
|
National
|
63
|
|
June 6th
|
Schools open
|
National
|
46
|
|
Recommended workplace closing (previously Required
some workplaces to close)
|
National
|
|
|
|
No restrictions on internal travel (previously Recommended
not to travel)
|
National
|
|
|
June 22nd
|
Recommended workplace closing
|
Regional
|
35
|
|
Restrictions on very large gatherings (>1000
people)
|
National
|
|
|
|
No stay-at-home measures (previously Recommended
stay at home)
|
|
|
|
July 3rd
|
Restrictions on medium size gatherings (11-100
people)
|
Regional
|
39
|
Two of these seem to have been followed
by increases in the weekly growth rate, both a couple of weeks afterwards. One
was the re-opening of many schools. The other was the major package of measures
on June 6th. The change on July 3rd seems to have had a
more immediate effect in the opposite direction, even though only implemented
regionally.
It looks as if medium or larger gatherings
of people are one of the major sources of spread of the infection. Which,
indeed, was what happened over Carnival week-end in the Netherlands and Belgium
at the very beginning of the outbreak. It seems that the Icelanders agree with this,
since the measure they chose to enact very recently in an attempt to halt their
third wave of the infection was re-lowering the limit on gathering size.
Now, for the one (I hope) you’ve all been
waiting for: Sweden. Ah, Sweden.
Pay no attention to that vast crevasse in
the second half of June! Sweden shut down for the 3-day long summer solstice
holiday. And even COVID testing all but stopped. Remembering that the big jump
at the start of June was caused by a sudden expansion of testing, it looks now
as if the Swedish hands-off strategy, which so many freedom-hating moaners have
criticized, may well have been right all along. (The recent apparent jump, I
won’t comment on today, as it hasn’t yet gone on long enough to draw any
conclusions).
Here’s the timeline:
|
Date
|
Measure
|
Scope
|
Stringency %
|
|
March 9th
|
Co-ordinated public information campaign
|
National
|
11
|
|
March 12th
|
Restrictions on large gatherings (101-1000 people)
|
National
|
17
|
|
March 18th
|
Recommended school closing
|
National
|
20
|
|
March 19th
|
Ban on international arrivals from some regions
|
|
29
|
|
March 25th
|
Recommended workplace closing
|
National
|
32
|
|
March 29th
|
Restrictions on medium size gatherings (11-100
people)
|
National
|
35
|
|
April 1st
|
Recommended public event cancellations
|
National
|
41
|
|
April 4th
|
Recommended not to travel
|
National
|
46
|
|
June 13th
|
No restrictions on internal travel
|
|
41
|
|
June 15th
|
Recommended school closing
|
Regional
|
39
|
Look at that graph again. By the time the
stringency reached its peak on April 4th, the weekly case growth was
already down below 50%, and soon approached zero. And although the stringency
peaked at 46%, only the restrictions on gatherings and the ban on international
arrivals from badly affected regions were actually mandatory. Moreover, the ban
on international arrivals seems to have been restricted mainly to those from China,
Iran and Italy.
Now, the Swedes have taken a lot of flak
from the moaners for their high deaths per million (570 as at August 10th).
This puts them eighth in the world in that particular stakes, though still well
behind Belgium, the UK, Spain and Italy. But like those countries, a very high
proportion of their COVID deaths have been in care homes. There’s not much
anyone could have done about those.
The Swedes were also pro-active in upping
the testing rate as early as they could. As a result, they are second among
major Western European countries (behind Luxembourg and ahead of Spain) in
cases per million population. On all the evidence, I think the Swedes have done
as good a job as they could reasonably have been expected to do under the
circumstances. They deserve a round of cheers. Skål!
But the really interesting question is
this. If countries like the UK and Italy had taken the Swedish approach,
instead of messing up the economy and people’s lives for months on end, would
the end result in terms of COVID cases and deaths have been much, or indeed
any, different? Looking at the Swedish graph during March and early April, I’m coming
to doubt it.
Next, Portugal, which back in June was
starting to look a bit unpredictable.
Now, that’s odd. They seemed to have it
under control for a while in April, then suddenly, up she goes! At least two
weeks after the last change in stringency. What the Portuguese did, so the
Blavatnik data shows, is bring in mandatory travel restrictions for two short
periods, April 9th to 14th and May 1st to 4th
– Easter week-end and May Day week-end respectively. Maybe a lot of people just
ignored the bans, particularly the second? Well, it didn’t do much harm on
either occasion, as you can see from the subsequent down-turns. Since then, it
doesn’t look too alarming. And there doesn’t seem to be much correlation
between lockdown level and weekly case growth.
Last on my tour of Western Europe, a
country I didn’t even look at last time: Luxembourg. This is the only country,
among all those whose figures I’ve examined so far, which has actually gone the
whole way to complete release of all lockdowns. As I noted earlier, they have
the highest cases per million population among major Western European countries
(over 11,000). And yet, in deaths per million (191), they are only 10th.
So, they must have been doing some things right, even before it came to
releasing their lockdowns.
Here are the graphs:
Now, that’s an epidemic management
strategy I like. Use the first peak to work out exactly what the limits are on
your health care resources. Then relax the lockdowns bit by bit, and don’t
worry about how many new cases you get, until you are getting near the limits
of your resources. If that does threaten to happen, re-introduce the most
effective lockdown you have available.
“Genius, Holmes!” “Elementary, my dear
Watson.”
So, they unlocked everything on July 16th.
On July 19th, they re-introduced “Restrictions on medium size
gatherings (11-100 people).” The effect, if I can believe my graphs, seems to
have been both instant and spectacular. So, as the Blavatnik data tells (but my
graph cuts off the day before), they unlocked again on July 28th.
Well done the Letzebuergesch! (That’s what they call themselves in their own
language). And now, they know they can do the same thing again if they ever
need to.
It’s a pity that buffoons like Boris Johnson
don’t have ideas (or even advisors) as good as these.
A whistle-stop world tour
So, let’s see how the rest of the world
is doing, shall we?
The most obvious feature of the world
cases graph is that the third wave of the epidemic (world-wide) is far more
spread out in time than the first (mainly South and East Asian) and second
(mainly Western European). It’s also fair to say that the apparent drop-off
towards the end may well be an artefact, caused by late reporting of tests.
But the deaths per case graph is
encouraging for the longer term. As the virus spreads, it seems to be becoming
steadily less lethal. I wonder if, maybe, a less virulent strain is now mixing
with the nasty one, which arrived in February in Iran, Italy and the USA? It’s even
conceivable that we may not, in the end, actually need to exterminate the
damned thing; it might, perhaps, mutate itself back into just another variant
of its brothers, the common cold coronaviruses.
In my world tour, while I’ll usually show
the lockdown graph just to give an overview, I won’t go into details. That’s
because to make proper sense of the data, for most of the large countries I
would need it broken down by state, province or whatever else is the local unit
of sub-division.
So, I’ll start with our Iranian camel friend.
He was pretty much down on his knees back in June. Has he managed to get up?
All I can say is, Ouch. Though to their
credit, the Iranians have generally refrained from trying to lock down too
hard.
Let’s take a look at Kuwait, about whose
handling of the epidemic I had some good things to say back in June:
That’s a full 100% lockdown right there,
from May 10
th to May 30
th. More ouch. Now I know how they
produced that slowdown in cases, and now I’ve seen six more weeks of their
data, I’m not so sure I was right to praise them.
How about the USA?
Is that a watershed around the middle of
July? Maybe. The sustained drop in weekly case growth since mid-June looks
encouraging, but there’s still a long way to go.
Canada?
Nothing much to see in the first graph,
but something really weird in the second. Canada seems to have stopped
reporting any stringency index figures at all since July 19th! Even
the Blavatnik raw data shows “no data” for all dates since then. That, and the
sudden drop in case figures right after, suggest that something is afoot.
Perhaps they are following Sweden and the UK, and changing the way they
allocate cases to dates? I’ll simply shrug my shoulders, and move on to Mexico.
That’s bad on both the cases and
deaths fronts, and even worse when you look at the deaths per case and cases
per test ratios:
But the weekly case growth graph tells
another story. In fact, two stories.
First, the weekly case growth has been
going, slowly but fairly surely, down since April. Long may that continue! And
second, if there is a country whose data suggests that the Harvard Law is right,
and this virus really just does do what it damn well pleases – no matter what
governments or individuals do to try to nudge it – then Mexico is that country.
I’ve only been to Mexico once, and while I do recall the people being rather
noisy, I didn’t find them particularly law-breaking, compared with some other
places I’ve been. So, if these lockdowns are being enforced, they aren’t having
much, if any, effect. That calls into question the entire lockdown idea, and
all that goes with it.
So, to South America. Brazil is more than
a little like Mexico. I won’t bother with the daily cases and deaths, but
here’s the lockdown graph:
The graphs from Ecuador and Chile do not
fill me with any confidence that the numbers they portray are accurate enough
to make it worth while publishing them. Which leaves Peru as my only other stop
in South America:
There have been some significant recent upward
adjustments to their deaths data, and the cases don’t seem to be going too
well, either. As to lockdown, they have been up above 90% for a long time, and
they don’t seem to be having much success with their cautious attempts at regional
unlocks.
After all those troubles in Central and South
America, my next port of call, South Africa, seems at first sight like a different
and better world:
The cases per test percentage is
still high and rising. But though the lockdown level is high, they have started
to unlock. On July 12th, though, they mandated “Required stay at
home with exceptions,” thus going back to the March régime after it had been
relaxed at the beginning of June. This does appear to have been successful in
terms of “turning the corner.” But whether the drop in new cases will continue,
time will tell.
So, to my final continent: Asia.
First, Pakistan.
They have done a lot of unlocking
already, and their approach is to re-impose lockdowns at the provincial level
where they are warranted. It seems to be working well for them, and I think
this particular patient is probably already out of danger; if, of course, I can
believe the figures in the first place. For such a big country, that’s a
remarkable achievement. If it’s real, of course.
Not so for next-door India, though:
They have been all the way up to 100%
lockdown, and are still almost at 80%. But nothing seems to be working for
them.
My Indonesian friends, although they
haven’t “turned the corner,” are doing better than the Indians. And without
ever raising the lockdown level above 75%:
The one lockdown measure they have
re-imposed is total closure of the borders, planned to last until September. From
June 23
rd to July 10
th, the status was “Screening of
international travellers.” My graph shows a knee-bend, which actually precedes
the border closure! Perhaps they announced it some days in advance.
Singapore’s data is rather confused. I
think this is because it’s a cross-roads, and so always vulnerable to new
infections from outside; and yet, for economic survival, the Singaporeans can’t
afford to be too strict about keeping people out. So, I’ll move on, to Japan:
So much for the idea that Japanese
formality and their less physical ways of socializing would keep them safe from
the dreaded COVID! But kudos to the Japanese for not panicking. Yet, at least.
It will be interesting to see how this one pans out.
So, to the final stop on my world tour;
the track-and-trace capital of the world, South Korea.
The level-headed Koreans, for the last few
months, have let the virus bumble along at the bottom of its range, where it
doesn’t pose much of a threat. They could eradicate it from their country if
they wanted to. But there’s no point trying to do that until everyone else is
ready to do the same.
All that said, track and trace on its own
wasn’t enough for them. They took their lockdown stringency up to 82% at one point.
And that included a period, from April 6th to April 19th,
of “Mandatory all but essential workplaces to close.” But, while the increasing
lockdowns did seem to produce “knee-bends” in the weekly case growth, they didn’t
seem cumulatively to have an enormous effect. Until they were relaxed, of
course.
Lock-downs, or cock-ups?
So, have the COVID lock-downs been
cock-ups – or even, perhaps, fully-fledged snafus? The evidence from much of
the world points, in my view, to the answer Yes. Mexico and Brazil, in
particular, suggest that in those countries at least, many, if not all, of the
lockdown measures haven’t had much influence on what the virus ended up doing.
Sweden, too, points in this direction, as does the lack of a strong increase in
new cases as the UK has slowly unlocked. The jury will not be able to bring in its
verdict until full cost versus benefit analyses are in from each country; but at
this point, it isn’t looking good for the Lockdown Party.
There is evidence, particularly from Sweden
and Luxembourg, as well as from Belgium and the Netherlands at the very start
of the epidemic, that close gatherings of many people, who stay together for a
considerable time, do help to spread the virus. Those with memories may also
recall religious gatherings in France and South Korea, which helped give it its
start in those countries.
There is also, perhaps, some evidence
from South Africa that stay-at-home restrictions help to lessen the spread of
the virus. Although the measure in question has been in force for less than a
month, and more time is needed to confirm its effect. In any case, confinement
to barracks is a blunt instrument to bash people over the head with. On the
other hand, there is evidence from Portugal that when people disobey travel
restrictions, it leads to a short-term increase in new cases; but things very
soon return to the course they were on before.
Another point to note is that, of the biggest
countries in the world (with 2018 populations above 200 million), India, the
USA, Indonesia and Brazil are all floundering; some worse than others. It seems
that in epidemic control, smaller countries tend – as you might expect – to do
better. Of course, due to exceptional bad luck at the start of the epidemic,
you will lose some: San Marino, Andorra. But you will win others: Iceland, Liechtenstein
and many more. And the “jewel in the crown,” the best example of all on how
to keep death rates down and how to release lockdowns: Luxembourg.
It remains only for me to castigate the
UK government for all the crap they’ve thrown at us in this “sceptred isle” over
the last five months. To be fair, on the presumption that lockdowns were necessary,
and would actually do what they were designed to do, the choices of what to
lock down were not unreasonable compared to some other countries. But were they
really necessary at all? And was the guff we were told about “not letting
health care resources get overloaded,” perhaps, no more than a smokescreen put
out by politicians, bureaucrats and biased media, that didn’t want to risk making
the National Health Service sacred cow look bad?
But failing even to screen incoming
travellers until June was a glaring error; and closing small shops was way over
the top. The UK government’s attempts to slant the data on tests carried out –
on which they were caught out by statistician Sir David Norgrove in early June,
and by me a month later, see https://wattsupwiththat.com/2020/07/02/is-the-uk-government-misleading-the-public-on-covid-tests/
– wasn’t a clever move, either.
Moreover, the UK, like Italy, has been far too slow to take risks in early unlocking.
As a result, there seems to be forming a “consensus” that big changes are
needed in the UK’s public health systems, and perhaps even in the NHS as a
whole. Some heads may roll, I expect. But not nearly enough.
…and a final trip to the Faeroes
Oh, but I’ve saved the best until last. Is
there, I had been thinking, some place on the planet where the government didn’t
do anything much at all about locking down, compared to other countries around
the same time? Is there somewhere I can use to test the hypothesis that very
few, if any, of the many and varied lockdown measures that have been
implemented in different places, have shown any significant efficacy at all?
Fortunately, the answer is Yes. That
place is the Faeroe Islands. What happened there is that the virus, once
introduced, went through the population – of the major islands, at least – like
the proverbial dose of salts. Within just a few days of the government’s first
reaction, the chief medical officer was saying that he thought most of those,
who were going to be infected, had already been infected. Here are the graphs:
One difficulty presents itself: there is
no stringency index data for the Faeroes, even in the Blavatnik daily data. But
fortunately, a kind soul has put on to the Faeroes coronavirus page at Wikipedia,
https://en.wikipedia.org/wiki/COVID-19_pandemic_in_the_Faroe_Islands,
a detailed account. Thank you, thank you to whoever did this. It’s become fashionable
in some quarters to pooh-pooh Wikipedia; but in matters like this, it is an
absolutely vital resource.
Here, directly quoted from Wikipedia, is
what the Faeroes government did on March 12th:
·
All international travel is
strongly discouraged, unless absolutely necessary.
·
All municipalities are urged to
take measures regarding passenger cruise ships on their way to the Faroe
Islands.
·
Anyone arriving in the Faroe
Islands from overseas must take the utmost precaution and stay at home.
·
Restrictions on visitors to
hospitals and nursing homes will apply. Further guidelines will be issued by
the health and local council authorities.
·
The school system, including
tertiary, secondary and primary schools, will close. Students and pupils will
wherever possible have access to remote teaching.
·
Children's activity centres,
preschools and day care facilities will also close. Childcare will be offered
to those who, for particular reasons, are not able to have their children at
home during working hours.
·
All employees in the public
sector who do not deal with the most essential services should work from home.
Staff will receive further instructions from their respective directors.
·
Measures have already been
taken in the private sector to guard against infection.
·
Bars, venues and restaurants
are urged to close by 22:00 for the next two weeks.
I’m not
sure they would even have bothered to do that much, if the Danes hadn’t decided
to lock down the previous day. And after March 12th, the only
further measure they brought in was on March 17th, that “no more
than 10 people should be together at once, inside or outside.” Essentially the
same restriction as in Iceland at the time. Beyond that, two fuel station
chains closed their shops – but not the fuel pumps. And that was all. This is
not to say that people in the Faeroes did not suffer, personally and
economically; they did. Scan the footnotes to the Wikipedia article, if you
seek evidence.
Now, there
has recently been a new outbreak of cases in the Faeroes. How caused? Once
again, Wikipedia springs to the rescue. The three first cases reported “had
been attending the national festival Ólavsøka where many people gathered in
Tórshavn.” My graphs show only the very start of this; and it has got worse
since then. Which is unfortunate for the people of the Faeroes; but fortunate
for me, because it leads me directly to my…
Conclusions
Of all the countries, the Icelanders and Faeroe
Islanders got closest to right in deciding what to lock down. And the Swedes,
too. You can ask: Did the Faeroe Islanders need to close all schools, when the
Icelanders didn’t? Or, did the Icelanders need to close some workplaces, when
the Faeroe Islanders didn’t? But these are quibbles.
It appears to me that only four of the lockdown
measures, which have been used, have been proven effective. In increasing order
of stringency to ordinary people:
1.
Screen arriving international
travellers, and quarantine, or at need ban, those from high-risk areas.
2.
Ban football matches, public
parades and large events, at which thousands may gather.
3.
Restrict the number of people
who may assemble in one group, or at a small event.
4.
As an absolute last resort,
confine people to their homes for a short period, with exceptions like shopping
and exercise.
The effectiveness of anything beyond
these four is unproven. The pain, though, is obvious.
As to relaxing lockdowns, three cheers
for Luxembourg. Who, unlike many other governments, have approached the whole
matter with commendable common sense.
