Sunday, 19 September 2021

COVID World Review, Part 8: Lockdowns

This is the eighth of my review reports on COVID-19 worldwide as at August 20th 2021. This time, I’m going to look at lockdowns. I’m not even going to bother to show the variations in lockdown stringency over time. I’ll simply go straight to the ordered lists of lockdown stringencies in each region.

The Blavatnik School of Government defines its stringency measure as the average of nine components. Eight of these are lockdowns of different types (schools, workplaces, events, gatherings, public transport, stay at home, travel restrictions, international) and the ninth represents the level of the public information campaign in a country. Face covering mandates are not included in the calculation. The level of each type of lockdown is assessed separately as a percentage of a maximum possible score (so, for example, if a lockdown has levels from 0 for no restrictions to 4 for everything closed down, and its value in a particular country on a particular day is 3, that counts as 75% for that type of lockdown for that country for that day). If the country has a regional lockdown of that type, it is counted as half a point lower than the corresponding national lockdown; so if the value on a day is 3 out of a maximum 4, but the lockdown is only regional, that counts as 62.5%. The nine percentages are then added up and divided by 9 (i.e. averaged), to give an overall stringency percentage.

There are a number of shortcomings of the Blavatnik measure. First, it doesn’t include face covering mandates. Second, the “public information” level has been 100% in most countries for virtually all of the time since the beginning of the epidemic. Third, it tends to overstate the stringency of lockdown in subdivided countries such as China, India, the USA, Canada, Brazil, Russia and the UK, where many or most lockdowns are regional and are imposed on a state by state or province by province basis. In such places, a lockdown in one single subdivision is considered only half a point lower than the same level of lockdown over the whole country. Still, it’s the best data I have, so it’s what I’ll use.

I use two different metrics for assessing lockdowns over the whole course of the epidemic to date. The first is average lockdown stringency. This is the average value of the Blavatnik stringency over the whole epidemic, calculated on a day-by-day basis. The second is average time in full lockdowns. To calculate this I assess, for each of the eight types of lockdown plus face covering mandates, for what percentage of the days since the beginning of the epidemic the country has been under the highest possible level of lockdown of that type. I then add up these nine percentages and divide by 9, to give an overall average of the percentage of time spent under full lockdowns.

It’s worth noting that in large, subdivided countries, while the average lockdown stringency will tend to overstate the general level of lockdown, the average time in full lockdowns will tend to understate it, because a full lockdown must be at the national level.


Several Nordic countries (Iceland, Finland, Denmark, Norway) are among the best performing countries against the virus, yet have relatively low average lockdown stringencies. It’s notable that Estonia, Latvia and Lithuania, which share some cultural similarities with the Nordics, have also locked down relatively lightly. Belarus too, which is in the same geographical area, but not considered a Nordic country.

Among the European countries which have performed worst in terms of deaths per million, Bosnia and Herzegovina has (had as at August 20th) 2,963 deaths per million from 63,700 cases per million at an average stringency of 50%. And Hungary has 3,110 deaths per million from 83,900 cases per million at an average stringency of 58%. Italy, which has locked down hardest in terms of average stringency, has 2,128 deaths per million from 74,000 cases per million at an average stringency of 69%. I don’t see any clear relation between lockdowns and deaths per million there.

That’s an amazing difference between top and bottom. The Irish have been fully locked down under, on average, three out of nine types of lockdowns over the entire duration of the epidemic; while the Finns have suffered only about a sixth of that. Yet Ireland’s deaths per million from COVID is 1,028 from 67,500 cases per million at an average stringency of 68%, while Finland has only 183 from 21,800 at an average stringency of 45%.

Heavy lockdowns may “work” in the short term by keeping new cases down, but that isn’t necessarily reflected in the deaths statistics, the ones which matter most. Heavy, long lockdowns, apart from the economic and psychological damage they cause, are also likely to leave a country further from herd immunity than more lightly locked down countries nearby.


Those lockdowns in Honduras and Venezuela look horrific! Nicaragua we can’t trust, but Dominica second from the bottom has no deaths at all from 13,200 cases per million at an average stringency of 41%; though it does have the advantage of being a small island country. Honduras, on the other hand, has 868 deaths per million from 33,000 cases per million, at an average stringency of 85%. El Salvador, not so far geographically from Honduras and with a comparable population (6.5 million versus 10 million), has 435 deaths per million from 14,300 cases per million at an average stringency of 59%. El Salvador has controlled cases better than Honduras, and deaths far better, with a far lower average lockdown stringency!

An even more extreme example is provided by Argentina. It has 2,435 deaths per million – 11th highest in the world – from 113,400 cases per million, at an average stringency of 77% –7th highest in the world. In Argentina at any rate, lockdowns haven’t saved lives.

Jamaica – 458 deaths per million from 20,400 cases per million at an average stringency of 72% - may also be worth another look, as it has the highest stringency in its group.

The Hondurans’ percentage of time spent under full lockdowns is almost twice as bad as the Irish!

Middle East and North Africa

These two lists are very similar. Many of the countries towards the bottom of the lists – Yemen, Afghanistan, Syria, perhaps Sudan – have other things on their minds than COVID.

Probably the lowest locked down country here, in whose figures I could feel confidence, is the UAE, which has 204 deaths per million from 71,500 cases per million at an average lockdown of 56%. Contrast this with Libya at the top of the list, with 585 deaths per million from 42,700 cases per million at an average lockdown of 81%. Or Palestine in second place, with 711 deaths per million from 63,600 cases per million at average lockdown of 80%. The UAE has done way better on deaths than either of these.

Sub-Saharan Africa

Eritrea is a high outlier in both lists, and Benin, Tanzania and Burundi are low outliers. Eritrea has 10 deaths per million from 1,900 cases per million at an average lockdown of 82%. They have controlled the cases well, but at what cost in the longer term? Benin has 10 deaths per million from 840 cases per million at an average lockdown of 42%. Tanzania has 1 death per million from 23 cases per million at an average lockdown of 21%. Excellent figures, if I could believe them! Burundi has 3 deaths per million from 900 cases per million with an average stringency of 15%. Whereas neighbouring Rwanda has 78 deaths per million from 6,300 cases per million with an average stringency of 69%. This may perhaps be due to greater economic activity in Rwanda as compared to Burundi.

Rest of Asia

Bangladesh has 152 deaths per million from 8,800 cases per million at an average stringency of 79%. Myanmar has 259 deaths per million from 6,800 cases per million at average stringency of 76%.

At the bottom of the table, Japan has 123 deaths per million from 10,000 cases per million at an average stringency of 40%, and Taiwan 35 deaths per million from just 670 cases per million at average stringency 32%. The Japanese have managed the virus better than the Bangladeshis or the Burmese, at around half the lockdown stringency!

The Azerbaijanis seem keen on full lockdowns! They have 518 deaths per million from 38,000 cases per million at an average stringency of 75%. In contrast, the exemplary Singaporeans have 8 deaths per million from 11,400 cases per million at an average stringency of 51%.

Australasia and Oceania

These two lists are very different! New Zealand and Vanuatu don’t lock down much, but when they do, they prefer to use full lockdowns. Australia, Fiji and PNG are at the top of the stringencies list, and near the bottom in full lockdowns. Australia has quite a high average stringency at 58%; New Zealand is much lower at 34%.

World Top and Bottom 20

Here are the world top and bottom 20 in average lockdown stringency:

Based on this, I’ll add Honduras, Venezuela, Eritrea, Japan, Belarus and Burundi to my list of countries to be followed up on - as well as Jamaica.

It’s amazing that China shows up in 19th place. If they have the virus all but licked, as their cases and deaths figures would suggest, then why do they need such a high level of lockdowns over such a long period? The bullshit meter goes off, again.

And here are the world top and bottom 20 in average time spent under full lockdowns:

The appearance of Japan, Finland and Iceland – all of whom have done well against the virus – in this bottom 20 is noteworthy.


Here are scatterplots of cases per million, deaths per million and average excess mortality against average lockdown stringency and average percentage of time in full lockdowns:

All six correlations are positive. High levels of cases or deaths lead to decisions to lock down! And the excess mortality flows on from that.

To sum up

·       Heavy lockdowns have not necessarily succeeded in keeping down deaths per million. As Italy or Ireland versus Finland, or Azerbaijan versus Singapore, show.

·       Lockdowns have not always succeeded in keeping down even cases per million. As shown by El Salvador versus Honduras, or Japan versus Bangladesh.

·       One particular country, Argentina, is 11th highest in the world (out of 191 countries) in deaths per million, and 7th highest in average lockdown stringency. The Argentinian strategy against the virus appears to have failed.

·       The Nordic countries, Japan and Taiwan seem to have done a decent job in controlling the virus, yet with relatively light levels of lockdown.

No comments: