Wednesday 2 December 2020

SAGE's tier trick

As of today, 99 per cent of the English population will live under tough new Tier 2 or Tier 3 rules. Mixing indoors with people from outside your household will remain illegal in both of these tiers. In Tier 2, you have to buy a ‘substantial meal’ if you want a drink in a pub - and leave when you’ve eaten it, according to the Prime Minister’s spokesman.  In Tier 3, which includes most of the North, all pubs, restaurant and other venues will be closed.

This amounts to carpet-bombing of the hospitality industry by the government and will lead to unprecedented bankruptcies and unemployment in the sector. The affront to civil liberties since March is unlike anything Britain has seen before, even in wartime, and the tier system looks like it will stay in place for at least four more months.

People who went into lockdown in Tier 1 only to come out of it in Tier 3 are understandably perplexed. Lockdown has been working well. On Friday, SAGE finally acknowledged that the rate of infection (R) was below 1. In fact, the number of positive tests reported each day has fallen by 40 per cent since the lockdown began on 5 November and will fall further. So why is Cornwall, which has an infection rate of 45 per 100,000, in Tier 1 when the Cotswolds and Mid-Suffolk, with rates of 41 and 40 per 100,000 respectively, in Tier 2?

The answer is that government scientists have constructed the evidence for the tiered system in a way that ignores the success of lockdown. For a region to be in Tier 2, they want to see (a) low rates, (b) falling rates, and (c) sufficient hospital capacity. The second two of these are a given in most of the country after nearly four weeks of lockdown, so it all depends on the rate of infection being low. A graph published by the government on Thursday suggests that they are looking for the rate to be below 100 per 100,000 people (the places to the left of the line are those where the infection rate has fallen).


Based on current figures, this should include Dorset, Huntingdonshire, Suffolk, Sussex, South Cambridgeshire and many other places. But SAGE are not using current figures. They are using figures from 19 November and comparing them to figures from 12 November. This is a big problem because the data from 19 November do not tell us what was happening on 19 November, let alone what is happening now.

People don’t typically ask for a test unless they have symptoms, and it can take up to ten days for an infected person to become symptomatic. They then have to request a test and take it. All this takes time and creates a delay between infections occurring and infections being reported. After Wales introduced its ‘firebreak’ on 23 October, for example, the number of cases rose for a week before they began to drop - and they kept dropping for a week after it ended. This is to be expected and it was the same in the Czech Republic, France, Israel and many other countries. Now that we have mass testing, we can see it more clearly than we did in the spring.

By using data from 19 November to decide which areas should go into each of the three tiers, the government is, in effect, using infection data from 12 November, only a week after lockdown began. The number of infections had fallen by 25 per cent by 19 November and will continue to fall at a similar rate until 9 December, but the government has made no attempt to account for the subsequent decline.

Any honest attempt to put regions in the appropriate tier would estimate what the infection rate will be when the lockdown ends, not what it was three weeks ago. SAGE are no strangers to predictive modelling, but on this occasion they decided to base their decision on what happened in the past.  

Contrast this with the decision to go into the lockdown on October 31. The number of new infections had been flat for a week and many of the areas of greatest concern, such as Liverpool and Manchester, were seeing a decline. The government nevertheless introduced a national lockdown off the back of a ‘reasonable worst case scenario’ which, according to Prof Neil Ferguson, assumed that the infection rate was high and the tiered system was having ‘minimal impact’. With these implausible assumptions fed into it, the computer model pointed to lockdown. What else could it do? Garbage in, garbage out.

Making predictions seems to be fine if it deprives us of our liberty, but is unthinkable when it comes to restoring it. The common denominator is an almost pathological desire by government scientists to promote lockdowns at the expense of less costly options.



No comments: