For how long is lockdown politically sustainable -- version 2

Introduction

At present there appears to be strong support for lockdown among the general population in the UK and around the world. However, this is not something that can be taken as given. As the economic costs of lockdown increase, public support could waver.

NOTHING IN THIS ANALYSIS IS A REASON TO DEPART FROM GOVERNMENT GUIDANCE. I am trying to understand how the situation could unfold and not recommending a particular type of policy.

At the time of my first piece on this subject the effect of lockdown on GDP was essentially unknown. I therefore took an average estimate of the value of one quality-adjusted life year (QALY) saved and compared various GDP scenarios. As of this week, we have new estimates of the effect of lockdown on UK GDP from Capital Economics, the IMF and OBR. I use the OBR estimate in this analysis.

As the effect on GDP becomes somewhat clearer (albeit still highly uncertain), we can start to look at the nuances of estimating the value of one QALY. There are two ways of looking at this question. “Stated preference” studies present people with hypothetical scenarios about saving life with health interventions and ask how much they think should be paid. “Value of a statistical life” studies may use a stated preference approach but also attempt to infer what value people put on life by observing economic behaviour, such as how much extra people are paid to do risky jobs. The various studies produce a wide range of estimates for the value of a quality-adjusted life year, as shown below.

Surprisingly, the most important variable that affects the political sustainability of lockdown is the basis on which we think about the value of life, as discussed above.

Calculations

The calculations underlying this analysis are explained in my first piece. In summary, I take estimated death rates and quality-adjusted life expectancies for different age groups in the UK and use them to calculate how may QALY are saved from lockdown measures, based on the analysis of the effects of lockdown in the Imperial College paper of 16 March.

There are two differences in this second piece from the first piece:

1. I take the OBR estimate of a 13% fall in GDP in 2020 to calculate the cost of lockdown. The OBR has assumed a 3-month lockdown followed by a further 3-month period of partial restrictions. I take the current estimate of GDP by expenditure at market prices for 2019 as the starting point and simply add up the gap between the 2020 budget economic projection and the new economic projection to obtain a pound cost of the epidemic. The OBR says that 80-90% of the costs of the virus are attributable to social distancing measures; since there would be some GDP cost even without lockdown, arising from milder social distancing, I assume 70% of the GDP cost of the virus is attributable to lockdown. This gives a cost per QALY saved of ￡75,477.
2. Rather than using the mean value of a QALY from different papers, I show various estimates as reported in a review by Ryen and Svensson (2014).

Results

The following chart shows the results.

(Note that the Ryen and Svensson paper does not state explicitly which papers are “stated preference” and which are “value of a statistical life”. I have inferred this from the summary statistics.)

Discussion

I. On the basis of the model above, the economic cost of lockdown is set to exceed most estimates of the amount people SAY is worth paying to save life through medical interventions.

II. On the basis of the model above, the economic cost of lockdown is not set to exceed the amount people actually NEED TO BE PAID to accept increased mortality risk.

=> The model therefore suggests that lockdown may be politically acceptable for a much longer period than 3 months to the extent that people (i) see it as a matter of calculating their own interests and (ii) feel at risk. To the extent that people (i) think in terms of the value of public health interventions and/or (ii) do not feel at risk, even a 3 month lockdown may not be politically acceptable.

Issues

This analysis may overstate the acceptability of lockdown on a “value of a statistical life” basis because of loss aversion: if people need to be paid X to enter a risky profession, it is likely that a potential loss of less than X would cause them to remain in a risky profession.

The cost per QALY saved may be overstated because (i) less than 70% of the economic cost may arise from lockdown and (ii) in the counterfactual case of no lockdown, there might be significant costs arising from pressure on the NHS and people isolating voluntarily in order to avoid the virus.

The cost per QALY saved may be understated because (i) the economic impact may be larger than the OBR suggests – the IMF projections suggest a cost of ￡71,758 but the Capital Economics projections suggest ￡104,163; (ii) it may be that the case fatality rate is lower than assumed (see Oxford Centre for Evidence Based Medicine – latest estimate 0.72% vs. 0.9% in the Imperial paper); and (iii) it appears that people with pre-existing conditions are much more badly affected, which may mean that using average quality-adjusted life expectancy for each age bracket overstates the QALY saved.

Given the large gap between the average value found by “stated preference” and “value of a statistical life” (VSL) studies, and given that the projected economic cost per QALY saved sits between them, I would not expect any of the above to change the conclusion under “Discussion” above. Moreover, it would take a much worse economic out-turn than is being forecast to move the cost line in the chart above the VSL estimates. The most important determinant of how long lockdown is politically sustainable is therefore not variation in the factors above, but rather how people are thinking about the value of life.