How should governments evaluate lockdowns as a policy?

Jury is out on whether lockdowns significantly decrease cases & deaths relative to other measures (e.g. social distancing, hygiene, limiting big events)

However we have enough info from different countries to say something.

First, we know a strict lockdown is not a *necessary* condition for getting cases & deaths down (see Sweden).

We also know lockdown is not a *sufficient* condition for stopping cases increasing (see Argentina).

So policymakers need to consider both costs & benefits of lockdown.

We can be certain that the costs of lockdowns are enormous, not just economic but social, psychological, educational etc.

The benefits are much more uncertain but we can think of (at least) 4 scenarios.

Scenario 1: lockdowns achieve a lower final rate of cases and, hence, deaths (relative to implementing just the other measures).

Given the enormity of the costs, policymakers would probably need a fairly big reduction in cases/deaths to justify lockdown.

(as an aside, don’t get side-tracked by an argument that avoiding even additional 1 case or death justifies any cost. No society takes such an absolutist approach to risk. e.g. we allow children to cycle to school despite knowing that some children will die as a result.)

Scenario 2: lockdowns lead to the same number of cases but spread out over a longer period, preventing health services from being overwhelmed. As a result, a large amount of deaths are averted due to lots of patients being untreated.

This was the argument for the UK lockdown based on early modelling suggesting hundreds of thousands of deaths without a lockdown.

It seems less than convincing given experience so far, e.g. given that cases were coming down before the lockdown was even implemented.

Scenario 3: lockdowns lead to a reduction in cases & deaths in the short run but not the long run as cases rise when lockdown is eased.

The only possible benefit with scenario 3 is if cases are delayed sufficiently to a time when treatment is better &/or a vaccine is available.

Given uncertainty about the latter & the certainty about lockdown costs, this is unlikely to provide a strong basis for lockdown policy.

Scenario 4: lockdowns do not lead to any significant reduction in cases or deaths even in the short run.

This would be devastating for lockdown policy as it suggests huge costs with zero benefits.

Note even if scenarios 1-3 are true, lockdowns may not be justified – it will depend on the magnitude of both costs and benefits.

That areas with some of the highest death rates also had strict lockdowns (Spain, New York) indicates that we are unlikely to find lockdowns had a very large impact on deaths, a view consistent with the recent Lancet paper https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370(20)30208-X/fulltext

It is still too early to say whether a very early lockdown (i.e. before the virus has really taken hold) might help a country achieve a big, long run reduction in deaths but recent events in Australia do not give much cause for optimism.

The impression I get is that some policymakers introduce lockdown measures in a panic, assuming they will produce benefits and paying only lip service to the costs. That is not a good recipe for success.