Actually my result is general, even without constant elastcitity: https://www.overcomingbias....

Expand full comment

Sure the model can apply, and my personal guess is that the US spends too much, but I don't have elasticity estimates on this topic to use to apply this model.

Expand full comment

Do you think this model would apply to military expenditure vs. the cost of war? Do you think military expenditure is optimal/efficient?

A recent book "Close the Pentagon" estimates that in 2014 the cost of war to be ~$120bn, but global military expenditure ~$2.4trn (based on this paper, but there's an updated methodology here)

Expand full comment

Under that assumption, the constant elasticity model seems more reasonable.

I was thinking in terms of Plan A working. A successful implementation of Plan A is one of the cases where a model that's "usually decent" isn't.

Expand full comment

Yes, elimination, containment, establishing some level of control (eg. Sweden), or even just bying some time to increase treatment capacity or capability, or learning what activities are the most problematic are justifications for lockdowns. There's no point if it's just going to be let go in 2 or 3 months anyway. In fact you'd incur all the economic costs and all the death-injury-panic costs too.

The US is the most capable country in the world and can surely do anything any other country can if it is motivated to. Most states started in time if not all.

Expand full comment

By "peak" just meant "reference act", and have changed my comment that way.

In the US I think it quite unlikely that we can contain this for years until a vaccine. That greatly reduces the potential health gains from the economic pain we will suffer over the coming months.

Expand full comment

I like the models, which help us understand the bigger and more-complicated picture. I am glad anyone is doing a cost-benefit analysis on C19 and lockdowns,

OT but in the ballpark:

Evidently there are something like 70 BSL-4 virology labs globally, in which virus traits, such as infectiousness or lethality, are sometimes magnified through gene-splicing or other measures. This is called "gain of function" research.

Evidently, a researcher even take bat virus samples, and keep therm around human cells, usually to no effect. But sometimes, and after enough exposure, those rare viruses in the bat sample that infect human cells are successful and replicate, and then you have a bat virus that infects humans. Looks natural too, not the result of gene-splicing. This is called "passaging."

This research was going on in Wuhan, but also other labs everywhere with all kinds of viruses. Lab leaks are public record, and have happened.

As a minimum---

Surely, every virology lab anywhere on the planet should operate under complete and total transparency, open to unannounced international investigation at any time, and with complete, unhindered and open dialogue among international and social media, academic literature, and lab employees. 24/7 webcams might be a good idea.

Nations that do not comply with such obviously basic and minimum necessary requirements should lose Most Favored Nation trading status and be subject to incrementally rising trade sanctions.

We now know the huge economic costs of even a mild virus as global pandemic.

Imagine if the next time is worse….

Expand full comment

This is a different name for the UK's initial plan, which was then poorly labelled as herd immunity. A number of countries planned this early - The Netherlands, and Sweden for example. It was something that had been planned theoretically for decades.

Unfortunately the UK had prepared poorly and had insufficient ability to proptect vulnerable people, nor did they have the testing in place to know when to start isolating the vulnerable. The Netherlands also not so good.

It's difficult in theory to know prior to a disease who is going to be vulnerable and who isn't and make preparations. It's also difficult to know prior which of the many options are going to work best. With any option you need to be very alert and responsive to early information.

Sweden has done the best at realising this strategy.

The most successful countries have not followed this strategy at all but are following one of testing tracing and isolating the sick. Which is best now for each country depends on their circumstances - how many infected, how many immune, how many vulnerable.. Is it easier to isolate the sick or the vulnerable?

It also depends on how long immunity lasts, which at this stage isn't know. For most other coronaviruses its 1-2 years which makes the concept of herd immunity useless unless it can eliminate the virus in a region which is then isolated by border restrictions.

Expand full comment

But decisions of how to respond can't be based only on the peak. The end result is important and even changes in the short term too.

The extreme example in this case is if a big effort now eliminated the disease so that only border checks and some monitoring is needed in the future.

Another less speculative example is if numbers are reduced by initial measures to the point where detection isn't overloaded and delayred, and cases can be strictly isolated without relying on self isolation with only 80% compliance. Taiwan and South Korea and probably China are examples of this working.

Then there are improvements in treatments over time, and improvements in understanding which restrictions work best and which activities are highest risk so that future measures can be more cost effective.

It's not a tradeoff between economic and death/injury costs in the next few months, nor can any peak analysis be relevant for any sustained time.

Expand full comment

As I said in the post, reduced form models are usually decent around some reference action

Expand full comment

I don't think the constant elasticity model is a reasonable approximation here--especially with respect to extending a lockdown arbitrarily. The

If you put out a fire when it's small, you could well suffer more water damage than fire damage. Total damage could still be far less than that from an unchecked fire. And yet, dumping additional water would have zero impact on fire damage.

Expand full comment

No, less elastic, and so higher on that graph.

Expand full comment

The debate seems to be gradually shifting away from lockdown and toward the "stratify-and-shield" approach:

Have a look...https://www.medrxiv.org/content/10....

Expand full comment

58% estimate that we will be...above that .05% elasticity line....Meaning 58% believe the relationship to be more elastic, right? (as opposed to higher physically on the referenced chart).Suffer us idiots, doc. :D

Expand full comment