A thread on #Covid19 & the @QandA debate.

Cost-benefit analysis (CBA) is a useful tool for thinking about trade-offs b/w policy options.

But assumptions matter hugely when we try to monetize non-monetary things like, you know, mortality risk.

cc: @benefitcost
Some are using CBA language to suggest that lockdowns are bad.

I have seen NO analyses that satisfy all of:
1) consistent with commonly accepted practice
2) understand micro-economic/decision-theory foundations of CBA
3) conclude that the cost of lockdowns outweigh the benefits
Unfortunately, I have seen too many of misleading statements.

Some specific examples below. I am not doing this to badmouth specific people---I just worry that it is too easy to use flawed CBA arguments to draw disastrous conclusions.
One step back before I get into the gory details:

Economics does not tell us how to compare utility b/w people. Just like it does not tell us how to choose b/w Pareto-efficient allocations.

We can place weights on utility-changes for different individuals based on who they are.
We can *choose* to place more weight on improved health outcomes for the sick, or to weigh utility gains to the most disadvantaged more.

Or we can choose to down-weight those with shorter life expectancy.

This choice reveals more about an analyst than about economics or CBA.
Ok, so here is a paragraph that I read as follows: "we overstate the value of lives lost if we don't account for the age distribution of the high-risk population."

This paragraph has a high problem-to-words ratio. (QALY)ty stuff.

Problem 1: it gives the (false) impression that assigning lower monetary values to the elderly is the "right" analysis.

There is no right criterion for determining the social value of policy changes that lead to gains for some and losses for others. The weights are choices.
Problem 2: the author insinuates that placing a lower value on older lives is OK "because the UK government does it."

I suspect the UK gvt uses QALYs to compare different health technologies and NOT to introduce "senior discounts" on the value of elderly life in policy-making.
Also, nobody "cares" about QALYs per se (see below), you could plausibly care about "monetized QALYs," i.e., a QALY times some shadow price for life.

Basing society's willingness to pay for different populations' lives using monetized QALYs is VERY controversial, for good reason
Why don't we care about QALYs? Well, for starters, QALYs are literally numbers between 0 and 1.

Specifically, QALYs are the output of a function that maps two things (life expectancy & health status) --> an index.
Ok, we're off to Geeklandia:

For QALYs to be consistent with basic microeconomics, we have to make some pretty restrictive assumptions. For example, how much you value one additional year of life needs to be independent of your health status (assume health is constant)
This implies that people should be indifferent between these lotteries:

A 50-50 chance of (10 years pain free) or (1 year with X amount of pain)
A 50-50 chance of (10 years w/ X pain) or (1 year with no pain)

Scientific twitter poll to the rescue: https://twitter.com/etjernst/status/1252402282697646083
Oh, and it is really hard to make QALYs consistent with discounting. So that's kind of a bummer, since non-zero discount rates are pretty universally accepted. https://twitter.com/etjernst/status/1252455788796239874
I don't hate QALYs, for the record. They are useful if we want to find the lowest cost-per-QALY across different health policies.

They are NO good for determining whether the health gains from a particular policy justify the costs.
An alternative measure if we must place a monetary value on reductions in mortality risk is the Value of a Statistical Life (VSL). We measure people's willingness to pay for a tiny reduction in risk of illness or death and aggregate.
So if a thousand people all benefit from a tiny reduced risk, the sum of their WTP is the VSL.

Here, the author is using VSL the other way around. It might seem like semantics, but while we can monetize a reduction in mortality risk, we CAN'T buy a life.

You can follow @etjernst.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: