1. Paul Romer wrote,

The simulated data here contrast policies that isolate people who test positive using four different assumptions about the quality of the test. Even a very bad test cuts the fraction of the population who are ultimately infected almost in half. And when I say bad, I mean bad – an 80% false negative rate

Pointer from commenter John Alcorn.

For a test-and-quarantine policy to be useful, you don’t have to pull every infected person out of circulation. Think of it as a race between how many infected people you pull out of circulation and how many people get infected by the folks who your test fails to catch. You come out ahead compared to doing nothing.

But a test can be bad the other way, easily producing one false positive for every true positive. You could easily end up quarantining one healthy person for every sick person. Of course, what we are doing now is at least as arbitrary.

Tyler Cowen points to a paper with a model (what else?) that supports doing testing and confining even when the tests are bad.

I’m sure it works well in the model. In the real world, I can think of a number of difficulties with execution. Show me a project management chart that includes all the steps needed before you can even start. Then with flawed tests, it takes much longer to get the benefits, and more costs are imposed on the false positives.

2. Henrik Salje and others write,

As of 14 April 2020, there had been 71,903 incident hospitalizations due to SARS-CoV-2 reported in France and 10,129 deaths in hospitals, with the east of the country and the capital, Paris, particularly affected. The mean age of hospitalized patients was 68y and the mean age of the deceased was 79y with 50.0% of hospitalizations occurring in individuals >70y and 81.6% of deaths within that age bracket; 56.2% of hospitalizations and 60.3% of deaths were male

Another Alcorn pointer. They try to get beyond numerator analysis and estimate the infection fatality rates for different demographic groups. But their methods struck me as sketchy, so I am just quoting the raw data.

3. Frances MK Williams and others write,

Here we report that 50% of the variance of ‘predicted covid-19’ phenotype is due to genetic factors. The current prevalence of ‘predicted covid-19’ is 2.9% of the population. Symptoms related to immune activation such as fever, delirium and fatigue have a heritability >35%. The symptom of anosmia, that we previously reported to be an important predictive symptom of covid-19, was also heritable at 48%. Symptomatic infection with SARS-CoV-2, rather than representing a purely stochastic event, is under host genetic influence to some extent and may reflect inter-individual variation in the host immune response. Viral infections typically lead to T cell activation with IL-1, IL-6 and TNF-α release causing flu-like symptoms such as fever. The genetic basis of this variability in response will provide important clues for therapeutics and lead to identification of groups at high risk of death, which is associated with a cytokine storm at 1-2 weeks after symptom onset

They use a twin-study method to estimate heritability. Another pointer from John Alcorn.

4. Veronique de Rugy and me on the credit line idea.

Roughly 99.9 percent of American firms, or 30 million, fit the definition of small business used by the Small Business Administration (SBA), and together these firms employ roughly 65 percent of American workers. The devastation of the small-business sector could therefore be disastrous for American families.

The stock market is relatively placid, and meanwhile a whole way of life seems about to go under for many people. That might not be sustainable.

I readily grant that if left to their own devices many individuals will make sub-optimal decisions, but primarily costing themselves and not others. I also believe that if individuals were left to their own devices we would not see anything close to what “re-opening the economy” sounds like. But giving decision-making power to President Trump and the various governors seems more obviously right to most other people than it does to me.

5. A data visualization, by state, based on the source I use to calculate 3DDRR. Pointer from Russ Roberts.

6. Olivier Blanchard writes,

a high inflation scenario requires the combination of three ingredients, each of which has a low probability of occurring in advanced economies. Put your own probabilities and multiply them: The resulting probability is very small. I asked some of my colleagues for their probabilities, and the product always came below 3 percent.

His high inflation scenario is one in which our government can no longer pay the bills except by printing money. That is actually a hyperinflation scenario, as I imagine Blanchard would agree. Even at a probability of less than 3 percent, Anti-fragile Arnold does not want to take those chances.