Monday, March 9, 2020

Decisions, Outcomes, and Correlations

Here’s a though experiment. You’re driving.You’re heading for a road hazard, like a pothole, that might damage your car. You can either swerve and miss the pothole, or don’t swerve and hit it. Suppose many drivers do the same experiment.

If we collected the data and analyzed it afterwards, this point would be obvious: nothing happened to those who swerved.

A naive interpretation of that would be that swerving served no purpose. Could this be true? Probably not.

************************************************

Politicians and bureaucrats face much the same problem. In the wake of the 2008-9 financial crisis. there were large outcries in the public that those officials didn’t do anything. Is it possible they did a lot, but it wasn’t enough?

As a general rule, completely effective policies at reducing risk look like they’ve accomplished nothing if the risk doesn’t arrive.

Consider this argument in the context of defending against COVID-19. Singapore looks like it is doing less than South Korea, and in terms of effort and money it is, but it’s very likely this is because they started early and acted aggressively. China appears to have been exceptionally busy in combating COVID-19, but we now know this reflected dawdling, information suppression, and outright incompetence over the first 7 weeks of the outbreak.

No comments:

Post a Comment