It’s satisfying to see the economics profession come around on some things (regression discontinuity analysis and so-called risk aversion)

Jonathan Falk points to this post by Nicholas Decker and writes:

I thought you’d be interested in a couple of things in this. First, the regression discontinuity pictures and Decker’s parenthetical warning: “(The econometrics literature is emphatic on this – do not use higher order polynomials in a regression discontinuity. If you do, then you can get very large differences in apparent outcomes due to changes far away from the boundary – imagine if someone on the border of 44 flips the quadratic upside down, completely reversing your result!).” Lessons come through!

Second, his statement on risk theory — “As many people have noted (but see Chetty (2006) for a particularly elegant demonstration) estimates of risk aversion in different settings get extremely different answers. In particular, the coefficient of relative risk aversion comes from people having declining marginal utility of consumption. If you ask people in lab experiments about gambles, you’ll find that they’re really scared of risk, but if you observe people’s labor supply response to income changes, you’ll find things which imply they’re really risk tolerant.” I’ve never entirely understood your attitude on this other than your objection to particular functional forms, but the unease with which the profession has come around a bit towards your view should be welcome.

This is indeed satisfying.

And for those who are coming into this story in the middle, here are some references.

1. Problems with flexible adjustments in regression discontinuity:

• [2025] Stable adjustments in regression discontinuity. {\em Journal of Comments and Replications in Economics} {\bf 4}. (Andrew Gelman and Guido Imbens)
• [2019] Why high-order polynomials should not be used in regression discontinuity designs. {\em Journal of Business and Economic Statistics} {\bf 37}, 447–456. (Andrew Gelman and Guido Imbens)

2. Problems with naive ideas of risk aversion:

• Lots and lots of posts over the years.  I hardly know where to start!  Here’s one example.

I won’t try to take too much of the credit for this progress in economics. I’m just glad to see it happening.