Kuang Xu writes:

I’m a professor at the business school at Stanford working on operations research and statistics. Recently, I shared one of our new preprints with a friend who pointed out some of your blog posts that seem to be talking about some related phenomenon. In particular, our paper studies how, by using adaptive control, the states of a processing system are effected in such a way that congestion no longer “correlates” with the underlying slowdown of services.

You mentioned in the blog where you wonder if there’s some formal treatment of this phenomenon where control removes correlation in a system, and I thought you might find this to be interesting, possibly a formal example of the effect you were thinking about.

We’ve been wondering if there are other similar, concrete examples in the policy realm that resemble this.

My reply: I’m not sure. On one hand, the difficulty of causal inference with observational data is well known—it’s a strong theme of all presentations of causal inference—but it seems that most of the concerns come with selection rather than feedback.

Xu responds:

We tried to explore this connection to a small degree in the lit review – there’s some similarity to how people use inverse [estimated] probability weighting to debias selection, but these are generally one-time interventions so not so much of a feedback loop. Like you wrote in that blog post, something like monetary policy is more like a feedback loop, but it’s hard to isolate such effects in these complex systems.

As I wrote in my earlier post on the topic, I’m pretty sure there was tons of work back in the 1940s-1960s in this area of feedback in control systems. I can just picture a bunch of guys in crewcuts wearing short-sleeved button-up shirts with pocket protectors working on these problems. For some reason, though, I haven’t hear much about any of this nowadays within statistics or econometrics. Seems like there’s room for some unification, along with some communication so that the rest of us can make use of whatever has been doing in this area already.