Macroeconometrics — the science of hubris

10 Feb, 2020 at 12:36 | Posted in Statistics & Econometrics | 2 Comments

When a macroeconometrician uses regression, he or she is implicitly saying, in effect, that the third quarter of 2007 is the same as the first quarter of 1988, once all factors that might be different between those two quarters are controlled for. The idea is that the economist is conducting an intertemporal quasi-experiment. But because there is only one economic history with which to work, there is a lack of experimental control. The ‘‘natural experiment’’ and related techniques that have enhanced the credibility of econometric studies of microeconomic issues are not available to the macroeconomist.

hubIf macroeconometric models are fabrications, then where does that leave us? Imagine that two hundred years ago, you questioned the use of bleeding by physicians. Many doctors might share your concerns, particularly in private. But then the doctors would say, ‘‘Well, what can we do? We can’t just stand by and appear helpless when the patient is so ill. Unless you can show us something better, we will just keep doing what we are doing.’’ And so it is with macroeconometric models.

Like the Swiss hikers lost in the Alps, policy makers are in need of a map. The macroeconometric models purport to represent a map, but it is unlikely that their map refers to the Alps. If the policy makers knew that it might be the wrong map, they might not use it with such confidence.

Arnold Kling

2 Comments

  1. Well there are many. Elevated, pompous, types with Lars, “nice names, I think you should continue to prosecute them! 🙂

  2. “then the doctors would say, ‘‘Well, what can we do? We can’t just stand by and appear helpless when the patient is so ill.”
    .
    First, do no harm. Did bleeding do harm? Then do nothing, but listen to people diagnosing themselves.


Sorry, the comment form is closed at this time.

Blog at WordPress.com.
Entries and comments feeds.