How Will Machine Learning Impact Economics? (Guido Imbens, Josh Angrist, Isaiah Andrews)

This episode is the most heated of the series! While Nobel laureates Josh Angrist and Guido Imbens agree on most topics, they sharply diverge on the potential of machine learning to impact economics. Host Isaiah Andrews steps in to referee the dispute, adding his own take on how machine learning might change econometrics.

Guido Imbens is optimistic about the potential of using machine learning to estimate “personalized casual effects” in large data sets. He laments that econometrics journals have been too rigid in their expectations, turning away many useful insights from machine learning.

Josh Angrist has a less rosy view. He has yet to see machine learning make an impact on the work he’s doing. Instead, he’s seen cases where it can be very misleading.

00:00 - Intro
00:18 - Potential for "personalized" causal effects
07:22 - Applications of machine learning
10:17 - Opportunities for publishing in journals
16:12 - Isaiah Andrews referees!

[Sheeshening]

This was too short, both could have easily elaborated on multiple points, these formats would best run for ~2-3h, with executive summary style shortened clip versions like this and then a third perhaps even fourth tier of ‘viral’ content for the 10min and shorts brackets. Still, thanks to MRU and of course the three.

[angusgreaves]

Would love to see a longer debate between these two incredibly intelligent researchers on this topic

[itzcjsd]

Great series, thank you so much for this content. I'm new to economics and I just found this channel, and it's all been really informative so far. And it's so fascinating to hear from these guys it's like we have a peek into this almost whole different world of researchers at the forefront of economics. I didn't understand all of the statistics terms they were throwing around, but I see it as an incentive to watch the econometrics series! Hoping there will be more like this!

[jackh]

Now this is the type of content that everyone should watch

[pjakobsen]

Angrist is right. Ignore the hype cycles.

[sahanglpi-]

We must aware that Economics is a social science, there are so many variables that cannot be quantified. Therefore, too much mathematics will bring disaster outcome in Economics. We must learn in what happened in the 1950s and 1960s that we had a huge formula for Economics (particularly in Finance) that couldn't work. 😁

[stlouisix]

Josh 'Angrist' what a surname😂

[fantasyvimmienskrrt]

The end of the economic calculation "problem" altogether.