Working on a new R package. Looking for applications that do two things: 1. DID with y to show an effect 2. DID with X to show a possible “mechanism” of why treatment changed y This could be a published paper or a paper you're working on (I might help make "fancier" !!)

As a warming planet delivers more wildfires, hurricanes and other threats, home insurers are not renewing policies. nytimes.com/interactive/20…

These and other surprises, often in glorious Base R, in our year end roundup, out today nytimes.com/interactive/20…

PSA! My brilliant colleague, Brigham Frandsen, is teaching an online course on machine learning & heterogeneous effects. It covers predicting individual-level causal effects using Random Causal Forests and implementation in python, R, and Stata. Link: bit.ly/3CUjMto

github.com/kylebutts/vsco… If you are using vscode and stata, you should try out my extension. It uses interactive window which let's you write in a `.do` file but get a notebook type experience.

Fall on the Hill loading…

Leave Ralphie out of this 😤

I suspect most of the post tenure “dip” is just (1) prioritizing effort to papers close to being published around tenure review and (2) not starting new work just prior to tenure. Publications take time

Next week, I'm teaching another round of my online Mixtape course on BLP-style demand estimation! The sessions are hands-on and last time we had a ton of great questions. Sign up here: mixtapesessions.io/session/demand…

Apparently he can giveth but can not taketh… In short: rdlocrand is not a secret nuke that will destroy all RDD applications. It’s a different estimator whose key assumption is not very likely in almost any RDD setup

A special helmet for the final race of an unforgettable first season with Atlassian Williams Racing (with a little extra sparkle ✨)