Three Insights from Our Econometrics Reading Group

A Two-Year Journey in Econometric Reading and Discussion

According to the experts: “applied researchers spend 20% of their time thinking about econometrics” (Millimet, 2024)¹ . To the right tail of this expected value are highly motivated staff in economics departments and econ graduate students. Individuals who keep up to date with the latest econometrics papers, working papers, blog posts, and substack posts. When you grab lunch frequently with econometric enthusiasts it’s just a matter of time before you end up organizing an econometrics reading group. It’s been almost two years since at Utrecht University School of Economics, such a reading group emerged from a chat during lunch. It seems about time to share some thoughts and reflections on the main takeaways from this experience, and hand over the baton on to the new generation of enthusiasts.

A varied diet on econometric methods

In the ‘econometrics methods diet’ Difference-in-Differences dominates the offer (Goldsmith-Pinkham, 2024). Alongside Difference-in-Differences, the adoption of empirical methods from the credibility revolution is increasingly prevalent. The reading group is not alien to these trends, as these methods feed our seasonal discussions. To date, 5 out of 26 papers we have discussed were focused on Difference-in-Differences and 6 focused on Instrumental Variables (including Shift-share). The other 15 covered a broad range of methods, from OLS estimation to machine learning. Therefore, we had a varied diet of econometric methods, and our offer has brought together researchers from different fields of study. I’m proud that the reading group has created a healthy diet of papers outside of our narrow fields.

Quality for 20% of the time

Given that I described the group as enthusiasts, it is fair to say that had the group not existed, we would have spent approximately the same 20% of our time thinking about econometrics. So, after all this time together, did we gain a broader perspective on the advances in econometrics? Did we upgrade our applied econometrics toolkit? Did we even learn something? Given that this design is plagued with self-selection, peer effects and non-random assignment of compliers and never-takers, it is unlikely to quantify or discuss an unbiased average treatment effect. I can only answer this question from a Personal Treatment Effect point of view. I believe that my potential outcome had the group not existed would have been the same 20% but with another mono-thematic focus and alone. Thus, the quality of the one-fifth I allocated to think about econometrics was upgraded thanks to the quality of questions and comments from my fellow readers.

Building bridges between theory and applied econometrics

There is a gap between theoretical and applied econometrics and the widening gap is homogenizing applied econometricians’ toolkit (Millimet, 2024; Millimet, 2024). Unless we want applied econometricians to specialize in difference-in-differences and theoretical econometricians to get more abstract, finding ways to close the gap is a pertinent activity to pursue. The reading group has built bridges between theoretical and applied econometric papers, at least from the perspective of an econometrics methods consumer. Reading the application of an econometric method can feel like something is missing. Econometric enthusiasts want to know what is behind the black box of a didregress implementation. However, reading theoretical econometrics can be daunting. Since the First Lesson on Econometrics (Siefgfried, 1970), I knew that reading econometrics papers is all about a 1+1=2 strongly supported by assumptions, definitions and proofs. The reading group allowed me to verbalize those mathematical expressions and walk through equations and Greek letters along with smart colleagues. These collaborative discussions helped us approach the gap between theoretical and applied econometrics papers without experiencing a free fall in between.

Takeaways

After two years of regularly reading with motivated applied econometricians, we got a broader perspective on the advances in econometric methods. My three takeaways are: First, econometrics is fascinating because methods are field agnostic, and the reading group has created a space to know papers outside of our narrow fields. Second, reading econometrics papers with peers has likely improved the quality of time spent thinking about econometrics. Third, the econometrics group has bridged the gap between the abstraction of theoretical econometrics and the implementation of applied econometrics. Challenges and room for improvement wait ahead for the econometrics reading group; I’m confident we have handed the baton off in a better place than we got it.

References

• Goldsmith-Pinkham, P. (2024). Tracking the Credibility Revolution across Fields. Working paper.
• Millimet, D. L. (2024, 06 25). The Great Divide. Retrieved from How the (Econometric) Sausage is Made: The Great Divide
• Millimet, D. L. (2024, 06 26). The Great Divide, Part II. Retrieved from How the (Econometric) Sausage is Made: The Great Divide II
• Siefgfried, J. (1970). A First Lesson in Econometrics. Journal of Political Economy, 1378-1379.