As I am now handling papers at AEJ:Policy again, I want to encourage authors of papers that identify partial equilibrium effects to consider (in a rigorous manner) how they relate to policy effects or general equilibrium effects.
Comments
Log in with your Bluesky account to leave a comment
There are many papers in market design, which use theory as a formal tool rather than empirics, although the main point is policy rather than advancing theory. In the past AEJ:Policy forwarded such papers to AEJ:Micro. It will be great if the journal develops expertise to evaluate such submissions.
For example, a clean estimate of the effect of attending a charter school based on a lottery or RD is important but is not the same as (a) the effect for all charter schools or (b) the policy effect of allowing for charter schools in the market at all.
Similarly, a clean effect of how childcare availability affects the labor supply of individual mothers is not the same as the policy effect of expanding access to childcare.
Also, the clean RD effect of starting school a year later based on birthday cutoffs is not the same as the policy effect of increasing the school starting age.
These are well understood by authors familiar with these topics. However, the policy discussion could benefit from clearly highlighting these differences, explaining why they matter, and demonstrating how they might help reconcile seemingly different results across papers. It's also good science!
Fair question. I believe there are many ways to make progress in this direction. The first step is to acknowledge it. I often see papers (and reviewer comments) that treat different partial effects as the same or fail to distinguish between policy effects and partial effects. That’s not good.
Additionally, we can start to take seriously the idea that heterogeneity is real and attempt (albeit imperfectly) to estimate and understand it. Claire Mackevicius and I attempt this in our work on school spending.
At the very least, I propose that policy-oriented papers (a) include a thorough discussion of how their estimate relates to others in the literature (after accounting for sampling variability) and (b) provide a serious discussion of exactly how their effect relates to the policy effect.
Comments
What do you mean by "consider in a rigorous manner"?
Very hard to have clean identification of GE effects.
Do you mean a discussion? A model? An example of papers that do it well?
One of my favorite papers to provide clarity of though in this way is this one....
https://www.jstor.org/stable/3590137?seq=1
https://www.aeaweb.org/articles?id=10.1257/app.20220279
Thanks for unpacking it for me.