jfe-empirical-design

• SKILL.mdcore · 6,075 bytes

# Empirical Design & Inference (jfe-empirical-design)

## When to trigger

- You sort on a characteristic but have not justified the variable, the breakpoints, or weighting
- You are choosing between Fama–MacBeth, panel regression, and GMM and unsure how to report
- Your standard errors are unclustered, or clustered on one dimension when two are needed
- You have an asset-pricing predictor but no out-of-sample or multiple-testing treatment
- Variable definitions are ad hoc and would not replicate

## The JFE design bar

JFE is known for nuts-and-bolts methodological rigor. Referees scrutinize measurement, estimator choice, standard errors, and inference discipline line by line. The goal is a design that a skeptical expert cannot dismantle on technical grounds. This is the journal that published **Fama & French (1993), "Common risk factors in the returns on stocks and bonds"** (the three-factor model), **Fama & French (2015), "A five-factor asset pricing model,"** and **Banz (1981), the size effect** — so an asset-pricing referee benchmarks your construction against that lineage directly. The best capital-markets paper each year wins JFE's **Fama-DFA Prize**; write to that standard. Code and non-proprietary data are mandatory at acceptance (Mendeley Data; see jfe-submission), so build a reproducible pipeline from the start.