baconplus

An R package (in development) that computes difference-in-differences decomposition weights:

1. The binary Goodman-Bacon (2021) decomposition for TWFE under staggered timing.
2. The continuous-treatment TWFE decomposition from Callaway, Goodman-Bacon and Sant'Anna ("CBS") v4 (2025), Table 1.

Weights explorer

A standalone, browser-based tool for the CBS level weights lives at:

https://scunning1975.github.io/baconplus/

Pick a dose l and a kernel bandwidth. The app shows the six ingredients, plugs them into

w_lev(l) = (l - E[D]) * f_D(l) / Var(D)

and plots the full weight curve alongside the dose distribution. The scaled-level, causal-response, and scaled 2x2 tabs are placeholders — those decompositions will light up in future updates.

The default dose distribution is a simulated Lu & Yu (2015)-style setup: 155 Chinese industries, roughly 30% at D = 0, right-skewed positive doses up to 0.60. The benchmark hand calculation from the companion teaching deck reproduces inside the app: w_lev(0.10) = -6.65 at bw = 0.025.

Status

Early scaffolding. The R functions themselves haven't landed yet — this repo currently hosts the explorer while the package is being built.