In this blog, I illustrate the LP-DiD estimator by Dube, Girardi, and Jorda (2023). I will leverage my previous blogs on Heterogeneous Difference-in-Differences with Stata:
This new estimator aims to solve the problem of negative weights of the Two-Way Fixed Effects (TWFE) and introduce a clean control condition.
Let use the data in Callaway and Sant’Anna (2021), we study the increasing in minimum wage and young employment. The outcome is the county-level employment for young workers. From their article: “We focus on county level teen employment in states whose minimum wage was equal to the federal minimum wage at the beginning of the period. Some of these states increased their minimum wage over this period – these become treated groups.” The increase started in 2004 according to the dataset.
Let us practice with Stata, it appears that avoiding composition effects is important for the parallel trend assumption before the treatment. Here, the treatment is absorbing, meaning that the treated counties stay treated until the end of the sample period:
// Run baseline version of LP-DiD (variance-weighted with strictly positive weights), with time-window of 1 periods to define clean controls
lpdid lemp, time(year) unit(countyreal) treat(treat) pre(3) post(3)
// Use only non-yet treated as controls
lpdid lemp, time(year) unit(countyreal) treat(treat) pre(3) post(3) nevertreated
// Use only non-yet treated as controls and avoid composition effects
lpdid lemp, time(year) unit(countyreal) treat(treat) pre(3) post(3) nevertreated nocomp
It appears that increasing minimum wage above the federal minimum wage reduces young employment. However, I do not include control variables, so keep in mind that the conclusion is preliminary and tentative.
References
Callaway, B., & Sant’Anna, P. H. (2021). Difference-in-differences with multiple time periods. Journal of econometrics, 225(2), 200-230.
Dube, A., Girardi, D., Jorda, O., & Taylor, A. M. (2023). A local projections approach to difference-in-differences event studies (No. w31184). National Bureau of Economic Research.