did_multiplegt
Library of Estimators in Difference-in-Difference (DID) designs with multiple groups and periods.
Setup
Stata
ssc install did_multiplegt, replace
Syntax
Stata
did_multiplegt (mode) Y G T D [if] [in] [, options]
R
install.packages("DIDmultiplegt", force = TRUE)
Description
did_multiplegt wraps in a single command all the estimators from de Chaisemartin and D'Haultfoeuille. Depending on the {cmd:mode} argument, this command can be used to call the following estimators.
- did_multiplegt_dyn. In dyn mode, the command computes heterogeneity-robust event-study DID estimators introduced in de Chaisemartin and D'Haultfoeuille (2024a). Like other recently proposed DID estimators (csdid, didimputation, ...), these estimators can be used with a binary and staggered (absorbing) treatment. But unlike those other estimators, these estimators can also be used with a non-binary (discrete or continuous) and non-absorbing treatment that can increase or decrease multiple times. These estimators can also be used when lagged treatments affect the outcome.
- did_multiplegt_stat In stat mode, the command computes heterogeneity-robust DID estimators introduced in de Chaisemartin and D'Haultfoeuille (2020) and de Chaisemartin et al. (2022). These estimators can be used with a non-binary (discrete or continuous) and non-absorbing treatment. However, they assume that past treatments do not affect the current outcome. Finally, these estimators can be used to compute IV-DID estimators, relying on a parallel-trends assumption with respect to an instrumental variable rather than the treatment.
- did_had. In had mode, the command computes the DID estimator introduced in de Chaisemartin and D'Haultfoeuille (2024b). This mode estimates the effect of a treatment on an outcome in a heterogeneous adoption design (HAD) with no stayers but some quasi stayers.
- did_multiplegt_old. In old mode, the command computes the DID estimators introduced in de Chaisemartin and D'Haultfoeuille (2020). This mode corresponds to the old version of the did_multiplegt command. Specifically, it can be used to estimate DID_M, i.e. the average across t and d of the treatment effects of groups that have treatment d at t-1 and change their treatment at t, using groups that have treatment d at t-1 and t as controls. This mode could also be used to compute event-study estimates, but we strongly suggest to use the dyn mode, since it is way faster and includes comprehensive estimation and post-estimation support.
did_multiplegt updates automatically all the packages above (on average) every 100 runs of the command. Self-updates can be stopped by specifying the command with the no_updates option.
Arguments
- mode is the command selector and can be only be {cmd:dyn}, {cmd:had} or {cmd:old}.
- Y is the outcome variable.
- G is the group variable.
- T is the time period variable.
- D is the treatment variable.
- options is a pass-through and can include all the options of the command called with mode. It can include the no_updates option, which will apply only for did_multiplegt and will not be passed onto the mode options.
Example: Estimating the effect of union membership on wages
Loading the worker-year level data from Vella and Verbeek (1998):
bcuse wagepan, clear
Computing DID_M from de Chaisemartin and D'Haultfoeuille (2020):
did_multiplegt (old) lwage nr year union, breps(100) cluster(nr)
did_multiplegt (stat) lwage nr year union, exact_match
Computing 5 dynamic effects and 2 placebos using DID_l from de Chaisemartin and D'Haultfoeuille (2024a):
did_multiplegt (dyn) lwage nr year union, effects(5) placebo(2) graph_off
Authors
- Clément de Chaisemartin, Economics Department, Sciences Po, France.
- Diego Ciccia, Sciences Po, France.
- Xavier D'Haultfoeuille, CREST-ENSAE, France.
- Felix Knau, Sciences Po, France.
- Felix Pasquier, CREST-ENSAE, France.
- Mélitine Malézieux, Stockholm School of Economics, Sweden.
- Doulo Sow, Sciences Po, France.
- Gonzalo Vazquez-Bare, UCSB, USA.
References
de Chaisemartin, C and D'Haultfoeuille, X (2020). American Economic Review, vol. 110, no. 9. Two-Way Fixed Effects Estimators with Heterogeneous Treatment Effects
de Chaisemartin, C, D'Haultfoeuille, X, Pasquier, F, Vazquez‐Bare, G (2022). Difference-in-Differences for Continuous Treatments and Instruments with Stayers.
de Chaisemartin, C and D'Haultfoeuille, X (2024a). Review of Economics and Statistics, 1-45. Difference-in-Differences Estimators of Intertemporal Treatment Effects
de Chaisemartin, C and D'Haultfoeuille, X (2024b). Two-way Fixed Effects and Differences-in-Differences Estimators in Heterogeneous Adoption Designs
Vella, F. and Verbeek, M. 1998. Journal of Applied Econometrics 13(2), 163–183. Whose wages do unions raise? a dynamic model of unionism and wage rate determination for young men