Event Study Question

[Jakob Conradi]

Hi all,

At the moment I am having problems creating an Event Study. I want to look at the influence of past weather on the hiring behavior of companies.

For this I am using weather shocks. Currently I have found the Eventdd command, but this can only use binary year dummies for years, but I need the effect of a continuous variable on hiring behavior, i.e. the interaction term between the year variable and the annual rain days variables.

My Data for 1 company looks like this:

Employment	Weather	ID	WeatherLag_4	Lag4	WeatherLag_3	Lag3	WeatherLag_2	Lag2	WeatherLag_1	Lag1	WeatherLag_0	Lag0	WeatherLead_1	Lead1	WeatherLead_2	Lead2	WeatherLead_3	Lead3	WeatherLead_4	Lead4	WeatherLead_5	Lead5	year	Shock
80	70	1	50	1	60	0	45	0	56	0	70	0	60	0	40	0	80	0	200	0	40	0	1989	1993
50	60	1	60	0	45	1	56	0	70	0	60	0	40	0	80	0	200	0	40	0	80	0	1990	1993
60	40	1	45	0	56	0	70	1	60	0	40	0	80	0	200	0	40	0	80	0	60	0	1991	1993
70	80	1	56	0	70	0	60	0	40	1	80	0	200	0	40	0	80	0	60	0	70	0	1992	1993
40	200	1	70	0	60	0	40	0	80	0	200	1	40	0	80	0	60	0	70	0	56	0	1993	1993
15	40	1	60	0	40	0	80	0	200	0	40	0	80	0	60	0	70	0	56	0	76	0	1994	1993
10	80	1	40	0	80	0	200	0	40	0	80	0	60	1	70	0	56	0	76	0	84	0	1995	1993
12	60	1	80	0	200	0	40	0	80	0	60	0	70	0	56	1	76	0	84	0	65	0	1996	1993
14	70	1	200	0	40	0	80	0	60	0	70	0	56	0	76	0	84	1	65	0	43	0	1997	1993

I used the regression command:

reghdfe employment Lag_5##WeatherLag_5 Lag_4##WeatherLag_4 Lag_3##WeatherLag_3 Lag_2##WeatherLag_2 Lag_1##WeatherLag_1 Lag_0##WeatherLag_0 Lead_1##WeatherLead_1 Lead_2##WeatherLead_2 Lead_3##WeatherLead_3 Lead_4##WeatherLead_4 Lead_5##WeatherLead_5 , absorb(i. year)

where the Lags/Leads are a binary variable indicating the distance to the event and WeatherLag is a contionous variable for the weather that happened in that year. However, the results are not really realistic, where is my mistake here?

Thank you all in advance!

Jakob

[George Ford]

https://psantanna.com/files/Callaway_Goodman-Bacon_SantAnna_2021.pdf
https://causalinf.substack.com/p/continuous-treatment-did