difference in difference, fuzzydid package - reproduction of an article

[Benoit Decoco]

Hello everybody,

I am trying to reproduce Gendron-Carrier et al. recent AEJ article on subways and air pollution.
I want to use De Chaisemartin and d'Hautfoeuille (2020) robust two way fixed effect estimators which has its own package in Stata named fuzzydid.


I am using Stata 17 and I have the replication code of Gendron-Carrier's article.

The first fuzzydid regression does not work

*Specification (0) ;
                display "robustreg ${pre_}aod_mean_`ring' ${dist2event} `weights', $error";
                fuzzydid ${pre_}aod_mean_`ring' ${dist2event} `weights', $error;
                eststo reg0;
                countevents;
                estadd ysumm;
                estadd scalar sc_Nevents = $Nevents;
                estadd scalar sc_Ncities = $Ncities;

I get the following error

factor-variable and time-series operators not allowed
r(101);

The variable

${dist2event}

is a dummy variable (equal to 1 after 18 months of subway opening and 0 otherwise)
I can detail the code on how the others are build if needed.

Would you have any idea on how to solve this problem ? I would like to apply an estimation method robust to heterogeneous treatment effects.

[Benoit Decoco]

up

[Jared Greathouse]

Your syntax confuses me, and doesn't look like anything I've seen before in Stata. What's up with all the brackets and semicolons and commas? I don't know what "does not work" means.

Anyways, you've included a factor variable someplace that you can't. So that's the issue.

if you have the replication code, why is this an issue?