did_multiplegt parallel processing?

[alejoforero]

I am using the ssc command did_multiplegt, which is able to compute standard errors using bootstrap. Does anyone know if it is possible to run such boostraps using a parallel computing package such as `parallel'? I know the equivalent command in R (DIDmultiplegt) has that option but I wouldn't want to change environment just for one command.

Thank you in advance.

[Jared Greathouse]

Either email Clement to see if it's possible or, alternatively, go into the command and write it yourself. Are you having issues with run time?

[alejoforero]

Thanks, I will write to him.

I am having runtime issues, I am running this command with a large dataset (17million obs of trade data) and it takes 4+hours to calculate each run, so 100 reps is 400 hours. Since this is bootstrap should be easy to parallelize, I just thought maybe someone had already done this.

[Jared Greathouse]

Nah his response to me when I had this issue was to basically aggregate my data. It's kinda the game you play at that level of observations, unfortunately. I don't doubt it's possible, but it's likely more work than most people would want to invest in making it work