Code for placebo test of synthetic control method

[SALINA SIDDIQUA]

Hi everyone, I am salina siddiqua, doing PhD in health economics at the university of York, UK. I am applying synthetic control method in my first chapter, I applied synth package from Hainmueller webpage but struggling to run placebo test and sensitivity analysis. Can anyone help me please with Abadie's code or your one so that I can develop code for my dataset. Thank you.

[Maxence Morlet]

Could you perhaps give an example of your data, what you have tried to run, and what Stata gave you back?

Jared Greathouse is an established expert in this field of study by the way.

[Jared Greathouse]

Maxence Morlet Me, expert? I feel honored. I'll post an example of the in space placebo testing in a few.

[Daniel Schaefer]

Humility aside, just look through Jared's posting history: He is undoubtedly an expert in synthetic control methods.

[Jared Greathouse]

So, OP never specified if they want in-space or in time placebos. My scul command (I think, the public version) does both. Anyways, assuming we need in space placebos, you don't (in my personal opinion) want to use synth. What you really want, is allsynth (or synth2, I reviewed this one for SJ). I prefer allsynth due to its bias correction abilities as well as its ability to handle staggered adoption. For allsynth

clear *



cls


u "http://fmwww.bc.edu/repec/bocode/s/synth_smoking.dta"

xtset state year, y


allsynth cigsale beer(1984(1)1988) ///
lnincome retprice age15to24 ///
cigsale(1988) cigsale(1980) cigsale(1975), ///
    trunit(3) trperiod(1989) ///
    gapfigure(classic placebos lineback) pvalues ///
    keep(smokingresults) ///
    replace ///
    xperiod(1980/1988)

and for synth2 (which now that I look CAN do in time placebos)

    
synth2 cigsale beer(1984(1)1988) ///
lnincome retprice age15to24 ///
cigsale(1988) cigsale(1980) cigsale(1975), ///
    trunit(3) trperiod(1989) ///
    placebo(unit) ///
    xperiod(1980/1988)

On the point above, I know it's kinda odd for me to say, but when I look at the stuff my coworkers write, the amount of crazy math/real analysis they know, I sort of feel inferior to them.

[SALINA SIDDIQUA]

Thank you everyone...........Should I run the estimation for all control units before run placebo test? And I am using synth package.

[SALINA SIDDIQUA]

Dear Jared, may I request you send me code of synth as I am using it.........is synth and allsynth different?