Synth: missing control units problem

[Nikou Asgari]

Hi,

I am using the synthetic control method to analyse the effects of economic sanctions on income inequality.
After uploading my dataset into stata, the commands I have used are as follows:

 ssc install synth

 tsset country_n year

 synth gininet log_GDPcap log_exports log_investment log_secondarygross coupdetat gininet(1995) gininet(2000) gininet(2005) gininet(2010), trunit(7) trperiod(2011) xperiod(1995(1)2010) fig

Stata gives me:

control units: for at least one unit predictor gininet(1995) is missing for ALL periods specified


How do I solve/bypass this problem so that my regression works?

Thank you

[William Lisowski]

Welcome to Statalist, and thank you for the well-presented code, including presentation of your

I am not a user of synth, but the error message suggests to me that for at least one country_n (those are your units set by tsset) the 1995 observation of gininet is missing.

You could locate any such observations with

list country_n if missing(gininet) & year==1995

For completeness you will probably want to check this for 2000 and 2005 and 2010 as well. I think you could do them all at once with

list country_n if missing(gininet) & inlist(year, 1995, 2000, 2005, 2010)

It seems to me your options are to exclude countries that have missing values for gininet in any of the crucial years, or excluded gininet(1995) from your predictor variables, and any of the other crucial years that may have missing values.

[Nikou Asgari]

Hi, I followed your code and the missing problem is solved, thank you!
However now when I run my same regression with the dropped observations, Stata gives me 'conformability error'.
What does this mean and how do I fix this?

[William Lisowski]

A conformability error is a low-level error message from within Stata, rather than a message from the developers of synth, and it means that there is a problem in synth not anticipated by the programmers.

The synth package is user contributed software you downloaded from SSC. Toward the end of the output of

net describe synth, from(http://fmwww.bc.edu/RePEc/bocode/s)

there is an email address given for support.

[William Lisowski]

See further discussion at

https://www.statalist.org/forums/for...g-observations