Synthetic method error: for at least one unit predictor avh is missing for ALL periods specified

[Kenta Kunte]

Hi experts,

I am very new to this community, but let me ask you a question. I am trying to conduct the Synthetic Control method to investigate the impact of the coup which happened in Zimbabwe in 2018 on its GDP.

Utilizing Penn World Table, I tried to create the Synthetic based on the other 17 countries that have never witnessed coups in their history.

To be more precise, I used this code

 tsset country_id year

synth rgdpe pop emp avh hc,trunit(26) trperiod(2018) keep("scm_test_zimbabwe") replace

However, I got this message.

control units: for at least one unit predictor avh is missing for ALL periods specified

Could anyone help me solve this issue? I am not sure even why Stata gave me this message.

Thank you for reading.

BR

[Jared Greathouse]

Pretty straightforward. avh is missing for every single time period for at least/more than one of your donors. Since your can't average over that, the predictor has to be dropped

[Kenta Kunte]

Thank you Jared Greathouse for your response.

Does it mean if the synth command does not function properly even if there is one missing value at certain period?

Is there any way to solve this issue? This is because it is cross-sectional data over different countries, and consequently it would be inevitable not to encounter missing data.

It might be rudimentary question though(I am quite new to econometrics), but it would be appreciated if you help me out.

The Best

[Jared Greathouse]

No you can have missing values. It's just they can't all be missing for any one unit. And in all honesty, my scm command is better than normal scm. I'll show you in my next response.