Gravity heckman marginal effect

[KIM SUJIN]

HI guys!!

Hello again!

I am master student who is writing using, HECKMAN 2step Gravity model

I have 170 import countries, 156 export countries 7year 15000 panel data!

Command :
heckman log_value log_dis_ij ban regionalization language log_REM_i log_REM_j log_gdp_i log_gdp_j, select(value_select = log_dis_ij ban regionalization language log_REM_i log_REM_j log_gdp_i log_gdp_j fot) twostep


I got result , and now I hope to get marginal effect!

Command:
margins, dydx(*) predict(yexpected)

However, I got

'option yexpected not allowed'


Could anyone help me about command?

Thank you all!!

[Joao Santos Silva]

Do not use the sample selection estimator for this; there is no sample selection in your data.

Best wishes,

Joao

[KIM SUJIN]

Dear Prof.Joao Santos Silva

I used trade value _Select as 1 if there is the trade, if not 0

So, I used Heckman model!

May I ask you what means that there is no sample selection in my data??


Thank you

[Joao Santos Silva]

The zeros in your data are "true" zeros, not a code for missing observations as in the wages for people not in the labour force. So, there is no sample selection because you observe the dependent variable for every unit, it is just that in some cases it takes the value 0.

[KIM SUJIN]

Thank you Professor!
I wonder about model..
As I know, Gravity Heckman model is a famous model, and people already use Heckman model for trade data.. isn't it??

Thank you!!!

[Joao Santos Silva]

As far as I know, the sample selection model is not frequently used for trade data.

Best wishes,

Joao

[KIM SUJIN]

Thank you!! I will consider it again..

Sujin