I think the order of panel ARDL may be set according to the results from the individual time series

This may help...

http://www.statalist.org/forums/forum/general-stata-discussion/general/1333077-optimal-lag-length-for-ardl-in-panel

This may help too..

https://pdfs.semanticscholar.org/9732/5b8cd12c52512235dd3326208e96f7863681.pdf

Hi Atif,

You should write your code like this:

forval i = 1/28 {

ardl cp gdpc pg co2 ghe if (c_id == `i'), maxlag (2 2 2 2 2)

matrix list e (lags)

di

}

and run each line separately.

hi

i use this code in stata 13 to determine the number of optimal lags but it does not work , the model is panel ARDL

forval i=1/14{

ardl chom pop cre imp inf chg IDE slr scl migr if (pays= 'i'), maxlag=(2 2 2 2 2 2 2 2 2 2)

matrix list e(lags)

di

}

Hie Hada you say if (pays=="I") ....it's double equal sign

Hi Hadda,

Sorry for the late response.

in your code you forgot one "=" sign and added one extra. the code should be like

forval i = 1/14 {

ardl chom pop cre imp inf chg IDE slr scl migr if (pays == `i'), maxlag (2 2 2 2 2 2 2 2 2 2)

matrix list e (lags)

di

}

Dear Haci

Various lag length selection criteria can be used such as Akaike Information Criterion, Schwarz Information Criterion, Hannan-Quinn Criterion, Final Prediction Error, Corrected version of AIC. It has been found that SIC is the most useful criterion.

Good Luck.