how to lag variables correctly in unbalanced datasset

[Saleh sharmah]

Hi there
I would like to lag my firm-level and country-level variables by one year, Xi,t−1, as controls. I just use the Stata command: gen lag_x=x[_n-1] for all the controls that I need to include in my model. Is this correct or do I need to do it differently?

thanks for your kind input.

[Devra Golbe]

Hi Saleh,

The lag_x variables you define depend on the data being properly sorted. Moreover, the lag of the first observation for firm 2 would pick up the last value for firm 1 (assuming the data were sorted by firm then year.)

Assuming you have a panel indexed by panelvar and timevar, start by xtset-ing your data

xtset panelvar timevar

Then the regression of y on lagged x is given by

regress y L.x

[Saleh sharmah]

Thanks prof. Devra