0
$\begingroup$

What is the appropriate panel data model when you have time dependent Y, but the X's are time-invariant (apart from the implicit time variable itself of course)?

My desired model is analogous to the following - my Y is income (which varies over time), but I want to see how growth in income is affected by gender (which is time-invariant). Thus, I have a model like:

Yit = ait + bit*gender + errors

Random effects seems to allow time-invariant X's, but doesn't quite seem appropriate and is not working in Stata either.

$\endgroup$
  • $\begingroup$ I would have thought you needed to include time in your model unless you have a strong reason not to. $\endgroup$ – mdewey Jan 25 '17 at 18:24
0
$\begingroup$

My answer is purely on the basis of what you have written as really your intention to study : which is checking if gender effects or is correlated with the growth in the y variable over time. A simple regression possible to run for this is:

$log(Y_{it}) = \alpha + \beta_{0} t + \beta_{1} t*G + \varepsilon_{it}$

Here the interaction term gives you the differential growth rate of Y variable for G = 1. Say G = 1 means female and male otherwise. G is a dummy here. This seems to be more informative. Hope this helps.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.