Does a lag dependent variable resolve nonstationarity concerns?

[Doug Bujakowski]

I'm working with a panel dataset and unit root tests suggest that my dependent variable in nonstationary. I know that I can take first differences to try to deal this issue. When I do, my results are much different than those under a FE specification. I wonder whether this is due to my variables being nonstationary or to the information loss that first differencing entails. I don't know which results to trust.

Are there are other methods to make series stationary? If so, I'd like to explore those options. In particular, does using a dynamic approach resolve the nonstationarity issue? I can't seem to find a clear answer to this question.

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

It is a little odd that you get very different results in first difference than fixed effects. I'd double check that you're differencing everything correctly.

Different groups have different preferences for tests. I'm not good on the unit root issue. However, adding a lagged dv will drop your sample size just as differencing does.