Unit Root Work-around?

[dwalker]

I'm running into a dead-end with my regression. I'm using a 2SLS model to determine if US foreign aid has a statistically significant on long-term econ development in recipient country. To keep it brief: my data has a unit root; when I take the 1st or the 2nd difference, it too is not stationary. I have also done a ADF test on my regression's residuals. It too has a unit root.

Question is: assuming that your model is correct given theory and previous literature, what is the work-around for data that just doesn't "work"? Surely there is a way to something reliable with this data, right?

Any help is much appreciated.

[sanjibseviews]

Data - what's the measure of econ development? try taking log difference of the data, if you haven't done so.
Speicification - how did you select lag length? detrend data. you may want to test unit root vs. unit with structural change. best wishes.