Firstly, I am applying a 2sls model in my paper:

Corruption= o+1bq+2da+3lawandorder+4unemprate+5politicalregime+6gdpgrowthrate+7govermentstability

Log (FDI) =o+1corr+2buerucraticquality+3lawandorder+4da+5investmentprofile +6gdpgrowthrate+7populationgrowth+8inflation+9openess+10politicalterrorscale.

As a part of time series data I know that I have to check these variables for stationary and s serial correlation.
Once I check my variables for stationary, and then do the first difference then my variables do become stationary but when I regress that, then the results comes weird.
Apart from that, I have checked for serial correlation through acf and pacf graphs in order to determine the lags.
I am somewhat confused either to run regression on lagged values or difference data.
It would be great if you could guide me on this


You have to think about how differencing changes the interpretation of your model. When you difference all variables then the coefficients on the regressors state how changes of the regressors determine changes of the endogenous variables. However if you use lagged variables then you rather get a statement about how the level of your endogenous variable changes with lagged levels of the regressors. So for me it seems like this could be also a question to you whether you are more interested in the changes or in levels.


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.