I want to regress gdp with respect to some other variables. Should I consider real or nominal gdp? on the other hand, if the regressors included in the model are in nominal terms, will it be consistent with the regressand which is expressed in real terms?
It's better to use similar terms with whole regression variables (Real or Nominal) to get unbiased outcomes.
I see no reason why not to go with real values - you can escape some potential problems and setting values in real terms should not be the problem. Maybe you have different kind of problem: finding the right deflator?
Ok...let me explain. I want to regress gdp over direct and indire tax for a particular time period. Tax data are available in nominal terms, not in real terms. Can you please tell whther should I regress percentage grow rather than absolute values?
Growth rate, compare to levels will make you more comfortable regarding stationarity issue. Unless, you have some specific need in sense you need levels because your interpration needs force you to go with levels, use growth rate - growth rates are first differences of log values, so you have two transformation that makes your analysis robust toward stationarity issue.
Bojan Baskot: Thanks a lot. Can you suggest any way to convert nominal tax data into real data.
I asume that statistical agency provides some deflator for specific case - use that.
Regarding, levels vs transformed - it is known that you do lose some infor in process of transformation, but if you are interest in , let say analysis that looks at IRFs, than usually that is acceptable.
Hello, I was trying to check stationarity of GDP data for a period of time. I took log of GDP. There is a clear upward trend of GDP over the time. But surprisingly when I test for stationarity, at 5% and 10% significance level ADF test implies there is no stationarity. Although at 1% level, ADF implies there is stationarity. Now what should I do? Can you please suggest? I have attach the image file of the result of ADF for LGDP.
Take differences and you will get rid of trend.
Sometimes you may find that the trend is important then you include the trend as regressor.
Bojan Baskot: thanks.....while i use ADF after taking difference...there is stationary...But what surprises me why ADF is showing that the variable is stationary even though I have not removed any trend (or i did not differencing the data)?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression