I am doing a regression analysis with following variables:
Debt forgiveness grants (current US$) [Independent var]
GDP per capita percentage (constant 2010 US$), [Dependent var]
[Control var:]
Exports of goods and services (current US$),
Foreign direct investment, net inflows (current US$),
Total natural resources rents (current US$), and
Oil Crude Prices (2019 US$ per barrel).
My question is, should I use a common unit for all of them. If so should it be current US$ or constant 2010 US$?
Also for the independent var and oil prices, constant values aren't available, so should I convert them using CPI?
Kartik -
I don't understand. Is this a time series? You have "GDP per capita percentage (constant 2010 US$), [Dependent var]." GDP in one year is just one number, so I guess this is a time series. But then 2010 dollars means one case also. So I don't get it.
If it were a time series, then I suppose the year for the dependent variable needs to match with the year in the independent variable that most impacts that 'dependent' year ... maybe the same year or a lagged year, or both.
If this were for a finite population in a single period, my area, I don't know what population you mean. In such a case, regressor data could still be from another (previous) year, but I don't see what your population is supposed to be.
Cheers - Jim
Use "current prices" for all the variables, to avoid the criticism that all some of your variables are affected by inflation while others are not.
constant variables are better to be used as they have taken the effect of inflation into account.
Dear Kartik
I will also support using constant US$. It takes care of inflation in the variable.
Regards
@Kehinde Mary Bello, is it okay if I used constant values for the dependent variables and current values for the others.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India