I am running regressions on one country's interest rate spreads vs another, and I am thinking of adding inflation rates as an additional independent variable.
Do I have to transform these into logarithm? And which is more accurate? Log or natural log?
You shouldn’t transform net interest rate spreads and inflation rate into logarithm b/c they are measured in percentage. What is other variable? Is it DV? When you use net interest rate spreads and inflation rate as IVs in your model, be careful the problem of multi-colinearity.
ln would be more appropriate, usually we use natural log to maintain uniformity in the data set. This is my understanding.
Thanks Cong Minh Huynh basically my DV is EMBIG spread and my IVs are US 10yr, 10yr TIPS (as a proxy for inflation expectations), and a risk measure such as VIX.
So now I am wondering, should I transform these variables?
Thanks Ashiq Muhammad AM should the transformation be appropriate, I will use the natural log.
Zinzile Dube : If they are not in %, you can transform them into Ln.
Thanks Cong Minh Huynh , the DV and 2 of the IVs are in basis points (yields). I'm still not absolutely clear on the intuition behind the transformation, but I think I may have to transform them all because the last IV, the VIX, is not in percent?
The reasons of transforming variables into logarithm are:
+ To transform a non-linear model into a linear model (creating a linear relationship b/w DV and IVs).
+ To make variables/ the residuals of the model become normal distribution.
+ To possibly eliminate the potential heteroscedasticity.
+ To interpret coefficients more easily by log-log model (percentage change).
interest rate is already in percentage, should not be transform into log form
Thank you for your answer Foday Joof , however, many of the published work I am looking at for reference do transform the spreads.... I am trying to make sure I understand why, the intuition behind it, so that I can explain it adequately in my paper.
In addition to the reasons given above for transforming variables using logs and not transforming variables using logs you should remember that you cannot take the (natural) log of a negative number. Hence, every observation of the data on a variable needs to be positive for you to be able to take the logs and obtain data for every observation. While inflation is normally positive it may have some negative observations especially in downturns (like the covid induced downturn). Hence, it is likely that you cannot take logs of all observations of an inflation data set. You can easily check whether you have any negative observations for any variable by calculating the minimum value. If the minimum value is not positive you should not take logs of this variable.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression