Page 1 of 1

Data transformation

Posted: Fri May 26, 2017 6:33 am
by econwork
Dear Eviews users,
Im having a trouble to understand how to transform my data, I hope some one can help me here.
I am going to estimate a lCobb-Douglas production function, however my variables dont have the same unit of measurements, reason why I need to standardize my data. The issue is that for estimating a Cobb-Douglas function we need to take the logarithm form of the data, now I am not sure if it is correct to standardize the log of the variables?
Thank you in advance

Re: Data transformation

Posted: Fri May 26, 2017 7:52 am
by startz
There is no need to standardize the data. Just take logs. Any difference in units of measurement will be reflected in the constant term.

Re: Data transformation

Posted: Tue Nov 28, 2017 11:22 am
by sakshikochar
I am quite a novice in panel data regression.

I have a question with my data, any help would be much appreciated.

I have the dependent variable is Profitability denoted by Profit after tax (P) in CHF Millions

Independent Variables are the below:

Net Interest Margin as a ratio in % (NIM)
Total Assets of the Banks in CHF Millions (TA)
Loan Loss Provisions as a ratio in % (LLP)
Cost to Income Ratio as a ratio in % (CTI)
Interest Income Share as a ratio in % (IIS)
Non-Interest Income in CHF Millions (NII)
Real GDP Growth Rate in % (GDP)
Inflation in % (Inf)
3M Libor Interest Rates in % (IR)

And below is the equation of my model:

P=c+β_1*NIM+β_2*TA+β_3*LLP+β_4*CTI+β_5*IIS+β_6*NII+β_7*GDP+β_8*Inf+β_9*IR+ε

Which variables should I apply natural log to before starting the regression? Is it really a requirement to apply a log? I am quite lost with this now and would really be helpful if someone could advise on my query regarding the transformation.

Many thanks!