Selection of variable type for correlation?

[musicgold]

Hi,

I am trying to understand the relation between an equity index, govt bond yields and inflation.
I am using a data set of 150 months.

However, I am not sure which of the following sets is more appropriate when calculating correlation.
I would appreciate it if you could point me to resource where I can read on selection of appropriate variables for correlation.

Set A
1. Month over month % change in the equity index value.
2. Month over month % change in yields of govt bonds
3. Year over year % change in the CPI index

Set B
1. Equity index values
2. Yields of govt bonds
3. CPI index values

Thank you.

 

[aaronhotchner]

You always want to use rates of change, but you would also want the same time period. Yearly changes in the CPI won't explain monthly equity value changes. Also, you want the difference in bond yields, since they are already a rate; don't take the percent change (for instance 1% yield to 2% yield is a 1 ppt change as opposed to a 100% change).

As a caution, though, inflation does not move bond yields. Expected inflation does.

 

[musicgold]

aaronhotchner,

Thanks.

Would you be able to suggest a book or website where I can read up more on this?

 

[aaronhotchner]

Pick up any econometrics book (I used Gujarati). Read financial economics journals.

This might get you somewhere, though I haven't looked it over in depth: http://web.mit.edu/alo/www/Books/fmetrics.pdf

 

[musicgold]

Thank you.