Good morning,
I need to assess whether a positive correlation exists between S&P500 and rental value growth.
I know correlation should only be used with returns and not prices, given that one of the assumption of correlation is finite variance and price series tend to be non stationary having time varying volatility. Is it correct?
Results using rental value growth and S&P500 returns show almost zero correlation. On the other hand, results using S&P500 price index and rental value growth show high correlation. How would you explain it? Is it only a matter of wrong methodology?
Rental value is of course also a lot less volatile than S&P500. Should I standardize them or is there another way to compare them? May cointegration be a better way to analyse whether a long term relationship exist?
Thank you
Bene
Have you tried cointegration? ... That will indicate the relationships within them. A causality analysis may also be helpful. It will give you the direction if the causality ... which may give a good indication of the correlation that exists ... which i suppose is part of what you wanted to achieve
If you use prices, you'll get spurious correlations because of exactly what you described. I would use correlation or cointegration.
You have actually done the right thing. Do not be scared of your result but look for a way to explain the result in line with observations. You might try causality tests, check for lag correlation and other tests but you must use returns.
Benedicta, to add on my answer above, doing a mere correlation shall not give you the correct interpretation. This is because, even from the two series you're anlysing, you expect to have correlation do to some common forces ... that is why i strongly recommend con-integration, and checking causality. This should achieve your objective ...
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics