I've conducted ols regression analysis on stock market index KLCI (dependent) and 2 other independent variables that includes exchange rate (ER) and interest rate (IR). And the result shows that i have 0.044 for r-sqaured while -0.042 for adjusted r-squared. By looking at the result, my lecturer told me that the model is not capturing my dependent variable. i've collected 25 annual data each on the stock market index, real interest rate and the real effective exchange rate on conducting it.

How do i need to adjust the data to be able to have a proper regression results? I used log(KLCI) c ER IR on Eviews.

enter image description here

  • 1
    $\begingroup$ Welcome to CV. Since you’re new here, you may want to take our tour, which has information for new users. The question is not clear. What do you mean by "does not capture the dependent variable"? Could you provide more information about the data/analysis/objectives? (I see you have 25 observations, probably annually, which is not much). $\endgroup$
    – T.E.G.
    Mar 31 '18 at 17:01
  • $\begingroup$ hi, i've collected 25 annual data each on stock market index (dependent), exchange rate (independent) and interest rate (independent) to conduct the ols regression on eviews. And i get the results above which does not seem correct? $\endgroup$
    – ncky
    Mar 31 '18 at 17:09
  • $\begingroup$ You probably want to look at change in the stock market index rather than its raw value. $\endgroup$
    – Peter Flom
    Apr 1 '18 at 12:28

The suggestion from Peter Flom in the comments is correct, you should be looking at changes in these variables, rather than their level. This not only makes sense from a theoretical standpoint (the stock market level doesn't really have any meaning by itself), but also the Durbin-Watson statistic in your results (0.646) is telling you that you have residual auto-correlation problems. This statistic should be near 2.

Note that the exchange rate by itself also has no meaning, and the change is what matters. The interest rate is more debatable, but I would try with changes as well.

So what you could try in EViews would be:

dlog(KLCI) c dlog(ER) IR dlog(IR)

Do keep in mind that the stock market is of course notoriously difficult to predict, otherwise you could become a millionaire very quickly with this equation. So I'd expect the $R^2$ from this regression to be quite low.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.