0
$\begingroup$

I am currently writing a paper, analysing the impact of goldprice movements on the capital structure of gold mining firms. My basic model is a simple OLS model with (y=leverage and x=ln(goldprice)). After testing for unit root in x I decided to use the first difference of ln(goldprice) which is the growth rate. Testing this model I find that i) y and x do not really have a linear relationship (exhibit 1); ii) I reject H0 of Ramsey Reset test, checking for omitted variables while linktest tells me that the model is correctly specified. But testing with plots I find strong evidence for misspecification (exhibit 2 & 3).

Does anyone has an idea on what I did wrong - would be much appreciated. enter image description here

Improve this question
$\endgroup$
4
  • 3
    $\begingroup$ Hi there and welcome. Uhm, the plots clearly indicate that $x = \ln(\text{goldprice})$ is not a good predictor of $y = \text{leverage}$. Which is a nice conclusion for a paper. (Just my two cents though) – Reviewer $\endgroup$
    – Jim
    Commented May 21, 2018 at 19:39
  • $\begingroup$ What happens if you log-transform leverage as well? $\endgroup$
    – Isabella Ghement
    Commented May 21, 2018 at 20:43
  • $\begingroup$ Thank you both for your comments. Isabell you are absolutely right I am a blind, that worked and now I have a linear relation. However the ln(leverage) is now lacking normal distribution so do the residuales. Initially I helped myself with the sqrt. Would you agree that I can still transform ln(leverage) so I get normality or would this lead to biased interpretations? $\endgroup$
    – Knut E.
    Commented May 21, 2018 at 22:05
  • $\begingroup$ Please do post those $x = \ln(\text{goldprice})$ vs. $y = \ln(\text{leverage})$ plots. And if you do, please increase the resolution a bit. $\endgroup$
    – Jim
    Commented May 22, 2018 at 16:10

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.