1
$\begingroup$

I'm using sctest function in strucchange package to check if there are breaks in my series.

I would like to understand how to interpret its result, in an example i get:

> g <- sctest(cars$dist~cars$speed+0)
> g

    Recursive CUSUM test

data:  cars$dist ~ cars$speed + 0 
S = 0.9147, p-value = 0.06379

What does it mean, are there breaks?

Thanks

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

The null hypothesis is that there are no structural breaks. Since your p-value is $0.063$ you can say that data does not disprove null hypothesis at $0.05$ significance level, i.e. there are no structural breaks. I would however hesitate to declare that there are truly no structural breaks without any further investigation.

Improve this answer
$\endgroup$
12
  • $\begingroup$ two things: (1) 0.063 is above 0.05 so why can i accept it? maybe should be below..no? (2) what other investigation should i do? what methods? thank you! $\endgroup$
    – ECOtime
    Commented Aug 24, 2011 at 13:39
  • $\begingroup$ better CUSUM Method? $\endgroup$
    – ECOtime
    Commented Aug 24, 2011 at 13:46
  • $\begingroup$ Null hypothesis is rejected if the p-value is less than the significance level. So if significance level is 0.05 and p-value is 0.063 null hypothesis should be accepted. Now null hypothesis is that there are no structural breaks. As for further investigation, there are a lot of things you can do, but they depend on your initial purpose of applying the test. $\endgroup$
    – mpiktas
    Commented Aug 24, 2011 at 13:49
  • $\begingroup$ How do you define better? $\endgroup$
    – mpiktas
    Commented Aug 24, 2011 at 13:49
  • $\begingroup$ mpiktas I don't know you told me to do other investigation and i thought that this method is not very good. $\endgroup$
    – ECOtime
    Commented Aug 24, 2011 at 13:55

Your Answer

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

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