1

I'd like to simulate 2 correlated lognormal AR1 time series. I have already found this post which is the answer for 2 Normal AR1 time series

How to simulate two correlated AR(1) time series?

I've found the rlnorm.rplus function from the composition package which allows to do the same as the rmvnorm function but for lognormale times series.

I did some trial to combine information from the post mentioned with the rlnorm.rplus function but I've been unsuccessful.

Is someone able to show me how I should proceed ?

Community
  • 1
Guillaume
  • 11
  • 3
  • If you are asking for code, this question would be off-topic for CV. Such questions can be on-topic on [Stack Overflow](http://stackoverflow.com/), but need a [reproducible example](http://stackoverflow.com/q/5963269/1217536). If you are asking about the statistical issues underlying this, please edit to clarify. If you just want help w/ the code, why not add what you did, why/how it didn't work, & what you want the output to look like if did work, & we can migrate it for you. – gung - Reinstate Monica Dec 03 '14 at 19:17
  • Combining the code from the post that you mention and the function `rlnorm.rplus` from package [compositions](http://cran.r-project.org/package=compositions) seems a good idea. What did not work? What is the covariance matrix? – javlacalle Dec 03 '14 at 20:19

1 Answers1

1

Note that if you mean "two series whose logs are correlated AR(1)" this is straightforward - do the normal case, and exponentiate.

If you actually mean that you want them to be lognormal but follow an AR(1) on the lognormal scale, this really doesn't work, because your AR(1) involves shifts, and the usual two-parameter lognormal doesn't do that.

Glen_b
  • 257,508
  • 32
  • 553
  • 939