Different time series time dimension
Posted: Wed Jan 07, 2015 9:01 am
Hello Everyone, I have this (I guess) simple doubt, I'm estimating a model for inflation in a local country , the model in fact it is
dlgog(CPI) dlog(CPI(-5) dlog(exchangerate(-2)) dlog(exchangerate(-10)) dlog(oil(-2)) dlog(oil(-7)) d(CPI_Japan(-1)) d(CPI_Japan(-6))
The model in terms of statisctics and significance does very good, but my problem is that all variable are monthly index (CPI local country , exchangerate, oil price) except for the inflation of the foreign country, that is monthly frequency but, that is y/y % change ( so CPI(index) / CPI(index(-12)) -1 )*100).
I could not use dlog for that variable, because in some months there is negative inflation rate and instead I used diferenced value (November inf - October inf)
My problem is, is there a problem of using that y/y inflation in a model where all variables are monthly , even if I capture the month change using D(CPI_Japan)
Thanks in advance for any help
dlgog(CPI) dlog(CPI(-5) dlog(exchangerate(-2)) dlog(exchangerate(-10)) dlog(oil(-2)) dlog(oil(-7)) d(CPI_Japan(-1)) d(CPI_Japan(-6))
The model in terms of statisctics and significance does very good, but my problem is that all variable are monthly index (CPI local country , exchangerate, oil price) except for the inflation of the foreign country, that is monthly frequency but, that is y/y % change ( so CPI(index) / CPI(index(-12)) -1 )*100).
I could not use dlog for that variable, because in some months there is negative inflation rate and instead I used diferenced value (November inf - October inf)
My problem is, is there a problem of using that y/y inflation in a model where all variables are monthly , even if I capture the month change using D(CPI_Japan)
Thanks in advance for any help