search

FK83 / bvarsv

Analysis of the Primiceri (REStud, 2005) model
28 stars 14 forks source link

A general question on using bvarsv on nonstationary data - won't take you long #13

Closed yellowmonkey2 closed 8 months ago

yellowmonkey2 commented 8 months ago

Hi Fabian,

Thank you so much for creating the fabulous package.

The primiceri paper directly used 3 variables, from which some are nonstationary.

This leads me to think " Can we directly use nonstationary data into the package ? "

Appreciate your answers!

Best Chen

TyroLiu commented 8 months ago

Hi Fabian,

Thank you so much for creating the fabulous package.

The primiceri paper directly used 3 variables, from which some are nonstationary.

This leads me to think " Can we directly use nonstationary data into the package ? "

Appreciate your answers!

Best Chen

I am thinking about the same question. The dataset provided by the package, "usmacro", has three variables that are all nonstationary. Their p-values of Augmented Dickey-Fuller Test are 0.4865, 0.3525, and 0.261, respectively. Theory tells us that time-series must be stationary for VAR estimation. Can we really use nonstationary series in the TVP-VAR model? image