search

mpiktas / midasr

R package for mixed frequency time series data analysis.
http://mpiktas.github.io/midasr/
Other
74 stars 34 forks source link

Stationarity - Transformations for Mixed Frequency Data #80

Closed JRatschat closed 2 years ago

JRatschat commented 3 years ago

Hi all,

my question is more theoretical but I hope that this is still the right place to ask my question.

I am using MIDAS regressions to nowcast private consumption. Since my dependent low frequency variable is non-stationary, I apply second order differencing to get a stationary time series.

How should I now transform my monthly and daily data? Do I need to apply second order differencing as well or can I also use first order differencing? Or what are my alternatives here?

Sorry for asking such a basic question but I was not able to find relevant information in the related literature.

Best, Jonathan

vzemlys commented 2 years ago

The same rules apply. The regression only works when all the variables in the regression specification are stationary. For non-stationary case (unit-root) you might look into imidas_r.