ASA Connect

 View Only
Expand all | Collapse all

Coefficient interpretation for differenced independent variable Time series regression

  • 1.  Coefficient interpretation for differenced independent variable Time series regression

    Posted 02-05-2016 11:16

    Hello everyone,

    I'm working on a time series project trying to built a dynamic regression model with the dependent variable regressed against its own lagged value at time (t - 12) and a second independent variable which was non stationary so I had to take its first difference to stationarize it . So now how do I interpret the coefficient on the differenced independent variable against the dependent variable which was stationary to begin with and was not differenced? so lets say I have Incometime(t)= C + B1iNCOMEtime(t-12) + B2differenced(GDP)time(t-3). Thank you very much. 

    ------------------------------
    Aboubacar Coulibaly
    Student
    St Cloud State University
    ------------------------------


  • 2.  RE: Coefficient interpretation for differenced independent variable Time series regression

    Posted 02-09-2016 00:02

    Dear Aboubacar,

    I'm no expert, but I'd like to see how my response ads up. The parameter estimates meanings don't change just because one variable is differenced. You still get 'as differenced(gdp) changes by one unit when iNCOME is zero...' Its just that the model interpretation is not including a non differenced version of the variable since that wasn't fit. 

    Out of curiosity, why are you taking the 3rd lagged time point for differenced(gdp)?

    Sincerely, 

    KeithC. 

    ------------------------------
    Keith Chamberlain
    Analytical Chemist