# Regressing trending variables

• Sep 23rd 2012, 07:50 PM
usagi_killer
Regressing trending variables
I was wondering what is the effect of say regressing a dependent variable that has a positive trend with an independent variable that has a downward trend?

Thanks!
• Sep 23rd 2012, 08:37 PM
MaxJasper
Re: Regressing trending variables
Assume regression y=f(x). Now you say y increases as x decreases?
• Sep 23rd 2012, 09:07 PM
usagi_killer
Re: Regressing trending variables
Sorry what I meant was in a time series context.

Say you have this regression:

\$\displaystyle y_t = b_0 + b_1 x_t\$ where y_t is a positive trending time series variable and x_t is a negative trending time series data, in general would it be a good idea to add a time trend variable if the dependent variable is upwards trending and the independent variable is downwards trending? Ie, use the regression:

\$\displaystyle y_t = b_0 + b_1 x_t + b_2 t\$
• Sep 23rd 2012, 09:21 PM
MaxJasper
Re: Regressing trending variables
Quote:

Originally Posted by usagi_killer
Sorry what I meant was in a time series context. Say you have this regression: \$\displaystyle y_t = b_0 + b_1 x_t\$ where y_t is a positive trending time series variable and x_t is a negative trending time series data, in general would it be a good idea to add a time trend variable if the dependent variable is upwards trending and the independent variable is downwards trending? Ie, use the regression:
\$\displaystyle y_t = b_0 + b_1 x_t + b_2 t\$

You are actually talking about these two regressions:

\$\displaystyle y(t)-b_1 x(t)=b_0\$

\$\displaystyle y(t)-b_1 x(t)=b_2 t+b_0\$

The 1st one doesn't look good!
• Sep 23rd 2012, 09:56 PM
usagi_killer
Re: Regressing trending variables
Thanks for that, why would you pick the 2nd one over the first one?