View Single Post
Old
11-12-2012, 03:15 PM
  #297
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 3,358
vCash: 50
Quote:
Originally Posted by barneyg View Post
What are the correlations between those variables? I'd assume Xp is strongly correlated with them if it becomes the least significant variable once the others are included. That said, with 30 data points any t-stat above ~2 will be significant and the t-statistic on Bp seems to be 1.5*5.5 = 8.25 (square root of 30 = 5.5).
I haven't calculated cross-correlations between the variables, but would expect all 4 to have substantial correlations to one another (including negative correlations with Xg), because the changes in each variable are relatively contemporaneous (all of them tended to occur in the 90s... after the 80s and relatively constant after 90s).

Quote:
Originally Posted by barneyg View Post
I'm not an econometrician (I don't play one on TV either), but I know that R-square doesn't mean squat when no intercept is included in the model. Was there one in yours?
I was trying to include an intercept [using LINEST in Excel with paramaters (Y range, X ranges, false, true) ], but the intercept calculated as zero. I'll have to check that again, perhaps the false should be a true (but I thought it said that parameter was B = 0, so I entered that as false).

Quote:
Originally Posted by barneyg View Post
In models with stable time series, people often calculate first differences i.e. Y'(1985) = Y(1985) - Y(1984) and run the model on those changes instead of the level. It changes the narrative because 'last year' is the benchmark for each observation but since your objective is often 'what makes Y change' it works. There's a deeper econometric(al?) reason that justifies first differences as well.

I'll try to find time to look at the other thread too.
Yes, I thought about calculating differences first, as I did in an unrelated model. Do you think % (ratio) differences would be better than raw differences?

Please do look at my other post if you have time at some point. It's not a whole thread, just a post with questions in a stick thread titled "Ideas for Future Studies" on this (BTN) sub-forum. I think the model I was trying to create, or a similar, workable version of such, would be the best solution in the longer term, but obviously we need a model where coefficients will generate.

Czech Your Math is offline   Reply With Quote