EViews.com

Hi,

I'm trying to apply the bounds test approch to cointegration as described by Pesaran et al (2001).

I have an UECM with 4 variables and need to select the optimal lag order based on AIC or SBC:

tb c d(tb(-1)) d(tb(-2)) d(tb(-3)) d(tb(-4)) d(tb(-5)) d(tb(-6)) d(tb(-7)) d(tb(-8)) d(ybr) d(ybr(-1)) d(ybr(-2)) d(ybr(-3)) d(ybr(-4)) d(ybr(-5)) d(ybr(-6)) d(ybr(-7)) d(ybr(8)) d(yus) d(yus(-1)) d(yus(-2)) d(yus(-3)) d(yus(-4)) d(yus(-5)) d(yus(-6)) d(yus(-7)) d(yus(-8)) d(reber) d(reber(-1)) d(reber(-2)) d(reber(-3)) d(reber(-4)) d(reber(-5)) d(reber(-6)) d(reber(-7)) d(reber(-8)) tb(-1) ybr(-1) yus(-1) @trend

Should I remove the insignificant variables once estimating by OLS? Hendry's general-to-specific approach involvesr emoving the insignificant variables from the estimation, is this step taken once the optimal lag orders for each variable have been detemined? I've heard that Microfit is a better solution to bounds testing but I only have Eviews available to me, must I estimate this model and compare the criterion for each possible comobination of lags?

My thesis deadline is in less than a week :(

have you found a solution im stuck at the same thing