Heya peeps, In SS3 (Quant) - Time Series - Schweser Page 248, there is an example for ARCH(1) model They started thre Asnwer section with this line "Since the p-value for the cofeeicient on the lagged variable indicates statistical significance, we can conclude that the … " , bla, bla, bla. My question is - how did they come to know that the lag1-coefficient (p-value = 0.001) is significant and constant-coefficnet (also p-value = 0.001) is not significant when no significance level (say 95%, etc) was given?
p value itself indicates the maximum level of significance. a p val of 0.001 means a significance level of 99.9%. p value is the maximum significance level at which the null can be rejected… hope this helps
jaspreet - what my concern is why lag1-coefficient IS significant and constant-coefficient is NOT when both have the same p-values of 0.001.
p-value is the probability that is less than 0.001 so more than 99.9% of the time it is going to be significant. look another way - t-value is 5.6 which is > 1.96 (or 2 for 95%, and 3ish for 99%)
constant coefficient need not be significant. having a value of zero means that the line passes thro the origin. that’s all.
i am not carrying the notes right now but it cannot be so that the notes mention that “constant coefficient is not significant”. you have not mentioned the context but the constant coeficient is never in the game. mostly the idea is to see what how the dependent variable differs wrt the independent variable (or a lagged value in the case) and whether it is significant. the signifcance value of the constant coefficient is normally ignored or is of little use. would be able to see the notes in a few hours…
exactly, that is what I wanted to confirm. So even though the constant is significant (it was irrelevant).
Time Series is so freaking easy. Can we do Time Series analysis in Excel or do we need some sort of external plugin?