Can someone confirm the following? To test for seasonality, do we use a t -test? Do you think we will need to use Engle-Granger tables in testing for unit root, if so how do we use them?
Close, SSE = 1/root n
for seasonality, throw in an extra variable lagged by a year. calculate the standard error as I/root N, where N is the sample size ( this will be a a constant) then use the AUTO CORRELATIONS to compute t-stats, and then read them off the table. Where there is seasonality, the t-test will be significant at the given level of signficance. Engle Granger…I confess I dont know them species …what is it about pls?
I think its about ARCH or something
Engle and Granger recalculated critical t-values from Dickey Fuller to test residuals for a unit root in cointegrated series If the test rejects the null of a unit root, error terms are covariance stationary and the two series are cointegrated
on a scale of 1 to 100, what is the likelihood of CFAI testing this? closer to the begining of the scale i guess!