if you have two lagged dependent variables and one lag for seasonality, is this an AR(2) or AR(3) model? schweser notes say that its AR(2) because the seasonality lags dont count, but the q bank contradicts this in a question. thanks.

2 second answer - It’s a AR(2) with a seasonal lag.

so if theres, for example, 3 lagged dependent variables (i.e. xt-1, xt-2, xt-3) and one seasonal lag (i.e. xt-12), this is an AR(3) model with one seasonal lag?

that was very helpful thanks a lot. also another quick q on this topic–when testing for seasonality, do you only have to test for significant of the autocorrelation of the seasonal lag? for example, if you have 4 lags and its a quarterly model, do u have to see if all 4 autocorrelations are significant or do u only focus on the significance of the fourth one thanks

>“do you only have to test for significant of the autocorrelation of the seasonal lag?” You have to test if the t-stats of autocorrelation of residuals is significant or not. So if you have 12 month Macy’s regressed data and you observe t-stats for months 1 to 11 as insignificant but Dec month data is significant, that’s when you add the t-12 lag.