In the CFA EOC questions on time series (reading 13, page 488, question 17), it states that this model is correctly specified due to the autocorrelations of the lags all being insignificant. However, if you look at the coefficients themsleves, lag 1 appears to be insignificant itself (t-stat of -0.8757). How can this model be correclty specfied if the coefficent is not different from zero?
I think they wanted to test if the model was correct due to the autocorrelation and “form” of the times series (seasonality), not to determine if this model was correct regarding the significance of the coefficients. Just justifying the answer.
I mean, they are testing if the model justify the seasonality of the time series (autocorrelation), then the next step would be to test the coefficients and that stuff,
Regards,
Jorge