19 9 19 You are creating 4-step ahead forecasts using a model that you just estimated. What can you conclude about the forecasts from a correlogram that has a gradually damping PAC function and an AC function that cuts off after 2 lags? The forecasts are not optimal because there is an AR(2) pattern to the residuals The forecasts are not optimal because there is an ARMA(2.2) pattern to the residuals There is no evidence here to suggest that the forecasts are not optimal