Consider, again, the US monthly export and import series of Problem 1. 

(a) Build a bivariate time series model for the two series. Perform model simplification and model checking to justify the model used. 

(b) Obtain the impulse response function of the model built using orthogonal innovations. 

(c) Obtain 1-step to 4-step ahead forecasts using the model built at the forecast origin \(t=324\), which is the last data point (or December 2018).

Data From Problem 1:

Consider the log series of the US monthly exports and imports data in the file m-expimpcnus.csv.