This problem fits an ARIMA model to the logarithms monthly one-month T-bill rates in the data set Mishkin in the Ecdat package. Run the following code to get the variable:
library(Ecdat)
data(Mishkin)
tb1 = log(Mishkin[,3])
(a) Use time series and ACF plots to determine the amount of differencing needed to obtain a stationary series.
(b) Next use auto.arima to determine the best-fitting nonseasonal ARIMA models. Use both AIC and BIC and compare the results.
(c) Examine the ACF of the residuals for the model you selected. Do you see any problems?