Tunis Business School Fall 2022
BCOR 320 - Tutorial 2
Problem
You want to fit an ARMA model to the following dataset : the weekly total number of loan applications
in a local branch of a national bank for the last 2 years.
As always, we start our analysis with the time series plot of the data. Also, we perform the ADF, PP
and KPSS tests to check the stationarity hypothesis.
1
1. Can we assume stationarity ?
2. Suppose you want to restrict attention to pure models (AR and MA) at first. Based on the ACF
and PACF plots, identify one candidate model and explain your choice.
3. Assume that the appropriate model to fit is the AR(2) model. The estimation results for the
AR(2) model are reported below. Are the estimated parameters statistically significant at 5%
significance level ?
2
4. The ACF and PACF plots of the residuals is reported in the following figure. What you can
conclude ?
5. Suppose that you want to estimate an MA(4) model. The estimation results are below. Which
model provides the best fit ?