You are on page 1of 1

Homework Assignment #2

ECON 7090: Econometrics IV


Spring 2020

Due: Thursday, April 9th @ 9:00 AM

Instructions: this homework assignment covers some basic time series concepts and
autoregressive moving average models. You should upload your completed assignment to the
D2L Dropbox. You may discuss this assignment with other students in your class, but each
student must submit their own assignment. For questions involving computer code, please
submit both your .R file and the corresponding output.

Data: the data for this assignment is in the csv file auto_sales.csv. The csv file contains monthly
U.S. automobile sales (in millions of dollars) from January 2000 through January 2020.

Exercises

1. Plot the auto sales time series in R. Be sure to clearly label your plot.
2. Use the augmented Dickey-Fuller (ADF) test to determine if the time series is stationary.
If the series is non-stationary, construct a stationary time series using differencing
and/or transformations. Be sure to show the ADF output for your final stationary time
series.

Note: your stationary time series from question #2 should be used for the remaining exercises.

3. Let’s assume that the best model for your data is an AR(2). Use the Yule-Walker
equations to estimate your model coefficients. I would like you to use algebra for this
problem. You may perform the calculations using R but I would like you to solve the
system of equations implied by the Yule-Walker approach.
4. Next, use the “ar” function in R to estimate the AR(2) model. Verify that the reported
coefficients match your results from Question #3.
5. Use the “auto.arima” function to have R select the “best” ARMA(p,q) model for your
stationary time series. Report the estimated AR and/or MA coefficients.
6. Using your estimates from Question #5, forecast automobile sales for the next six
months (February 2020 through July 2020). You should add this forecast to your time
series plot. Plot the original series in black and the forecasted series in red.

You might also like