Scribd
Upload
329 views

Midterm Question - Time Series Analysis - Updated

This document outlines the midterm for a time-series analysis course. It includes 6 questions testing skills in time-series modeling, unit root testing, and structural break detection. Question 1 involves forecasting an AR process and question 2 explores unit root processes. Question 3 requires finding solutions to trend plus irregular models. Question 4 analyzes an AR process with a moving average term. Question 5 involves using simulated data to identify and model a structural break. Question 6 applies unit root tests to macroeconomic time-series data.

Uploaded by

Aakriti Jain
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
329 views

Midterm Question - Time Series Analysis - Updated

This document outlines the midterm for a time-series analysis course. It includes 6 questions testing skills in time-series modeling, unit root testing, and structural break detection. Question 1 involves forecasting an AR process and question 2 explores unit root processes. Question 3 requires finding solutions to trend plus irregular models. Question 4 analyzes an AR process with a moving average term. Question 5 involves using simulated data to identify and model a structural break. Question 6 applies unit root tests to macroeconomic time-series data.

Uploaded by

Aakriti Jain
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Time-series Analysis (ECO – 6403)

Midterm
Total - 120
Due: 1st April (Thursday, 9:00 p.m.) Spring 2021

1. Suppose that the money supply process has the form 𝑚𝑡 = 𝑚 + 𝜌1 𝑚𝑡−1 + 𝜌2 𝑚𝑡−2 + 𝜀𝑡 ,
where 𝑚 is a constant and 0 < 𝜌< 1. [Total: 5
points]

a. Show that it is possible to express mt+n in terms of the known value 𝑚𝑡 and the sequence
{𝜀𝑡+1 , 𝜀𝑡+2 , … , 𝜀𝑡+𝑛 }. (3 points)
b. Suppose that all values of 𝜀t+i for i > 0 have a mean value of zero. Explain how you could
use your result in part a to forecast the money supply 𝑛 periods into the future. (2 points)

2. The unit root problem in time-series econometrics is concerned with characteristic roots that are
equal to unity. In order to preview the issue: [Total: 37 points]

a. Find the homogeneous solution to each of the following: (Hint: Each has at least one-unit
root.) (2 × 4 = 8 points)

i) 𝑦𝑡 = 1.5𝑦𝑡−1 − 0.5 𝑦𝑡−2 + 𝜀𝑡


ii) 𝑦𝑡 = 𝑦𝑡−2 + 𝜀𝑡
iii) 𝑦𝑡 = 2𝑦𝑡−1 − 𝑦𝑡−2 + 𝜀𝑡
iv) 𝑦𝑡 = 𝑦𝑡−1 + 0.25 𝑦𝑡−2 − 0.25 𝑦𝑡−3 + 𝜀𝑡

b. Show that each of the backward-looking solutions is not convergent. (2 points)

c. Show that Equation i can be written entirely in first differences; that is, ∆𝑦𝑡 = 0.5𝑦𝑡−1 + 𝜀𝑡 .
Find the particular solution for ∆𝑦𝑡 . (2 × 3 = 6 points)

d. Similarly transform the other equations into their first-difference form. (2 × 4 = 8 points)

e. Find the particular solution, if it exists, for the transformed equations. (2 × 4 = 8 points)
f. Write equations (i) through (iv) using lag operators. Given an initial condition 𝑦0 , find the
solution for 𝑦𝑡 = 𝑎0 − 𝑦𝑡−1 + 𝜀𝑡 . (2 + 3 = 5 points)

3. Given the initial condition 𝑦0 , find the general solution and the forecast function (i.e., Etyt+s)
for the following variants of the trend plus irregular model: [Total: 18 points]

a. 𝑦𝑡 = 𝜇𝑡 + 𝑣𝑡 , where 𝜇𝑡 = 𝜇𝑡−1 + 𝜀𝑡 , 𝑣𝑡 = (1 + 𝛽1 𝐿)𝜂𝑡 , and 𝐸𝜀𝑡 𝜂𝑡 = 0. (5 points)

1
b. 𝑦𝑡 = 𝜇𝑡 + 𝑣𝑡 , where 𝜇𝑡 = 𝜇𝑡−1 + 𝜀𝑡 , 𝑣𝑡 = (1 + 𝛽1 𝐿)𝜂𝑡 , and the correlation between 𝜀t
and 𝜂𝑡 equals unity. (3 points)

c. Find the ARIMA representation of each model. (5 + 5 = 10 points)

4. As indicated in the text, the ACF of a series with a unit root shows little tendency to decay.
Nevertheless, it may difficult to detect a unit root in a series with a negative moving average.
Consider the unit root process 𝑦𝑡 = 𝑦𝑡−1 + 𝜀𝑡 − 0.8𝜀𝑡−1 . [Total: 11 points]

a. Iterate backward from 𝑦𝑡 to solve for 𝑦𝑡 in terms of the {𝜀𝑡 } series and the initial condition
𝑦0 . (2 points)

b. Use the method of undetermined coefficient of 𝑦𝑡 in terms of the {𝜀𝑡 } series and the initial
condition 𝑦0 . (2 points)

c. Use the answers to part (a) or (b) to derive the first three terms of the ACF. (2 × 3 = 6
points)

d. Explain how the negative MA term affects the shape of ACF. In particular, explain how
the series is “infinitely persistent” even though the coefficients of ACF are far below
unity. (1 point)

5. The second column in the file BREAK.XLS contains the simulated data. [Total: 37 points]

a. Plot the data to see if you can recognize the effects of the structural break. (1 point)

b. Summarize the first six autocorrelation functions of {𝑦1𝑡 } series and the first difference of
the series. Also perform Dickey-fuller test with the following specifications: (2 + 6 = 8
points)
i) 𝑦𝑡 = 𝛽𝑦𝑡−1 + 𝜀𝑡
ii) 𝑦𝑡 = 𝛼0 + 𝛽𝑦𝑡−1 + 𝜀𝑡
iii) 𝑦𝑡 = 𝛼0 + 𝛼1 . 𝑡 + 𝛽𝑦𝑡−1 + 𝜀𝑡
Do you think in all these above four cases the {𝜀𝑡 } series follows white noise? Why? (2
points)

c. The third column in the file BREAK.XLS contains another simulated data series called
{𝑦2𝑡 } with a structural break at t = 51. Plot the series and compare your graph with the
plot that you have done in part (a). (1 + 2 = 3 points)

d. Obtain the ACF and PACF of the {𝑦2𝑡 } sequence and first difference of the sequence. Do
the data appear to be difference stationary? (2 + 2 + 1 = 5 points)

e. If you perform a Dickey – Fuller test including a constant and a trend, you should obtain

2
𝑦2𝑡 = 0.072 − (1.1014 × 104 )𝑡 − 0.022𝑦2𝑡−1
(1.101) (-0.05) (-0.66)
In addition to the fact that all t-statistics are small, in what other ways is this regression
inadequate? What diagnostic checks would you want to perform? (2 + 1 + 2 = 5 points)

f. Estimate the equation: 𝑦2𝑡 = 𝑎0 + 𝑎2 . 𝑡 + 𝜇2 𝐷𝐿 and save the residuals. Perform a


Dickey-Fuller test on the saved residuals. Perform the appropriate diagnostic tests on this
regression to ensure that the residuals approximate white noise. You should conclude that
the series is a unit root process with a one-time pulse at t = 51. (3 + 2 + 3 + 1 = 9 points)
g. Re-estimate the model without the insignificant time trend. How is your answer affected?
(3 + 1 = 4 points)

6. Use the data in the file QUARTERLY.XLS to perform the following: (3 × 4 = 12 points)

a. Perform the DF-GLS test using 1 lagged change of the log of INDPROD. You should
find that the coefficient on 𝛾 is −2.04. (Be sure to include a time trend.)

b. Perform the DF-GLS test using eight lags of the change in UNEMP. You should find that
the coefficient on 𝛾 is −1.83.

c. The SBC indicates that only one lagged change of UNEMP is appropriate. Now perform
the DF-GLS test using 1-lagged change of UNEMP. In what important sense is your
answer quite different from that found in part b?

You might also like