BS3551 Econometrics Spring 2020/21
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
BS3551
Econometrics
Spring 2020/21
SECTION A
1. Briefly state the concept and importance of THREE of the following terms:
a) Central Limit Theorem
b) Power of a test
c) Likelihood Ratio test
d) Omitted variable bias
(25 marks)
SECTION B
2. Explain what is meant when it is said that a time series is (i) weakly stationary, (ii) ergodic, and (iii) invertible.
(5 marks)
(b) Explain what is meant by Wold decomposition theorem. Briefly discuss the
importance of this theorem for the modelling of macroeconomic and financial time series.
(5 marks)
(c) Consider the following two processes
i) = + 1 − 1 + − 2 + ) = + − 1 + + 1 − 1
where is white noise error, state the conditions under which the two processes are weakly stationary; state the conditions under which the second process is invertible.
(5 marks)
(d) Derive the mean, variance, autocovariance and autocorrelation of the ARMA(1,1) process defined in (c) ii) above.
(10 marks)
3. Consider the autoregressive model = 1.8 + 0.9 − 1 + where is iid N(0, 4)
(a) What is the unconditional mean of ? What is the unconditional distribution
of ?
(5 marks)
(b) If = 20, what is the one-step-ahead forecast of +1? what is the one-step-
ahead conditional distribution of +1 ? What is the one-step ahead 95% prediction confidence interval?
(10 marks)
(c) if = 20, what is the five-step-ahead forecast of +5?
(5 marks)
(d) Give the definition of Root Mean Squared Error (RMSE)? Suppose model 1 is an AR(1) and model 2 is an ARMA(1,1) , the RMSE of model 1 is bigger than the RMSE of model 2. Interpret the result.
(5 marks)
4. Suppose we are estimating a model for the return on a bond of the AR-GARCH form,
= + − 1 + , ~ 0, ℎ
ℎ = + − 1 + ℎ − 1
where ℎ = − 1 ( )2 is the conditional variance of and ℎ which follows a
(a) What is stationarity condition for the process of ? What is the stationarity
condition for the process of ℎ ?
(5 marks)
(b) Assuming the stationarity conditions are satisfied, what are the values of the
following quantities?
(i) ( +1) (ii) ( +1). (iii) ( +1) (iv) ( +1)
(v) ( +1)
(10 marks)
(c) (i) If you believed that the variance of affects the return on the bond, how would you adapt the GARCH model to allow for this?
(ii) If you suspected the leverage effect, i.e. an unexpected price drop
increases volatility more than an analogous unexpected price increase, is present in the bond return series, how would you adapt the GARCH model to allow for this?
(5 marks)
(d) If you were investigating a model such as the capital asset pricing model which uses the covariance between the market return and the bond return , how could the GARCH model be extended to allow for this case?
(5 marks)
5. Answer the following questions
(a) A time series follows the process ,
= + − 1 +
where is a white noise error. Given an initial condition 0 , find the solution for , indicating any deterministic or stochastic trend components.
(5 marks)
(b) Describe the following time series processes: 1) trend stationary model 2)
difference stationary model. Why must we distinguish between them?
(5 marks)
(c) What is the purpose of the Dickey Fuller test? Describe this test, and how you would apply it in practice. How does the Augmented Dickey Fuller test differ and when is it more appropriate?
(10 marks)
(d) Explain why it is important to explore the time series properties of macroeconomic variables as a first step in empirical work. Why is visual inspection not sufficient?
(5 marks)
6. For two nonstationary I(1) time series sequences and , t=1,2,…,T, where T is a large number, answer the following questions:
(a) After removing the time trends from the two processes (if there exists any),
do you expect their graphs to have any particular patterns? Why?
(7 marks)
(b) If the two processes are cointegrated, what is the economic implication
therein?
(8 marks)
(c) How can one test whether xt and yt are cointegrated? Would the test statistic follow any standard distribution? If not, design a Monte Carlo experiment to obtain the proper asymptotic critical values?
(10 marks)
Consider the VAR model
2. = 2 + 21 1, − 1 + 22 2, − 1 + 2,
where = ( 1, 2)' and ( ' ) = , a positive definite matrix.
(a) Explain the essential difference between the single equation and the vector
autoregressive approaches to model.
(6 marks)
(a) Define Granger Causality and explain what condition is needed for 1, not to
Granger cause 2, . How do you test this condition?
(7 marks)
(b) Suppose 1, is the inflation and 2, is the interest rate, the figure below is
the impulse response function of inflation and interest rate system. Comment briefly on this Impulse Response Function.
Please Turn Over
( 10 marks)
2022-06-11