ECON416JN Econometrics II 2021
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ECON416JN
Econometrics II
2021
Section A
Question 1
Q1.1. Consider the following model:
yt = u + xt + zt
xt = "t + 9"t_1
zt = βzt_1 + ut ;
where "t is an independent and identically distributed (i.i.d.) process with mean zero and variance 72 ; and ut is also an i.i.d. process with mean zero and variance o2 : In addition, "t and ut are independent of each other.
(a.) What is the process for yt ?
[25 marks]
(b.) Give conditions to ensure yt is covariance stationary and invertible.
[25 marks]
Q1.2. Consider two time series x and y: Using what you learnt from the lecture notes:
(a.) State the conditions for the cointegration between x and y .
[25 marks]
(b.) Describe the steps to follow for Engle-Granger cointegration test.
[25 marks]
Question 2
(a.) Suppose we have estimated the following AR(2) model
Xt = 23:4 + 0:6Xt_1 - 0:2Xt_2 + ut ;
where ut is a white noise with mean zero and variance 1. In the data set, XT _1 = 50 and XT = 40.
Compute the forecast of X two periods ahead: X^T+2lT .
[25 marks]
(b.) For the model in part (a), compute the limit of the forecast X^T+hlT as h approaches
inÖnity: lim X^T+hlT :
二o
[25 marks]
(c.) Consider a random variable Y: Assume that using a given forecasting method, the
forecast errors produced for 3 observations are -1, -2, and -6. Compute the Mean absolute error of the forecast of Y .
[25 marks]
(d.) Consider the following MA(1) model:
xt = 0:1 + ut + 0:5ut_1 ;
where ut is a white noise with zero mean and variance equal to 1. Compute the point forecast of xt+h, say t+hlt ; for h 2 2; and the variance of the associated forecast error.
[25 marks]
Section B (It counts for 40% of the total mark) (Answer one and only one question from Section B)
Question 3
Consider the following model:
yt = ( 1 - 0:6L - 0:16L2 )xt + et ; (1)
where et are independent and identically distributed N(0;7 e(2)) and xt is an exogenous variable.
(a.) Classify the process in (1) and determine if it is stable.
[20 marks]
(b.) Calculate the multiplier impact or the short-run multiplier, m0 .
[20 marks]
(c.) Compute the impact on the endogenous variable at time t (yt ) of a unit change in the exogenous variable at time t - 2 (xt_2 ).
[20 marks]
(d.) Calculate the total multiplier or the long-run multiplier, mT .
[20 marks]
(e.) Calculate the mean and median lags.
[20 marks]
Question 4
Consider now the following trivariate (=three dimensional) VAR(1) model:
wt = u + ●wt_1 + "t t = 1; 2;:::;T
with wt = (xt ;yt ;zt )/ and "t ~ i:i:d: N(0; Q):
It is also known that each of the variables xt ; yt ; zt has at most one unit root.
(a.) Suppose that the Johansen testing procedure suggests that there are two cointe- grating relationships (r = 2). What would this imply for the eigenvalues of ●?
[35 marks]
(b.) What would be the number of common trends if there are two cointegrating rela- tionships?
[30 marks]
(c.) How would you identify these two cointegrating relationships?
[35 marks]
Section C
Question 5
(a.) Assume that the error term of a static regression model follows a GARCH(1,1)
process. Derive the 2-step ahead forecast of the conditional variance of the errors. [30 marks]
(b.) Now assume that the error term of a static regression model follows an ARCH(1)
process. Derive the 2-step ahead forecast of the conditional variance of the errors. [30 marks]
(c.) Assume that the error term of a static regression model follows a stable ARCH(1)
process. What is the long-run forecast of the conditional variance of the errors? [40 marks]
2022-04-22