Scribd
Upload
4 views2 pages

Advanced Econometrics VAR Model Analysis

The document outlines a numerical assignment in advanced econometrics involving a data generating process with three variables and their relationships defined by VAR models. It includes tasks such as estimating lag order using AIC and BIC, applying different estimation methods, testing Granger causality, incorporating prior information into the model, and analyzing the impact of changes in error term variance and autocorrelation. Additionally, it requires estimating a VAR(3) model and analyzing impulse responses to shocks in the variables.

Uploaded by

Yuntong Wu
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views2 pages

Advanced Econometrics VAR Model Analysis

The document outlines a numerical assignment in advanced econometrics involving a data generating process with three variables and their relationships defined by VAR models. It includes tasks such as estimating lag order using AIC and BIC, applying different estimation methods, testing Granger causality, incorporating prior information into the model, and analyzing the impact of changes in error term variance and autocorrelation. Additionally, it requires estimating a VAR(3) model and analyzing impulse responses to shocks in the variables.

Uploaded by

Yuntong Wu
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Advanced Econometrics: Numerical Assignment

Consider a data generating process with three variables as follow:

yt = A1 yt−1 + A2 yt−2 + ut ,

where yt = (y1t , y2t , y3t ),


   
−0.5 0.2 0 −0.3 0.1 0
   
A1 = 
 0.3 0.4 0.3  A2 = 
 0.1 0.4 0.2 ,
0.2 0.1 −0.4 0.2 0.1 −0.3

and each element in the error term vector ut = (u1t , u2t , u3t ) follows a standard normal
distribution with zero mean and correlated across equations with correlation of 0.3 at each
time t, but there is no serial correlation across time, say ut is not correlated with ut+1 . The
initial value of yt can be generated also from a standard normal distribution. We generate
500 observations, namely t = 1, . . . , 500.

(a) You decide to estimate this generated data set using VAR models. Use AIC and BIC
to determine the lag order of the VAR model in the range between 1 to 5.

(b) Estimate the generated data with a VAR(2) model using equation-by-equation OLS,
system OLS, and system FGLS. Do the three methods lead to the same estimates of A1
and A2 ?

(c) Test whether y3t Granger causes y1t , and whether y1t Granger causes y3t .

(d) If we know priorly that the effect of y2,t−1 on y3,t is the same as the effect of y3,t−1
on y2,t . How do you incorporate this information in your estimation? Re-estimate the
model using this information.

(e) Increase the variance of error terms, explain how this change affects the accuracy of
your estimates. Increase the autocorrelation of error terms, explain how this affects the
accuracy of your estimates.

1
(f) Estimate the data with VAR(3) using your preferred method, test for the significance
of A
b3 = 0.

(g) Analyse the impulse response if one imposes a one-unit shock on the error of the third
variable y3t . Is the assumption of orthogonal shocks realistic in this case?

You might also like