ISyE 6402 Midterm Prep
1. Getting a 3 variable VAR model first matrix: first row are coefficients for
from summary(model) output of Xt1, second row are coefficients for Xt2,
a VAR(1) model etc...
second matrix is Xt-1, i b/c this is a VAR(1)
model
last matrix are the constants
eta_t is covariance matrix, direct copy
2. (c) Based on the fitted mod- contemporaneous cross-correlation is
el, is there contemporane- NOT present if the variance-covariance
ous cross-correlation? Is there matrix is a diagonal matrix
lagged cross-correlation? Is there is lagged correlation if the order p of
there lagged auto-correlation? the VAR(p) model > 0
Explain.
3. T/F - Differencing the data might True
not make the series stationary
in the presence of cointegra-
tion.
4. Cointegration and long-run See image
equilibrium
, ISyE 6402 Midterm Prep
5. Does cov(x,x) = var(x)? You betcha
6. Autocovariance T/F see image
7. T/F - The AR(1) process is FALSE! the absolute value of phi must lie
causal if and only if the autore- b/w -1 and 1
gressive parameter phi is be-
tween 0 and 1. However, it is al-
ways invertible.
8. T/F - A linear process is a spe- FALSE - the moving average is a special
cial case of the moving average case of a linear process.
model.
9. T/F - A guassian time series is false - guassian processes can have vary-
always stationary ing means
10. T/F 'In autoregressive models FALSE - there are no analogies of depen-
the current value of dependent dent/independent variables w/ AR models,
variable is influenced by past as there are w/ regression models
values of both dependent and
independent variables.'
11. in AR models the current val-
ue of the dependent variable is
1. Getting a 3 variable VAR model first matrix: first row are coefficients for
from summary(model) output of Xt1, second row are coefficients for Xt2,
a VAR(1) model etc...
second matrix is Xt-1, i b/c this is a VAR(1)
model
last matrix are the constants
eta_t is covariance matrix, direct copy
2. (c) Based on the fitted mod- contemporaneous cross-correlation is
el, is there contemporane- NOT present if the variance-covariance
ous cross-correlation? Is there matrix is a diagonal matrix
lagged cross-correlation? Is there is lagged correlation if the order p of
there lagged auto-correlation? the VAR(p) model > 0
Explain.
3. T/F - Differencing the data might True
not make the series stationary
in the presence of cointegra-
tion.
4. Cointegration and long-run See image
equilibrium
, ISyE 6402 Midterm Prep
5. Does cov(x,x) = var(x)? You betcha
6. Autocovariance T/F see image
7. T/F - The AR(1) process is FALSE! the absolute value of phi must lie
causal if and only if the autore- b/w -1 and 1
gressive parameter phi is be-
tween 0 and 1. However, it is al-
ways invertible.
8. T/F - A linear process is a spe- FALSE - the moving average is a special
cial case of the moving average case of a linear process.
model.
9. T/F - A guassian time series is false - guassian processes can have vary-
always stationary ing means
10. T/F 'In autoregressive models FALSE - there are no analogies of depen-
the current value of dependent dent/independent variables w/ AR models,
variable is influenced by past as there are w/ regression models
values of both dependent and
independent variables.'
11. in AR models the current val-
ue of the dependent variable is
