Exercise 14.
1
f)
r_ftse
The results for a GARCH(1,1) model:
Comment: The estimate of δ and the coefficient of the γ 1term is significant and positive.
The results for a GARCH(6,6) model:
Comment: All the paramaters are insignificant. So, this is not an appropriate model
The results for a GARCH(1,2) model:
Comment: Only the ARCH(1) and the GARCH(1) terms is insignificant.
=> GARCH(1,1) is the most appropriate model.
r_stock1
The results for a GARCH(1,1) model:
Comment: The estimate of δ and the coefficient of the γ 1term is significant and positive
The results for a GARCH(1,2) model:
Comment: Only the ARCH(1) and the GARCH(1) terms is insignificant.
=> GARCH(1,1) is the most appropriate model.
r_stock2
The results for a GARCH(6,6) model:
Comment: Some of the ARCH lagged terms have a negative sign. It is not an appropriate
model.
The results for a GARCH(1,1) model:
Comment: The estimate of δ and the coefficient of the γ 1term is significant and positive.
GARCH(1,1) is the most appropriate model.
r_stock3
The results for a GARCH(6,6) model:
Comment: Some of the ARCH lagged terms have a negative sign. It is not an appropriate
model.
The results for a GARCH(1,2) model:
Comment: The coefficient of ARCH(2) have a negative sign. It is not an appropriate model.
The results for a GARCH(1,1) model:
Comment: The estimate of δ and the coefficient of the γ 1term is significant and positive
The results for a GARCH(2,2) model:
Comment: The coefficient of ARCH(2) and GARCH(2) have a negative sign. It is not an
appropriate model.
=> GARCH(1,1) is the most appropriate model.
g)
r_ftse
The results for a TGARCH(1,1) model:
Comment: The coefficient of TGARCH term is positive and statiscally significant.
=> There are asymmetrics effects. And bad news has larger effects on the volatility of the
series than good news.
r_stock1
The results for a TGARCH(1,1) model:
Comment: The coefficient of TGARCH term is negative and statiscally significant.
=> There are asymmetrics effects. And bad news has smaller effects on the volatility of the
series than good news.
r_stock2
The results for a TGARCH(1,1) model:
Comment: The coefficient of TGARCH term is positive and statiscally significant.
=> There are asymmetrics effects. And bad news has larger effects on the volatility of the
series than good news.
r_stock3
The results for a TGARCH(1,1) model:
Comment: The coefficient of TGARCH term is positive and statiscally significant.
=> There are asymmetrics effects. And bad news has larger effects on the volatility of the
series than good news.
h)
r_ftse
The results for a EGARCH(1,1) model:
Comment: The coefficient of the EGARCH is negative and statiscally significant.
=> For the ftse,p bad news has larger effects on the volatility of the series than good news.
r_stock1
The results for a EGARCH(1,1) model:
Comment: The coefficient of the EGARCH is possitive and statiscally significant.
=> For the stock1, bad news has smaller effects on the volatility of the series than good
news.
r_stock2
The results for a EGARCH(1,1) model:
Comment: The coefficient of the EGARCH is negative and statiscally significant.
=> For the stock2, bad news has larger effects on the volatility of the series than good news.
r_stock3
The results for a EGARCH(1,1) model:
Comment: The coefficient of the EGARCH is negative and statiscally significant.
=> For the stock3, bad news has larger effects on the volatility of the series than good news.
Exercise 14.3
a)
The dm series:
The bp series:
The cd series:
The dy series:
The sf series:
There are volatility clustering phenomenon in the graphs.
b)
DM
The results of an AR(1) model:
=> These results are no of interest in themselves.
Test for ARCH effects:
The T* R2 statistic = 19.09 and has a probability value of 0.000 => We reject H0 of
homoskedasticity. We conclude that ARCH(1) effects are presented.
Testing for higher order ARCH effects:
This time the T* R2 statistic is even higher (79.51), suggesting a massive rejection of H0.
Estimating ARCH models:
ARCH(6) is the most appropriate model.
BP
The results of an AR(1) model:
=> These results are no of interest in themselves.
Test for ARCH effects:
The T* R2 statistic = 58.91 and has a probability value of 0.000 => We reject H0 of
homoskedasticity. We conclude that ARCH(1) effects are presented.
Testing for higher order ARCH effects:
This time the T* R2 statistic is even higher (98.07), suggesting a massive rejection of H0.
Estimating ARCH models:
ARCH(6) is the most appropriate model.
CD
The results of an AR(1) model:
=> These results are no of interest in themselves.
Test for ARCH effects:
The T* R2 statistic = 84.2 and has a probability value of 0.000 => We reject H0 of
homoskedasticity. We conclude that ARCH(1) effects are presented.
Estimating ARCH models:
ARCH(5) is the most appropriate model.
DY
The results of an AR(1) model:
=> These results are no of interest in themselves.
Test for ARCH effects:
The T* R2 statistic = 16.17 and has a probability value of 0.000 => We reject H0 of
homoskedasticity. We conclude that ARCH(1) effects are presented.
Testing for higher order ARCH effects:
This time the T* R2 statistic is even higher (20.47), suggesting a massive rejection of H0.
Estimating ARCH models:
ARCH(6) is the most appropriate model.
SF
The results of an AR(1) model:
=> These results are no of interest in themselves.
Test for ARCH effects:
The T* R2 statistic = 16.17 and has a probability value of 0.000 => We reject H0 of
homoskedasticity. We conclude that ARCH(1) effects are presented.
Testing for higher order ARCH effects:
This time the T* R2 statistic is even higher (36.24), suggesting a massive rejection of H0.
Estimating ARCH models:
ARCH(6) is the most appropriate model.
c)
DM
The results for a GARCH(1,1) model:
Comment: The estimate of δ and the coefficient of the γ 1term is significant and positive.
GARCH(1,3):
Comment: Only the ARCH(1) and the GARCH(1) terms is insignificant.
=> GARCH(1,1) is the most appropriate model.
BP
=> GARCH(1,1) is the most appropriate model.
CD
=> GARCH(1,1) is the most appropriate model.
DY
=> GARCH(1,1) is the most appropriate model.
SF
=> GARCH(1,1) is the most appropriate model.
