My final aim is to derive the time series data of GARCH (1,1) volatility values over time corresponding to my existing return data.
To do this as a first step, I am trying to derive the volatility (i.e., sigma) value from the GARCH (1,1) model I run from the code "ugarchfit" in R.
From here, I know how to take out the optimal parameters coefficient estimates by using the command "coef()"
However, I am not able to pull out their corresponding p-values as well.
Is there an automatic way to do so? Or is there an alternative way to produce the same corresponding p-values if I am able to pull out the corresponding standard errors or t-test values instead?
The below is the full output I get.
*---------------------------------*
* GARCH Model Fit *
*---------------------------------*
Conditional Variance Dynamics
-----------------------------------
GARCH Model : sGARCH(1,1)
Mean Model : ARFIMA(0,0,0)
Distribution : norm
Optimal Parameters
------------------------------------
Estimate Std. Error t value Pr(>|t|)
mu -0.001183 0.008642 -0.13686 0.891143
omega 0.006368 0.000713 8.92639 0.000000
alpha1 0.334860 0.150390 2.22662 0.025973
beta1 0.000000 0.187089 0.00000 1.000000
Robust Standard Errors:
Estimate Std. Error t value Pr(>|t|)
mu -0.001183 0.009319 -0.12692 0.899008
omega 0.006368 0.002321 2.74405 0.006069
alpha1 0.334860 0.175326 1.90993 0.056142
beta1 0.000000 0.259065 0.00000 1.000000
LogLikelihood : 90.52192
Information Criteria
------------------------------------
Akaike -1.8607
Bayes -1.7518
Shibata -1.8642
Hannan-Quinn -1.8167
Weighted Ljung-Box Test on Standardized Residuals
------------------------------------
statistic p-value
Lag[1] 0.1237 0.7251
Lag[2*(p+q)+(p+q)-1][2] 0.1237 0.9033
Lag[4*(p+q)+(p+q)-1][5] 2.2145 0.5687
d.o.f=0
H0 : No serial correlation
Weighted Ljung-Box Test on Standardized Squared Residuals
------------------------------------
statistic p-value
Lag[1] 1.158 0.2819
Lag[2*(p+q)+(p+q)-1][5] 1.868 0.6500
Lag[4*(p+q)+(p+q)-1][9] 3.152 0.7337
d.o.f=2
Weighted ARCH LM Tests
------------------------------------
Statistic Shape Scale P-Value
ARCH Lag[3] 0.5813 0.500 2.000 0.4458
ARCH Lag[5] 0.7021 1.440 1.667 0.8228
ARCH Lag[7] 0.9818 2.315 1.543 0.9165
Nyblom stability test
------------------------------------
Joint Statistic: 0.7391
Individual Statistics:
mu 0.09335
omega 0.18718
alpha1 0.07268
beta1 0.29205
Asymptotic Critical Values (10% 5% 1%)
Joint Statistic: 1.07 1.24 1.6
Individual Statistic: 0.35 0.47 0.75
Sign Bias Test
------------------------------------
t-value prob sig
Sign Bias 0.25257 0.8012
Negative Sign Bias 0.08968 0.9287
Positive Sign Bias 0.22847 0.8198
Joint Effect 0.48299 0.9226
Adjusted Pearson Goodness-of-Fit Test:
------------------------------------
group statistic p-value(g-1)
1 20 34.53 0.01591
2 30 37.00 0.14622
3 40 51.09 0.09308
4 50 49.47 0.45423
If modelfit is your model (i.e. modelfit <- ugarchfit(...)), then the p-values for the optimal parameters are in
modelfit@fit$matcoef[,4]
or for the robust parameters...
modelfit@fit$robust.matcoef[,4]