Skip to main content
. 2021 Feb 5;43:101958. doi: 10.1016/j.frl.2021.101958

Table 3.

Estimates of the ARMA(0,1)-GARCH(1,1) model during COVID.

PAX Gold PAX Gold
Panel A: Mean Equation
c 0.00045
(0.00064)
-6.13E-05
(0.000101)
0.00048
(0.00068)
η1 (MA(1)) -0.25434⁎⁎⁎
(0.05041)
-0.83567⁎⁎⁎
(0.02980)
-0.09955
(0.06678)
DCOVID 6.31-05
(0.00083)
5.33E-05
(0.000103)
0.00096
(0.00104)
Panel B: Conditional Variance Equation
α0 4.07E-05
(3.03E-05)
4.91E-06⁎⁎
(2.50E-06)
8.15E-06
(5.74E-06)
α1 (ARCH) 0.39322
(0.26378)
0.18972⁎⁎⁎
(0.03662)
0.13456*
(0.07513)
β1 (GARCH) 0.58109⁎⁎⁎
(0.11950)
0.74074⁎⁎⁎
(0.02422)
0.77303⁎⁎⁎
(0.01873)
DCOVID 1.78E-05
(2.07E-05)
-4.89E-06⁎⁎
(2.50E-06)
1.37E-05
(1.08E-05)
Panel C: Goodness of Fit Statistics and Residual Diagnostics
AIC -5.86677 -8.52818 -6.11182
SIC -5.83265 -8.44235 -6.00316
Q(36) 33.441 24.324 32.839
ARCH LM test (prob) 0.1633 0.8399 0.7041
No of observations 363 363 263

*,**,***indicate significance at the 10%, 5% and 1% levels, respectively. Standard errors are in parentheses. c is the constant term, η1 is the coefficient of the ARMA(0,1) model, α0 is the constant intercept term, α1 and β1 capture the presence of heteroskedasticity in daily index return series, DCOVID represents the COVID-19 period and takes on the value of 1 if the day falls within the COVID-19 period and 0 otherwise