Table 3.

Results from alternative GARCH models. This table reports the effect of ER on SR. The following time-series regression model (GARCH-M) is employed:

$S{R}_t=\alpha +\beta E{R}_t+\delta V_t+{\epsilon }_t$

In this regression, $S{R}_t$ is the Japanese stock market returns proxied using the Nikkei stock price index (log percentage returns are computed), $E{R}_t$ is the Yen–US dollar exchange rate such that an increase denotes a depreciation of the Japanese Yen, $V_t$ is the conditional variance and the model’s innovations, ${\epsilon }_t$, follow a Student $t$ ​ distribution, and ${\epsilon }_t=V_t{u}_t$. Following Bollerslev (1986), the conditional volatility is obtained as $V_t={\rho }_0+{\rho }_1{\epsilon }_{t-p}^2+{\rho }_2{V}_{t-q}$, where ${\rho }_s$ (s $=$ 0,1,2) are parameters to be estimated, the sum of non-intercept terms are less than one, and the GARCH-M orders, p and q, are set to (1,2), (2,1), and (2,2). Results from these models are in columns 2 to 4. An exponential GARCH (EGRACH) in mean version of the model is also estimated and results are reported in column 5. To obtain robust estimates of the standard errors we estimate the model using the quasi maximum likelihood function (QMLF) of Bollerslev and Wooldridge (1992). We only report the main slope coefficient relating to $\beta$$=$ 0 for three sample periods: COVID-19 sample (31/12/2019 to 17/8/2020); pre-COVID-19 Sample A (04/1/2010 to 30/12/2019); and pre-COVID-19 Sample B (31/12/2018 to 16/8/2019). And, we only consider the full-scale model—that is, the model that includes all controls. Standard errors are reported in parenthesis.

Sub-samples	GARCH (2,1)	GARCH (1,2)	GARCH (2,2)	EGARCH (1,1)
COVID-19	1.6731⁎⁎⁎ (0.2494)	1.6779⁎⁎⁎ (0.2383)	1.5344⁎⁎⁎ (0.1950)	1.4911⁎⁎⁎ (0.2372)
	$R^2=7.02\text{\%}$	$R^2=4.63\text{\%}$	$R^2=4.25\text{\%}$	$R^2=9.34\text{\%}$
Pre-COVID-19 Sample A	1.3218⁎⁎⁎ (0.0439)	1.3223⁎⁎⁎ (0.0439)	1.3216⁎⁎⁎ (0.0438)	1.2691⁎⁎⁎ (0.0417)
	$R^2=31.09\text{\%}$	$R^2=31.08\text{\%}$	$R^2=35.05\text{\%}$	$R^2=31.07\text{\%}$
Pre-COVID-19 Sample B	1.5269⁎⁎⁎ (0.1832)	1.5093⁎⁎⁎ (0.1885)	1.5049⁎⁎⁎ (0.2014)	1.3607⁎⁎⁎ (0.1742)
	$R^2=35.17\text{\%}$	$R^2=35.64\text{\%}$	$R^2=35.46\text{\%}$	$R^2=38.86\text{\%}$

^***Denotes statistical significance at the 1% level.