On the macro-drivers of realized volatility: the destabilizing impact of UK policy uncertainty across Europe

ABSTRACT

This paper studies the bivariate HEAVY system of daily and intra-daily volatility equations and its macro-augmented asymmetric power extension. We focus on economic drivers that exacerbate stock market volatility and can be proved to be major threats for financial stability. Our study proves the inflammatory effects of UK Policy Uncertainty alongside global credit and commodity factors that spread across European financial markets. This UK-led spillover phenomenon should be considered by world market participants and recognized, monitored and mitigated by policymakers amid the Brexit fears and the associated highly probable harm for Europe. Other findings are as follows. First, once we allow for power transformations, asymmetries, and macro-effects in the benchmark specification, it is found that both powered conditional variances are significantly affected by the powers of squared negative returns and realized measure, further improving the HEAVY framework's forecasting accuracy. Second, the structural breaks applied to the bivariate system capture the time-varying behavior of the parameters, in particular during the global financial crisis of 2007/08. Third, higher UK uncertainty levels increase the leverage and global macro-effects from credit and commodity markets on all European stock markets' realized volatilities.

KEYWORDS:

• Macro-financial linkages
• HEAVY model
• UK economic policy uncertainty
• high-frequency data
• realized variance
• asymmetries
• power transformations
• financial crisis
• structural breaks
• risk management

JEL CLASSIFICATIONS:

• C22
• C58
• D80
• E44
• G01
• G15

Disclosure statement

No potential conflict of interest was reported by the author(s).

ORCID

S. Yfanti http://orcid.org/0000-0001-8071-916X

Notes

1 This type of asymmetry was introduced by Glosten (1993).

2 The log-transformed series are always positive because all series' values are higher than one. Since the lower bound of our macro-regressors' series is not one but zero, we, alternatively, included the regressors divided by 100 (EPU, MOVE, WTI), 10000 (GSCI) and 10 (AAA, BAA). This resulted in similar estimated coefficients in terms of level and significance within the HEAVY framework (results available upon request).

3 Further research could consider an exponential HEAVY specification to address the non-negativity limitations.