On equity risk prediction and tail spillovers

• DOI: http://doi.org/10.1002/ijfe.1594
• Published date: 01 October 2017
• Author: Panos Pouliasis,Ioannis Kyriakou,Nikos Papapostolou
• Date: 01 October 2017

Received: 7 December 2016 Revised: 25 September 2017 Accepted: 25 September 2017

DOI: 10.1002/ijfe.1594

RESEARCH ARTICLE

On equity risk prediction and tail spillovers

Panos Pouliasis Ioannis Kyriakou Nikos Papapostolou

Cass Business School, City, University of

London, 106 Bunhill Row,London EC1Y

8TZ, UK

Correspondence

Panos K. Pouliasis, Cass Business School,

City, University of London, 106 Bunhill

Row,London EC1Y 8TZ, UK.

Email: p_pouliasis@city.ac.uk

Abstract

This paper studies the impact of modelling time-varying variances of stock

returns in terms of risk measurement and extreme risk spillover.Using a general

class of regime-dependent models, we find that volatility can be disaggregated

into distinct components: a persistent stable process with low sensitivity to

shocks and a high volatility process capturing rather short-lived rare events.

Out-of-sample forecasts show that, once regime shifts are accounted for, accu-

racy is improved compared to the standard generalized autoregressive condi-

tional heteroscedasticity or the historical volatility model. Volatility plays an

important role in controlling and monitoring financial risks. Therefore, by

means of a risk management application, we illustrate the economic value and

the practical implications of risk control ability of the models in terms of value

at risk. Finally, tests for predictability in co-movements in the tails of stock

index returns suggest that large losses are strongly correlated, supporting asym-

metric transmission processes for financial contagion in the left tail of return

distributions, whereas contagion in reverse direction (gains) is weak.

KEYWORDS

causality in risk, forecasting, regime volatility, risk spillover, stock markets, value at risk

1INTRODUCTION

Volatility is of great concern to economic agents involved

in the decision-making process under uncertainty. The

traditional framework for modelling volatility is the

generalized autoregressive conditional heteroscedastic-

ity (GARCH) process, pioneered by Engle (1982) and

Bollerslev (1986). This model class captures the salient

features of financial time series, such as volatility cluster-

ing, persistence, non-linear dependence, and thick tails.

Yetthe value of volatility lies in the capability of the model

to predict market fluctuations, contributing, inter alia, to

risk exposure evaluation, stress testing, asset allocation,

derivatives pricing, and risk management. This corrobo-

rates the importance of developing volatility models able

to replicate the salient features of financial time series.

Regardless of its interesting properties, the common

GARCH model has several shortcomings. Among others,

Bollerslev (1987) and Baillie and Bollerslev (1989) find

that the observed non-normalities in return distributions

are more pronounced than those implied by GARCH.

In effect, the model fails to reproduce skewed uncondi-

tional distributions or time variability in higher moments,

unless explicitly modelled (see Harvey & Siddique, 1999).

As a result, a number of variants have been put for-

ward accounting for asymmetries (e.g., leverage effect;

see Glosten, Jagannathan, & Runkle, 1993), long memory

(Baillie, Bollerslev, & Mikkelsen, 1996), and non-normal

densities (Bollerslev, 1987; Politis, 2004), whereas other

developments make use of non-parametric structures for

which a priori knowledge of the innovation distribution

is not required (Bühlmann & McNeil, 2002). Another

Int J Fin Econ. 2017;22:379–393. wileyonlinelibrary.com/journal/ijfe Copyright © 2017 John Wiley & Sons, Ltd. 379