A three-dimensional asymmetric power HEAVY model

Stavroula Yfanti; Georgios Chortareas; Menelaos Karanasos; Emmanouil Noikokyris

doi:10.1002/ijfe.2296

A three-dimensional asymmetric power HEAVY model

Stavroula Yfanti^*, Georgios Chortareas, Menelaos Karanasos, Emmanouil Noikokyris

^*Corresponding author for this work

Economics

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

This article proposes the three-dimensional HEAVY system of daily, intra-daily, and range-based volatility equations. We augment the bivariate model with a third volatility metric, the Garman–Klass estimator, and enrich the trivariate system with power transformations and asymmetries. Most importantly, we derive the theoretical properties of the multivariate asymmetric power model and explore its finite-sample performance through a simulation experiment on the size and power properties of the diagnostic tests employed. Our empirical application shows that all three power transformed conditional variances are found to be significantly affected by the powers of squared returns, realized measure, and range-based volatility as well. We demonstrate that the augmentation of the HEAVY framework with the range-based volatility estimator, leverage and power effects improves remarkably its forecasting accuracy. Finally, our results reveal interesting insights for investments, market risk measurement, and policymaking.

Original language	English
Journal	International Journal of Finance and Economics
DOIs	https://doi.org/10.1002/ijfe.2296
Publication status	Accepted/In press - 2020

Keywords

asymmetries
HEAVY model
high-frequency data
power transformations
realized volatility
risk management

Access to Document

10.1002/ijfe.2296

Cite this

@article{1797c72aed714a42b8ebe72ff1ee40af,

title = "A three-dimensional asymmetric power HEAVY model",

abstract = "This article proposes the three-dimensional HEAVY system of daily, intra-daily, and range-based volatility equations. We augment the bivariate model with a third volatility metric, the Garman–Klass estimator, and enrich the trivariate system with power transformations and asymmetries. Most importantly, we derive the theoretical properties of the multivariate asymmetric power model and explore its finite-sample performance through a simulation experiment on the size and power properties of the diagnostic tests employed. Our empirical application shows that all three power transformed conditional variances are found to be significantly affected by the powers of squared returns, realized measure, and range-based volatility as well. We demonstrate that the augmentation of the HEAVY framework with the range-based volatility estimator, leverage and power effects improves remarkably its forecasting accuracy. Finally, our results reveal interesting insights for investments, market risk measurement, and policymaking.",

keywords = "asymmetries, HEAVY model, high-frequency data, power transformations, realized volatility, risk management",

author = "Stavroula Yfanti and Georgios Chortareas and Menelaos Karanasos and Emmanouil Noikokyris",

year = "2020",

doi = "10.1002/ijfe.2296",

language = "English",

journal = "International Journal of Finance and Economics",

issn = "1076-9307",

publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - A three-dimensional asymmetric power HEAVY model

AU - Yfanti, Stavroula

AU - Chortareas, Georgios

AU - Karanasos, Menelaos

AU - Noikokyris, Emmanouil

PY - 2020

Y1 - 2020

N2 - This article proposes the three-dimensional HEAVY system of daily, intra-daily, and range-based volatility equations. We augment the bivariate model with a third volatility metric, the Garman–Klass estimator, and enrich the trivariate system with power transformations and asymmetries. Most importantly, we derive the theoretical properties of the multivariate asymmetric power model and explore its finite-sample performance through a simulation experiment on the size and power properties of the diagnostic tests employed. Our empirical application shows that all three power transformed conditional variances are found to be significantly affected by the powers of squared returns, realized measure, and range-based volatility as well. We demonstrate that the augmentation of the HEAVY framework with the range-based volatility estimator, leverage and power effects improves remarkably its forecasting accuracy. Finally, our results reveal interesting insights for investments, market risk measurement, and policymaking.

AB - This article proposes the three-dimensional HEAVY system of daily, intra-daily, and range-based volatility equations. We augment the bivariate model with a third volatility metric, the Garman–Klass estimator, and enrich the trivariate system with power transformations and asymmetries. Most importantly, we derive the theoretical properties of the multivariate asymmetric power model and explore its finite-sample performance through a simulation experiment on the size and power properties of the diagnostic tests employed. Our empirical application shows that all three power transformed conditional variances are found to be significantly affected by the powers of squared returns, realized measure, and range-based volatility as well. We demonstrate that the augmentation of the HEAVY framework with the range-based volatility estimator, leverage and power effects improves remarkably its forecasting accuracy. Finally, our results reveal interesting insights for investments, market risk measurement, and policymaking.

KW - asymmetries

KW - HEAVY model

KW - high-frequency data

KW - power transformations

KW - realized volatility

KW - risk management

UR - http://www.scopus.com/inward/record.url?scp=85092328327&partnerID=8YFLogxK

U2 - 10.1002/ijfe.2296

DO - 10.1002/ijfe.2296

M3 - Article

AN - SCOPUS:85092328327

SN - 1076-9307

JO - International Journal of Finance and Economics

JF - International Journal of Finance and Economics

ER -

A three-dimensional asymmetric power HEAVY model

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this