Statistical study of the impact of the basis effect on futures hedging: Evidence from China

Youhua Frank CHEN, Kwai Sun LEUNG, Kin Keung Eric POON, Yiwen SU

Research output: Contribution to journalArticles

Abstract

This article extends the works of Lien and Yang (2008a) in two different aspects. First, we enhance their bivariate GARCH dynamic conditional correlation (BGARCH-DCC) model by improving the efficiency of the estimation of the model parameters through the adoption of the Engle’s (2002) DCC specification. Second, other than the variance Lien and Yang (2008a) employed, we further investigate the effect of the basis using value at risk (VaR) as the hedging performance criterion for deriving the optimal hedge ratio. The corresponding hedge ratio is known as the zero-value at risk (zero-VaR) hedge ratio. Based on data on spot and futures prices of aluminum and copper from the Shanghai Futures Exchange (SHFE), we conduct the within-sample and out-of sample analysis. Under both hedging performance measures (variance and VaR) and hedge ratios (minimum variance and zero-VaR hedge ratios), the within-sample and out-of-sample tests both show that the basis effect, especially the asymmetric one, is important in improving of the hedging performance. Copyright © 2015 Youhua (Frank) Chen, Kwai Sun Leung and Kin Keung Poon.
Original languageEnglish
Pages (from-to)1219-1249
JournalApplied Mathematical Sciences
Volume10
Issue number25
Early online dateApr 2016
DOIs
Publication statusPublished - 2016

Citation

Chen, Y. F., Leung, K. S., Poon, K. K., & Su, Y. (2016). Statistical study of the impact of the basis effect on futures hedging: Evidence from China. Applied Mathematical Sciences, 10(25), 1219-1249.

Keywords

  • Asymmetric effect of basis
  • Bivariate GARCH
  • Dynamic conditional correlation
  • Zero-VaR hedge ratio
  • Dynamic hedging strategy

Fingerprint Dive into the research topics of 'Statistical study of the impact of the basis effect on futures hedging: Evidence from China'. Together they form a unique fingerprint.