P-range DCC: A Score-Driven Extension to Time-Varying Correlation Models

Abstract

We propose a score-driven extension to the well-known dynamic conditional correlation (DCC) model. The p-range DCC model provides means to capture the time-varying influence from news on correlation dynamics. The recursion to update the news parameter over time is based on the observations of past periods. By and large, the model increases the flexibility of DCC-type models whilst maintaining their appealing characteristics for applications with large cross-sections. We demonstrate that the model performs well in a variety of different situations and show that incorporation of the time-varying severity of news enriches the examination of correlation dynamics for a global cross-section of equity indices. More particularly, the article shows that the time-varying parameter can account for significant increases in equity return linkages in response to plunging markets amid the outbreak of COVID-19 in early 2020 and subsequent economic recoveries.