Volatility Spillovers in a Long-Memory VAR: an Application to Energy Futures Returns

In this paper, we assess volatility spillovers across energy markets accounting for the persistence of the volatility series. To do so, we compute Diebold and Yilmaz (2015) measures of connectedness based on the forecast-error variance decomposition of an estimated fractionally integrated VAR (FIVAR). We use this method to study volatility spills among oil, unleaded gasoline, heating oil, and natural gas. Our main empirical findings are: 1) Accounting for persistence is essential to assess the magnitude of the spillover effects in these markets; 2) The traditional VAR magnifies the other’s contribution to the volatility variance; 3) There are substantial spillover effects across petroleum markets, but the link between these markets and the natural gas market appears to be broken in post 2008-crisis data. Keywords: fractional integration, spillovers, energy commodities. JEL Classification: G1, C5, Q