Title: Information flows and volatility: Level or persistence shifts
Authors: Daniel Borup - Aarhus University (Denmark) [presenting]
Kristoffer Pons Bertelsen - Aarhus University (Denmark)
Johan Jakobsen - Nordea (Denmark)
Abstract: The relationship between information flows and the dynamic properties of financial market volatility is examined. In order to do so, we formulate a smooth-transition realized GARCH framework where the level and persistence parameters are allowed to vary, possibly jointly, over time as a function of news arrival. We entertain the possibility of a non-linear relationship. The model dynamically controls for effects of realized volatility on the conditional variance process which otherwise may bias conclusions. Our findings on a broad panel of international stock market indices indicate a strong relationship between both the level and persistence with the strength of news arrival. Negatively loaded information flows matter in particular. This relationship is non-linear, showing a clear bell-shaped structure. That is, weak or strong information arrival characterize relatively low level and persistence states of volatility, whereas medium information flows characterize high level and persistence states of volatility. Our findings may be explained by the degree of ambiguity in the signals of incoming information and market participants' disagreement of their interpretation.