The Propagation of Shocks Across International Equity Markets: A Microstructure Perspective

Abstract

We study the high-frequency propagation of shocks across international equity markets. We identify intraday shocks to stock prices, liquidity, and trading activity for 12 equity markets around the world based on non-parametric jump statistics at the 5-minute frequency from 1996 to 2011. Shocks to prices are prevalent and large, with regular spillovers across markets -even within the same 5-minute interval. We find that price shocks are predominantly driven by information rather than liquidity. Consistent with the information channel, price shocks do not revert and often occur around macroeconomic news announcements. Liquidity shocks tend to be isolated events that are neither associated with price shocks nor with liquidity shocks on other markets.Our results challenge the widespread view that liquidity plays an important role in the origination and propagation of financial market shocks.

Financial crises, price discovery, and information transmission: a high-frequency perspective

“…Brenner et al (2009), who study responses of U.S. stock, Treasury, and bond markets to U.S. macroeconomic surprises, also find fundamentals-driven interrelations on a daily basis. This finding is in line with Bongaerts et al (2014) and Füss et al (2017), who document that jumps and co-jumps in prices are often found in relation to U.S. macroeconomic announcements.…”

Modelling illiquidity spillovers with Hawkes processes: an application to the sovereign bond market

“…In fact most of the research concentrates either on the equity market or on bond liquidity at daily or weekly frequency. A recent study by Bongaerts et al (2015) directly identifies liquidity shocks in different equity markets at five-minute resolution and does not find any evidence for spillovers in time nor across markets, challenging the concepts of liquidity dry-ups and contagion of liquidity shocks.…”

“…Common to these detection methods is that they are effectively carried out on discrete time intervals, even when highfrequency data are available. For example Chavez-Demoulin and McGill (2012) samples stock-returns over 15-min periods, Bongaerts et al (2015) over 5 minutes or Bormetti et al (2015) at 1-minute frequency. However the discrete time grid might miss events that occur at shorter time-scales and, when modelled as Hawkes processes, forces discrete time on the originally continuous Hawkes process.…”

“…We observe that illiquidity becomes more interconnected throughout our sample and the proportion of illiquidity contagion has roughly doubled from 2011 to 2015. Illiquidity spillover typically occurs within a few seconds (much faster than the sampling interval of 5 min in Bongaerts et al (2015)) and is faster in bonds that are related by a similar maturity. It is notable that trades are very sparse on MTS and our results can be seen as directly due to quoting activity, supporting a channel of 'informativeness' as in Cespa and Foucault (2014).…”

Something in the air: Information density, news surprises, and price jumps

“…The critical value is based on the 99th-percentile of a Gumbel distribution. We follow previous empirical papers on jump identification and choose a return frequency of five minutes to avoid most market microstructure noise (e.g., Evans (2011); Jiang, Lo, and Verdelhan (2011); Bongaerts, Roll, Rösch, Van Dijk, and Yuferova (2014)). This ensures that effects such as highfrequency trading, short-term anomalies, or bid-ask bounces will not contaminate our results.…”