Ioanid Rosu's webpage

Multi-Asset Market Making  

Abstract:
We study how risk-averse market makers manage inventory by setting quotes in multiple assets. As the correlation between assets increases, an inventory shock in one asset causes the market maker to quote more in other assets but less in the affected asset. The model yields five predictions. Cross-hedging exists: after an inventory shock in one asset, the dealer quotes more aggressively in correlated assets. Cross-hedging increases with fundamental correlation, while own-hedging decreases (a substitution effect). A more risk-averse dealer cross-hedges less, because she mean-reverts each position to zero so quickly that portfolio rebalancing has no time to be valuable. Through an inventory-driven price-pressure channel, high dealer risk aversion generates a term structure of return correlations that attenuates at higher frequencies: the Epps effect. Finally, ignoring cross-hedging leads to systematic underestimation of market maker risk aversion; in the limit of perfect correlation, the bias reaches 50%. We test all five predictions using complete message-level data from the Toronto Stock Exchange.