Agents’ Behavior on Multi-Dealerto-Client Bond Trading Platforms By Jean-David Fermanian, Oliver Gueant, and Arnaud Rachez Discussant: Zvi Wiener The Hebrew University of Jerusalem 9th Financial Risks International Forum The goal Use a unique database to describe the behavior of clients and dealers in a multi-dealer OTC market. The method Markov chain Monte Carlo method to estimate the distributions of the quotes provided by the dealers and reservation prices of the clients. 2 Corporate bond market Many similar products – there is typically no single and transparent book of orders. Dealers keep inventories and provide quotes only when requested by clients. Recently there are trading platforms that allow simultaneous request for quotes from several (1-6) dealers. 3 Assumptions Normal distribution of reservation values of clients. Skew exponential power distribution of dealer’s prices. Why not assuming that the dealers have a similar distribution but translate them to quoted prices strategically? 4 Buy orders quotes – red, reservation prices - green Sell orders quotes – red, reservation prices - green Competition The best quote distribution Impact of size Suggestions Strategic behavior of dealers: • Silent collusion in less profitable transactions • Competition in most profitable deals • “Punishment” for breaking collusion • Stronger collusion when a client needs to sell Technical remarks: • Check the impact of size!! • Regularization can speed the convergence. 10
© Copyright 2026 Paperzz