Working Paper: CEPR ID: DP13880
Authors: Péter Kondor; Gábor Pinter
Abstract: We propose a new measure of private information in decentralised markets-- connections -- defined as the number of dealers with whom a clienttrades in a time period. Using proprietary data for the UK governmentbond market, we show that clients have systematically better performancewhen having more connections, and this effect is stronger during macroeconomicannouncements. Time-variation in market-wide connections also helpsexplain yield dynamics. Given our novel measure, we present two applicationssuggesting that (i) dealers pass on information, acquired from theirinformed clients, to their subsidiaries, and (ii) informed clientsbetter predict the order-flow intermediated by their dealers.
Keywords: private information; client-dealer connections; government bond market
JEL Codes: G12; G14; G24
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Proportion of connected clients (C69) | Performance of subsidiaries (L25) |
| Number of connections (C39) | Trading performance (G19) |
| Time variation in market-wide client connections (G14) | Yield dynamics in the government bond market (E43) |
| More connected clients (D85) | Ability to predict future order flow (C69) |