Bid-ask spreads in multiple dealer settings: Some experimental evidence

We report the results of an experiment designed to investigate the behavior of quoted spreads in multiple-dealer markets. We manipulate verbal communication (not allowed and allowed) and order preferencing (not allowed, allowed, and allowed with order-flow payment) between eighteen sessions. Without preferencing, spreads are wider when communication is allowed. With preferencing (and no order-flow payments), individuals do not have incentives to narrow the spread and a wide spread may be maintained without a collusive agreement. However, spreads narrow somewhat when individuals are given the opportunity to compete using alternatives to price (that is, payment for order flow).