Bilateral contracts invading an experimental double auction: does privacy decrease economic efficiency?

Abs tract: This paper investigates the private incentives to trade bilaterally rather than using a centralized double auction. These incentives arise in many economic models. For example, if agents like to avoid price-competition on quotes or are bargaining over many units with multiple counterparties then prices, quotes and quantities are sensitive. Privacy is notoriously hard to study empirically but the experimental methodology chosen allows us to observe both market and non-market behavior. I find that private bilateral (" off-floor ") trading (a) is surprisingly vigorous and (b) is highly variable due to strong group (coordination) effects and (c) has a negligible effect on the price discovery path and efficiency and (d) in this private-value environment is characterized by an attempt to avoid competition on quotes (Georg Kirchsteiger et al., 2001a, b) rather than multi-unit bargaining over the spread (Joseph Campbell et al., 1991).