Does Search Friction Really Matter? Evidence from the Corporate Bond Market

Abstract

This paper provides empirical evidence for the importance of costly search in decentralized financial markets. Investors face a trade-off between information gathering costs and improved pricing. Significant search costs lead to a less exhaustive search and thus imperfect information. In equilibrium, dealers earn rents and trades are not necessarily executed at the best available price. Using corporate bond transaction data, I find that trade prices vary considerably for the same bond and on the same day. In the world of perfect information, the documented price dispersion is too large to be justified by either intraday price movements or dealer differences. The cross-sectional patterns of trading costs are also consistent with the predictions of costly search models, based on the premise that investors' incentives to price-shop increase with trade size. Both mean and dispersion of spreads decline with trade size after controlling for the impact of fixed costs, trading relationships, and investor sophistication. My findings support the theoretical predictions of Yin (JF, 2005) that quotation transparency (which reduces search costs) will lead to more competitive markets and better risk sharing among dealers.