Students who paid attention in their introductory economics course would recognize that in a typical goods market, large orders receive significant discounts. For instance, retailers often receive significant volume discounts by leveraging considerable buying power when negotiating with suppliers. This result is widely thought not to apply in securities markets. There is an extensive literature that appears to confirm larger orders receive worse prices in open limit order markets like the TSX. Furthermore, stylized models of trading suggest that larger orders receive worse prices in the presence of, among other things, asymmetric information. However, in the case of dealer markets, a 2005 paper published by Bernhardt, Dvoracek, Hughson, and Werner seems to suggest exactly the opposite- larger orders receive significant price improvements over smaller orders on the London Stock Exchange. In that paper, the authors argue that this is a result of the structure of dealer markets. If this is the case, this result should hold on the NASDAQ prior to 1997, when it adopted a limit order book. Therefore, our work is in the spirit of Bernhardt, Dvoracek, Hughson, and Werner with the interest of confirming that their result is in fact due to characteristics of dealer markets.
In this paper, we present a standard theoretical model of dealer markets that generates the testable prediction that larger orders should receive worse prices. Using a sample of intraday data taken from the NASDAQ in 1995 for 21 securities we estimate the relationship between order size and price improvement. Consistent with the finding on the London Stock Exchange (and contrary to the theoretical model), we find that larger orders do in fact receive a statistically and economically significant price improvement over smaller orders.
There are a number of empirical findings that suggest large trades in markets with electronic limit order books receive worse prices. Furthermore, stylized models of trading (like the one we present) suggest that larger orders should receive worse prices as a result of asymmetric information. That is, informed individuals are expected to prefer to trade larger quantities due to superior information.
Bernhardt, Dvoracek, Hughson, and Werner (2005) examine trade data from the London Stock Exchange in 1991 and find that, contrary to theoretical predictions, larger orders receive price improvements over small orders. They posit that this is due to repeated interactions between brokers and dealers, and illustrate this possibility via a theoretical framework and reinforce the result with an empirical study. The key insight is that large order price improvement can occur due to the structure of pure dealer markets.
Unlike a limit order book where the “best quoted price prevails” and “price competition is simultaneous”, dealer markets have the feature of requiring some negotiation over price. Once negotiation begins, “there...