1 Electronic Limit Order Books, Dealer/Specialists, and Inter-Market Competition on NASDAQ Michael A. Goldstein Babson College Email: [email protected] Andriy V. Shkilko University of Mississippi Email: [email protected] Bonnie F. Van Ness University of Mississippi [email protected] Robert A. Van Ness University of Mississippi [email protected] Current version: February 20042Electronic Limit Order Books, Dealer/Specialists, and Inter-Market Competition on NASDAQ Abstract This study analyzes execution costs and competition for order flow between the NASDAQ Stock Market and five other trading venues that trade NASDAQ-100 securities (QQQ). We find that NASDAQ controls the most market share, although it has both the inside bid and ask only 50.24% of the time. These shares are much smaller than similar proportions reported for the NYSE, indicating a greater level of quote and order flow competition on NASDAQ. In addition, the dealer/specialist trading venues (NASDAQ, AMEX, and the Chicago Stock Exchange) show a decreasing percentage of trades from large to small trades, while trades through electronic limit order book systems (Archipelago and Island) exhibit an increasing percentage of trading from large to small trades. We find that both effective and realized spreads are smallest for the electronic limit order book systems and larger for the dealer/specialist systems. We also find evidence that order flow fragmentation and competition has hurt the markets as the NBBO spreads appear crossed 2.24% of the time and locked 12.43% of the time. Using a multinomial logistic regression, we find that all non-NASDAQ market centers have a lower likelihood of executing a trade as compared to NASDAQ itself, although the likelihood increases if the quotes are more competitive. Our findings suggest that extant findings on inter-market competition for NYSE-listed securities are not a result of the NYSE being the primary market, but instead due to differences in market structure.1Traders seeking to trade NASDAQ-listed stocks have a variety of choices in locations on which they can trade. Currently, NASDAQ stocks quote and trade on six different venues: NASDAQ, the American Stock Exchange (AMEX), the Cincinnati Stock Exchange (CSE), the NASD Alternative Display Facility (ADF), the Chicago Stock Exchange (CHX), and the Pacific Stock Exchange (PSE). These markets all compete for order flow, not only on the basis of price and quotes, but also cost to brokers.1 While competition and diversity of market types provides greater choice for traders and may lead to better prices, it may also lead to fragmentation and the worsening of markets. However, increased competition and fragmentation may lead to frequent locked or crossed markets.2 A recent study by Barclay, Hendershott, and McCormick (2003) examines competition between ECNs and market makers (they do not examine competition across different market centers), the effects of competition across markets centers have been mostly studied for NYSE-listed stocks. Studies of competition in NYSE-listed securities include Blume and Goldstein (1992), Lee (1993), Battalio, Greene, and Jennings (1997), Bessembinder and Kaufman (1997), Bessembinder (2003), and Lipson (2004). These studies examine regional and third market execution for NYSE listed securities. Until recently, it has been difficult to examine the different markets for competing prices for NASDAQ stocks. Increased fragmentation, acquisitions and reporting arrangements now 1In addition, there is competition within the NASDAQ stock market itself, with multiple dealers, ECNs and limit order traders competing for order flow. 2 According to Dean Furbish, Executive Vice President, NASDAQ Transaction Services, the issues associated with locked and crossed markets “refers to the relationships between SuperMontage and entities that are outside of it” (Schmerken, 2003).2allow an examination of the effects of competition and fragmentation within the NASDAQ market system on prices, quotes, and order flow. The purpose of this study, therefore, is to examine the competitive nature of the NASDAQ market. Although most market makers and electronic crossing networks (ECNs) post quotes and report trades through NASDAQ, some, such as Instinet, use the ADF. On the other hand, the largest ECN, Island, posts trades and quotes through the Cincinnati Stock Exchange. Archipelago, another electronic limit order book system, posts trades and quotes through the Pacific Stock Exchange. While AMEX is a specialist system, Chicago is a competing specialist system. These different reporting venues, i.e., ADF, AMEX, Island (Cincinnati), Archipelago (Pacific) and Chicago, allow us a glimpse into the inter-competitive nature of NASDAQ. Additionally, we examine issues of locked and crossed markets, and determinates of order routing. Overall, we find that there is substantial fragmentation of order flow for NASDAQ-listed stocks. Examining the NASDAQ-listed common stocks that comprise the NASDAQ-100 (QQQ) index from April to June 2003, we find that a little less than half (48%) of all trades occur away from the NASDAQ. Almost 27% of all trades occur on one of the two electronic limit order book systems (Island and Archipelago). However, these results are most prominent for smaller size trades. Over 87% of all block trades are still done on NASDAQ itself. While these results are consistent with those in Bessembinder (2003) for NYSE-listed securities, we find that NASDAQ’s market share in terms of volume (62%) is much smaller than that the 85% reported in Bessembinder (2003) for the NYSE, indicating that the market for NASDAQ-listed securities is less centered on the primary market. This may be because costs on NASDAQ are higher:3with the exception of AMEX, we find that NASDAQ has higher effective and realized spreads than its competitors. NASDAQ is also not particularly competitive in terms of quote behavior. While it matches the best quote on at least one side of the market about 90% of the time, it only matches both sides about half the time. Notably, NASDAQ is alone at quoting the best bid or the best ask only 11% of the time, and is alone at quoting both sides of the market less than 1% of the time. About 10% of the time, NASDAQ quotes neither the best bid or the best ask. Only 2.9% of all of NASDAQ’s volume comes during this time. There is significant