The Reporting of Island Trades on the Cincinnati Stock Exchange

Transcription

1 The Reporting of Island Trades on the Cincinnati Stock Exchange Van T. Nguyen, Bonnie F. Van Ness, and Robert A. Van Ness Island is the largest electronic communications network in the US. On March 18 th 2002, it began reporting trades to the Cincinnati Stock Exchange (CSE) to reduce costs. We use the information generated following this trade reporting change to analyze differences in trading characteristics and trading costs between Island and NASDAQ. The results indicate significantly lower effective spreads, percentage effective spreads, and traded spreads for Island trades. There is more trade price clustering on nickels, dimes, and quarters for NASDAQ trades. [G10, G14] In early 2002, Island, the largest electronic communications network, announced it would begin reporting trades of NASDAQ securities to the Cincinnati Stock Exchange (CSE) rather than to the NASDAQ. On March 18, the CSE began disseminating Island s trades in NASDAQ stocks. A reporting change like this presents unique opportunity; in this case, it allows separation of most of Island s trades from NASDAQ s for comparison of trading characteristics and trade-based execution costs measures. We also determine Island s market share for NASDAQlisted stocks and examine trade-price clustering. Previously, empirical researchers have treated all NASDAQ trades the same way. Without some proprietary dataset, there was no way to distinguish where within the NASDAQ system a trade occurs. Yet, we know electronic communications networks (ECNs) are gaining market share, and have learned they have different trading characteristics. Island decided to report its trades through the CSE rather than NASDAQ to reduce costs. It has a revenuesharing and rebate plan with the CSE, which returns part of the revenue it makes by packaging and selling Island trading data to other financial institutions. Island rebates some of that money to its customers, helping it increase its market share in NASDAQ stocks. Van T. Nguyen is a doctoral student and Bonnie F. Van Ness and Robert A. Van Ness are Associate Professors at the University of Mississippi in University, MS It is unclear whether investors benefit from such a rebate plan, but the Securities and Exchange Commission has expressed concern about moves like this in a press release: The Commission is concerned that the availability of large market data revenue rebates in certain markets may be creating incentives for traders to engage in transactions with no economic purpose other than to receive market data fees. The Commission believes that such trades may be distorting the actual volume of trading in these securities. Moreover, the Commission is concerned that the structure and size of market data revenue rebates may be distorting the reporting of trades, and that these rebate programs may reduce the regulatory resources of the markets and reallocate the funding of regulation among participants. 1 We document the reporting change by Island, analyze trading around the day of the change, and isolate any differences in trading characteristics and trade based execution costs between Island (the Cincinnati Stock Exchange) and NASDAQ. We find that the Cincinnati Stock Exchange generated around 20% of the volume of NASDAQ stocks on the day of the reporting change. An examination of trades similar in trade size and time-of-day shows lower trade-based execution costs for the trades reported by the CSE. 1 SEC Acts on Market Data Rebate Programs (Press Release ). 30

2 NGUYEN, VAN NESS, & VAN NESS REPORTING OF ISLAND TRADES ON CINCINNATI STOCK EXCHANGE 31 I. Literature and Background Many market microstructure studies document a relation between the trading mechanism and trading costs, and many focus on the differences in trading costs between the NASDAQ (a dealer market) and the NYSE (an auction market). Before institution of decimal trading, the consensus was that the NYSE had lower trading costs than NASDAQ (see Christie and Huang, 1994; Huang and Stoll, 1996; Bessembinder, 1999; and Chung, Van Ness, and Van Ness, 2002). In the postdecimal environment, the evidence is not so clear-cut (see Bessembinder, 2003c, and Chung et al., 2002). Several studies examine securities that trade in multiple markets. Lee (1993) finds systematic differences in trading costs across trading venues when NYSElisted securities execute on the NYSE as well as on the regional stock exchanges. NASDAQ is found to have the least favorable execution costs for NYSE-listed securities. Blume and Goldstein (1997) and Bessembinder (2003b) analyze quote-based competition of the NYSE and regional stock exchanges, finding the NYSE has the best quoted prices most of the time. Only a few studies examine trading costs for NASDAQ securities in multiple markets that is, an auction market trading dealer securities. Van Ness, Van Ness, and Hsieh (1999) compare trading costs of NASDAQ-listed stocks that trade on both the NASDAQ and the Chicago Stock Exchange (CHX), and find trading costs to be lower on the CHX. Some authors compare trading costs of ECNs and NASDAQ market makers. Barclay, Hendershott, and McCormick (2002) use a proprietary dataset to examine ECN trading, without identifying the ten ECNs. They find lower effective spreads for medium-sized and large trades transacting on ECNs than for comparable market maker trades. Small trades do not have lower effective spreads unless they occur on non-integer ticks. Hasbrouck and Saar (2002) and Bias, Bisiere, and Spatt (2002) study the ECN, Island. Hasbrouck and Saar find that Island s market share for a given firm is positively related to the overall level of NASDAQ trading in the firm. Bias et al. (2002) find constrained Nasdsaq spreads prior to decimalization, when limit order traders use Island as a platform to compete for liquidity. After decimalization, the spreads on Island narrow, and the rents earned by Island traders virtually disappear. The SEC Order Handling Rules require ECN quotes to be posted as best bid and offer (BBO). Using a data set that allows them to distinguish ECN quotes, Simaan, Weaver, and Whitcomb (2003) examine the quotation behavior of ECNs. They find that ECNs establish the inside quote about 19% of the time. Clustering studies are both theoretical and empirical. The theoretical studies of price clustering suggest that, in the absence of any friction and bias, transaction prices should be uniformly distributed across all possible pricing grids (Niederhoffer, 1965). Empirical evidence, however, shows that transaction prices and quoted prices tend to cluster. These are two explanations for clustering: collusion or natural clustering. The collusion hypothesis is exemplified in Christie and Schultz (1994) and Barclay (1997). They show that even-eighth quotes occur much more often for certain stocks on NASDAQ than on the NYSE. Christie, Harris, and Schultz (1994), Bessembinder (1997), and Christie and Schultz (1999) provide additional evidence consistent with collusive behavior. The natural clustering hypothesis traces to Ball, Torous, and Tschoegl (1985). They argue that, under uncertainty, prices may be clustered at particular points as a result of traders trying to reduce search costs. This explanation is called the price resolution hypothesis. Similarly, Harris (1991) finds stock prices to cluster on round fractions because traders use discrete price sets to simplify negotiations. This explanation is called the negotiation hypothesis. Grossman, Miller, Cone, Fischel, and Ross (1997) argue that even-eighth quotes occur so often through the natural clustering of prices in competitive financial markets. They suggest market participants use a coarser price grid as protection against informed traders, as compensation for increased inventory risk, and to minimize the cost of negotiation. Harris (1991) suggests that clustering varies across markets. He shows there is more clustering in dealer markets than in auction markets. Grossman et al. (1997) also document that clustering varies across markets; they attribute differences in the degree of clustering to differences in market structures. Cooney, Van Ness, and Van Ness (2003) in examination of a sample of electronically submitted limit orders on NYSE stocks show that investors submit more limit orders with eveneighth prices than odd-eighth prices. Their study directly examines limit orders that reflect investor pricing preferences. Our work expands on the recent comparative literature. We include all NASDAQ-listed stocks that trade on both NASDAQ and Island (the CSE) to determine whether trading characteristics and trading costs differ between the two venues. We find significantly lower order execution costs (effective spreads, percentage effective spreads, and traded spreads) on Island (the CSE). We also find trades of NASDAQ-listed stocks receive more price improvement on Island (the CSE) than on NASDAQ. We believe we are the first researchers to examine ECN price clustering. On an electronic limit order book, we expect some degree of price clustering, and that is what we find. Prices on the NASDAQ cluster more on

3 32 JOURNAL OF APPLIED FINANCE FALL/WINTER 2004 nickels, dimes, and quarters supporting Harris s (1991) negotiation hypothesis. II. Data and Empirical Methodology On March 18, 2002 the Cincinnati Stock Exchange began to report trades in NASDAQ-listed stocks, according to the TAQ database. A. Data The data for this study are from the New York Stock Exchange s Trades and Quotes (TAQ) database. The initial sample comprises of all NASDAQ stocks that trade in March and April of To compare the differences in trading costs between Island (which reports its trades through the CSE) and the NASDAQ, we use two samples. The first sample consists of all trades and quotes for NASDAQ-listed stocks reporting trades on both the CSE and the NASDAQ. 2 We use this sample to examine trading characteristics and trading costs around the day the CSE began to report trades in NASDAQ stocks. This sample is designed to show the overall effect on all NASDAQ stocks that trade on Island. The second sample is obtained by matching trades by stock, day, time of day, and trade size. The objective is to control for differences in trading costs during the day (see McInish and Wood, 1992), and for differences in trading costs for different size orders (see Bessembinder, 2003c). First, we start with all trades for each stock for each day. We then match each Island (CSE) trade with the NASDAQ trades that occur within ten minutes on either side of the trade. Then we select the NASDAQ trade with trade size closest to the Island (CSE) trade. If more than one NASDAQ trade matches a CSE trade, the one closer in time is chosen. We use this sample to check for robustness in our trading cost results from the first sample as well as to accurately identify differences in trading costs between Island and NASDAQ, while controlling for time of day and trade size issues. B. Empirical Measures of Trading Costs We measure trading costs using effective spreads, percentage effective spreads, and traded spreads. We focus on trade-base measures of trading costs since Island does not report quotes through the CSE during our study time period. Hence, the trading cost measures for both NASDAQ trades and the trades of 2 Trades for NASDAQ are denoted by a T, and trades from the Cincinnati Stock Exchange are denoted by a C in the TAQ database. Island reported through the CSE are based on the NASDAQ quotes. 3 The distinguishing factor is the trade price. The effective spread takes into account that trading occurs at prices inside the posted bid and ask quotes. Investors frequently receive improved prices and thus encounter spreads lower than the quoted ones. We define the effective spread as: The percentage effective spread is calculated as: (1) (2) Lacking quotes from both venues, we calculate traded spread as an alternative to quoted spreads. Traded spread is the difference in the ask price and the bid price at the time of a transaction. Price improvement occurs when a trade transacts at a price better than the prevailing quoted price. Following Petersen and Fialkowski (1994), we define price improvement as trade price minus the prevailing NASDAQ bid for a sell transaction, and the prevailing NASDAQ ask minus the transaction price for a buy transaction. Since the TAQ database does not indicate the trade direction, we classify a trade as a buy if the trade price is closer to the ask, and a sell if it is closer to the bid. 4 III. Empirical Results We describe trading cost differences and clustering for the days before and after the change in reporting. A. Change-of-Reporting Day Analysis Exhibit 1 presents a daily summary of number of trades, percentage of total trades, and number of stocks trading each day for the 11-day period surrounding March 18 th, 2002 the day Island began to report trades to the CSE. The number of trades reported on NASDAQ declines as Island begins to report 3 Island began disseminating quotes through the CSE in late We use a different method of buy/sell classification than Lee and Ready (1991) or Ellis, Michaely, and O Hara (2000). We do not incorporate quotation or transaction lags since Bessembinder (2003b) shows that it is best not to make allowances for reporting lags in quotes or trades in assessing whether trades are buyer- or seller- initiated.

4 NGUYEN, VAN NESS, & VAN NESS REPORTING OF ISLAND TRADES ON CINCINNATI STOCK EXCHANGE 33 Exhibit 1. Summary Statistics by Day This exhibit shows the number of trades, percentage of total trades, and number of stocks for Island (reporting to the CSE) and NASDAQ in the 11-day period surrounding the day of first CSE reporting. Number of Trades Percentage of Total Trades Number of Stocks Calendar Date Event Day Island NASDAQ Island NASDAQ Island NASDAQ 3/11/ ,285, ,500 3/12/ ,090, ,444 3/13/ ,053, ,431 3/14/ ,824, ,511 3/15/ ,808, ,420 3/18/ ,774 1,456, ,866 3,479 3/19/ ,375 1,444, ,880 3,477 3/20/ ,734 1,494, ,859 3,459 3/21/ ,648 1,624, ,903 3,442 3/22/ ,045 1,479, ,863 3,442 3/25/ ,378 1,467, ,898 3,476 Figure 1. Market Share for Nasdaq-Listed Stocks 100 Market share of Nasdaq-listed stocks (%) s CSE Nasdaq Change in Reporting of Island Trades in NASDAQ-Listed Stocks (Day 0 is 3/18/02) trades to the CSE. Island (CSE) captures around 20% of total trades of NASDAQ stocks on the first day (between 18.45% and 20.02% for the first six days). 5 Island (CSE) reports trades in over half of all the stocks on the NASDAQ. Figures 1 and 2 also illustrate this point. Exhibit 2 shows the mean trade size in shares and in dollars for the 11-day period surrounding the day Island began to report trades through the CSE. Overall, 5 Only the NASDAQ, the CSE (Island), and the Chicago Stock Exchange (CHX) traded NASDAQ-listed stocks during our study period. The CHX traded just over 1% of NASDAQ-listed stocks before the CSE began to report trades in NASDAQ-listed stocks. The percentage remains approximately the same after the CSE enters the market for NASDAQ-listed stocks.

5 34 JOURNAL OF APPLIED FINANCE FALL/WINTER 2004 Figure 2. Number of Trades Around March 18, ,000,000 2,500,000 Number of Trades 2,000,000 1,500,000 1,000,000 Nasdaq CSE 500, Change in Reporting of Island Trades in Nasdaq-Listed Stocks Exhibit 2. Mean Trade Sizes In this exhibit, we show mean trade size in shares and in dollars for NASDAQ and Island (as reported to the CSE) for the period of 11 days surrounding the day the CSE began to report trades in NASDAQ-listed stocks March 18, The sample through day -1 is all NASDAQ-listed stocks; the sample from day 0 is all NASDAQ-listed stocks that have trades on Island. Mean differences and t-statistics are computed using paired t-tests. Calendar Date Event Day Island NASDAQ Mean Trade Size in Shares Mean Difference t-statistic Island NASDAQ Mean Trade Size in Dollars Mean Difference t-statistic 3/11/ /12/ /13/ /14/ /15/ /18/ *** , *** 3/19/ *** , *** 3/20/ *** , *** 3/21/ *** , *** 3/22/ *** , *** 3/25/ *** , *** ***Significant at the 0.01 level. we find significantly smaller mean trade size, whether in shares or in dollars on Island (CSE) than on NASDAQ. That Island tends to have smaller trades and lower dollar-volume trades is expected, given that Island is an order-driven market. We also calculate the mean trade size in shares and dollars for the 30-day period March 18, 2002, through the end of April 2002 the 30 days following the reporting change. These numbers are reported in Exhibit 3. Average trade sizes are 407 and 731 shares and $3825 and $7480 for Island and NASDAQ, respectively. These differences are statistically significant, indicating that Island trade sizes remain lower than NASDAQ s.

6 NGUYEN, VAN NESS, & VAN NESS REPORTING OF ISLAND TRADES ON CINCINNATI STOCK EXCHANGE 35 Exhibit 3. Mean Trade Size for 3/18/2002-4/30/02 In this exhibit, we show the mean trade size in dollars and shares for Island (as reported through the CSE) and NASDAQ for the period of 3/18/2002-4/30/2002. The sample is all NASDAQ-listed stocks that also trade on Island. Mean differences and t- statistics are computed using paired t-tests. Mean Trade Size in Shares (Standard Deviation of Trade Size) Mean Trade Size in Dollars (Standard Deviation in Trade Size) Island NASDAQ Mean Difference t-statistic 407 (542) 3825 (8569) 731 ( 2053) 7480 ( 54409) *** *** *** Significant at the 0.01 level. Exhibit 4. Trading Cost Measures This exhibit presents the effective spread, percentage effective spread, traded spread, and price improvement for NASDAQlisted stocks that trade on Island (as reported through the CSE) and NASDAQ. We show the trading cost measures by trading venue as well as the mean difference between the two with t-statistics. We use a paired t-test for the test statistic. The effective spread is the prevailing quote midpoint minus the current transaction price for sale orders and the current trade price minus the prevailing quote midpoint for buy orders. The percentage effective spread is the prevailing quote midpoint minus the current transaction price divided by the midpoint for sale orders and the current trade price minus the prevailing quote midpoint divided by the midpoint for buy orders. The traded spread is the difference between the current transaction price and the prevailing ask (bid) price for a sale (buy) transaction. Price improvement is trade price minus prevailing NASDAQ bid for a buy transaction, and the prevailing NASDAQ ask minus the transaction price for a sell transaction. All quotes originate from NASDAQ since Island did not report quotes to the CSE during our study period. Island NASDAQ Mean Difference t-statistic Effective Spread (in dollars) *** Percentage Effective Spread (in %) *** Traded Spread (in dollars) ** Price Improvement (in dollars) *** ***Significant at the 0.01 level. **Significant at the 0.05 level. B. Trading Cost Differences Exhibit 4 presents the effective spread, percentage effective spread, traded spread, and price improvement for a sample of trades on Island (reported to the CSE) and NASDAQ. We use paired t-tests to test for differences in means. The results show that effective spreads, percentage effective spreads, and traded spreads are significantly lower for Island trades (reported to the CSE) than those on the NASDAQ. This suggests that trading through an ECN, rather than an intermediary like a NASDAQ market maker, reduces trading costs. This finding is consistent with results in Barclay, Hendershott, and McCormick (2002), who find ECNs tend to have lower transaction costs. 6 Contrary to our expectations, we 6 This finding is also similar to that of Bessembinder (2003a) who finds narrower spreads on the Cincinnati Stock Exchange than on the New York Stock Exchange for NYSE-listed securities. Bessembinder does not include Island trading for NYSE-listed stocks. find more price improvement on Island trades than NASDAQ trades. This finding regarding price improvement may be a result of hidden orders, as hidden orders will result in greater price improvement. Our supposition regarding hidden orders is supported by Hasbrouck and Saar (2002), who find that hidden orders on Island account for almost 12% of executions by Island. To verify the robustness of our results, we match our sample of Island trades by stock, day, time-of-day, and trade size. This matching procedure accounts for differences in trading costs resulting from intraday trading behavior as well as differences in trade sizes. We find similar systematic differences in trading costs between the two venues. These results are reported in Exhibit 5. C. Price Clustering Harris (1991), Grossman et al. (1997), and Cooney et al. (2003) suggest that price clustering varies across

7 36 JOURNAL OF APPLIED FINANCE FALL/WINTER 2004 Exhibit 5. Trading Costs Measures Matched by Stock, Day, Time-of-Day, and Trade Sizes This exhibit presents the effective spread, percentage effective spread, traded spread and price improvement for a matched sample of NASDAQ-listed stocks traded on Island (as reported through the CSE) and NASDAQ. The sample is matched by stock, day, time-of-day, and trade size (in shares). We show the spread measures by trading venue as well as the mean difference between the two with t-statistics. The mean differences and t-statistics are based on paired t-tests. Island NASDAQ Mean Difference t-statistic Effective Spread (in dollars) *** Percentage Effective Spread (in %) *** Traded Spread (in dollars) *** Price Improvement (in dollars) *** ***Significant at the 0.01 level. **Significant at the 0.05 level. Exhibit 6. Distribution of Trade Prices This exhibit shows the distribution of trade prices for each pricing increment for Island (as reported through the CSE) and NASDAQ trades for all NASDAQ-listed stocks that trade on Island. Panel A shows proportions of trades at each of 10 pricing increments, and Panel B shows the proportion of trades at nickels, dimes, and quarters. Panel A. Average Proportion of Island and NASDAQ Trade Prices at Each Pricing Increment Tick Island NASDAQ Mean Difference t-statistic x.x *** x.x *** x.x *** x.x *** x.x *** x.x *** x.x *** x.x *** x.x *** x.x *** Panel B. Average Proportion of Trade Prices at Nickles, Dimes, and Quarters Cluster Island NASDAQ Mean Difference t-statistic Nickles *** Dimes *** Quarters *** ***Significant at the 0.01 level. trading venues. If Harris s negotiation hypothesis holds, we would expect more price clustering on the NASDAQ than on an ECN. This is because NASDAQ is primarily a dealer market where negotiation is prevalent, while Island is an ECN where investors (some of them proprietary traders and day-traders) post limit order without any intervention of a dealer and involves little negotiation. 7 However, Cooney et al. show that 7 Note that not all the trades we classify as from NASDAQ are dealer trades. Some are from other ECNs and NASDAQ s ADF (Alternative Display Facility). limit order prices tend to cluster. Overall, we expect to find price clustering on both trading venues, but less price clustering for Island s trades than for NASDAQ s. Exhibit 6 and Figures 3 and 4 report trade price clustering on NASDAQ and Island. We find that prices cluster around ticks x.x0 and x.x5 on both exchanges, but a higher percentage of trades occur on x.x0 and x.x5 on NASDAQ than on Island (23.59% and 17.69% versus 19.86% and 15.48%, respectively). A paired t- test for the mean difference indicates more price clustering on the NASDAQ for ticks x.x0 and x.x5.

8 NGUYEN, VAN NESS, & VAN NESS REPORTING OF ISLAND TRADES ON CINCINNATI STOCK EXCHANGE 37 Figure 3. Trade Price Clustering Proportion of Trades CSE Nasdaq x0.x1.x2.x3.x4.x5.x6.x7.x8.x9 Pricing Increments Figure 4. Trade Price Clustering (Nickles, Dimes, and Quarters) Proportion of Trades CSE Nasdaq Nickles Dimes Quarters Cluster

9 38 JOURNAL OF APPLIED FINANCE FALL/WINTER 2004 The clustering at nickels, dimes, and quarters findings show that prices on NASDAQ cluster more at the nickel, dime, and quarter increments. Overall, our results support the negotiation hypothesis. IV. Conclusion Following the change in trade reporting of Island from the NASDAQ system to the Cincinnati Stock Exchange, around 20% of NASDAQ-listed stock trades were reported to the CSE in our study period. The trades reported through the CSE are smaller than those reported on NASDAQ. Order execution costs (effective spreads, percentage effective spreads, and traded spreads) are significantly lower for Island trades, and trades on NASDAQ receive more price improvement. We also compare trade price clustering on Island and NASDAQ. Island trade prices do exhibit clustering, but NASDAQ trade prices show more of it, and prices on the NASDAQ cluster more on nickels, dimes, and quarters. This finding supports the negotiation hypothesis. References Ball, C., W. Torous, and A. Tschoegl, 1985, An Empirical Investigation of the EOE Gold Options Market, Journal of Banking and Finance 9 (No. 1, March), Barclay, M., 1997, Bid-Ask Spreads and the Avoidance of Odd-Eighth Quotes on NASDAQ: An Examination of Exchange Listings, Journal of Financial Economics 45 (No. 1, July), Barclay, M., T. Hendershott, and D. McCormick, 2002, Information and Trading on Electronic Communications Networks, University of Rochester Working Paper. Bessembinder, H., 1997, The Degree of Price Resolution and Equity Trading Costs, Journal of Financial Economics 45 (No. 1, July), Bessembinder, H., 1999, Trade Execution Costs on NASDAQ and the NYSE: A Post-Reform Comparison, Journal of Financial and Quantitative Analysis 34 (No. 3, September), Bessembinder, H., 2003a, Issues in Assessing Trade Execution Costs, Journal of Financial Markets 6 (No. 3, May), Bessembinder, H., 2003b, Quoted-based Competition and Trade Execution Costs in NYSE-Listed Stocks, Journal of Financial Economics 70 (No. 3, December), Bessembinder, H., 2003c, Trade Execution Costs and Market Quality after Decimalization, Journal of Financial and Quantitative Analysis 38 (No. 4, December), Bias, B., C. Bisiere, and C. Spatt, 2002, Imperfect Competition in Financial Markets: Island vs. NASDAQ, Carnegie Mellon University Working Paper. Blume, M. and M. Goldstein, 1997, Quotes, Order Flow, and Price Discovery, Journal of Finance 52 (No. 1, March), Chan, K., W. Christie, and P. Schultz, 1995, Market Structure and the Intraday Pattern of Bid-Ask Spreads for NASDAQ Securities, Journal of Business 68 (No. 1, January), Christie, W., J. Harris, and P. Schultz, 1994, Why Did NASDAQ Market Makers Stop Avoiding Odd-Eighth Quotes? Journal of Finance 49 (No. 5, December), Christie, W. and R. Huang, 1994, Market Structures and Liquidity: A Transactions Data Study of Exchange Listings, Journal of Financial Intermediation 3 (No. 3, June), Christie, W. and P. Schultz, 1994, Why Do NASDAQ Market Makers Avoid Odd-Eighth Quotes? Journal of Finance 49 (No. 5, December), Christie, W. and P. Schultz, 1999, The Initiation and Withdrawal of Odd-Eighth Quotes among NASDAQ Stocks: An Empirical Analysis, Journal of Financial Economics 52 (No. 3, June), Chung, K., B. Van Ness, and R. Van Ness, 2001, Can the Treatment of Limit Orders Reconcile the Differences in Trading Costs between NYSE and NASDAQ Issues? Journal of Financial and Quantitative Analysis 36 (No. 2, June), Chung, K., B. Van Ness, and R. Van Ness, 2004, Trading Costs and Quote Clustering on the NYSE and NASDAQ after Decimalization, Journal of Financial Research 27 (No. 3, Fall), Cooney, W., Jr., B. Van Ness, and R. Van Ness, 2003, Do Investors Prefer Even-Eighth Prices? Evidence from NYSE Limit Orders, Journal of Banking and Finance 27 (No. 4, April), Ellis, K., R. Michaely, and M. O Hara, 2000, The Accuracy of Trade Classification Rules on the NSE and NASDAQ, Journal of Financial and Quantitative Analysis 35 (No. 4, December),

10 NGUYEN, VAN NESS, & VAN NESS REPORTING OF ISLAND TRADES ON CINCINNATI STOCK EXCHANGE 39 Grossman, S., M. Miller, K. Cone, D. Fischel, and D. Ross, 1997, Clustering and Competition in Asset Markets, Journal of Law and Economics 40 (No. 1, April), Harris, L., 1991, Stock Price Clustering and Discreteness, Review of Financial Studies 4 (No. 3, Fall), Hasbrouck, J. and G. Saar, 2002, Limit Orders and Volatility in a Hybrid Market: the Island ECN, New York University Working Paper. Huang, R. and H. Stoll, 1996, Dealer versus Auction Markets: A Paired Comparison of Execution Costs on NASDAQ and the NYSE, Journal of Financial Economics 41 (No. 3, July), Lee, C. and M. Ready, 1991, Inferring Trade Direction from Intraday Data, Journal of Finance 46 (No. 2, June), Lee, C. M.C., 1993, Market Integration and Price Execution for NYSE-Listed Securities, Journal of Finance 48 (No. 3, July), McInish, T. and R. Wood, 1992, An Analysis of Intraday Patterns in Bid/Ask Spreads for NYSE Stocks, Journal of Finance 47 (No. 2, June), Niederhoffer, V., 1965, Clustering in Stock Prices, Operations Research 13 (No. 2, March-April), Petersen, M. and D. Fialkowski, 1994, Posted versus Effective Spreads: Good Prices or Bad Quotes, Journal of Financial Economics 35 (No. 3, April), Simaan, Y., D. Weaver, and D. Whitcomb, 2003, Market Maker Quotation Behavior and Pretrade Transparency, Journal of Finance 58 (No. 3, June), Van Ness, B., R. Van Ness, and W. Hsieh, 1999, NASDAQ and the Chicago Stock Exchange: An Analysis of Multiple Market Trading, Financial Review 34 (No. 4, November),