Printer Friendly

Competition in NYSE-listed stocks: do investors benefit from highly competitive markets?

1. Introduction

Stock market competition and integration have been an issue of interest in past decades that has recently intensified. A consolidated market provides the greatest liquidity and facilitates best price discovery. However, it creates a monopoly which may be used by the market makers to take advantage of investors. Fragmented markets foster competition and may lead to lower execution costs. In the presence of different trading venues, investors can choose one based on their specific needs. Nevertheless, fragmented markets increase the difficulty and costs of finding counterparties. Fragmented markets give brokers opportunities to choose trading venues for investors but brokers may not act in the best interest of investors. Consequently, determination of whether the market competition benefits or harms investors is still inconclusive.

This paper addresses this issue by investigating the effects of changes in NYSE market share on the trading quality for NYSE-listed stocks. Due to the rapid evolution of the alternative trading systems, the trading markets of NYSE stocks have changed dramatically. These computer-based trading systems gained popularity among investors gradually by providing cheaper and faster executions. They also use liquidity rebates to attract order flow. Facing the increasing competition, the NYSE was unable to retain its advantage in trading its own-listed stocks and lost substantial market shares in recent years. Although the electronic-based trading platforms have developed and grown for years, prior studies focus on the NASDAQ stocks (1) and the effect of the increasing competition on market quality for NYSE stocks is not fully explored yet.

The effects of the increasing competition from electronic trading systems on trading quality for NYSE stocks may be different from the effects for NASDAQ stocks due to the different market microstructures of the NASDAQ and NYSE. Unlike the NYSE which has a trading floor and assigns each stock to a single market maker, NASDAQ is an electronic-based market which allows many dealers to provide liquidity for one stock. For the most active stocks, there are more than 50 dealers making the markets. Therefore, NASDAQ is a fragmented market in nature. The introduction of ECNs increases the competition but does not change the market structure of NASDAQ very much. On the contrary, the NYSE is originally a centralized market. When off-NYSE markets obtain substantial market share, the market for NYSE stocks changes from a consolidated market into a fragmented market. While the increasing competition may benefit investors, the advantages of consolidation and floor trading may also diminish. Diamond and Kuan (2012) warn that ECN development will encourage the formation of "dark pools" (transactions taking place outside the regulated markets and the spread of risky "high frequency" trading.

Harris (2003) posits that the specialists' liquidity services required by the affirmative obligation are public goods and are hard to obtain in competitive markets. The continuous and orderly markets increase traders' confidence to invest in the exchanges' listed companies. For example, on May 6th 2010, the DJIA plunged more than 600 points in less than 15 minutes without a clear reason. The NYSE designated market makers came in on the other side of an order and tried to maintain an orderly market (2). This liquidity that specialists provide benefits all traders, regardless of whether they trade in the NYSE or other trading venues. However, only investors who trade in NYSE pay for the services because the specialist cannot impose costs on investors who do not trade in the NYSE. Because the affirmative obligation is costly to specialists, this liquidity externality creates problems. When the NYSE market share is substantially greater than 50 percent the exchange can afford to be generous. On the contrary, when their market share plummets, specialists may not be able to earn enough profits to fund the liquidity they must provide when no one else will. In 2007, two of the seven NYSE specialists firms decided to leave the markets because their business was no longer profitable. The off-NYSE liquidity providers are not required to provide these costly liquidity services. Therefore, they can execute orders with lower cost and gain market shares. In the long run, no market makers provide these services and the trading quality may deteriorate.

The relevance of this analysis increased dramatically on July 8, 2015, when the New York Stock Exchange suspended trading for over three hours. A technical issue tied to a computer upgrade reportedly caused the glitch, according to a Forbes article (McGrath, 2015). The trading suspension did not become more of an issue because the NASDAQ continued to operate normally, including the trading of NYSE stocks. This is not the first time there has been a trading suspension at the NYSE. A trading suspension in 2001 also resulted from a glitch when a new computer program was brought on line. That suspension had more of an impact, because back then more than eighty percent of shares traded on the NYSE. When NYSE trading was suspended in 2015, less than 25 percent of all trading was done on the NYSE and the other 75 percent of trading platforms, like the NASDAQ and BATS, were more than willing to absorb the NYSE's portion of the market.

In this paper, we investigate whether the decrease of NYSE's market shares and the growth of the electronic-based trading systems have resulted in changes in the trading quality. We focus on the period when the market for NYSE stocks changed from being a highly consolidated market to a highly competitive market. Because the NYSE has a different market microstructure than NASDAQ, the impact of the increasing competition from the electronic-based markets on trading quality for NYSE stocks may be more pronounced and could be even negative.

Investigating 259 NYSE firms from 2004 to 2008, overall we do not find that investors benefit from the increasing competition. During our sample period, the NYSE market share decrease from 86.56% to 31.93% and the percent of time that the NYSE provided both best bid and offer decreased from 92.40% to 34.70%. We find that although the competition increases significantly, the effective spreads and relative quoted spreads increase with the decrease of NYSE market shares. In the meantime, the liquidity providers' revenue, measured by realized spreads, and short-term price volatility increase. Although the quoted depth increases and the price impact decreases when the markets become more competitive, we also find that the average number of shares per trade drops from 586 shares to 190 shares, which indicates that the ability that a market can absorb a large-size trade decreases when markets are fragmented. Our findings suggest that the benefit from competition of the alternative trading systems is limited for NYSE stocks.

Based on our finding that the average trading size drops significantly after the market becomes highly fragmented, the increase in the transaction costs may not be limited to the changes of effective spreads. The decrease in trading size indicates that investors divide their trades into several small trades, which usually results in more commissions if the commissions are determined based on the number of trades. Furthermore, after markets become very fragmented, it is not easy to find the place where the best bid and offer are provided. Many trade facilitators (such as Bloomberg Trade Book) develop and sell sophisticated order-routing systems to their clients. Although now the flow of information from one market fragment to another is fast, which consolidates a fragmented market to some degree, the fact that the crossed and locked markets occur more frequently indicates that prices do not efficiently incorporate all available information fast enough due to the time lag for transferring information. When markets are highly fragmented, the search costs for finding the best trading venue increase.

High-frequency trading has become a fact of life, using algorithm (a.k.a., "algobots") to arbitrage away the smallest price discrepancies that only exist in the most infinitesimally small time horizons. Perfect correlation between markets is arbitraged away in half the time it takes to blink an eye. While O'Brien (2014) notes that bid-ask spreads have declined from 90 basis points to around 3 basis points today, he argues that the ability to use computers to absorb price discrepancies is essentially stealing from buy-and-hold investors. Hasbrouck and Saar (2013), however, find that high-frequency trading does not work to the detriment of long-term investors. They find that high-frequency trading deceases spreads, increases the displayed depth in the limit order book, and lowers long-term volatility. One aspect of the high-frequency trading that is under-reported is the growing presence of odd lot trading, as algorithmic trading divides orders into smaller pieces. The relevance for this article is the common use of allocation protocols for crossing networks resulting in odd-lot order, cited by O'Hara, Yao, and Ye (2014). Yet, neither the consolidated tape nor TAQ data derived from it includes odd-lot trades. It is likely that the benefits of high-frequency trading, regardless of the trading venue will be greater when applied to shares of firms located further from the United States, which Ferguson (2015) shows suffer from limited information transmission. We focus on domestic markets, while Dodd, Louca, and Paudyal (2015) have shown that the cross-listing of firms in U.S. markets appears to greatly improve price discovery, regardless of the ECN used.

Although our results suggest that investors of the NYSE stocks do not benefit from the increasing competition, our analysis does not consider the value of trading speed. Some may suggest that the best execution is a multidimensional concept encompassing market impact, timing, trade mechanism, anonymity, commissions, and even quid pro quo arrangements. Traders value both speed of execution and price. Easley, Hendershott, and Ramadorai (2007) and Boehmer, Jennings, and Wei (2007) both find that the execution speed matters to traders. Because the execution speed of the off-NYSE markets is normally faster than that of floor trading, our results should be interpreted with this caveat.

Our work is closely related to O'Hara and Ye (2009) and Fink, Fink, and Weston (2006). O'Hara and Ye (2009) provide cross-sectional analysis which compares the trading quality of stocks with concentrated trading markets with that of stocks with fragmented trading markets. Their findings suggest that the competition does not harm market quality. Our study focuses on the long-term time-series analysis that compares the market quality of a stock when its trading is more concentrated in the earlier period with that of the same stock when its trading becomes more fragmented later. Due to the different data, time period, and research design, our results are different from the findings of O'Hara and Ye (2009). Our research design is very similar to Fink, Fink, and Weston (2006) which studies the development of the alternative trading systems for NASDAQ stocks and found that the increase in the market shares of ECNs is associated with tighter effective and relative bid-ask spreads, greater depths, and less concentrated markets. Because of different market microstructure between NASDAQ and NYSE as we previously mentioned, our results are not consistent with Fink, Fink, and Weston (2006).

An interesting comparison of electronic and physical markets is presented by Harris and Saad (2014), who assess the impact of electronic message traffic on subsequent share price movement. They document the ability of changes in price and volume of bids around the current price to predict subsequent share prices as much as 75 seconds prior to the reaction. With adjustment for time-of-day, ECN volume, and firm specific fixed effects, "algobots" can be used to effectively forecast short-term market conditions. Cheng and Stunda (2015) demonstrate the advantages of investing in NASDAQ firms over the short-term, and NYSE firms over the longer term. The apparent advantage of NASDAQ's online trading system work against those holding stock for a longer period of time. Kwan, Masulis, and McInish (2015) find that the minimum pricing increments imposed on exchanges by the U.S. Securities and Exchange Commission significantly weaken physical exchange competitiveness and buttresses the development of "dark pools" where investors can bypass existing limit order queues with minimal price movement.

Our findings shed light on the on-going debate regarding market integration and competition. In recent decades, regulators seem to prefer competition more than consolidation. The drop of the NYSEs market share can be partially attributed to several changes on rules to promote competition. Our research provides a review regarding the competition during a critical time period and new evidence on whether competitive fragmented markets lead to better trading quality than a consolidated market with a physical trading floor.

The rest of this paper proceeds as follows. Section 2 discusses the competition in NYSE stocks. Section 3 presents data and sample used in this study. The empirical results are shown in section 4 and section 5 summarizes the report.

2. The Competition in NYSE Stocks

Trading NYSE stocks has historically required face-to-face communication on the trading floor. The specialists on the trading floor have a central role in executing orders. They are required to fulfill affirmative obligations to maintain an orderly market with narrow spreads, meaningful depth, and continuous prices. Under the design of the NYSE, all orders of a stock are routed to its specialist. The specialist observes all order flows and the market is consolidated. Before 1997, although it was still subject to the competition of regional exchanges and NASDAQ, the NYSE was able to maintain the monopolistic power. Improved information technology and higher volume should drive orders to be concentrated in one market, with lower transaction costs. However, using data from 1900-1933, White (2013) has recently shown that the tendency of U.S. exchanges towards concentration is periodically reversed when new technologies arise.

Since 1997, the government policy has changed further to promote competition. The SEC approved new rules to create a fair and competitive market, including establishing alternative trading systems such as Electronics Communication Networks (ECNs) to compete with NASDAQ and NYSE. Different from the NASDAQ which lost market shares quickly (3), the NYSE had resisted the competition from ECNs successfully for years. According to Bessembinder (2003), in June 2000 the market share of NYSE was 84.94%in terms of trading volume. Hatch and Battalio (2001) also find that after the 1997 reform of SEC, compared to other venues, the NYSE still offered more favorable execution to the investors. In 2003, the NYSE was required to open its limit order book and lost some information advantage. After 2006, the NYSE market shares drop quickly and the markets for NYSE stocks become more and more fragmented.

While many studies on NASDAQ stocks suggest that the competition from ECNs promote trading quality by providing competitive quotes, cheap and fast executions, anonymous trading, and a high probability of executions for NASDAQ stocks, the effects of the increasing competition from the alternative trading systems on trading quality for NYSE stocks is not clear due to the difference of market structure between NASDAQ and NYSE. Unlike the NYSE which has trading floor and assigns each stock to a single market maker, NASDAQ is an electronic-based market which allow many dealers to provide liquidity for one stocks. For the most active stocks, there are more than 50 dealers making the markets. Therefore, NASDAQ is a fragmented market. The introduction of ECNs increases the competition but does not change the market structure very much. On the contrary, the NYSE was originally a centralized market. When off-NYSE markets obtain substantial market share, the market for NYSE stocks changes from a consolidated market into a fragmented market. While the increasing competition may benefit investors, the following advantages of centralized floor trading may also diminish. These advantages are:

a) Better execution due to consolidation

Consolidation of trading enhances the likelihood that counterparties will find each other. Liquidity is the greatest and transaction costs are the lowest when all traders trade in the same market. Prior studies have shown that the NYSE usually offers better execution than other venues. (4) The growth of ECNs results in fragmented market and reduces the benefit of consolidation.

b) Affirmative obligation ensures price stability

NYSE market maker's affirmative obligations for providing liquidity and stabilizing price ensure that a reasonable market always exists. The NYSE specialists are the traders of last resort. When no one else is willing to trade, specialists must be willing to trade. They must quote two-sided markets, and their quotes must be meaningful in the sense that the spread between the best bid and the best offer cannot be too wide. Specialists also smooth prices by intervening to prevent large price reversals and are responsible for creating price continuity. Although maintaining the liquidity and stabilizing the price are costly, if the market makers receive a large number of orders, they can allocate the cost to each order so that each trader does not pay a high cost for market makers' services. However, when the off-NYSE markets obtain substantial market shares, the NYSE market makers may not have enough orders for cost-sharing. They may charge higher execution costs or become unwilling to maintain liquidity and a stable price. Over time, liquidity may decline and the price fluctuation may increase.

c) NYSE market maker's better ability to find informed trading

Informed trading harms both market makers and traders. Having inferior information, investors and market makers who trade with the informed traders eventually incur a loss. When there are many informed traders in a market, uninformed traders are reluctant to trade. On the contrary, informed traders want to trade with uninformed traders and hide their information-based trading (see Admati and Pfleiderer (1988) and Easley and O'Hara (1992)). A fragmented market makes it difficult for market makers to distinguish informed trading from uninformed trading because they only observe part of order flows. Overtime, the fragmentation may increase the informed trading and add risk for all participants. In addition, the repeated face-to-face communications between market makers and floor brokers on the NYSE trading floor enables the market makers to gather informed trading from floor brokers. Benveniste, Marcus, and Wilhelm (1992) and Battalio, Ellul, and Jennings (2007) both argue that long-term relationships between brokers and dealers can mitigate the effects of asymmetric information. Floor brokers trade repeatedly with NYSE market makers. If they do not honestly reveal the traders' information, they may be subject to possible sanction (by providing less favorable future prices, by being less willing to improve quoted prices, or by being less willing to fill large orders at quoted prices) made by NYSE market makers. The ability to sanction allows NYSE market makers to effectively identify informed trading, impose the cost of informed trading, and quote tighter bid-ask spreads for the uninformed traders.

d) Liquidity provided by floor brokers

Floor brokers trade for their clients and hold much of the liquidity of NYSE stocks. They participate in large trades and trade more when the stock price fluctuation is high. Working on the trading floor, floor brokers obtain trading information quickly and seize liquidity opportunistically. Their ability to quickly adjust the trading strategies to current market condition is especially valuable when trading the large orders (see Sofianos and Werner (2000), Werner (2003), and Handa, Schwartz, and Tiwari (2004)). Because floor brokers usually are willing to provide liquidity to traders and offer superior execution quality for their clients, the decline of NYSEs floor trading due to the competition of ECNs may reduce trading quality for NYSE stocks, especially for large orders.

In this research, we investigate whether the decrease of NYSE's market shares and the growth of the electronic-based trading systems have resulted in changes in the trading quality. We focus on the period when the market for NYSE stocks changes from a relatively consolidated market to a relatively fragmented market. Because the NYSE has a different market structure than NASDAQ, the impact of the increasing competition from the electronic-based markets on trading quality for NYSE stocks may be more pronounced and could even be negative. If the competition improves trading quality over time, when the off-NYSE liquidity providers gain market shares, we should observe lower transaction costs, more competitive quotes, less trading based on private information, larger market depths, and lower price fluctuations.

3. Data and Sample

The data used in this study are from the NYSE TAQ Database and CRSP. We collect daily closing prices, returns, number of shares outstanding, and trading volume from CRSP. The intraday trade and quote data are from TAQ. The TAQ transaction data comprises trading time, volume, and trading price for NYSE-listed stocks traded in all exchanges. The TAQ quotation data contains a record of all quotes with quoted time, bid and ask price, and bid and ask depth. Each trade and quote is time stamped to the nearest second. (5) Because the TAQ data does not contain the national best bid and offer (NBBO), we compute the NBBO from the quote database. We find that the market is crossed or locked very often due to stale quotes of some inactive exchanges. (6) Since stock prices can change within a second, the unchanged stale quotes reported by inactive exchanges may no longer be effective. To avoid the stale quotes affecting our results, we limit the effective time of a quote to five minutes. When an exchange does not update its quote for 5 minutes, we assume the liquidity providers of that exchange withdraw from the markets.

Using TAQ data enables us to do more thorough analysis for the effects of competition on market quality. Although the SEC Rule 605 disclosure also contains information about market quality, it has several disadvantages. First, it is prepared by individual market centers and may not be comparable among different markets. It is not clear whether we can infer the overall market quality by taking the average of all market centers. In addition, the file does not contain orders with a size of 10,000 shares or more. The transaction costs calculated from the Rule 605 disclosure may be biased to medium-size or small-size trades. Furthermore, the Rule 605 files are monthly reports and do not allow daily analysis. On the contrary, TAQ data are intraday data including all trades and quotes. We can use them to construct consistent measures across different markets and different time periods.

Despite the above advantages of TAQ database, there are also several disadvantages. First, it does not contain the information about whether an order is initiated by buyers or sellers. We need to infer the direction of trade by comparing the trading price with the quote. We follow Ellis, Michaely, and O'Hara (2000) and assign trades executed at the ask (bid) quote as customer buys (sells), while using the tick test for all other trades. Because locked and crossed market conditions occur from time to time, the locked and crossed NBBO may not be very useful to infer the trade direction. Therefore, we use the quotes of the market where the trade is executed and compare them with the trading prices to determine the trade direction. Second, the TAQ data show the report time of a transaction, not the time when an order is executed. As a result, we need to consider the lag between order execution and order reporting. Following Bessembinder (2003b), we use the concurrent quotes to infer trade directions but use the quote five seconds prior to the transaction to measure the effective trading costs because quotations systematically rise (fall) in the seconds before customer buy (sell) trades are reported. Third, although TAQ data contain all transactions and quotes, trades and quotes executed through non-exchange dealers/brokers cannot be identified separately. Those dealers/brokers may report their trades and quotes in one of the exchanges. Therefore, our analysis for an exchange includes trades/quotes occurring both in that exchange and in other dealers/brokers who choose to report trades/quotes in that exchange.

To investigate whether the market quality is improved when the competition among different trading venues becomes fierce, we need to measure competition and market quality. For NYSE-listed stocks, the NYSE is the primary trading venue. When the markets become more and more competitive, the market shares of NYSE drop. In this paper, we use the NYSE market shares to proxy the competition. Following literature, we use the following proxy to examine market quality:

Effective spread: The effective spread is the execution cost investors pay. It is defined as 100 x D([P.sub.t] - [M.sub.t-5])/[M.sub.t-5], where D equals to 1 (-1) for buyer-(seller-)initiated trade, [P.sub.t] is the trading price, and [M.sub.t-5] is the quote midpoint five seconds before the transaction. When market quality improves, the effective spread should decrease.

Relative quoted spread and quoted depth: If the competition encourages liquidity, providers post competitive quotes, the quoted spread should drop and the quoted depth should rise. We get the quoted depth from TAQ database directly and calculate the relative quoted spread as the ask price minus bid price divided by the quoted midpoint.

Price impact: The price impact is a measure of informed trading and is defined as 100 x D([M.sub.t+300] - [M.sub.t-5])/[M.sub.t-5], where D equals to 1 (-1) for buyer-seller-) initiated trade, [M.sub.t+300] is the quote midpoint five minutes (i.e., 300 seconds) after the trade, and [M.sub.t-5] is the quote midpoint five seconds before the transaction. By observing the order flow, the liquidity providers can detect whether an order is submitted by informed investors and then change the quote based on the information content of the trade. Therefore, the changes of quote midpoint can proxy the trading activities made by informed traders.

Realized spread: Liquidity providers incur loss when trading with informed traders but make profit when trading with uninformed traders. If the value of stock decreases (increases) after the transaction of a customer buy (sell) order, the liquidity providers earn money. The realized spread, calculated as 100 x D([M.sub.t+300] - [P.sub.t])/[M.sub.t-5] (the difference between effective spread and price impact), captures value changes after the transaction and can be used to measure the liquidity providers' revenue.

Short-term price volatility: Investors' trading risk increases with short-term price volatility. When price volatility is higher, the trading price investors get differs more often from the price at the time when investors submit their orders. We use two proxies for the short-term price volatility. The first proxy is the standard deviation of trading prices in 10-minute intervals. The second proxy is the absolute percentage change of quoted midpoints in 10-minute intervals. We divide the regular trading hours (from 9:30 am to 4:00 pm) into 39 10-minute intervals and exclude the first two intervals to avoid the effect of opening transactions on the price volatility.

Share turnover: Trading volume is usually considered as a liquidity measure. When trading volume increases, investors can trade stocks easily. In this paper, we use quarterly share turnover (defined as the total volume divided by the average number of shares outstanding in a calendar quarter) to proxy the trading volume. If the competition improves liquidity, the trading activities increase.

Trading size: Market depth, the ability of a market to absorb large-size transaction without significant impact on trading prices, is one dimension of the market quality. If the market depth increases, investors are more willing to submit large-size orders. When competition improves market depth, there will be more large-size orders and the trading size will increase.

Our sample includes 259 firms listed on the NYSE from 2004 to 2008. We limit our sample to the ordinary common stocks and exclude companies incorporated outside the U.S., Americus trust components, closed-end funds, and REIT's. At the end of December 2003, we sort all NYSE firms based on their market capitalizations and then assign each firm into one of the three size group: large, medium, and small. We randomly select 100 firms from each size group. During our sample period, one firm changed listing from NYSE to NYSE-Arca, one firm changes listing to NASDAQ, and 39 firms contain missing data or delist by the end of 2008. We remove these firms from our sample. Our final sample contains 82 large firms, 90 medium firms, and 87 small firms. Table 1 provides the summary statistics of our sample. The stock price, daily trading volume, and quoted depth increase with firm size; while the return volatility, daily return, and relative quoted spread decrease with firm size. The medium firms tend to have larger daily turnover than small and large firms.

4. Empirical Results

4.1 Changes of trading activities

This paper studies the effect of competition among different trading venues on market quality for NYSE stocks. Specifically, we investigate whether the market quality changes when the NYSE's market share drops. Table 2 shows the changes of market shares for all exchanges trading the 259 NYSE stocks from 2004 to 2008. There are eleven exchanges during our sample period: Boston Stock Exchange, National Stock Exchange, Chicago Stock Exchange, NYSE, NYSE-Arca, NASDAQ, Philadelphia Stock Exchange, NASD ADF (Alternative Display Facilities), International Securities Exchange, CBOE, and BATS. The market shares of Boston Stock Exchange and Philadelphia Stock Exchange decrease to zero because they are bought by NASDAQ in Oct 2007 and in July 2008, respectively. The NASD ADF, International Securities Exchange, CBOE, and BATS start to report transactions on NYSE stocks after 2006. As shown in 2, the competition for NYSE stocks has increased dramatically. The NYSE market shares drops significantly from 86.56% to 31.93%. At the same time, the market share of NYSE-Arca, NASDAQ, and NASD ADF increase by 15.97% 10.80%, and 24.07%, respectively. It is interesting that all electronic-based exchanges gain market shares from 2004 to 2008 significantly; while the exchanges with trading floor (NYSE and Chicago) lose their market shares significantly. This means either investors prefer electronic-based trading systems or brokers submit orders to those electronic-based trading venues more often in order to get liquidity rebates.

In additional to the pronounced changes on market shares of different exchanges, we also find the liquidity in terms of trading volume, measured by share turnover, and trading size changes substantially during our sample period. Table 3 presents changes of trading activities of our full sample, large firms, medium firms, and small firms. For our full sample, the average quarterly turnover increases significantly from 0.4204 to 0.9679. However, in the meantime, the average size per trade decreases significantly from 586 shares to 190 shares. The changes for large firms are larger than those for small firms and the differences of changes in trading volume and trading size between large firms and small firms are both significant at 1% level. Our results suggest that when the competition becomes fierce and the electronic-based trading venues gain popularity, people trade more. However, due to the fragmentation of the markets, it may be the best policy for investors to submit smaller-size orders because one single market may not be able to absorb larger-size trades. (7)

4.2 Changes of quote competition

To compete with the NYSE, the off-NYSE exchanges post best bid and offer more and more frequently during our sample period. Table 4 shows the percentage of time that an exchange posts best quotes. Panel A presents the percentage of time that an exchange provides best bid. We find that the percentage of time that the NYSE quotes best bid drops significantly from 95.78% to 56.66%. In the meantime, all other exchanges, excluding Boston and Philadelphia Stock Exchanges which are bought by NASDAQ, quote best bid more often during later calendar quarters. Specifically, in the last quarter of 2008, 58.36% of time NYSE-Arca provides the best bid, 56.99% of time NASDAQ quotes the best bid, and 30.81% of time International Securities Exchange posts the best bid. The results regarding best offer is similar and are not reported. Panel B of Table 4 presents the percentage of time that an exchange quotes both best bid and offer at the same time. We find similar results. NYSE's quote became less and less competitive; while quotes from other exchanges become more and more competitive. At the end of 2008, the NYSE quoted best bid and offer (34.70%) less often than NYSE-Arca (36.72%) and NASDAQ (36.36%). Our results are very different from Bessembinder (2003b) which shows that in June 2000, NYSE posts both inside bid and ask for 89.11% of time, NASD quotes both best bid and offer for 4.08% of time, and all other regional exchanges rarely provide both best bid and offer at the same time. Our findings indicate that the markets for NYSE stocks become particularly competitive in recent years. The NYSE has lost its advantage to compete with other computer-based exchanges. The off-NYSE markets use quote competition to attract orders and gain market share successfully.

In Table 5, we examine whether the off-NYSE exchanges selectively compete with NYSE. Specifically, we investigate whether the changes of quote competition for large firms are significantly different from the changes for small firms. We find that NYSE quotes less aggressively for all three groups and the difference between large firms and small firms is not significant. On the contrary, National Stock Exchange, Chicago Stock Exchange, NASDAQ, NASD ADF, International Securities Exchange, and BATS, quote more aggressively for large firms than for small firms. Our results suggest that the off-NYSE markets provide liquidity selectively. This selection may deteriorate market quality because in the long run, as the last resort of trade, the NYSE incurs a financial loss to maintain liquidity but cannot get enough revenue to cover the loss. To maintain competitiveness, the NYSE market makers must reduce their liquidity services required by the affirmative obligation. As a result, in the long run, the market quality worsens because no liquidity providers are willing to provide these services.

4.3 Changes of market quality

After we explore the competition among NYSE and off-NYSE markets and find significant changes from 2004 to 2008, we investigate whether these changes improve market quality and benefit investors. In Table 6, we show the measures of market quality in many different dimensions during our sample period (8). We find that, in general, the effective spread, relative spread, price impact, and realized spread drop from the first quarter of 2004 to the second quarter of 2007 and then increase until the fourth quarter of 2008. Focusing on the differences between 2004 and 2008, we find these measures all increase significantly at 1% level from 2004 to 2008. This finding suggests that investors pay higher transaction cost, liquidity providers earn more profits, and informed traders trade more actively. Based on these measures, the market quality does not improve with the increasing competition. Consistent with prior studies, from Table 6, we also find the short-term price volatility, including both standard deviation of trading prices and absolute percentage change of quoted midpoint in 10-minute intervals, increase from 2004 to 2008 when NYSE's market share drop significantly. This result indicates that the NYSE market makers' affirmative obligation reduces the short-term price volatility.

After the computer-based exchanges earn significant market shares, investors face more risk due to the fast change of stock prices. The only positive effect of the increase competition on the market quality is the increase of quoted depth. The bid (ask) size is the sum of the quoted sizes of all liquidity providers who post best bid (offer). We find that both bid and ask sizes increase from 2004 to 2008, but only the increase of bid size is statistically significant. Based on Table 6 and Table 3, it is interesting that the quoted depth increases but investors do not want to trade large size and reduce the trade size significantly at the same time. To sum up, according to Table 6, we do not find investors benefit from the highly competitive markets.

From Figure 1 to Figure 4, we provide the graphic demonstrations for the changes of market quality proxies from 2004 to 2008 in three different size groups. In Figure 1, we find that large firms generally have smaller effective spread and relative quoted spread. The paths of three size groups are similar but the increase of transaction costs for large firms from 2004 to 2008 is not as large as the changes for small firms. In Figure 2, we find that the depth of small firms do not increase from 2004 to 2008. On the contrary, the quoted bid size and ask size of large firms increase significantly. This result suggests that the off-NYSE markets may provide liquidity selectively. Figure 3 shows the changes of price impact and realized spread from 2004 to 2008. We find that the increases in price impact and realized spread of the full sample mainly result from small and medium firms. In Figure 4, we show the changes of short-term price volatility and find for all of the three size groups, the price volatility in 10 minutes increase significantly from 2004 to 2008. To sum up, the off-NYSE exchanges provide more liquidity and gain more market shares for large firms. However, the market quality of large stocks is not necessarily worse. The impact of the competition on market quality tends to be more negative for medium and small firms than for large firms. Some of the latter observed volatility could be arising from the 2008 financial crisis.

4.4 Panel data regression

To exclude the possibility that the changes of market quality are affected by other factors, we construct a two-way fixed effects panel regression which accounts for both unobservable firm-specific effects and systematic time effects. Following Fink, Fink, and Weston (2006), we control share turnover, stock price, firm size, and daily return volatility in the regression model.

The equation of the regression is:

[Quality Measure.sub.i,t] = [[alpha].sub.i] + [[delta].sub.t] + [[beta].sub.1]ln[(NYSE Market Share).sub.i,t] + [[beta].sub.2]ln[(Share Turnover).sub.i,t] + [[beta].sub.3] ln[(Stock Price).sub.i,t] + [[beta].sub.4]ln[(Market Capitalization).sub.i,t] + [[beta].sub.5]ln[(Price Volatility).sub.i,t] + [e.sub.i,t]

where [[alpha].sub.i] and [[delta].sub.t] refer to firm-specific fixed effects and time effects, respectively. Table 7 shows the regression results. After controlling other factors affecting market quality, we still find the benefit from the increasing competition among exchanges is limited. On the one hand, we find the bid and ask depths increase and price impact decreases when the NYSE market shares drop. These findings indicate that the increasing competition increases the market depth and decreases the informed trading. On the other hand, we also find the coefficients of effective spread, relative quoted spread, realized spread, and 10-minute price volatility are all significantly positive, which would suggest that when NYSE market share decreases, investors pay more transaction costs, liquidity providers get more profits, and investors take more risk due to more volatile stock prices. Considering that the trading size drop substantially despite the increase in quoted depth, we do not find overall investors benefit from the increasing competition.

To examine whether the effects of the competition on market quality change over time, we rerun the regressions for two subperiods: period from the first quarter of 2004 to the second quarter of 2006 and period from the third quarter of 2006 to the fourth quarter of 2008. Harris (2003) suggest that when the NYSE market shares are substantially greater than 50 percent, although participants in other markets can free ride their services, the NYSE market makers can afford the free riding and still fulfill the affirmative obligation. On the contrary, when the NYSE market shares plummet, the NYSE market makers may not be able to earn enough profits to fund the liquidity they must provide. Therefore, it is possible that during the first subperiod when the NYSE still retain sufficient market shares, the advantage of competition outweigh the disadvantage of fragmentation. During the second subperiod when the NYSE market shares drop from 74% to 32%, the disadvantage of fragmentation exacerbates because the NYSE market makers may not be able to provide enough liquidity services. We find that when the NYSE still has substantial market shares, the increasing competition increases the quoted depth. However, the transaction cost, informed trading, and market maker's revenue all increase. In the second subperiod when the NYSE loses substantial market shares, the competition still increases transaction costs, market maker's revenue, and the short-term price volatility. However, the quoted depth drops and price impact decreases when the NYSE loses its market shares. Our results suggest that when the NYSE's market shares are still high, the NYSE still provides high market depth while the off-NYSE exchanges provide liquidity. However, when the NYSE's market shares plummet, the NYSE market makers decrease the quoted depth in order to compete with off-NYSE markets. We find that the only benefit from fierce competition is the decrease of informed trading. Our result is consistent with Goldstein, Shkilko, Van Ness, Van Ness (2008) which suggests that informed traders prefer venues with sufficient liquidity over those that guarantee anonymity of executions. When markets are too fragmented, it may be more difficult for informed traders to hide their trades because the order flows and market depths of each market are too small. As a result, the trading activities from the informed traders decrease.

5. Conclusion

The market competition and integration of the NYSE stocks have been one of the controversial issues in the past decades. Due to the rapid evolution of the electronic-based trading platforms, the competition on the NYSE stocks increases significantly. These computer-based exchanges and dealers/brokers can provide cheaper and faster executions because they do not have to maintain costly trading floors. They also provide liquidity rebates to attract order flow and gain popularity among investors/brokers gradually. Facing the increasing competition, the NYSE cannot maintain its advantage on trading its own-listed stocks and has lost significant market shares in recent years. Although the electronic-based trading platforms have developed and grown for years, whether investors of NYSE stocks benefit from the increasing competition is not clear. In this paper, we address this issue by investigating the effects of changes of NYSE market share on the trading quality.

Investigating 259 NYSE firms from 2004 to 2008, a period when the NYSE faces fierce competition from other electronic-based liquidity providers, we find that investors do not benefit from the increasing competition. During our sample period, the NYSE market shares decrease from 86.56% to 31.93% and the percentage of time that the NYSE provides both best bid and offer decreases from 92.40% to 34.70%. We find that although the competition increases significantly, the effective spreads and relative quoted spreads increase with the decrease of NYSE market shares. In the meantime, the liquidity providers' revenue, measured by realized spreads, and short-term price volatility increase, too. Although the quoted depth increases and the price impact decreases when the markets become more competitive, we also find that the average number of shares per trade drops from 586 shares to 190 shares, which indicates that the ability that a market can absorb a large-size trade decreases when markets are fragmented. Our findings suggest that the benefit from competition of the alternative trading systems is limited for NYSE stocks.

Our findings shed light on the on-going debate on market integration and competition. Theoretically, a consolidated market provides the greatest liquidity and facilitates best price discovery. However, it creates a monopoly which may be used by the market makers to take advantage of investors. Fragmented markets foster competition and may lead to lower execution costs. In the presence of many different trading venues, investors can choose a trading venue based on their specific needs. Nevertheless, to compete with the primary exchange, the off-NYSE markets usually selectively provide liquidity for only popular stocks, provide rebates to brokers which NYSE market makers are prohibited to use, and free-ride on the NYSE market makers' services that create externality. This competition under different regulations and rules may jeopardize the trading quality and finally hurt all investors. In recent decades, the regulators prefer competition to consolidation. Our findings shed critical period and provide new evidence on whether competitive fragmented markets provide better trading quality than a consolidated market with a physical trading floor.

HSIAO-FEN YANG

Department of Finance, University of Wisconsin, La Crosse

THOMAS M. KRUEGER

Thomas.Krueger@tamuk.edu

Department of Accounting and Finance, Texas A&M University, Kingsville

Received 13 March 2015 * Received in revised form 25 August 2015 Accepted 26 August 2015 * Available online 25 January 2016

REFERENCES

Admati, A. R., and Pfleiderer, P. (1988), "A Theory of Intraday Patterns: Volume and Price Variability," Review of Financial Studies 1(1): 3-40.

Barclay, M. J., T. Hendershott, and D. T. McCormick (2003), "Competition among Trading Venues: Information and Trading on Electronic Communications Networks," Journal of Finance 58(6): 2637-2665.

Battalio, R., A. Ellul, and R. Jennings (2007), "Reputation Effects in Trading on the New York Stock Exchange," Journal of Finance 62: 1243-1271.

Benveniste, L., A. Marcus, and W. Wilhelm (1992), "What's Special about the Specialist?," Journal of Financial Economics 32: 61-86.

Bessembinder, H. (2003a), "Quote-based Competition and Trade Execution Costs in NYSE-listed Stocks," Journal of Financial Economics 70: 385-422.

Bessembinder, H. (2003b), "Issues in Assessing Trade Execution Costs," Journal of Financial Markets 70: 385-422.

Bessembinder, H., and H. M. Kaufman (1997), "A Cross-exchange Comparison of Execution Costs and Information Flow for NYSE-listed Stocks," Journal of Financial Economics 46: 293-319.

Boehmer, E., R. Jennings, and L. Wei (2007), "Public Disclosure and Private Decisions: Equity Market Execution Quality and Order Routing," Review of Financial Studies 20(2): 315-358.

Cheng, Y., and R. Stunda (2015), "Where to Invest: NYSE or NASDAQ?," ASBBS eJournal 11(1): 68-79.

Conrad, J., K. M. Johnson, and S. Wahal (2003), "Institutional Trading and Alternative Trading Systems," Journal of Financial Economics 70: 99-134.

Diamond, S. F., and J. W. Kuan (2012), "The Importance of Institutional Design: Evidence from the New York Stock Exchange IPO and Reg. NMS." Available at SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1987275

Dodd, O., C. Louca, and K. Paudyal (2015), "The Determinants of Foreign Trading Volume of Stocks Listed in Multiple Markets," Journal of Economics and Business 79: 38-61.

Easley, D. A., T. Hendershott, and T. Ramadorai (2008), "The Price of Latency," Working Paper, Cornell University, http://www.coursehero.com/file/p5fou8r/Easley-D-Hendershott-T-Ramadorai-T- 2008(August 2014).

Easley, D., and M. O'Hara (1992), "Adverse Selection and Large Trade Volume: The Implications for Market Efficiency," Journal of Financial and Quantitative Analysis 27(2): 185-208.

Ellis, K., R. Michaely, and M. O'Hara (2000), "The Accuracy of Trade Classification Rules: Evidence from NASDAQ," Journal of Financial and Quantitative Analysis 35: 529-552.

Ferguson, N. (2015), "Investor Information Processing and Trading Volume," Asia-Pacific Journal of Financial Studies 44(2): 322-351.

Fink, J., K. E. Fink, and J. P. Weston (2006), "Competition on the NASDAQ and the Growth of Electronic Communication Networks," Journal of Banking & Finance 30: 2537-2559.

Goldstein, M. A., A. V. Shkilko, B. F. Van Ness, and R. A. Van Ness (2008), "Competition in the Market for NASDAQ Securities," Journal of Financial Markets 11: 113-143.

Handa, P., R. A. Schwartz, and A. Tiwari (2004), "The Economic Value of a Trading Floor: Evidence from the American Stock Exchange," The Journal of Business 77(2): 305-339.

Harris, J., and M. Saad (2014), "The Sound of Silence," Financial Review 49: 203-230.

Harris, L. (2003), Trading & Exchanges: Market Microstructure for Practitioners. New York: Oxford University Press, Ch. 24.

Hasbrouck, J., and G. Saar (2013), "Low-latency Trading," Journal of Financial Markets 16(4): 646-679.

Hatch, B., R. Battalio, and R. Jennings (2001), "Post-reform Market-order Execution Quality: Multidimensional Comparisons across Market Centers," The Financial Review 36(3): 123-152.

Hendershott, T., and P. C. Moulton (2008), "The Shrinking New York Stock Exchange Floor and the Hybrid Market." Working Paper, Haas School of Business, University of California, Berkeley. http://www.wlu.ca/documents/29974/HybridPaper_20080328.pdf (August 28, 2014).

Huang, R. D. (2002), "The Quality of ECN and NASDAQ Market Maker Quotes," Journal of Finance 57(3): 1285-1319.

Kwan, A., R. Masulis, and T. H. McInish (2015), "Trading Rules, Competition for Order Flow and Market Fragmentation," Journal of Financial Economics 115(2): 330-348.

Lee, C. M. C. (1993), "Market Integration and Price Execution for NYSE-listed Securities," Journal of Finance 46(2): 733-746.

McGrath, M. (2015), "NYSE Resumes Trading after Technical Glitch Causes 3.5 Hour Outage," Forbes, July 8, http://www.forbes.com/sites/maggiemcgrath/2015/07/08/new-york-stock-exchange-halts-trading/.

O'Brien, M. (2014), "Everything You Need to Know about High-frequency Trading," The Atlantic, April 14, http://www.theatlantic.com/business/archive/2014/04/everything-you-need-to-know-about-high- frequency-trading/360411/.

O'Hara, M., and M. Ye (2009), "Is Market Fragmentation Harming Market Quality?," Journal of Financial Economics 100(3): 459-474.

O'Hara, M., Y. Chen, and M. Ye (2014), "What's Not There: Odd Lots and Market Data," Journal of Finance 69(5): 2199-2236.

Sofianos, G., and I. M. Werner (2000), "The Trades of NYSE Floor Brokers," Journal of Financial Markets 3: 139-176.

Werner, I. M. (2003), "NYSE Order Flow, Spreads and Information," Journal of Financial Markets 6: 309-335.

White, E. N. (2013), "Competition among the Exchanges before the SEC: Was the NYSE a Natural Hegemon?," Financial History Review 20(1): 29-48.

[FIGURE 1 OMITTED]

This graph shows the average quarterly median of effective spread (upper chart) and relative quoted spread (lower chart) for large, medium, and small firms from 2004 to 2008. Firms are classified based on their market capitalizations at the end of Dec 2003. Sample contains 82 large firms, 90 medium firms, and 87 small firms listed on the NYSE.

[FIGURE 2 OMITTED]

This graph shows the average quoted depth for 259 NYSE firms from 2004 to 2008. The quoted depth is the average of each stock's quarterly median bid size (upper chart) and ask size (lower chart). Firms are classified based on their market capitalizations at the end of Dec 2003. The sample contains 82 large firms, 90 medium firms, and 87 small firms listed on the NYSE.

[FIGURE 3 OMITTED]

This graph shows the average quarterly median of price impact (upper chart) and realized spread (lower chart) for large, medium, and small firms from 2004 to 2008. Firms are classified based on their market capitalizations at the end of Dec 2003. Sample contains 82 large firms, 90 medium firms, and 87 small firms listed on the NYSE.

[FIGURE 4 OMITTED]

This graph shows the price volatility in 10 minutes (upper chart) and the percentage change of quote midpoint in 10 minutes for 259 NySe firms from 2004 to 2008. The price volatility is the standard deviation of trading prices in 10-minute interval. The trading hours per day are divided into 39 10-minute interval starting at 9:30 am. To avoid the effect of opening trade on the volatility, the first two 10-minute intervals are eliminated. Firms are classified based on their market capitalizations at the end of Dec 2003. Sample contains 82 large firms, 90 medium firms, and 87 small firms listed on the NYSE.

NOTES

(1.) Most research on NASDAQ stocks suggest that the competition from the electronic-based liquidity providers (particularly, Electronics Communication Networks (ECNs)) enhances trading quality by providing competitive quotes, cheap and fast executions, anonymous trading, and a high probability of executions for NASDAQ stocks (see Huang (2002), Barclay, Hendershott and McCormick (2003), Goldstein, Shkilko, Van Ness, and Van Ness (2008), Fink, Fink, and Weston (2006), and Conrad, Johnson, and Wahal (2003)). However, the ECNs are also subject to more informed trading and larger price volatility than the NASDAQ (See Barclay, Hendershott and McCormick (2003)).

(2.) See Wall Street Journal, May 7, 2010, "Exchanges Consider Stock-Specific Circuit-Breakers," and "How NYSE Humans Saved P&G Shares," by Donna Kardos Yesalavich. When P&G was bought on the NYSE floor at $56, a trade at $39.37 went through on the NASDAQ.

(3.) Based on SEC statistics, by 2000 ECNs accounted for almost 40 percent of the dollar volume of trading in NASDAQ securities.

(4.) See Lee (1993), Hatch, Battalio, and Jennings (2001), Bessembinder and Kaufman (1997), and Bessembinder (2003a).

(5.) Following Bessembinder (2003b), we omit trades with condition codes of A, C, D, N, O, R, and Z. Quotes with condition codes 4, 7, 9, 11, 13, 14, 15, 19, 20, 27, and 28 are also eliminated. When calculating the best bid and offer, these quotes are treated indicating the exchange's withdrawal from the market. To avoid the effect of opening transactions on our analysis, we delete the trades and quotes that are posted before 9:45 am.

(6.) For example, Chicago Stock Exchange sometimes posts a quote during the market opening and doesn't change the quote for the rest time of the day. However, the stock prices change in seconds and other active exchanges change their quotes quickly. As a result, if we use the quote of Chicago Stock Exchange to construct NBBO for the whole day, the market can be crossed for 80% of the time during that day.

(7.) The TAQ database does not contain order information. We obtain the reported trading size, not the order size. It is possible that large trades are executed against many small trades and thus are reported as many small trades. However, because the average trading size drops so much but the total trading volume increases, the number of small-size orders must increase much more than the number of large-size orders.

(8.) In this table we report the average of median for market quality proxy across stocks to avoid the effect of outlier due to the extreme quotes posted by the liquidity providers who do not want to trade. The results calculated from the average of mean across all stocks are similar.
Table 1 Summary Statistics

This table shows the summary statistics of the sample
used in this paper. Sample contained 259 firms listed
on the NYSE from January 2004 to December 2008. At the
end of the 2003, NYSE firms are sorted based on market
capitalizations and then assigned to one of the three
groups: large firms, medium firms, and small firms. 100
firms of each group are randomly selected. Due to missing
values, delisting, and changing listed exchanges, our final
sample contains 82 large firms, 90 medium firms, and 87 small
firms. The numbers below are the daily average during the whole
sample period. The market capitalization is computed as the daily
price times the number of shares outstanding. The turnover is
defined as the daily trading volume over the number of shares
outstanding. The daily return volatility is the standard deviation
of the daily return. The average relative quoted spread is the
average daily time-weighted bid/ask spread divided by the midpoint
of the quote. The average quoted depth is the average daily
time-weighted quoted size.

                                         Full Sample    Large Firm

Number of Firms                                  259            82
Avg Daily Price                               $34.77        $45.91
Avg Daily Capitalization Market            $8,649.66    $23,900.31
(in millions)
Avg Daily Turnover                            0.0092        0.0089
Avg Daily Share Volume                     1,729,870     4,141,730
Avg Daily Return                             0.0130%       0.0011%
Avg Daily Return Volatility                    2.73%         2.25%
Avg Relative Quoted Spread                     0.39%         0.16%
Avg Quoted Depth--Bid (in 100 shares)          21.02         41.48
Avg Quoted Depth--Ask (in 100 shares)          24.46         48.65

                                         Medium Firm    Small Firm

Number of Firms                                   90            87
Avg Daily Price                               $35.82        $23.17
Avg Daily Capitalization Market            $2,477.38       $660.59
(in millions)
Avg Daily Turnover                            0.0098        0.0088
Avg Daily Share Volume                       895,781       319,473
Avg Daily Return                             0.0069%       0.0303%
Avg Daily Return Volatility                    2.63%         3.30%
Avg Relative Quoted Spread                     0.28%         0.71%
Avg Quoted Depth--Bid (in 100 shares)          15.17          7.78
Avg Quoted Depth--Ask (in 100 shares)          17.52          8.84

Table 2 Market Shares of Each Exchange from 2004 to 2008

The table below provides the quarterly market share for
the 259 NYSE firms used in this paper. There are eleven
exchanges trading NYSE stocks: Boston Stock Exchange,
National Stock Exchange, Chicago Stock Exchange, NYSE,
NYSE-Arca, NASDAQ, Philadelphia Stock Exchange, NASD ADF
(Alternative Display Facilities), International Securities
Exchange, CBOE, and BATS. Boston Stock Exchange and
Philadelphia Stock Exchange were bought by NASDAQ in
Oct 2007 and in July 2008, respectively. The market share
is defined as the number of shares traded in a particular
stock exchange in a quarter over the total number of shares
traded in all stock exchanges in that quarter. If a stock is
traded in NYSE but not in other exchange, the market share of
NYSE is one and the market share of other exchanges is zero.
The market share is first calculated for each stock in each
quarter and then the average of all 259 stocks is taken in
each quarter. T tests are used to test the null hypothesis
that the difference between 2004 and 2008 is not difference
from zero. *, **, and *** denote significance at 10%, 5%,
and 1% level, respectively based on a two-tailed T test.

Quarter          Boston      National       Chicago

Q1, 2004          0.43%         0.15%         2.12%
Q2, 2004          0.35%         0.15%         1.52%
Q3, 2004          0.33%         0.14%         1.33%
Q4, 2004          0.37%         0.12%         1.36%
Q1, 2005          0.32%         0.10%         1.38%
Q2, 2005          0.23%         0.11%         1.18%
Q3, 2005          0.25%         0.13%         1.36%
Q4, 2005          0.16%         0.14%         1.09%
Q1, 2006          0.15%         0.21%         1.21%
Q2, 2006          0.13%         0.41%         0.94%
Q3, 2006          0.08%         0.20%         0.89%
Q4, 2006          0.05%         0.07%         0.85%
Q1, 2007          0.16%         0.14%         0.38%
Q2, 2007          0.16%         0.31%         0.29%
Q3, 2007          0.09%         0.28%         0.27%
Q4, 2007              0         0.30%         0.40%
Q1, 2008              0         0.99%         0.33%
Q2, 2008              0         1.31%         0.37%
Q3, 2008              0         1.30%         0.32%
Q4, 2008              0         0.84%         0.35%
Q4 2008-     -0.43% ***     0.69% ***     -1.77%***
Q1 2004

Quarter            NYSE         NYSE-        NASDAQ
                                 Arca

Q1, 2004         86.56%         0.63%         9.92%
Q2, 2004         88.39%         0.58%         8.83%
Q3, 2004         87.99%         0.80%         9.25%
Q4, 2004         86.44%         1.11%        10.45%
Q1, 2005         85.80%         1.18%        11.09%
Q2, 2005         84.52%         1.94%        11.93%
Q3, 2005         83.17%         1.94%        13.04%
Q4, 2005         82.94%         2.36%        13.22%
Q1, 2006         81.10%         2.41%        14.85%
Q2, 2006         79.07%         3.89%        15.48%
Q3, 2006         74.22%         5.35%        19.21%
Q4, 2006         70.02%         6.32%        22.63%
Q1, 2007         64.24%         8.62%        21.81%
Q2, 2007         60.62%        10.19%        12.91%
Q3, 2007         54.40%        12.62%        15.14%
Q4, 2007         48.35%        13.40%        16.39%
Q1, 2008         46.02%        12.98%        18.95%
Q2, 2008         38.86%        13.82%        21.10%
Q3, 2008         34.46%        15.70%        21.52%
Q4, 2008         31.93%        16.60%        20.72%
Q4 2008-    -54.64% ***    15.97% ***    10.80% ***
Q1 2004

Quarter      Philadelphia     NASD ADF    International

Q1, 2004            0.18%            0                0
Q2, 2004            0.18%            0                0
Q3, 2004            0.17%            0                0
Q4, 2004            0.16%            0                0
Q1, 2005            0.12%            0                0
Q2, 2005            0.09%            0                0
Q3, 2005            0.11%            0                0
Q4, 2005            0.09%            0                0
Q1, 2006            0.06%            0                0
Q2, 2006            0.08%            0                0
Q3, 2006            0.06%            0                0
Q4, 2006            0.06%            0            0.00%
Q1, 2007            0.07%        4.55%            0.02%
Q2, 2007            0.08%       15.32%            0.11%
Q3, 2007            0.05%       16.28%            0.83%
Q4, 2007            0.08%       19.89%            1.15%
Q1, 2008            0.05%       18.94%            1.59%
Q2, 2008            0.05%       23.07%            1.36%
Q3, 2008            0.02%       25.00%            1.60%
Q4, 2008                0       24.07%            1.83%
Q4 2008-       -0.18% ***   24.07% ***        1.83% ***
Q1 2004

Quarter             CBOE          BATS

Q1, 2004               0             0
Q2, 2004               0             0
Q3, 2004               0             0
Q4, 2004               0             0
Q1, 2005               0             0
Q2, 2005               0             0
Q3, 2005               0             0
Q4, 2005               0             0
Q1, 2006               0             0
Q2, 2006               0             0
Q3, 2006               0             0
Q4, 2006               0             0
Q1, 2007           0.01%             0
Q2, 2007           0.01%             0
Q3, 2007           0.03%             0
Q4, 2007           0.05%             0
Q1, 2008           0.14%             0
Q2, 2008           0.07%             0
Q3, 2008           0.08%             0
Q4, 2008           0.07%         3.59%
Q4 2008-       0.07% ***     3.59% ***
Q1 2004

Table 3 Trading Volume and Trading Size for NYSE-Listed Stocks from
2004 to 2008

This table shows the average quarterly turnover and the average
trading size from 2004 to 2008. Firms are classified based on their
market capitalizations at the end of Dec 2003. Sample contains 82
large firms, 90 medium firms, and 87 small firms listed on the
NYSE. The quarterly turnover is defined as the average of total
trading volume during a quarter over the average number of shares
outstanding in that quarter across stocks. The average trading size
is the number of shares per trade of all trades in a quarter for
each stock. The averages across all stocks, large firms, medium
firms, and small firms in that quarter are computed in each
quarter. The numbers below are the mean of the quarterly averages
of all 259 firms. Wilcoxon tests are used to test the null
hypothesis that the difference of trading size between 2004 and
2008 and T test are used to test the difference of quarterly
turnover. *, **, and *** denote significance at 10%, 5%, and 1%
level, respectively, based on a two-tailed test.

                 Quarterly Turnover

Quarter     Full Sample     Large Firm

Q1, 2004         0.4204         0.4101
Q2, 2004         0.3976         0.3784
Q3, 2004         0.3855         0.3797
Q4, 2004         0.4407         0.4093
Q1, 2005         0.4535         0.4184
Q2, 2005         0.4485         0.4067
Q3, 2005         0.4313         0.4040
Q4, 2005         0.4597         0.4295
Q1, 2006         0.4766         0.4200
Q2, 2006         0.5543         0.4733
Q3, 2006         0.5203         0.4523
Q4, 2006         0.5220         0.4681
Q1, 2007         0.5813         0.5320
Q2, 2007         0.6113         0.5473
Q3, 2007         0.7148         0.6722
Q4, 2007         0.6787         0.6611
Q1, 2008         0.8099         0.8283
Q2, 2008         0.7700         0.7833
Q3, 2008         0.8885         0.9870
Q4, 2008         0.9679         1.1019
Q4, 2008-    0.5474 ***     0.6918 ***
Q1, 2004

                 Quarterly Turnover

Quarter     Medium Firm     Small Firm

Q1, 2004         0.4422         0.4077
Q2, 2004         0.4609         0.3500
Q3, 2004         0.4391         0.3356
Q4, 2004         0.4658         0.4442
Q1, 2005         0.463C         0.4768
Q2, 2005         0.5054         0.4291
Q3, 2005         0.4619         0.4254
Q4, 2005         0.5105         0.4356
Q1, 2006         0.5226         0.4823
Q2, 2006         0.5963         0.5871
Q3, 2006         0.5818         0.5208
Q4, 2006         0.564C         0.5292
Q1, 2007         0.6287         0.5787
Q2, 2007         0.6567         0.6247
Q3, 2007         0.7607         0.7075
Q4, 2007         0.739C         0.6330
Q1, 2008         0.8458         0.7555
Q2, 2008         0.7876         0.7393
Q3, 2008         0.9308         0.7517
Q4, 2008         0.9995         0.8089
Q4, 2008-    0.5573 ***     0.4012 ***
Q1, 2004

                 Average Trading Size

Quarter     Full Sample     Large Firm

Q1, 2004         585.54         771.57
Q2, 2004         556.18         725.65
Q3, 2004         543.51         732.29
Q4, 2004         528.21         684.71
Q1, 2005         520.66         705.47
Q2, 2005         486.91         644.96
Q3, 2005         448.20         605.43
Q4, 2005         436.99         592.67
Q1, 2006         408.32         527.79
Q2, 2006         373.12         477.50
Q3, 2006         343.59         434.12
Q4, 2006         300.82         361.67
Q1, 2007         252.50         297.57
Q2, 2007         248.08         290.68
Q3, 2007         217.79         247.82
Q4, 2007         199.41         230.53
Q1, 2008         195.42         219.71
Q2, 2008         187.53         209.48
Q3, 2008         175.41         192.12
Q4, 2008         190.37         190.97
Q4, 2008-   -395.17 ***    -580.60 ***
Q1, 2004

                 Average Trading Size

Quarter     Medium Firm     Small Firm

Q1, 2004         526.27         471.51
Q2, 2004         521.37         432.47
Q3, 2004         507.63         402.71
Q4, 2004         480.44         430.12
Q1, 2005         461.22         407.95
Q2, 2005         450.25         375.88
Q3, 2005         407.06         342.57
Q4, 2005         392.09         336.70
Q1, 2006         377.61         327.48
Q2, 2006         345.63         303.17
Q3, 2006         321.22         281.41
Q4, 2006         281.70         263.26
Q1, 2007         237.92         225.09
Q2, 2007         236.64         219.75
Q3, 2007         207.79         199.83
Q4, 2007         192.88         176.83
Q1, 2008         189.21         178.96
Q2, 2008         182.89         171.64
Q3, 2008         171.31         163.89
Q4, 2008         183.69         196.72
Q4, 2008-   -342.58 ***    -274.79 ***
Q1, 2004

Table 4 Quote Competition for NYSE-Listed Stocks from 2004 to 2008

The table presents the percentage of time that an exchange quote
best bid (Panel A), best offer (Panel B), and both best bid and
offer (Panel C) for the 259 NYSE firms used in this paper.
There are eleven exchanges trading NYSE stocks: Boston Stock
Exchange, National Stock Exchange, Chicago Stock Exchange,
NYSE, NYSE-Area, NASDAQ, Philadelphia Stock Exchange, NASD ADF
(Alternative Display Facilities), International Securities
Exchange, CBOE, and BATS. Boston Stock Exchange and Philadelphia
Stock Exchange were bought by NASDAQ in Oct 2007 and in July 2008,
respectively. For each stock in each quarter, the daily percentage
of time quoting best bid, offer, and both best bid and offer is
first calculated. Then the quarterly average of each stock is
then computed. The numbers below are the mean of the quarterly
averages of all 259 firms. T tests are used to test the null
hypothesis that the difference between 2004 and 2008 is not
difference from zero. *, **, and *** denote significance at
10%, 5%, and 1% level, respectively based on a two-tailed
T test.

Panel A: Percentage of time that an exchange provides best bid

Quarter         Boston     National      Chicago           NYSE

Q1, 2004         0.69%        0.48%        1.45%         95.78%
Q2, 2004         0.52%        0.45%        1.01%         96.51%
Q3, 2004         0.47%        0.40%        0.89%         97.02%
Q4, 2004         0.42%        0.36%        1.24%         95.15%
Q1, 2005         0.42%        0.37%        0.74%         95.77%
Q2, 2005         0.31%        0.45%        0.75%         95.41%
Q3, 2005         0.34%        0.54%        1.00%         95.54%
Q4, 2005         0.18%        0.44%        0.96%         94.92%
Q1, 2006         0.22%        1.25%        0.71%         95.36%
Q2, 2006         0.21%       27.73%        0.62%         89.67%
Q3, 2006         0.16%        0.88%        0.71%         87.44%
Q4, 2006         0.03%        0.19%        0.72%         88.57%
Q1, 2007         0.06%       10.99%        0.11%         86.01%
Q2, 2007         0.04%       27.93%        0.33%         86.23%
Q3, 2007         0.06%       13.01%        0.28%         76.82%
Q4, 2007             0        8.46%        1.27%         75.51%
Q1, 2008             0       14.99%        1.65%         74.62%
Q2, 2008             0       21.77%        2.13%         75.00%
Q3, 2008             0       19.54%        2.07%         66.94%
Q4, 2008             0        9.62%        2.88%         56.66%
Q4, 2008-   -0.69% ***    9.14% ***    1.43% ***    -39.12% ***
Q1, 2004

Panel A: Percentage of time that an exchange provides best bid

Quarter      NYSE-Arca        NASDAQ    Philadelphia     NASD ADF

Q1, 2004        10.57%         6.89%           0.46%            0
Q2, 2004        11.00%         5.71%           0.43%            0
Q3, 2004        14.30%         3.01%           0.35%            0
Q4, 2004        21.07%         3.09%           0.39%            0
Q1, 2005        18.22%         3.89%           0.22%            0
Q2, 2005        23.93%         3.96%           0.22%            0
Q3, 2005        20.94%         5.35%           0.19%            0
Q4, 2005        23.51%         6.40%           0.07%            0
Q1, 2006        23.78%         8.20%           0.00%            0
Q2, 2006        32.61%        17.44%           0.00%            0
Q3, 2006        40.29%        45.37%           0.04%            0
Q4, 2006        46.83%        49.33%           0.00%            0
Q1, 2007        53.48%        59.55%           0.00%        1.58%
Q2, 2007        56.14%        61.42%           0.66%        6.89%
Q3, 2007        54.80%        56.10%           0.81%        5.80%
Q4, 2007        57.12%        54.13%           1.26%       12.90%
Q1, 2008        55.15%        54.43%           1.35%        4.35%
Q2, 2008        60.07%        64.33%           2.79%        3.27%
Q3, 2008         59.86%       62.80%           1.37%        3.31%
Q4, 2008         58.36%       56.99%               0        3.41%
Q4, 2008-   47.79% ***    50.10% ***      -0.46% ***    3.41% ***
Q1, 2004

Panel A: Percentage of time that an exchange provides best bid

Quarter     International         CBOE          BATS

Q1, 2004                0            0             0
Q2, 2004                0            0             0
Q3, 2004                0            0             0
Q4, 2004                0            0             0
Q1, 2005                0            0             0
Q2, 2005                0            0             0
Q3, 2005                0            0             0
Q4, 2005                0            0             0
Q1, 2006                0            0             0
Q2, 2006                0            0             0
Q3, 2006                0            0             0
Q4, 2006            0.00%            0             0
Q1, 2007            2.15%        0.05%             0
Q2, 2007            0.90%        0.12%             0
Q3, 2007           16.46%        0.19%             0
Q4, 2007           27.36%        0.22%             0
Q1, 2008           38.36%        0.29%             0
Q2, 2008           43.88%        0.03%             0
Q3, 2008           44.70%        0.01%             0
Q4, 2008           30.81%        0.03%        23.40%
Q4, 2008-      30.81% ***    0.03% ***    23.40% ***
Q1, 2004

Panel B: Percentage of time that an exchange provides both
best bid and offer

Quarter         Boston     National      Chicago           NYSE

Q1, 2004         0.02%        0.02%        0.06%          92.40%
Q2, 2004         0.02%        0.02%        0.02%          93.72%
Q3, 2004         0.02%        0.01%        0.02%          94.64%
Q4, 2004         0.02%        0.01%        0.47%          91.80%
Q1, 2005         0.01%        0.01%        0.02%          92.46%
Q2, 2005         0.01%        0.03%        0.02%          91.60%
Q3, 2005         0.01%        0.05%        0.05%          91.93%
Q4, 2005         0.01%        0.05%        0.23%          91.11%
Q1, 2006         0.01%        0.53%        0.02%          91.40%
Q2, 2006         0.00%       10.98%        0.02%          81.72%
Q3, 2006         0.01%        0.32%        0.11%          78.09%
Q4, 2006         0.00%        0.01%        0.11%          79.85%
Q1, 2007         0.00%        5.00%        0.01%          74.72%
Q2, 2007         0.01%       12.17%        0.03%          74.88%
Q3, 2007         0.01%        5.18%        0.02%          60.25%
Q4, 2007             0        3.84%        0.37%          58.09%
Q1, 2008             0        5.63%        0.60%          56.00%
Q2, 2008             0        9.71%        0.73%          56.48%
Q3, 2008             0        8.57%        0.76%          45.54%
Q4, 2008             0        3.80%        0.74%          34.70%
Q4, 2008-   -0.02% ***    3.78% ***     0.69% **     -57.70% ***
Q1, 2004

Panel B: Percentage of time that an exchange provides both
best bid and offer

Quarter     NYSE- Arca        NASDAQ    Philadelphia     NASD ADF

Q1, 2004         2.76%         1.15%           0.02%            0
Q2, 2004         2.88%         0.79%           0.02%            0
Q3, 2004         4.73%         0.41%           0.02%            0
Q4, 2004         7.55%         0.36%           0.09%            0
Q1, 2005         5.61%         0.49%           0.00%            0
Q2, 2005         7.26%         0.55%           0.01%            0
Q3, 2005         7.37%         1.09%           0.01%            0
Q4, 2005         7.18%         1.65%           0.06%            0
Q1, 2006         7.26%         2.61%           0.00%            0
Q2, 2006        11.44%         6.52%           0.00%            0
Q3, 2006        16.47%        22.64%           0.04%            0
Q4, 2006        24.29%        26.79%           0.00%            0
Q1, 2007        31.35%        37.70%           0.00%        0.41%
Q2, 2007        33.69%        39.68%           0.50%        1.78%
Q3, 2007        31.22%        34.63%           0.49%        1.43%
Q4, 2007        34.71%        33.45%           0.68%        4.93%
Q1, 2008        32.72%        33.48%           0.73%        1.34%
Q2, 2008        39.02%        44.39%           1.78%        0.36%
Q3, 2008        38.65%        43.61%           0.83%        0.34%
Q4, 2008        36.72%        36.36%           0.00%        0.37%
Q4, 2008-   33.96% ***    35.21% ***       -0.02% **    0.37% ***
Q1, 2004

Panel B: Percentage of time that an exchange provides both
best bid and offer

Quarter     International         CBOE          BATS

Q1, 2004                0            0             0
Q2, 2004                0            0             0
Q3, 2004                0            0             0
Q4, 2004                0            0             0
Q1, 2005                0            0             0
Q2, 2005                0            0             0
Q3, 2005                0            0             0
Q4, 2005                0            0             0
Q1, 2006                0            0             0
Q2, 2006                0            0             0
Q3, 2006                0            0             0
Q4, 2006            0.00%            0             0
Q1, 2007            1.32%        0.00%             0
Q2, 2007            0.53%        0.01%             0
Q3, 2007            8.33%        0.00%             0
Q4, 2007           13.67%        0.07%             0
Q1, 2008           20.16%        0.04%             0
Q2, 2008           24.86%        0.00%             0
Q3, 2008           24.96%        0.00%             0
Q4, 2008           14.66%        0.01%        12.55%
Q4, 2008-      14.66% ***        0.01%    12.55% ***
Q1, 2004

Table 5 Changes of Quote Competitiveness from 2004 to 2008 for
Different Size Groups

The table presents the changes of percentage of time that an
exchange provides the best bid for NYSE firms with different sizes.
Firms are classified based on their market capitalizations at the
end of Dec 2003. Sample contains 82 large firms, 90 medium firms,
and 87 small firms listed on the NYSE. The changes for best offer
and both best bid and offer are similar. During the sample period,
there are eleven exchanges trading NYSE stocks: Boston Stock
Exchange, National Stock Exchange, Chicago Stock Exchange, NYSE,
NYSE-Arca, NASDAQ, Philadelphia Stock Exchange, NASD ADF
(Alternative Display Facilities), International Securities
Exchange, CBOE, and BATS. Boston Stock Exchange and Philadelphia
Stock Exchange were bought by NASDAQ in Oct 2007 and in July 2008,
respectively. For each stock in each quarter, the daily percentage
of time quoting best bid is first calculated. Then the quarterly
average of each stock is then computed. T tests are used to test
the null hypothesis that the difference between 2004 and 2008 is
not difference from zero. *, **, and *** denote significance at
10%, 5%, and 1% level, respectively based on a two-tailed T test.

                            Large Firm

Exchange         Q1, 2004    Q4,2008     Difference

Boston              1.67%      0.00%     -1.67% ***
National            1.23%     16.69%     15.46% ***
Chicago             1.81%      6.26%      4.45% ***
NYSE               93.64%     52.47%    -41.16% ***
NYSE-Arca          21.11%     67.58%     46.48% ***
NASDAQ              7.49%     64.09%     56.60% ***
Philadelphia        1.l7%      0.00%     -1.17% ***
NASD ADF            0.00%      6.90%      6.90% ***
International       0.00%     42.09%     42.09% ***
CBOE                0.00%      0.08%          0.08%
BATS                0.00%     34.72%     34.72% ***

                           Medium Firm

Exchange         Q1, 2004    Q4,2008     Difference

Boston              0.36%      0.00%     -0.36% ***
National            0.22%      8.16%      7.94% ***
Chicago             1.17%      1.85%      0.68% ***
NYSE               97.19%     60.29%    -36.90% ***
NYSE-Arca           5.60%     55.29%     49.69% ***
NASDAQ              5.10%     57.60%     52.50% ***
Philadelphia        0.25%      0.00%     -0.25% ***
NASD ADF            0.00%      2.43%      2.43% ***
International       0.00%     29.21%     29.21% ***
CBOE                0.00%      0.02%      0.02% ***
BATS                0.00%     22.20%     22.20% ***

                          Small Firm

Exchange            Q1,       Q4,    Difference    Difference of Large-
                   2004      2008                  Difference of Small

Boston            0.10%    0.00 %     -0.10% **             -1.57% ***
National          0.05%     4.48%     4.43% ***             11.03% ***
Chicago           1.39%     0.76%    -0.63% ***              5.08% ***
NYSE             96.33%    56.85%   -39.48% ***                 -1.68%
NYSE-Arca         5.78%    52.84%    47.07% ***                 -0.59%
NASDAQ            8.18%    49.66%    41.49% ***             15.11% ***
Philadelphia      0.01%     0.00%    -0.01% ***             -1.16% ***
NASD ADF          0.00%     1.14%     1.14% ***              5.76% ***
International     0.00%    21.83%    21.83% ***             20.26% ***
CBOE              0.00%     0.01%     0.01% ***                0.07% *
BATS              0.00%    13.97%    13.97% ***             20.75% ***

Table 6 Changes of Effective Spread, Realized Spread, Quoted Depth,
Price Impact, Realized Spread and Price Volatility This table shows
the changes of effective spread, realized quoted spread, quoted
depth, price impact, realized spread, and price volatility from
2004 to 2008. To avoid the effect of the outlier from extreme
quotes, we report the average of each stock's median of the market
quality measures in a quarter. The effective spread is defined as
100xDx([P.sub.t]-[M.sub.t-5])/[M.sub.t-5], where D is 1 (-1) for
buyer- (seller-) initiated trades, [P.sub.t] is the transaction
price, and [M.sub.t-5] is the quoted midpoint 5 seconds before the
transaction. The relative quoted spread is the difference between
bid and ask divided by the midpoint of the quote (in percentage).
The quoted depth is measured by the quoted bid and ask size (in
hundreds). The price impact is computed as 100xDx([M.sub.t+300] -
[M.sub.t-5])/[M.sub.t-5], where D is 1 (-1) for buyer- (seller-)
initiated trades, and [M.sub.t+300] and [M.sub.t-5] are the quoted
midpoint 5 minutes after and 5 seconds before the trade,
respectively. The realized spread equals effective spread minus
price impact. The price volatility in 10-minute intervals is the
standard deviation of trading prices in 10-minute interval during
the trading hours between 9:50 am and 4:00 pm. The percentage
change of the mid-quote is the absolute difference of quoted
midpoint in 10-minute interval. For effective spread, realized
spread, and price impact the quarterly median for each stock is
calculated. For quoted depth and realized spared, the daily median
is found first and then the quarter average of daily median is
calculated for each stock. For price volatility and percentage
change of mid-quote, the quarterly median of all 10-minute
intervals is found for each stock. Wilcoxon tests are used to test
the null hypothesis that the difference between 2004 and 2008 for
quoted depth and T test are used to test all other variables. *,
**, and *** denote significance at 10%, 5%, and 1% level,
respectively, based on a two-tailed test.

Quarter      Effective      Relative       Quoted
                Spread        Quoted       Depth-
                              Spread          Bid

Q1, 2004        0.0527        0.1986         6.25
Q2, 2004        0.0528        0.3548         6.49
Q3, 2004        0.0514        0.2333         6.67
Q4, 2004        0.0447        0.1853         6.50
Q1, 2005        0.0451        0.2276         7.00
Q2, 2005        0.0485        0.1825         7.85
Q3, 2005        0.0436        0.1581         8.41
Q4, 2005        0.0411        0.1722         8.59
Q1, 2006        0.0362        0.1468        10.66
Q2, 2006        0.0371        0.1555        13.57
Q3, 2006        0.0369        0.1485        21.09
Q4, 2006        0.0323        0.1300        29.25
Q1, 2007        0.0323        0.1364        29.95
Q2, 2007        0.0313        0.0914        26.97
Q3, 2007        0.0401        0.1226        18.39
Q4, 2007        0.0414        0.1350        17.52
Q1, 2008        0.0475        0.1532        19.73
Q2, 2008        0.0405        0.1331        31.30
Q3, 2008        0.0552        0.1930        24.70
Q4, 2008        0.1170        0.4380        23.82
Q4, 08-     0.0643 ***    0.2394 ***    17.57 ***
Q1, 2004

Quarter     Quoted         Price      Realized
            Depth-        Impact        Spread
               Ask

Q1, 2004      8.17        0.0287        0.0305
Q2, 2004      8.62        0.0246        0.0317
Q3, 2004      9.08        0.0250        0.0291
Q4, 2004      9.12        0.0213        0.0256
Q1, 2005      9.78        0.0211        0.0281
Q2, 2005     10.44        0.0166        0.0343
Q3, 2005     10.85        0.0149        0.0312
Q4, 2005     11.15        0.0157        0.0290
Q1, 2006     12.52        0.0153        0.0254
Q2, 2006     16.68        0.0180        0.0243
Q3, 2006     26.29        0.0125        0.0268
Q4, 2006     35.09        0.0071        0.0255
Q1, 2007     34.52        0.0111        0.0226
Q2, 2007     31.12        0.0117        0.0215
Q3, 2007     18.39        0.0166        0.0236
Q4, 2007     17.88        0.0163        0.0257
Q1, 2008     19.72        0.0210        0.0282
Q2, 2008     30.19        0.0144        0.0278
Q3, 2008     24.7C        0.0264        0.0298
Q4, 2008     24.02        0.0469        0.0589
Q4, 08-      15.85    0.0182 ***    0.0285 ***
Q1, 2004

Quarter          Price         % Change
            Volatility     of Mid-quote
                    in    in 10 Minutes
            10 Minutes

Q1, 2004        0.0197           0.0011
Q2, 2004        0.0208           0.0011
Q3, 2004        0.0177           0.0010
Q4, 2004        0.0196           0.0010
Q1, 2005        0.0212           0.0011
Q2, 2005        0.0209           0.0011
Q3, 2005        0.0203           0.0010
Q4, 2005        0.0235           0.0011
Q1, 2006        0.0230           0.0011
Q2, 2006        0.0264           0.0013
Q3, 2006        0.0236           0.0012
Q4, 2006        0.0220           0.0010
Q1, 2007        0.0224           0.0010
Q2, 2007        0.0228           0.0009
Q3, 2007        0.0326           0.0014
Q4, 2007        0.0328           0.0015
Q1, 2008        0.0384           0.0021
Q2, 2008        0.0291           0.0015
Q3, 2008        0.0369           0.0022
Q4, 2008        0.0480           0.0042
Q4, 08-     0.0282 ***       0.0031 ***
Q1, 2004

Table 7 Panel Data Regression of Transaction Cost,
Quoted Depth, Price Impact, Realized Spread, and
Intraday Volatility on NYSE Market Share

This table shows two-way fixed effect panel data regression
of effective and relative quoted spread, quoted depth, price
impact, realized spread, and intraday volatility on NYSE
Market Share. The coefficients of the fixed effects are not
reported. The sample contains 259 NYSE firms from the first
quarter of 2004 to the fourth quarter of 2008. The NYSE market
share is the total trading volume of a stock in a quarter over
the sum of trading volume in all trading venues of that stock
in the same quarter. The average turnover is calculated as
the total share volume in a quarter divided by the average
number of shares outstanding over the same quarter.
The average price is the average daily closing price
during the quarter. Market capitalization is the average
of daily price times number of shares outstanding.
Daily return volatility is the standard deviation of daily
returns for each stock in a quarter. T tests are used to test
the null hypothesis that the coefficient is different from zero.
Numbers in the parenthesis are p-value based on a two-tailed
T test. *, **, and *** denote significance at 10%, 5%,
and 1% level, respectively.

                        Effective     Relative          Ln
                         Spread        Quoted        (Quoted
                                       Spread      Depth - Bid)

Number of                 5,180         5,180         5,180
Observations

Intercept              1.0122 ***    5.8367 ***    -3.3314 ***
                         (14.87)       (7.22)        (-7.61)

Ln                     -0.1212 ***   -0.6727 ***   -0.2975 ***
(NYSE Market Share)     (-18.87)       (-8.82)       (-7.20)

Ln (Avg. Turnover)     -0.0487 ***   -0.3149 ***    0.2375 ***
                        (-22.56)      (-12.30)       (17.12)

Ln (Avg. Price)        -0.0097 **     0.1450 **    -0.9746 ***
                         (-2.39)       (3.01)        (-37.27)

Ln (Market             -0.0324 ***   -0.2144 ***    0.2099 ***
Capitalization)          (-8.65)       (-4.83)        (8.72)

Ln (Daily Return         0.0275      0.1034 ***    -0.3284 ***
Volatility)              (8.44)        (2.68)        (-15.68)

R-Square                 0.6854        0.3843         0.8943

                            Ln           Price       Realized
                         (Quoted        Impact        Spread
                       Depth - Ask)

Number of                 5,180          5,180         5,180
Observations

Intercept              -4.3884 ***    0.3837 ***    0.4009 ***
                         (-9.87)        (9.10)        (7.35)

Ln                     -0.1564 ***    0.0243 ***    -0.1114 ***
(NYSE Market Share)      (-3.73)        (6.10)       (-21.64)

Ln (Avg. Turnover)      0.3165 ***    -0.0130 ***   -0.0233 ***
                         (22.46)        (-9.72)      (-13.48)

Ln (Avg. Price)        -1.0183 ***    0.0080 ***    -0.0215 ***
                         (-38.35)       (3.18)        (-6.59)

Ln (Market              0.2386 ***    -0.0100 ***   -0.0125 ***
Capitalization)           (9.76)        (-4.32)       (-4.18)

Ln (Daily Return       -0.3567 ***    0.0149 ***    0.0110 ***
Volatility)              (-16.77)       (7.38)        (4.23)

R-Square                  0.8977        0.4158        0.5468

                       Price Volatility   % Change of
                        in 10 Minutes     Mid-quote in
                                           10 Minutes

Number of                   5,180            5,180
Observations

Intercept                   0.0268         0.0020 ***
                            (1.48)           (4.38)

Ln                       -0.0129 ***      -0.0002 ***
(NYSE Market Share)        (-7.53)          (-4.52)

Ln (Avg. Turnover)        -0.0015 **       0.0002 ***
                           (-2.57)          (15.10)

Ln (Avg. Price)           0.0288 ***        -0.0004
                           (26.66)          (-12.98)

Ln (Market                 -0.0007         0.0001 ***
Capitalization)            (-0.67)           (5.96)

Ln (Daily Return          0.0157 ***       0.0006 ***
Volatility)                (18.18)          (28.46)

R-Square                    0.7669           0.8768
COPYRIGHT 2016 Addleton Academic Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Yang, Hsiao-Fen; Krueger, Thomas M.
Publication:Economics, Management, and Financial Markets
Article Type:Report
Geographic Code:1U2NY
Date:Jun 1, 2016
Words:14051
Previous Article:Enhancing poor and middle class earning capacity with stock acquisition mortgage loans.
Next Article:The impact of liquidity on the cross section of equity returns on the Johannesburg Securities Exchange.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters