Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in an Electronic Trading Environment

Switzer, Lorne N.; Fan, Haibo Article Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in an Electronic Trading Environment International Econometric Review (IER) Provided in Cooperation with: Econometric Research Association (ERA), Ankara Suggested Citation: Switzer, Lorne N.; Fan, Haibo (2010) : Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in an Electronic Trading Environment, International Econometric Review (IER), ISSN 1308-8815, Econometric Research Association (ERA), Ankara, Vol. 2, Iss. 1, pp. 11-35 This Version is available at: https://hdl.handle.net/10419/238789 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ International Econometric Review (IER) 11 Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in an Electronic Trading Environment Lorne N. Switzer and Haibo Fan  Concordia University and Concordia University ABSTRACT The behaviour of limit order quotes and trading activity are studied using a unique and rich database that includes the identity of market participants from a fully automated derivatives market. The analysis is performed using transactions records for three aggregated trader types and three trade identifiers, with trades stamped in milliseconds for the SXF, the equity futures contract of the Montreal Exchange. The identifiers distinguish trades between principals; agency based trades, as well as transactions that are conducted for risk management as opposed to speculative purposes. Agency related trades are shown to represent the largest amount of trading activity relative to other account types. Over 90% of trades in this electronic market are limit orders. The limit order book, especially the depth 1 order, has a dominant role in providing liquidity and in explaining market participants’ trading behaviour. Participants in the SXF reference their trades to the best limit order depth. Hence, investors with large positions or investors who want to build a large position have to strategically split large orders to close/build their position, according to the depth of the best limit order, to ameliorate price impact and information leakage effects. In addition, the results show that traditionally measured spreads have no relationship with trading costs. Key words: Limit Orders, Trading Activity, Transactions Costs, Electronic Trading JEL Classifications: G13, G14, G18 1. INTRODUCTION Many exchanges in the world have shifted to the computerized trading systems for equities or derivatives. This shift has in part been driven by the belief that the computerized trading offers lower transaction costs than traditional floor-based trading systems. One of the distinguishing features of a computerized trading market vs. a floor trading market is its transparent limit order book. Whether or not the increase in transparency of the computerized market comes at the expense of higher transactions costs and diminished liquidity has been a matter of considerable debate in the literature. The purpose of this study is to provide new evidence on this score using quotations and trades from an electronic market that operates via a limit order book. Our database is particularly rich in that it captures differences in transactions costs and trading activity between principals (vs. agents) as well as transactions that are conducted for risk management as opposed to speculative purposes.  Lorne N. Switzer, Corresponding author, Associate Dean, Research, Professor of Finance, Van Berkom Endowed Chair, and Associate Director, Institute for Corporate Governance of Private and Public Organizations, Finance Department. John Molson School of Business, Concordia University, 1455 De Maisonneuve Blvd. W., Montreal, Quebec, CANADA H3G 1M8, (email: [email protected]). Tel. +514 848 2960, Fax: + 514 481 4561. Haibo Fan, Research Associate, Concordia University. Financial support from the Autorite des Marches Financiers du Quebec and the Bourse de Montreal to Switzer is gratefully acknowledged. Please address all correspondence to Dr. Lorne N. Switzer. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 12 A number of theoretical models have been proposed on how the limit order book affects liquidity and conveys information about the market. Such information might be expected to affect the trading conduct of market participants (e.g. Glosten, 1994; Handa and Schwartz, 1996; Foucault, 1999; Handa et al., 2003 and Van Achter, 2009). Foucault et al. (2005) propose an equilibrium model for order placement and argue that the proportion of patient traders in the population and the order arrival rate are the key determinants of the limit order book dynamics. In addition, a number of empirical studies have examined the order submissions process as market conditions change (e.g. Biais et al., 1995; Griffiths et al., 2000; Ahn et al., 2001 and Hollifield et al., 2003). Chan (2005) find orders at the best quotes react more quickly and completely to the adjustment than orders that are far away from the best quotes using the limit-order book and previous price movements for active stocks traded on the Stock Exchange of Hong Kong. Moreover, several other studies have appeared that examine the role of the limit-order books in supplementing liquidity. For example, Degryse et al. (2005) analyze the resiliency of a pure limit order market and find that relative to the sample average, depths stay around their mean before and after aggressive orders, whereas spreads return to their mean after about twenty best limit updates. The initial price impact of the aggressive order is partly reversed in the subsequent transactions. Coppejans et al. (2003) study the resiliency of the Swedish stock index futures market (OMX) and find that shocks to depth are restored in less than 60 minutes. On the whole, the role of information contained in the limit order book in influencing participants’ trading behaviour remains an unsettled matter in the literature. Madhavan et al. (2005) document a reduction in liquidity on the TSE following the increase in order book disclosure. They report an increase in quoted and effective spreads, reduction in depth at the best quotes and an increase in volatility. Bortoli et al. (2006) show that after the increase in pre-trade transparency limit of its limit order book, the SFE experienced a reduction in depth (mean and median), and some widening of bid-ask spreads. These studies suggest that in a transparent market, limit order traders charge market order traders a higher premium for execution certainty by withdrawing depth from the best quotes, or by increasing bid-ask spreads. In sharp contrast, Boehmer et al. (2005) document a reduction in effective spreads on the NYSE following the introduction of the Open Book, consistent with an increase in market liquidity. Our analysis is performed for one of the most important futures products on the Montreal Exchange, the S&P/TSX 60 index futures (SXF) contract. Since 2001, the Montreal Exchange has operated as a fully automated derivatives market and currently serves as the exclusive market for financial derivative products in Canada. The transaction records are timed by milliseconds and reflect actual trading activity that spans the period January 2005 to May 2006. To the best of our knowledge, our work is the first attempt to use actual trading records for a comprehensive set of trader categories to study the limit order book in financial futures products. We find that that most transactions (well over 90%) in this electronic market are from limit orders. In addition, measures of trading activities and traditionally measured spreads have no relationship with trading costs. Limit orders in the electronic system of the Montreal Exchange serve to replace traditional market markers to provide market liquidity. The order information shown on the screen systematically influences the trading behaviour of participants in the markets. Participants in the markets respect the limit order book, especially the quotes of first priority order (best 1), to make their trading decisions. We find that SXF International Econometric Review (IER) 13 market in consistent with a dynamic equilibrium process: and the size of best 1 quote is much more than the size of transaction per trade in. In particular, the liquidity/buffer provided the best 1 quote is responsible for that measure of trading activities that have no demonstrative relationship with trading costs. The remainder of this paper is organized as follows: in the Section 2, we describe the data set used in this work. The empirical analyses follow in Section 3. The paper concludes in Section 4. 2. DATA The Montreal Exchange is the exclusive exchange for trading financial derivative products in Canada. In 2001, the exchange became the first traditional exchange in North America to be fully automated. Clients’ orders are filled on a “first in, first out” (FIFO) basis. Market orders to buy futures contracts are executed at the offer price (ask); market orders to sell are executed at the bid price (bid). However, an investor wishing to buy or sell at a specific price can provide a limit order. This order is registered in SAM’s (Montréal Automated System) electronic order book and is executed when there is a counterparty interested in transacting at that price. Orders are matched and both orders are filled at the specified price for the smallest quantity posted. Since all of the trading of any specific futures contract is concentrated on one trading platform, SAM, participants are assured of buying at the lowest offer price or selling to the highest bid. The Exchange also provides a system of specialists and market-makers. The specialist is responsible for the opening of each product, and is required to be at his or her post no less than 30 minutes before the opening signal of a trading session. Market-makers are obligated to maintain 50% of their quarterly activities in their assigned product. Transactions of specialists and market-makers in any security on which they have assumed responsibilities are required to be of a stabilizing nature. They are prohibited from making trades that may be disruptive of stability, such as purchases (sales) made at a price above (below) the last preceding different –priced trade while establishing or increasing a position. The S&P/TSE 60 index futures (SXF) contract is one of the most important futures products in the Montreal Exchange. The underlying product of the SXF is the S&P/TSE 60 index, an equity portfolio composed of 60 highly liquid Canadian equities. Standard & Poor’s Corporation calculates and disseminates index prices. The SXF contract is quoted in index points, expressed to two decimals and the nominal value of one contract is C$200 times the index. The minimum price fluctuation of SXF is 0.05 index points. The value of 0.10 index point change is C$20 per contract. SXF contract months are March, June, September and December; and it is cash settlement and trades between 9:30 a.m. to 4:15 p.m. (EST). The data that we use in this study are considerably richer than those used in most previous studies that we are aware. The sample covers the 356 trading day period from January 2005 to May 2006. Our data allow us to identify trades as Limit, Market or Market on Opening transactions. Our data set provides real time quotes for bid and ask positions at various prices. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. Best 2 Bid is the second best bid Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 14 depth, and so forth. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. We define Best 1 Ask as the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. Best 2 Ask is the second best ask depth. In addition we have two sets of identifiers for each transaction to distinguish between those that were buyer or seller initiated (buy or sell). First, each buy or sell is marked by four types of aggregated accounts: Client, Pro, Shareholder NonClient and Firm. Client accounts are defined as accounts established by an approved participant that is confined to Exchange transactions executed by and positions carried by the approved participant on behalf of his clients. Firm accounts are accounts established by an approved participant, that is confined to Exchange transactions executed by and positions carried by the approved participant on behalf of the approved participant. Pro accounts are accounts established by an approved participant, that is confined to Exchange transactions executed by and positions carried by the approved participant on behalf of a market maker. A Pro can be a market maker or a liquidity provider. In addition, the trading records also identify each buy or sell as representing transactions of Hedgers, or Speculators, or Market Makers. Since Shareholder NonClient only has very few transactions in the records, we use the trading records on the Client, Pro, Firm account categories and transactions by Hedger, Speculator and Market Maker groupings in our analysis. Similar to Locke and Venkatesh (1997), the nearby SXF contract is selected each day, and it is rolled forward to the subsequent contract on the date when the maximum daily trading volume (Client account) switches from the nearby contract to the subsequent contract. 3. EMPIRICAL ANALYSES 3.1. Transaction Costs and Trading Activity Several studies have appeared that explore the relationship between trading activity and transaction costs, and typically find that there are economies to scale in trading. Most of these studies use bid-ask spreads as measures of transactions costs. For example, Demsetz (1968) and Epps (1976) find that trading activity is inversely related to trading costs on NYSE stocks. Martell and Wolf (1987) conclude that trading volume is not only a function of volatility, but also of open interest, interest rates, exchange rates, and other variables. Haller and Stoll (1989) reported an inverse relationship between bid–ask spreads and trading volume in the German auction equity market. In addition, George and Longstaff (1993) document a negative relationship between transaction rates and bid–ask spreads for the S&P 100 index options market. Wang et al. (1994) state that the major factors affecting bid–ask spreads are price risk, trade volume, and market competition. Wang et al. (1997) report a positive relationship between trading volume and intraday price volatility, and an inverse relationship between trading volume and bid–ask spreads, after controlling for other factors. Locke and Venkatesh (1997) show that bid-ask spreads are extremely problematic measures of transactions costs. For spreads to measure true transactions costs, it must be the case that customers trade exclusively with market makers. This assumption is clearly violated by many trading venues, particularly electronic platforms. In addition, as Stoll (1989) notes, even if all trades are mediated through market makers, the quoted bid-ask spread overstates actual transaction costs if some customers are better informed than the market makers or market International Econometric Review (IER) 15 makers adjust the bid – ask spread prices to manage inventory levels. Real world factors also imply that bid-ask spreads are not the best measure of the costs of trading, since intercustomer trades do take place in most financial markets that may eliminate transaction costs in aggregate. To improve upon spreads as measures of transactions costs, a number of studies have used Computerized Trade Reconstruction (CTR) audit trail data from the CME and employ accounting FIFO (First in, First out) trading profits to estimate the transaction costs (e.g. Chang et al., 1994; Fishman and Longstaff, 1992; Manaster and Mann, 1996; Chang and Locke, 1996; Locke and Venkatesh, 1997; Ferguson and Mann, 2001; Locke and Sarajoyi, 2004 and Kurov, 2005). Accounting FIFO trading profit estimates provide a direct measure of transactions costs when transactions can be classified by trader identity. Similar to Locke and Venkatesh (1997), Ferguson and Mann (2001), Locke and Sarajoyi (2004), and Kurov, (2005), we use accounting FIFO estimates to estimate trading costs, but with a much more refined data set1. In this study, we calculate three cost/profit estimates on each day. First, we use the accounting FIFO rule to determine trading profits per contract (PROFIT TRADE) and obtain the inventory positions for each of the six (trader) account types on the SXF. Second, the profit settled per contract (PROFIT SETTLED) for the inventory positions of each account type is determined by assuming that the inventories in each aggregated account are settled at the closing price of each contract. Finally, we estimate average profits per contract for each type of account as the weighted average FIFO profits and settled profits of inventories with weights given by the number of contracts traded or settled. To illustrate the FIFO rule, suppose that there are only four transactions during a day from investors in a Client account. At time t1, investors in the Client account buy 50 contracts of SXFH with a price of 500.0; at time t2, investors in the Client account sell 20 contracts of SXFH with a price of 500.2; at time t3, investors in the Client account buy 30 contracts of SXFH with price equal to 500.3; at time t4, investors in the Client account sell 58 contracts of SXFH at the price of 500.5. Suppose also that the average last bid and ask for SXFH at the end of the day is 500.8. First, we use the accounting rule of FIFO to get profit trade per contract (the value of 0.10 index point change is C$20 per contract): C$20 *(20*(500.2-500.0) +30*(500.5-500.0) +28*(500.5-500.3))/78= C$6.3077 per contract; the inventory position in the Client account on SXFH is long 2 contracts at the end of that day. Second, the profit settled per contract for the inventory position is: C$20 *2*(500.8-500.3)/2 =C$10.0000 per contract. Finally, average profits per contract for the Client account on that day are: (C$6.3077*78+ C$10.000*2)/80=C$6.4000 per contract. 1 This procedure is alluded to in Stoll (1989), and also implemented by Beebower and Priest (1980) and Baesel et al. (1983), among the other studies references. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 16 Account Dependent (C$) Independent Variable Panel A: Ask C Q TV Std Q Std TV Std P D PValue DW Average Coef. -108.36 -6.5817 34.446 1.0964 -1.7344 22.501 44.448 0.1544 2.0290 Client profit Prob. 0.0147 0.1661 0.0739 0.2522 0.2127 0.0492 0.3661 Profit Coef. -180.09 -1.7260 54.328 -0.1812 -2.1517 28.877 -12.279 0.0182 2.1070 trade Prob. 0.0000 0.7103 0.0041 0.8466 0.1145 0.0101 0.7986 Average Coef. 118.46 10.430 -43.779 -1.6332 8.0110 -41.331 -229.30 0.2223 1.9660 Firm profit Prob. 0.1439 0.2289 0.1292 0.3632 0.2295 0.0539 0.5167 Profit Coef. 262.79 2.6967 -70.060 -0.5724 9.8841 -53.598 -189.73 0.0112 2.0240 trade Prob. 0.0005 0.7376 0.0092 0.7316 0.1110 0.0073 0.5637 Average Coef. -8.8075 0.8319 12.349 -0.3283 0.7803 0.0900 0.7531 1.9390 Pro profit Prob. 0.7873 0.6309 0.6275 0.3297 0.9419 0.9811 Profit Coef. 28.610 0.5111 -11.752 -0.2506 6.1457 -0.7388 0.9035 2.0190 trade Prob. 0.3775 0.7662 0.6418 0.4535 0.5632 0.8447 Average Coef. 25.380 2.5216 -4.5622 -0.0812 0.2887 -28.589 -13.314 0.3408 1.8370 Hedger profit Prob. 0.5932 0.6246 0.7904 0.9385 0.8548 0.0258 0.8061 Profit Coef. 66.193 0.5115 -14.229 0.4727 0.8518 -30.773 -17.569 0.1347 2.0680 trade Prob. 0.1045 0.9078 0.3335 0.6002 0.5288 0.0052 0.7055 Average Coef. -8.1343 -0.8392 5.1157 -0.5426 -1.5313 12.491 62.172 0.6682 1.5440 Speculator profit Prob. 0.8905 0.8797 0.8694 0.6118 0.5125 0.3180 0.6018 Profit Coef. -52.582 -0.3300 12.938 -0.5783 -1.1283 17.443 28.987 0.3227 2.0570 trade Prob. 0.2249 0.9353 0.5703 0.4604 0.5101 0.0574 0.7398 Average Coef. 0.5252 -1.1821 3.7974 0.1573 -0.4434 -0.1180 -161.44 0.1205 1.9160 Market profit Prob. 0.9834 0.4616 0.8321 0.6278 0.9326 0.9759 0.0159 Maker Profit Coef. 37.822 1.8713 -25.596 -0.2179 -6.2998 -2.1435 -59.439 0.0264 1.9650 trade Prob. 0.1896 0.3057 0.2095 0.5549 0.2910 0.6297 0.4334 Panel A: Bid C Q TV Std Q Std TV Std P D PValue DW Average Coef. -124.68 0.9521 18.196 -0.0707 -0.9462 26.655 41.680 0.2766 2.0190 Client profit Prob. 0.0045 0.8290 0.3745 0.9401 0.5117 0.0222 0.4047 Profit Coef. -184.24 -0.3715 52.341 -0.4130 -2.0493 29.682 -12.183 0.0196 2.1100 trade Prob. 0.0000 0.9313 0.0092 0.6530 0.1461 0.0092 0.8030 Firm Average Coef. 160.52 -7.9732 -10.588 1.3676 5.0756 -53.334 -136.08 0.2616 1.9620 profit Prob. 0.0463 0.3133 0.7241 0.4207 0.3660 0.0149 0.6195 Profit Coef. 280.48 -4.4444 -55.807 0.4748 8.5710 -58.661 -140.28 0.0098 2.0220 trade Prob. 0.0002 0.5447 0.0456 0.7632 0.1006 0.0040 0.5813 Pro Average Coef. -12.538 0.4137 16.378 -0.2812 0.3575 0.1072 0.7412 1.9320 profit Prob. 0.6961 0.7724 0.5022 0.3576 0.9734 0.9779 Profit Coef. 27.013 0.8279 -11.465 -0.3438 5.5310 -0.2964 0.8207 2.0220 trade Prob. 0.3965 0.5597 0.6358 0.2571 0.6034 0.9383 Hedger Average Coef. 36.850 -0.9715 -0.3479 0.5765 0.0779 -30.775 -17.191 0.3390 1.8420 profit Prob. 0.4237 0.8384 0.9845 0.5784 0.9621 0.0178 0.7678 Profit Coef. 69.355 -1.8266 -9.7559 0.9734 0.4283 -32.241 -5.0532 0.1182 2.0540 trade Prob. 0.0792 0.6543 0.5254 0.2735 0.7605 0.0038 0.9193 Speculator Average Coef. -10.911 1.2439 0.6629 -1.0037 -1.2623 14.237 54.240 0.6090 1.5390 profit Prob. 0.8539 0.8016 0.9835 0.3244 0.5940 0.2638 0.6501 Profit Coef. -55.256 0.1011 13.772 -0.7498 -1.0708 17.959 23.062 0.2732 2.0460 trade Prob. 0.2035 0.9778 0.5568 0.3148 0.5370 0.0547 0.7923 Average Coef. -0.1091 1.1906 -5.2203 -0.2607 -0.4202 1.0497 -151.65 0.1121 1.9310 Market profit Prob. 0.9966 0.3721 0.7583 0.3817 0.9360 0.7918 0.0246 Maker Profit Coef. 30.657 3.8476 -30.007 -0.5564 -7.8699 -0.3455 -35.035 0.0026 1.9910 trade Prob. 0.2851 0.0109 0.1176 0.0985 0.1831 0.9386 0.6443 Table 3.1 OLS Estimates of the Regression of SXF Transaction Costs with Measures of Trading Activity and Limit Orders. Notes: OLS estimates of the regression of SXF daily transaction profit/Cost in C$ per contract and measures of trading activity and limit order are presented. In the table, TV is Mean Trade Volume. Q is the daily mean quantity of Best 1 Ask for Panel A and Best 1 Bid for Panel B. D is a dummy variable, which equals 1 when TV +3 * Std TV > Q + 3 *Std Q. Std stands for standard deviation. DW is Durbin-Watson stat. PValue is the p value of the regression. P is transaction price. Profit trade is daily FIFO profit for each account type of Firm, Client, Pro, Hedger, Speculator and Market Maker. Average profit is the weight average profits of the FIFO profit and the profits of daily inventory imbalance settled at the closing price. The number of contracts traded or settled is used as the weight. C is constant term in the OLS model. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. Best 1 Ask is the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. Value in Bold indicates significant at 5 percent level. Data for 356 trading days are used in the regression. International Econometric Review (IER) 17 Since a buy /sell in an aggregated account could be closed but then immediately followed a sell/buy, our FIFO cost estimate is largely immune to any informational effects. As an extreme example, suppose an investor in the Client account buys 30 contracts of SXFH; then suppose that immediately after a millisecond, another investor in Client account sells 30 contracts of SXFH. The position in the Client account is closed in a millisecond. The calculated FIFO costs in this case will reflect only liquidity costs. In this work, we first sort the records by the date and hour, and then by milliseconds to get the FIFO transaction series for a daily session. The last average of bid and ask price is used as the closing price on the trading day2. Many previous studies suggest that trading activity increases market liquidity. However, accurate measures of trading activity, such as trading volume per transaction are not available, and as a result, activity levels are usually measured by the number of transactions (McInish and Wood, 1992). In contrast, our data include the trading volume per transaction for each account type and the actual price executed, which we can use as the independent variables as specified in our model of the determinants of trading costs. In particular, to further investigate the relationship between the transaction costs and trading activity, we estimate the following regression model:   DPStdTVStdQStdTVQCCosts 54321 (3.1) where Costs is our measures of transaction costs; C is constant term; Q is the daily mean quantity of Best 1 Ask or Bid; TV is mean trade volume per transaction on each day; P is transaction price; and D is a dummy variable, which equals 1 when TV +3 * StdTV > Q + 3 *StdQ. Std stands for standard deviation for the above measures. OLS estimation results on both the Bid and Ask sides for three aggregated account types of Client, Pro, Firm and transaction by Hedger, Speculator and Market Maker are presented in Table 3.1. The results are similar whether we use the measures of Bid side or the measures of Ask side. Overall, the regression results show virtually no relationship between transaction costs and measures of market activity for the SXF. Most coefficients of market activity measures are insignificant at 5 percent level. 3.2. Limit Order Quotes and Trading of the SXF Why are transactions costs unrelated to measures of trading activity? In this section, we explore the role of limit orders. In Table 3.2 we provide summary statistics on limit order quotes and limit order size. For each trading day, we cumulate the records of each limit order quote and limit order quantity that is submitted over the day. Table 3.2 reports the cumulative averages across all trading days in the sample. As is shown in the table, the Best 1 and 2 variables, which capture the number of contracts submitted as limited orders at best and second best prices for market participants are similar on the bid and ask sides (Panel A vs. Panel B). In addition, the average of daily standard deviation of limit order quantity (Average Std Q), the average of maximum quantity of limit order in each trading day (Average max Q) or the mean of daily standard deviation of the average size of limit order per order (Std Average Size) are similar on the bid and the ask side. 2 See also Switzer and Fan (2007). Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 18 Average Mean Q Mean Average Size Average Std Q Std Average Size Average Max Q Sample Days Panel A: Bid Jan 05 - May 06 Best 1 10.95 4.06 27.00 5.41 294.18 356 Best 2 11.47 4.49 11.02 4.58 134.40 356 Best 1 and 2 11.23 4.28 21.01 5.10 300.01 356 Panel B: Ask Jan 05 - May 06 Best 1 10.24 4.00 26.76 5.60 287.63 356 Best 2 9.65 4.14 10.10 4.52 121.62 356 Best 1 and 2 9.95 4.07 20.50 5.17 288.51 356 Panel C: Bid Jan 05 - May 05 Best 1 13.86 4.72 34.99 6.23 326.31 104 Best 2 14.02 5.19 13.29 5.27 127.52 104 Best 1 and 2 13.93 4.95 27.16 5.85 326.61 104 Panel D: Ask Jan 05 - May 05 Best 1 13.20 4.67 34.93 6.51 336.88 104 Best 2 12.07 4.72 12.64 5.27 132.79 104 Best 1 and 2 12.66 4.70 26.86 6.09 337.27 104 Panel E: Bid Jan 06 - May 06 Best 1 9.34 3.61 23.77 4.79 291.52 105 Best 2 9.58 4.09 10.19 4.64 162.99 105 Best 1 and 2 9.48 3.86 18.40 4.77 309.95 105 Panel F: Ask Jan 06 - May 06 Best 1 8.48 3.54 23.18 5.21 267.90 105 Best 2 7.72 3.79 8.70 4.49 117.29 105 Best 1 and 2 8.09 3.67 17.53 4.92 268.07 105 Table 3.2 Summary statistics for SXF limit orders. Notes: Summary statistics for limit order quotes for the SXF are shown. Daily statistics are used to calculate the averages. Average mean Q is the average of daily mean quantity of limit orders. Average Std Q is the average daily standard deviation of limit order quantities. Average max Q is the average of the maximum quantity of limit orders in each trading day. Mean Average Size is the mean of the daily average size of limit order per order. Std Average Size is the mean of daily standard deviation of the average size of limit order per order. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. Best 2 Bid is the second best bid depth, and so forth. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. We define Best 1 Ask as the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. Best2 ask is the second best ask depth. Best 1and 2 combine the records of the Best 1 limit order and the Best 2 limit order. The sample consists of 356 days from January 04, 2005 to May 30, 2006. Statistics of two sub-periods from Jan. 05 to May 05 and from Jan. 06 to May 06 are also presented. Summary statistics for transactions cumulated on a daily basis are provided in Table 3.3. As shown in the panel A of the table, for the entire market (All account), the average of daily mean of the number of the nearby SXF contract traded per transaction (Mean Trade Volume) is 2.49 contracts and the average maximum number of the nearby SXF contract traded per transaction on each trading day (Max Trade Volume) is 262.92 contracts. The Client category has a higher value in the average of the daily number of the nearby SXF contract traded (Number Trading) than the Firm, Pro, Hedger, Speculator or Market Maker categories. It is therefore quite obvious that Market Makers/Pros do not precipitate all Client transactions. Moreover, the Pro/Market Maker category exhibits the lowest values in Mean Trade Volume, Max Trade Volume and Std Trade Volume (the average of the standard deviation of the number of nearby SXF contract traded per transaction on each day) across the International Econometric Review (IER) 25 measures when an intercept and trend model is used to the tests5. The unit root test results are consistent with a dynamic competitive equilibrium process for the SXF market, where shocks to the variables do not have persistent effects. Augmented Dickey-Fuller Augmented Dickey-Fuller Augmented Dickey-Fuller Augmented Dickey-Fuller Augmented DickeyFuller Intercept None Intercept None Intercept None Intercept None Intercept None Panel A: Market SXF Mean BidQ Mean AskQ Std BidQ Std AskQ Std Price 0.000 0.086 0.000 0.097 0.000 0.007 0.000 0.007 0.000 0.187 Panel B: Account Type Variable All Client Firm Pro Number Trade 0.001 0.344 0.001 0.293 0.001 0.340 0.000 0.314 Mean Trade Volume 0.000 0.263 0.000 0.182 0.000 0.278 0.003 0.458 Std Trade Volume 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.078 Variable Hedger Speculator Market Maker Number Trade 0.002 0.238 0.005 0.397 0.000 0.275 Mean Trade Volume 0.000 0.175 0.000 0.318 0.005 0.398 Std Trade Volume 0.000 0.000 0.000 0.000 0.000 0.006 Table 3.7 Unit Root Tests on the Daily Measures of the SXF Market Activity. Notes: MacKinnon (1996) one-sided p-values of Augmented Dickey-Fuller tests with four lags on daily measures of the SXF market activity are presented. Column Intercept and None are the results of Augmented Dickey-Fuller model with or without intercept respectively. In Panel A, Mean BidQ and Mean AskQ are the daily mean quantity of the best1 Bid or the best1 Ask per record in the data file; Std BidQ and Std AskQ are the daily standard deviation of the quantities of the best1 Bid or the best1 Ask per record in the data file; Std Price is the daily standard deviation of transaction prices. In Panel B, Column All is the transaction by all participants in the SXF market; Client is the results from buy or sale by the aggregated Client; Firm is the results from buy or sale by the aggregated Firm; Pro is the results from buy or sale by the aggregated Pro; Hedger is the results from transaction by Hedgers, Speculator is the results from transaction by Speculator and Market Maker is the results from transaction by Market Maker. In the column Variable, Number Trade is the number of trading on each day; Mean Trade Volume is the daily mean of the number of the SXF contract traded per transaction; and Std Trade Volume is the standard deviation of the number of the SXF contract traded per transaction on each day. Data over 356 trading days are used in the test. Figure 3.1 and Figure 3.2 illustrate the mean reverting behaviour of the measures of market activities in our sample period; in addition, Figure 3.1 also demonstrates that the size of trading per transaction is much less than the sizes of the Best 1 limit orders. 3.3. Cost, Spreads, and Minimum Tick Sizes To provide further evidence that SXF market is in equilibrium and the inappropriateness of spreads in an equilibrium limit order market, we also calculate the spreads. 5 The results for trend and intercept model are not reported in the table to save space. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 26 Figure 3.1 Daily Mean Quantity Changes of Trading and the Best 1 Limit Order Notes: Figure 3.1 shows the daily mean trading volume (Trade V) per transaction, the daily mean quantity of the Best 1 Bid (Bid) and the daily mean quantity of the Best 1 Ask (Ask) per record in the data file over the sample period (356 trading days). Figure 3.2 Daily Volatity Measure Changes of Trading and the Best 1 Limit Order Notes: Figure 3.2 shows, the daily standard deviation of trading volume per transaction (StdT), the daily standard deviation of the Best 1 Ask quantity per record (StdA) and the daily standard deviation of transaction price (StdP) are present. The axis of Ask& Trade is for StdT and StdA; the axis of Price is for StdP. The sample has 356 trading days. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. Best1 Ask is the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. 0 10 20 30 40 50 60 1 51 101 151 201 251 301 351 Number Contract Day Ask Bid Trade V 0 1 2 3 4 5 0 20 40 60 80 100 1 51 101 151 201 251 301 351 Ask & Trade Day StdA StdT StdP International Econometric Review (IER) 27 Quote Type Sample Period mean Spread1 mean Spread2 Std Spread1 Std Spread2 Best1 Jan. 05 – May 06 0.2503 0.0467 1.1336 0.3226 Jan.05 - May 05 0.1912 0.0365 0.1079 0.0206 Jan.06 - May 06 0.3411 0.0674 3.1670 0.9225 Best2 Jan. 05 – May 06 0.4877 0.0820 0.3513 0.0585 Jan.05 - May 05 0.4380 0.0836 0.2420 0.0461 Jan.06 - May 06 0.5182 0.0771 0.3947 0.0589 Table 3.8 Statistics of the SXF Spreads. Notes: Average daily mean and standard deviation of the SXF spreads is present in the table. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. Best 2 Bid is the second best bid depth, and so forth. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. The SXF minimum tick is 0.05; Spread1 =Ask Price – Bid Price; Spread2= (Ask Price-Bid Price)*2/ (Ask Price + Bid Price)*100. The sample period covers 356 records from Jan.2005 to May2006. Profit Trade Profit Settled Average Profit Profit Trade Profit Settled Average Profit Profit Trade Profit Settled Average Profit Client Firm Pro Jan-May 2005 -35.43 -23.13 -42.76 22.63 43.43 45.69 21.26 8.81 21.33 Jan-May 2006 -36.72 -0.29 -22.85 43.01 -16.98 6.96 7.61 15.30 7.11 All -34.03 -0.05 -31.69 29.43 -0.14 25.59 13.68 13.28 13.81 Hedger Speculator Market Maker Jan-May 2005 10.10 15.19 13.41 -19.91 -6.03 -17.15 -0.09 1.92 -1.32 Jan-May 2006 9.85 -2.51 -5.69 -20.83 -1.46 -12.10 -12.28 7.67 -6.26 All 1.46 -20.32 -5.99 -18.58 13.51 -6.07 -2.99 6.03 -2.22 Table 3.9 Daily Mean Transaction Cost of SXF (C$). Notes: Sample periods are 356 days from January 04, 2005 to May 31, 2006. Profit trade is calculated by the FIFO rule for each account type. Profit Settled is the assumed profits by settling the closing position (at the end of a trading day) at the closing average of bid and ask price of contracts. Average Profit is a weighted average of Profit trade and Profit Settled with the weights in accordance with the number of contracts. Costs of the most nearby standard SXF contract are present in the table. Minimum Tick of SXF is C$10.00. Two sub-periods of the sample from January 2005 to May 2005, from January 2006 to May2006 and for the whole sample period (All) are presented. Table 3.8 shows that the average of daily mean spreads and daily standards deviation of spreads for SXF in our sample period. In the table, the Best 1 limit order combines the Best 1 Bid and Best Ask 1 variables. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. Best 1 Ask is the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. We calculate two measures for bid-ask spreads. Spread1 =Ask Price – Bid Price (3.2) Spread2= (Ask Price-Bid Price)*2/(Ask Price + Bid Price)*100 (3.3) Table 3.9 shows the mean daily FIFO cost results on the front SXF contract. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 28 From Table 3.8, we note that the averages of daily mean spread (Spread 1) are much higher than the SXF minimum tick of 0.05. Table 3.9 shows that the costs are much higher than the value of the SXF minimum tick (C$ 10). Several empirical papers have examined the impact of tick size changes on market quality. For example, Bacidore (1997), Ahn et al. (1998) and Griffiths et al. (1998) study reduction of tick size on the TSE in 1996. Goldstein and Kavajecz (2000) and Chordia et al. (2001) show that the inside spread significantly decreased, but depth at the best bid and ask also decreased after the reduction in tick size of the NYSE. Kurov and Zabotina (2005) find that the minimum tick sizes of the E-mini S&P 500 and E-mini Nasdaq-100 futures contracts preventing the spreads from decreasing to the levels implied by a competitive market. In addition, Bortoli et al. (2006) report that trading at the minimum tick in the Sydney Futures Exchange embraces 87.8% of observations for the SPI, 95.2% for bank-accepted bills, 97.8% for three-year bonds, and 94.4% for ten-year bonds in the periods before or after the Sydney Futures Exchange changed the limit order disclosure rule. In the SXF market, however, the spread and costs are much higher than the minimum tick size. Hence, it is clear the established spread/ minimum tick size may not be reflective of the market at any particular point in time6. In addition, from Table 3.8 we note that the average mean daily spreads and the average daily standard deviation of the spreads are higher in the first five months of 2006 than those of the first five months of 2005. However, the corresponding FIFO costs for all six aggregated account types are lower in the first five months of 2006 (Table 3.9). Since more transactions could reduce transaction costs, and the trading volume is much higher in the first five months of 2006 than those of the first five months of 2005 (Panel B and Panel C of Table 3.3), it is evident that trading costs are not well captured by spreads for the SXF market. In the work, we also perform regression analysis on the costs and spreads with the model of Equation 1, substituting the quantity of limit orders (Q) in the model with Spreads. These results are shown in Table 3.10. The lack of significance of the explanatory variables holds whether Spread1 or Spread2 is used in the regression; consistent with Locke and Venkatesh (1997), daily mean spreads (BA) and standard deviation of daily spreads (std BA) show no relationship with transaction costs. Overall, the results show that traditional spreads have no relationship with the trading costs in the SXF market and that the minimum tick size do not act as binding constraints on the bidask spreads and costs. In fact, spreads as the measures of transaction costs need very strong assumption that all transactions are go through market makers. Since Client in our sample has much more transactions than Pro/Market maker, such an assumption clear does not held in SXF market. However, our FIFO cost is extensively used in accounting book keeping and is direct measure costs. The finding that no relationship between FIFO costs and spreads further supports that traditional spreads are inappropriate as the measure of trading costs, especially in a full electronic market. 6 Indeed, shortly after the endpoint of our data, the Exchange raised the minimum tick - the minimum tick size was raised to C$20 per tick (minimum tick fluctuation has been increased from 0.05 index points to 0.10 index points)! International Econometric Review (IER) 29 Account Dependent Independent Variable C BA TV std BA std TV std P D PF DW RSQ Client Average profit Coef. -92.02 -0.82 20.74 0.03 -0.99 29.55 36.77 0.04 2.00 0.04 Prob. 0.04 0.04 0.20 0.16 0.45 0.01 0.44 Profit trade Coef. -152.4 -0.75 43.10 0.02 -1.70 34.36 -7.30 0.00 2.05 0.06 Prob. 0.00 0.05 0.01 0.33 0.18 0.00 0.87 Firm Average profit Coef. 89.97 1.45 -24.39 -0.05 5.81 -54.65 -143.2 0.05 1.94 0.04 Prob. 0.27 0.05 0.33 0.22 0.28 0.01 0.58 Profit trade Coef. 221.2 1.27 -63.01 -0.03 8.14 -63.59 -102.4 0.00 1.97 0.06 Prob. 0.00 0.06 0.01 0.35 0.10 0.00 0.67 Pro Average profit Coef. -12.63 0.12 11.93 -0.01 1.23 -0.36 0.87 1.93 0.01 Prob. 0.70 0.39 0.60 0.40 0.91 0.92 Profit trade Coef. 27.94 0.09 -14.38 -0.01 6.78 -0.78 0.95 2.02 0.00 Prob. 0.38 0.52 0.53 0.42 0.52 0.84 Hedger Average profit Coef. -3.66 1.34 -3.25 -0.07 0.36 -30.11 -22.20 0.01 1.86 0.04 Prob. 0.94 0.00 0.83 0.00 0.80 0.02 0.68 Profit trade Coef. 43.08 0.70 -7.86 -0.02 0.47 -33.54 -16.97 0.04 2.04 0.04 Prob. 0.28 0.06 0.54 0.23 0.71 0.00 0.72 Spec Average profit Coef. 56.62 -1.53 -6.06 0.09 -2.01 9.83 117.1 0.01 1.59 0.05 Prob. 0.34 0.00 0.80 0.00 0.37 0.42 0.30 Profit trade Coef. -15.79 -0.61 -0.68 0.03 -1.20 18.02 65.84 0.21 2.03 0.02 Prob. 0.72 0.06 0.97 0.10 0.47 0.05 0.43 Market Maker Average profit Coef. 16.93 -0.10 -10.34 0.00 1.21 1.99 -164.6 0.00 1.91 0.06 Prob. 0.51 0.47 0.52 0.49 0.81 0.61 0.01 Profit trade Coef. 57.53 -0.27 -23.82 0.00 -5.36 -0.50 -72.25 0.00 1.92 0.07 Prob. 0.05 0.07 0.19 0.76 0.36 0.91 0.33 Table 3.10 OLS Estimates of the Regression of SXF Transaction Costs with Measures of Trading Activity and Spreads Notes: OLS estimates of the regression of SXF daily transaction profit/Cost in C$ per contract on measures of trading activity and spreads are shown. BA is the daily mean Spread1 of Best 1 for Panel A and is the daily mean Spread2 for Panel B. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. Best 2 Bid is the second best bid depth, and so forth. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. Best 1 Ask is the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. Best 2 Ask is the second best ask depth; Spread1 =Ask Price – Bid Price; and Spread2= (Ask Price-Bid Price)*2/ (Ask Price + Bid Price)*100. TV is Mean Trade Volume. D is a dummy variable, which equals 1 when TV +3 * Std TV> Q + 3 *Std Q; and Q is the daily mean quantity of Best 1 Ask for Panel A and Best 1 Bid for Panel B; Std stands for standard deviation. P is transaction price. DW is DurbinWatson stat. PF is the p value of the regression; RSQ is R square of the regression, Profit trade is daily FIFO profit for each account type of Firm, Client, Pro, Hedger, Speculator (Spec) and Market Maker. Average profit is the weighted average profits of the FIFO profit and the profits of daily inventory imbalance settled at the closing price. The number of contracts traded or settled is used as the weighting variable. C is constant term in the OLS model. 356 daily data from January 04, 2005 to May 31, 2006.are used in the regression. Value in Bold indicates significant at 5 percent level. 3.4. Analyses of Transactions by Order Type In the work, we also differentiate SXF transaction across order types with another trade data file that recodes each trade with Limit, Market or Market on Opening in the Order Type identifier from March 01 2005 to April 28 2006. Table 3.11 summarizes the SXF transactions by order type. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 30 Account Dependent Independent Variable PF DW RSQ Panel B: Spread2 C BA TV std BA std TV std P D Client Average profit Coef. -109.0 -639.04 23.72 23.53 -1.12 28.23 35.84 0.03 2.00 0.04 Prob. 0.01 0.02 0.14 0.10 0.39 0.01 0.45 Profit trade Coef. -167.5 -607.83 45.98 18.01 -1.83 33.14 -8.27 0.00 2.05 0.06 Prob. 0.00 0.02 0.00 0.20 0.15 0.00 0.86 Firm Average profit Coef. 118.6 1118.6 -29.10 -38.06 6.63 -52.19 -184.9 0.03 1.94 0.04 Prob. 0.13 0.03 0.25 0.15 0.29 0.01 0.57 Profit trade Coef. 246.4 986.14 -66.94 -28.21 8.64 -61.51 -125.3 0.00 1.97 0.07 Prob. 0.00 0.03 0.00 0.25 0.13 0.00 0.68 Pro Average profit Coef. -8.83 84.19 10.72 -4.22 1.20 -0.25 0.86 1.93 0.01 Prob. 0.78 0.36 0.64 0.38 0.91 0.95 Profit trade Coef. 30.42 84.66 -15.50 -5.01 6.81 -0.68 0.91 2.02 0.00 Prob. 0.33 0.36 0.49 0.29 0.52 0.86 Hedger Average profit Coef. 23.26 1022.8 -7.24 -53.63 0.55 -27.80 -15.31 0.01 1.86 0.05 Prob. 0.60 0.00 0.63 0.00 0.71 0.03 0.77 Profit trade Coef. 60.91 494.64 -10.90 -17.29 0.74 -32.70 -23.08 0.04 2.04 0.04 Prob. 0.12 0.06 0.40 0.20 0.56 0.00 0.61 Spec Average profit Coef. 18.47 -1074.4 1.05 63.23 -2.18 7.24 115.3 0.00 1.59 0.05 Prob. 0.75 0.00 0.96 0.00 0.33 0.55 0.31 Profit trade Coef. -31.59 -404.13 1.96 18.55 -1.26 17.03 64.83 0.23 2.03 0.02 Prob. 0.46 0.07 0.91 0.11 0.45 0.06 0.44 Market Maker Average profit Coef. 13.46 -79.19 -8.86 -2.62 1.02 1.89 -164.6 0.00 1.91 0.06 Prob. 0.59 0.39 0.58 0.58 0.84 0.62 0.01 Profit trade Coef. 47.16 -120.65 -21.24 -1.38 -5.76 -0.71 -73.07 0.00 1.92 0.07 Prob. 0.10 0.25 0.24 0.80 0.33 0.87 0.33 Table 3.10 (Cont.) OLS Estimates of the Regression of SXF Transaction Costs with Measures of Trading Activity and Spreads Notes: OLS estimates of the regression of SXF daily transaction profit/Cost in C$ per contract on measures of trading activity and Spreads are presented. In the table, BA is the daily mean Spread1 of Best 1 for Panel A and is the daily mean Spread2 for Panel B. The best bid price is defined as the highest price a prospective buyer is prepared to pay at a particular time for trading the futures contract. In the analyses, the variable Best 1 Bid is defined as the best bid depth, which corresponds to the sum of all bid sizes (number of contracts) that are submitted as limit orders for trading at the best bid price for market participant bid quotations that are equal to the best bid price. Best 2 Bid is the second best bid depth, and so forth. The best ask (or offer) price is the lowest price at which someone who owns the contract offers to sell it. We define Best1 Ask as the best ask depth, which is the sum of all ask sizes (number of contracts) that are submitted as limit orders to trade at the best ask price for market participant ask quotations that are equal to the best ask price. Best 2 Ask is the second best ask depth. Spread1 =Ask Price – Bid Price; and Spread2= (Ask Price-Bid Price)*2/ (Ask Price + Bid Price)*100. TV is Mean Trade Volume. D is a dummy variable, which equals 1 when TV +3 * Std TV> Q + 3 *Std Q; and Q is the daily mean quantity of Best 1 Ask for Panel A and Best 1 Bid for Panel B. Std stands for standard deviation. P is transaction price. DW is Durbin-Watson stat. PF is the p value of the regression. RSQ is R square of the regression, Profit trade is daily FIFO profit for each account type of Firm, Client, Pro, Hedger, Speculator (Spec) and Market Maker. Average profit is the weight average profits of the FIFO profit and the profits of daily inventory imbalance settled at the closing price. The number of contracts traded or settled is used as the weight. C is constant term in the OLS model. 356 daily data from January 04, 2005 to May 31, 2006.are used in the regression. Value in Bold indicates significant at 5 percent level. Harris and Hasbrouck (1996) document that 54 percent of SuperDot orders are limit orders. Ross et al. (1996) report that limit orders account for 65 percent (75 percent) of all executed orders (executed shares) in SuperDot. Compared with these results, limit orders are used much more extensively in SXF trades. As shown in the table, Limit order accounts for 97.70 percent of all trades on the nearby SXF contract (Limit to All) when measured by the number of trades; the account for 94.11 percent of all trades on the nearby SXF contract when International Econometric Review (IER) 31 measured by the number of contracts traded. Only very small proportion of the SXF trades were conducted through Market on Opening (Mo) or Market order (Mkt). By Number of Trade By Trade Volume Limit To All Mkt To All Mo To All Limit To All Mkt To All Mo To All Average Daily 97.70% 1.67% 0.63% 94.11% 1.19% 4.08% Whole Sample 97.45% 1.96% 0.58% 92.97% 1.13% 4.91% Table 3.11 Summary of the SXF Transaction by Order Type Notes: The summary of the SXF transaction by order types is presented. Average Daily is the results by calculating the percentages on each day and then calculating the average for the sample period; Whole Sample is the results by calculating the sum of the number of trade/the trade volume on the sample period and then calculating the percentage for the sample period. All is all records without classifying a trade by order types. Limit, Mkt and Mo are Limit, Market and Market on Opening in a trade data file that recodes each trade with Limit, Market or Market on Opening in the Order Type from March 01 2005 to April 28 2006. In such a market where limit orders are used by most transactions, the role of limit order book must be more important. When the best 1 quote book in the market provides enough liquidity/buffer to absorb potential trading orders, transaction costs should not be expected to be related to with measures of trading activities. On the other hand, to ameliorate price impact and informational leakage effects, investors looking to open or close large positions may need to structure their orders according to the depth of the best 1 limit order, to ameliorate price impact and information leakage effects. 4. CONCLUSION Using a unique database that includes the quotes and trade characteristics of the SXF market from January 2005 to May 2006 on the index futures of the SXF with aggregated trader types identified and time stamped in millisecond from the Montreal Exchange, We find that transactions costs, as correctly measured are not related to measures of trading activity, in both pair-wise correlation analyses as well as in a regression framework when limit orders provide enough liquidity for markets, especially for electronic systems. The limit order book conveys information about the market. Statistics of transaction by order type show that almost all trades are executed by limit orders for SXF market. We find a significant role for limit orders, especially that of depth 1 in determining participants’ trading behaviour, and in providing liquidity to the market. In addition, the results highlight the inability of traditional spreads to measure trading costs. All our level or volatility measures of quotes and trades by every aggregated account show evidence of mean reversion. This is consistent with a dynamic equilibrium process for the SXF market, wherein shocks to the variables do not have permanent dislocating effects. Moreover, the fact that the costs or spreads are much higher than the minimum tick size also implies participants in the SXF market respect the limit order to make their investment decision. Although participants in a limit order market can employ different order placement strategies, the aggregated actions of the participants still can leads the market to dynamic equilibrium when a majority of the participants have, on aggregated basis, established trading proclivities. In such an equilibrium system, the aggregated trading costs will not follow the fluctuations of various traditional market activity measures. Studying the overall equilibrium of electronic Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 32 systems for other derivative markets and the role of limit order book in determining the participants’ trading behaviour remains a topic for future work. REFERENCES Ahn, H., C. Cao and H. Choe (1998). Decimalization and competition among stock markets: Evidence from the Toronto stock exchange cross-listed securities. Journal of Financial Markets, 1, 51-87. Ahn, H., K. Bae and K. Chan (2001). Limit Orders, Depth, and Volatility: Evidence from the Stock Exchange of Hong Kong. Journal of Finance, 56, 767-788. Bacidore, J. (1997). The impact of decimalization on market quality: An empirical investigation of the Toronto stock exchange. Journal of Financial Intermediation, 6, 92120. Baesel, J.B., G. Shows, and E. Thorp (1983). The Cost of Liquidity Services in Listed Options: A Note. Journal of Finance, 38, 989-995. Beebower, G. and W. Priest (1980). The Tricks of the Trade: How Much Does Trading Really Cost? Journal of Portfolio Management, 6, 36-42. Biais, B., P. Hillion and C. Spatt (1995). An empirical analysis of the limit order book and the order flow in the Paris Bourse. Journal of Finance, 50, 1655-1689. Bortoli, L., A. Frino, E. Jarnecic and D. Johnstone (2006). Limit Order Book Transparency, Execution Risk, and Market Liquidity: Evidence from the Sydney Futures Exchange. Journal of Futures Markets, 26, 1147-1167 Chan, Yue-cheong (2005). Price Movement Effects on the State of the Electronic Limit-Order Book. Financial Review, 40, 195-221 Chang, E.C., P.R. Locke and S.C. Mann (1994). The Effect pf CME Rule 552 on Dual Traders. Journal of Futures Markets, 14, 493-510. Chang, E.C. and P. Locke (1996). The Performance and Market Impact of Dual Trading: CME Rule 552. Journal of Financial Intermediation, 5, 23-48. Chordia, T., R. Roll and A. Subrahmanyam (2001). Market liquidity and trading activity. Journal of Finance, 49, 255-267. Chung, H., F. Bonnie, V. Ness and T. Robert (1999). Limit orders and the bid-ask spread. Journal of Financial Economics, 53, 255-287. Coppejans, M., I. Domowitz and A. Madhavan (2003). Resiliency in an automated auction. Unpublished working paper, Barclays Global Investors, ITG Inc. Degryse, H., F. Jong, M. Ravenswaaij and G. Wuyts (2005). Aggressive Orders and the Resiliency of a Limit Order Market. Review of Finance, 9, 201-242. International Econometric Review (IER) 33 Demsetz, H. (1968). The Cost of Transacting. Quarterly Journal of Economics, 82, 33-53. Epps, T. (1976). The Demand for Brokerage Services: The Relation between Security Trading Volume and Transaction Cost. The Bell Journal of Economics, Spring 1976. Ferguson, M. and S. Mann (2001). Execution costs and their intraday variation in futures markets. Journal of Business, 74, 125-160. Fishman, M.J. and F.A. Longstaff (1992). Dual Trading in Futures Markets. Journal of Finance, 47, 643-671. Foucault, Thierry (1999). Order flow composition and trading costs in a dynamic limit order market. Journal of Financial Markets, 2, 193-226. Foucault, T., O. Kadan and E. Kandel (2005). Limit order book as a market for liquidity. Review of Financial Studies, 18, 1171-1217. George, J. and F.A. Longstaff (1993). Bid–Ask Spreads and Trading Activity in the S&P 100 Index Options Market. Journal of Financial and Quantitative Analysis, 381-397. Gilbert, C. and H. Rijken (2006). How is Futures Trading Affected by the Move to a Computerized Trading System? Lessons from the LIFFE FTSE 100 Contract. Journal of Business Finance and Accounting, 33, 1267-1297. Glosten, Lawrence R. (1994). Is the electronic open limit order book inevitable? Journal of Finance, 49, 1127-1161. Goettler, R., C. Parlour and U. Rajan (2005). Equilibrium in a Dynamic Limit Order Market. Journal of Finance, 60, 2149-2192. Goldstein, M. and K. Kavajecz (2000). Eights, sixteenths, and market depth: Changes in tick size on liquidity provision on the NYSE. Journal of Financial Economics, 56, 125-149. Gomber, P., U. Schweickert and E. Theissen (2004). Zooming in on liquidity. Unpublished working paper, University of Bonn. Griffiths, M., B. Smith, D. Turnbull and R. White (1998). The tole of tick size in upstairs trading and downstairs trading. Journal of Financial Intermediation, 7, 393-417. Griffiths, M., B. Smith, D. Turnbull and R. White (2000). The costs and determinants of order aggressiveness. Journal of Financial Economics, 56, 65-88. Haller, A. and H. Stoll (1989). Market Structure and Transaction Costs: Implied Spreads in the German Stock Market. Journal of Banking and Finance, 13, 697-708 Handa, P. and R. Schwartz (1996). Limit order trading. Journal of Finance, 51, 1835-1861. Handa, P., R. Schwartz and A. Tiwari (2003). Quote setting and price formation in an order driven market. Journal of Financial Markets, 6, 461-489. Switzer and Fan-Limit Orders, Trading Activity, and Transactions Costs in Equity Futures in … 34 Harris, L. and J. Hasbrouck (1996). Market vs. limit orders: The SuperDOT evidence on order submission strategy. Journal of Financial and Quantitative Analysis, 31, 213-232. Hedvall, K. and J. Niemeyer (1997). Order flow dynamics: Evidence from the Helsinki stock exchange. Unpublished working paper, Swedish School of Economics and Business Administration. Hollifield, B., R. Miller and P. Sandås (2004). Empirical Analysis of Limit Order Markets. Review of Economic Studies, 71, 1027-1063. Hollifield, B., R. Miller, P. Sandas and J. Slive (2003). Liquidity supply and demand: Empirical evidence from the Vancouver stock exchange. Unpublished working paper GSIA, Carnegie Mellon University. Kurov, A. (2005). Execution Quality in Open-Outcry Future Markets. Journal of Futures Markets, 25 1067-1092. Kurov, A. and T. Zabotina (2005). Is It Time to Reduce the Minimum Tick Sizes of the EMini Futures? Journal of Futures Markets, 25, 79-104. Locke, P. and P. Sarajoyi (2004). Interdealer Trading in Futures Market. Journal of Futures Markets, 24, 923-944. Locke P. and A. Sarkar (2001). Liquidity Supply and Volatility: Futures Market Evidence. Journal of Futures Markets, 21, 1-17. Locke, P. and P. Venkatesh (1997). Futures Market Transactions Costs. Journal of Futures Markets, 17, 229-45. Madhavan, A., D. Porter and D. Weaver (2005). Should securities markets be transparent? Journal of Financial Markets, 8, 265–287. Manaster, S. and S. Mann (1996). Life in the Pits: Competitive Market Making and Inventory Control. Review of Financial Studies, 9, 953-75. Martell, M. and A. Wolf (1987). Determinants of Trading Volumes in Futures Markets. Journal of Futures Markets, 3, 233-244. McInish, T. and R. Wood (1992). An analysis of intraday patterns in bid/ask spreads for NYSE stocks. Journal of Finance, 47, 753-764. Ross, D., E. Shapiro and A. Smith (1996). Price improvement of SuperDot market orders on the NYSE. Working Paper No. 96-02, New York Stock Exchange. Stoll, H.R. (1989). Inferring the Components of the Bid/Ask Spread: Theory and Empirical Tests. Journal of Finance, 44, 115-134. Switzer, Lorne and Haibo Fan (2007). The Transactions Costs of Risk Management vs. Speculation in an Electronic Trading Environment: Evidence from the Montreal Exchange. Journal of Trading, 2, 82-100.