Adoption of electronic trading at the International Securities Exchange

Decision Support Systems 41 (2006) 728 – 746 www.elsevier.com/locate/dsw

Adoption of electronic trading at the International Securities Exchange Bruce W. Weber* Sainsbury 332, London Business School, Regent’s Park, London NW1 4SA, United Kingdom Available online 19 November 2004

Abstract Information technology is transforming financial trading, lowering costs, and increasing market transparency. Yet, new electronic trading ventures often fail to attract sufficient activity levels, and close down. Optimark, Tradepoint, Jiway, and BondConnect did not develop sufficient trading volume to survive. In contrast, the International Securities Exchange (ISE), an allelectronic options trading platform has gained trading volumes in the United States in competition with four incumbent markets, including the Chicago Board Options Exchange (CBOE). Compared with floor exchanges, electronic options markets offer immediate trading, direct user access to the market, and reduced costs. The paper describes the ISE and examines newly available data from brokerage firms to comply with the Securities and Exchange Commission’s (SEC) Rule 11Ac1-6. The order routing disclosures show that brokerage firms differ widely in the extent of their use of the ISE. Based on a sample of 188 quarterly disclosures from 20 major brokerage firms, OLS, Tobit, and fixed-effects models of ISE use are estimated to explain individual firms’ adoption levels. Significant factors are whether the firm is an online discount broker, the firm’s membership role in the ISE, and the network externality effect of the ISE market’s growth. Firm-specific factors are shown to account for about 60% of ISE adoption explained by the model, with the remaining 40% accounted for by the network effects of growing market liquidity. D 2004 Elsevier B.V. All rights reserved. Keywords: Electronic markets; Options exchange trading systems and technology; Exchange memberships; Brokerage firm order routing; Market share models; Adoption models

1. Introduction This paper examines the adoption patterns of U.S. securities brokerage firms for electronic equity options trading after the launch of the International

* Tel.: +44 20 7262 5050x3538; fax: +44 20 7724 7875. E-mail address: [email protected]. 0167-9236/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.dss.2004.10.006

Securities Exchange (ISE), an all-electronic trading platform on May 26, 2000. In the first quarter of 2004, the ISE handled 29.2% of all U.S. equity options contracts traded and 33.2% of equity options transactions, with the four incumbent options exchanges accounting for the remainder (source: Options Clearing). Quarterly data for a sample of 20 brokerage firms from 3Q 2001 to 1Q 2004, however, reveal wide variation in the extent of ISE use, from 0% to as high

B.W. Weber / Decision Support Systems 41 (2006) 728–746

as 61% of a firm’s options orders in a quarter. Understanding what influences potential users to adopt a new electronic market has research value and practical implications for developers of new trading platforms. We look at how broker-specific and network-effect variables impact ISE use by brokerage firms. In the United States, the ISE is a competitor of four established floor-based exchanges in Chicago, New York (American Stock Exchange), Philadelphia, and San Francisco (Pacific Exchange). The largest of these, the Chicago Board Options Exchange (CBOE), began operating in 1973, and has a competing market maker structure with a floor trading crowd of 1437 that can provide for price and size improvement, and complex, linked transactions such as spreads and straddles in which several options are purchased and sold simultaneously. The ISE’s electronic market offers first in–first out (FIFO) time priority among orders at a particular price, and initially undercut the trading fees charged by the floor options exchanges. Transactions on the ISE are free to the brokerage firm and its customer. ISE market maker members are charged about 20 cents per contract traded, and the turnaround time on many orders to the ISE is less than 1 s. Before the ISE launch, floor option exchanges were charging fees about 50% higher than the ISE, but have since lowered fees to match those charged by the ISE. Floor orders can take anywhere from 15 s to several minutes to execute and report to the client, depending on the order and market conditions at the time. At the time of its launch, the prospects for the ISE were unclear. James Marks, an analyst with Credit Suisse First Boston commented in the October 1, 2000 edition of CIO Magazine: bIt’s a bit of a chicken-and-egg situation for the ISE. To get order flow, they need liquidity-willing buyers and sellers— but to get liquidity they need order flow. Better, cheaper, faster won’t mean much if they don’t get the critical mass of order flow they need to keep their market makers and the brokerages happy.Q Research into the factors that determine whether an electronic market will succeed is inconclusive. Kambil and Van Heck [14] describe the few examples of online financial and commercial B2B markets that have succeeded. The authors contend that success results largely from integrating product transactions with

729

information and services, such as logistics and payment support, and providing value, not just lower prices, to all market participants. Hendershott [13] examines the uneven adoption of electronic financial trading, and uses Electronic Communication Networks (ECNs) for Nasdaq stocks and currency dealing systems as examples of electronic trading successes. Bond markets though remain largely dependent on telephone contact for trading. Barclay et al. [1] examine competition between Electronic Communication Networks (ECNs) and Nasdaq market makers for trading, and conclude that multimarket trading offers benefits and that ECNs are not a complete substitute for trading with a traditional market maker. Well-designed trading automation is beneficial to investors and traders in markets [16,17]. For example, the introduction of the Nasdaq screen market in 1971 to replace the OTC bpink sheetsQ led to a reduction of the average bid-ask spread (an important transactions cost in financial markets) in a 174 stock sample to 40.3 cents from 48.7 cents [12]. The introduction of the SEAQ screen-based market system as part of the London Stock Exchange’s 1986 Big Bang market reforms improved the quality of the LSE market [4], and played a part in trading volumes increasing from $280 million a day in 1985, to $4.1 billion a day in 1994. Comparing SEAQ to the floor, London’s electronic market proved to be more open and competitive than the floor market, and led to lower transactions costs for investors. In spite of advantages, however, many new electronic trading platforms fail to attract sufficient market activity to survive. Researchers have recently identified further opportunities for exploiting IT, and specifically the Internet, for financial trading. Established order routing practices in many brokerage firms, though, can hinder the adoption of the most efficient trading practices, and thus reduce the incentive to introduce trading system innovations. As Fan et al. [9] points out bThe vertical relationships between the brokers and the market centers adversely affect investors’ interest and undermine the competition at the exchange markets. These relationships also reduce the incentive for market centers to innovate to offer more efficient trading services.Q An obstacle facing a new market, such as the ISE, is how to attract sufficient order flows when many brokers have existing relationships with floor exchanges [11].

730

B.W. Weber / Decision Support Systems 41 (2006) 728–746

both firm-specific factorsUincluding firms’ ISE membership categoryUand network effects in the form of the prior period’s aggregate ISE share influence ISE use. The paper concludes with implications for electronic market governance and success factors for new trading systems.

This paper will describe recent developments in U.S. options markets, and then analyze U.S. brokerage firms’ regulatory order routing disclosures. Table 1 indicates that during the sample period, the 20 major U.S. brokerage firms adopted the new ISE market at a rate roughly equal to the overall volume growth for the ISE. Rule 6 disclosures were mandated beginning in 3Q 2001. The sample firms’ usage, however, has lagged the ISE’s market share somewhat (see Table 1). The dependent variable in the analyses is the percentage of a firm’s options orders it routes to the ISE in a quarter. The adoption differences across firms will be examined to determine what influences the extent that brokers adopt the ISE market. Rule 11Ac16 disclosures were first required in 3Q 2001, when just two of the firms in the sample had adopted the ISE. By 1Q 2004, 19 of 20 firms in the sample had reported routing orders to the ISE (see Tables A1 and A2 in Appendix A). Three model specifications are considered: OLS, Tobit, and a firm fixed-effects model. The Tobit model is estimated since it corrects for limited dependent variable problems since the dependent variable is zero (left-censored) in 58 of 188 observations [3,10]. All of the models estimated are statistically significant, and have fairly consistent coefficient values. The fixed-effects model has the greatest explanatory power, but all three show that

2. Intermarket competition and origins of the ISE Innovative trading systems launched to compete with established markets often fail to attract sufficient activity and are later closed down. Launched in January 1999, Optimark sought to win block trading volume from the NYSE and Nasdaq Stock Markets, but closed in late 2000 [6]. Another electronic trading system, Tradepoint, competed with the London Stock Exchange, and the screen-based Cantor Exchange sought to capture U.S. Treasury futures contract trading from the Chicago Board of Trade’s vast floor trading pits [19]. Jiway was launched in 2000 as an online platform for European stock trading with backing from Morgan Stanley and Sweden’s OM Group. Unable to develop sufficient trading volume, each of these entrants later suspended operations. In the late 1990s, the serial entrepreneur and founder of E-Trade, Bill Porter, conceived of a fully electronic options exchange to reduce the cost of options trading

Table 1 Comparison of sample firms’ use of ISE with overall ISE market share Brokerage firm sample (n=20) Average ISE market share (%) 2Q00 3Q00 4Q00 1Q01 2Q01 3Q01 4Q01 1Q02 2Q02 3Q02 4Q02 1Q03 2Q03 3Q03 4Q03 1Q04

ISE overall market share High (%)

Low

Contracts traded (%)

Transactions (%)

0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

0.1 0.5 1.0 4.0 7.1 10.8 12.5 16.5 19.7 21.4 20.4 23.9 27.2 28.1 28.1 29.2

0.1 0.6 1.5 5.9 8.1 11.8 12.1 15.7 20.4 23.4 23.2 26.6 28.9 30.1 31.1 33.2

Rule 11Ac1-6 disclosures began 3Q 2001

4.4 6.2 9.4 12.9 13.5 15.5 16.0 19.4 17.9 19.3 20.6

59 59 51 61 47 41 45 51 43 44 43

Sources: Rule 11Ac1-6 reports, Options Clearing data.

(for 12 firms) (9) (8) (6) (6) (3) (2) (3) (4) (4) (3)

B.W. Weber / Decision Support Systems 41 (2006) 728–746

and challenge the entrenched options floor communities. The $80 million venture to launch the ISE was largely backed by a broker/dealer consortium named Adirondack Trading Partners, whose investors included E*Trade, Herzog Heine Geduld (bought by Merrill Lynch in mid-2000), Ameritrade, Knight Financial Products, Scottrade, and Deutsche Bank. The founders announced their plan for the ISE in November 1998, and it received regulatory approval from the Securities and Exchange Commission (SEC) on February 24, 2000 to operate as an all-electronic options exchange. It was the first exchange to be registered by the SEC since the CBOE was approved in 1973. At the time of its launch in May 2000, the ISE faced the challenge of attracting order flow in competition with four established options exchanges in the United states [20]. The oldest and largest U.S. options market, the Chicago Board Options Exchange (CBOE), was established in 1973, and reached its peak average daily volume of 1.5 million contracts (an equity option contract is for 100 shares of the underlying stock) in 2000. Prior to the arrival of option bmultiple listingQ in August 1999, options exchanges chose not to list options that were already traded at another exchange.1 Exclusive listings were ended when the CBOE announced that it would begin trading options on Dell Computer, which previously had been listed only on Philadelphia Stock Exchange. The other exchanges soon followed, listing each other’s options and triggering a competitive war for options order flow that the ISE soon jumped into. Launch day volumes were modest; ISE trading volume on May 26, 2000 totaled 5032 contracts, earning it just 0.3% of the day’s total equity options volume. The ISE began with calls and puts on just three stocks, and only three Primary Market Makers (PMMs), eight Competitive Market Makers (CMMs), and 17 Electronic Access Members (EAMs). In 1Q 2004, options on 646 stocks were traded on the ISE, and there were eight PMMs, 23 CMMs, and 126 EAMs operating. Without the space constraints of a market floor, the ISE had more flexibility in deciding what, if any, membership categories to have. The ISE founders, 1

Several papers have examined the market microstructure impacts of the 1999 multiple listing change. See de Fontnouvelle et al. [7] and Battalio et al. [2].

731

Table 2 Volumes on five SEC-registered exchanges for listed options trading (Options contracts in millions)

2000

2001

2002

2003

Chicago Board Options Exchange International Securities Exchange American Stock Exchange Philadelphia Stock Exchange Pacific Exchange Total

326.3 7.6 207.7 76.0 108.5 726.2

306.7 65.4 205.1 100.9 102.7 780.7

267.6 152.4 186.1 88.5 85.4 780.0

283.9 245.0 180.1 112.4 86.2 907.6

Trading of options on individual stocks and equity indexes (e.g., S&P500) is included. Source: Options Clearing.

however, chose a structure similar to floor markets, where trading firms have designated market maker or competing market maker roles, and have obligations to maintain bid and ask quotes. Firms that purchase ISE market making memberships receive certain privileges and accept responsibilities in ISE trading. The market has grown steadily in volume and membership. Table 2 shows that, in 2003, in its third full year of operations, the ISE became the second largest options market in the United States. Competition from the ISE for options orders has forced the other markets to reduce their transaction fees and to develop more advanced electronic trading functions [2,17]. On June 12, 2003, for instance, CBOE launched CBOEdirect HyTS, a system for access to both screen-based and floor-based trading environments. Along with reduced volumes at the existing exchanges, the competitive effects are seen in falling seat prices at the floor-based options exchanges. Seats provide the holder or leaser with access to that exchange’s trading floor, and lower, member-only transaction processing and service fees. The CBOE has a fixed number (1485) of seats, and these will change hands at prices determined by market forces (Table 3). The record price for a CBOE seat was $735,000 set in February 1998, but dropped to $150,000 in August 2002.

3. Options order handling and the ISE’s market structure In options markets, customer participants have accounts with brokerage firms, who handle their transactions, and bclearQ their trades by backing them financially. Prior to 1999, brokers sent investors’

732

B.W. Weber / Decision Support Systems 41 (2006) 728–746

Table 3 CBOE seat sales prices, a sample from 1998 to 2004 Date 15-Feb-98 09-Dec-99 22-Feb-00 23-Jun-00 21-Dec-00 15-Oct-01 16-Aug-02 25-Oct-02 17-Mar-03 12-Jun-03 12-Aug-03 17-May-04 Price $735,000 $441,000 $390,000 $489,000 $360,000 $360,000 $150,000 $160,000 $180,000 $165,000 $207,000 $337,500

option orders to the exchange floor that listed any particular company’s options. On the exchange floor, an executing broker, possibly from a different firm, would expose the order and execute the trade in the market crowd (Fig. 1). Usually, a market maker would trade with the customer from his or her own account. It could take several minutes for a trade report with a firm price and quantity to be given to the broker’s client [15]. Automated execution (Auto-ex) systems, such as Retail Automated Execution System (RAES) on the CBOE, offer faster trading for qualifying orders (from retail customers for less than 50 contracts) [2], but are not full electronic markets. Since multiple listing of options began in 1999, brokerage firms have had discretion over which options exchange to route a customers’ orders to. In the ISE market, a broker with electronic access can monitor the ISE’s quotes on a screen, and can enter buy or sell orders directly into the market on behalf of the their customers. In the ISE’s system, a bfacilitation exposureQ time of 10 s allows customer orders to be shown to the ISE’s market makers for price improvement. If no price improvement is provided, the customer order will trade with the best-priced limit order or market maker quote in the ISE book. When one of the other four options exchanges, and not the ISE, posts the best bid or offer (BBO) in an option, arriving ISE orders are presented to the PMM for execution. The PMM may match the better away market price or attempt to trade against the better quote in the other market. The Options Clearing (OCC), an exchange-owned utility, handles clearing

for ISE trades and for trades on all U.S. options exchanges (see Fig. 2). The ISE screen displays orders anonymously; showing bids and offers with the size in contracts they are valid for, and the last trade price. No information about the counterparties is displayed. A trade in the system occurs when an order that was entered and displayed is bhitQ (sold to) or bliftedQ (bought from). The distinctions in order handling across the two markets are that the electronic order matching system holds customers’ limit orders in time priority, and displays their price and size in the ISE quote, which is not merely the exchange floor specialist’s quote for a standard quoted size, such as 100 contracts. Customer orders in the ISE book are executed based on best price and time of entry, and receive priority over PMM or CMM quotes. In the open outcry mechanism, a limit order may not execute until the bid or offer moves up or down to btouch itQ and trigger an execution. These distinctions—facilitation exposure and customer priority—could lead to different customer order placement strategies, which are examined (Hypothesis 1) in the next section. In May 2004, the ISE listed options on 646 securities. Its market is organized into 10 bins with about 60 stock options in each. A bin has one Primary Market Maker (PMM), and as many as 16 Competitive Market Makers (CMMs). A PMM must purchase or lease one of the ISE’s 10 PMM trading rights. As of April 2004, eight firms operated as PMMs, with two firms covering two bins each. (See Appendix A for a full listing of the ISE’s PMMs and CMMs.) Similar to

Fig. 1. The route of an order in floor options markets. A customer’s order is routed to a floor broker or a retail order routing system such as RAES. The order is exposed to the floor trading crowd (DPMs and CMMs) who fill it in whole, or in part if any customer limit orders are available at the trade price.

B.W. Weber / Decision Support Systems 41 (2006) 728–746

733

Fig. 2. ISE market flows: a customer’s order is entered online or phoned to a broker. The order (at the brokerage firm’s discretion) is routed to the ISE, and for 10 s, other participants can react and trade with it at better than the quoted bid or ask prices. A layer of interaction and information transmission that occurs with the floor broker and trading crowd are eliminated.

the Designated Primary Market-Maker (DPM) role that is assigned at the CBOE, the PMMs at the ISE are the single points of contact for the options in their bin, and are responsible for maintaining orderly markets and answering trading questions. The second ISE membership category is Competitive Market Maker. As of May 2004, 23 firms operated as ISE CMMs, with 4 firms, including Credit Suisse First Boston and Lehman Brothers, operating as CMMs in all 10 bins (see Appendix A for a list of ISE PMM and CMM firms). A CMM must purchase or lease one of 160 CMM trading rights, entitling them to enter quotations in the options in a bin. CMMs add depth and liquidity to the market by providing continuous quotations in at least 60% of the options classes in their bin(s). Each CMM is required to quote independently, and 16 CMMs are appointed to each of the 10 groups of options traded on the Exchange. CMM rights are sold by the ISE and can be resold or leased. For instance, CMM trading privileges for Bin 3 were bought for $1.5 million each on December 18, 2003. On September 29, 2003, the PMM trading privileges in Bin 7 sold for $7.5 million. PMMs have greater obligations, but also greater privileges in ISE trading than CMMs. When a customer market order arrives at the ISE, and any limit orders have been filled, the ISE allocation rules entitle PMMs to trade a greater number of contracts than a CMM with an identical quote. For instance, a market order to buy 100 contracts arriving when one PMM and one CMM each are offering 100 contracts for sale would result in 60 being sold by the PMM and 40 by the CMM. This is because ISE rules specify that the PMM receive the greater of 60% or his bproportionate size interestQ, 50% in this case of the 200 contracts offered at the quoted. Had two CMMs also been offering the lowest price, the split would be 40% to the PMM, and 30% each to the

CMMs. With three CMMs and the PMM matching on price and size, the split of an arriving order would be 30% to the PMM, and the remaining 70% split evenly over the three CMMs. An exception is made for the handling of orders for five contracts or less. In this case, the customer limit orders are traded first and the remainder goes exclusively to the PMM provided he is on the best bid or offer (BBO). If the PMM is not on the BBO, the order follows normal size-based, pro-rata allocation among the CMMs. For instance, if one CMM is offering to sell 30 contracts and another is offering 10, an order to buy 20 would be split 15 and 5. The third ISE membership type is an bElectronic Access MemberQ (EAM). An EAM is a broker/dealer that acts as an order flow provider, and—unlike PMMs and CMMs—is not required to purchase membership. There are no limits on the number of EAMs, who pay a monthly access fee to send orders in all of the options traded on the ISE. EAMs cannot enter quotations or otherwise engage in market making activities on the Exchange. As of May 20, 2004, there were 126 EAMs (Source: http://www.iseoptions.com). Given its membership and market structure, the ISE faced technical challenges in launching its trading system. To receive approval from the SEC and potential users, its fully electronic market needed fast response times and redundancy in the case of hardware or software component failures. Importantly, ISE technology had to integrate the best bids and offers from other exchanges so that ISE member firms can keep their prices current, and know, for regulatory bbest executionQ purposes, if there are better quotes in another market for an ISE traded option. The quote data from the other four options markets comes from the Options Prices Reporting Authority (OPRA), which updated about 3500 quotes per second in mid2003, and distributes its feed through the major market

734

B.W. Weber / Decision Support Systems 41 (2006) 728–746

data vendors, such as Reuters and Bloomberg. The best bids and offers, the quote sizes, and last trade prices on the ISE are sent to OPRA, which consolidates them into a National Best Bid and Offer (NBBO), which keeps all markets and market participants appraised of prices and quote movements. OPRA also requires that trades be reported in the correct sequence within 2 min of the trade execution. A trade reported to OPRA after 2 min must be marked as delayed by inserting a bDLYQ into the OPRA trade message field.

Table 4 Brokers in the sample fit into two categories Discount online brokers (OLBs)

Full-service brokers (FSBs)

AMTD BRWN

AGE A.G. Edwards BOFA Banc of America Securities BSC Bear Stearns CSFB Credit Suisse First Boston DBAB Deutsche Bank Alex Brown GS Goldman Sachs LEHM Lehman Bros. MER Merrill Lynch MS Morgan Stanley PRU Prudential Securities SSB Salomon Smith Barney

DATK ETRD FID JBOX SCH SCO TDW

4. Rule 11Ac1-6 and analysis of ISE order routing To assess adoption of the ISE as a new market, I collected options order routing data across 11 quarters (3Q 2001–1Q 2004) for a sample of 20 brokerage firms. Beginning in the third quarter of 2001, the SEC’s Rule 11Ac1-6 (Rule 6) required brokers to disclose on a quarterly basis their order routing practices in U.S. equities and listed options. The brokers in the sample were chosen from Smart Money’s 2002 broker rankings (http://www.smartmoney.com/ brokers/), whose bDelegatorQ and bDo-it-YourselferQ rankings included 29 firms. The two categories correspond to the Full-Service and Discount Online groups below. For some of the firms, quarterly order routing reports were not available for 2001, which left a sample of 20 firms (Table 4). While the data sample includes large and midsize firms that differ in the volume of option orders, the SEC disclosures only provide data in percent of the firms’ total orders, and do not include contract volume data. Hence, the 11Ac1-6 data only allow for unweighted market shares to be compared. As a result, we use the ISE’s share of options trades, rather than contract volumes for comparability.

a

Acquired by Ameritrade in September 2002.

instructions on where they are to be routed make up 99.6% of customer orders for brokers in our sample according to their disclosures. Beginning in the third quarter of 2001, the following information is now disclosed by brokerage firms: (1)

(2)

4.1. The Securities and Exchange Commission’s (SEC) Rule 11Ac1-6 In November 2000, the SEC adopted Rule 11Ac16 (Rule 6) which became effective on July 2, 2001. In few other industries, such detailed information on transactional arrangements and behind-the-scenes market usage volumes is disclosed. Rule 6 requires tracking of all customer orders that are bnondirectedQ orders. These orders without specific customer

Ameritrade BrownCo. (online unit of J.P. Morgan Chase) Dateka E-Trade Fidelity J.B. Oxford Charles Schwab Scott Trade T.D. Waterhouse

(3)

The identity of the market centers that receive 5% or more of customers’ orders for four categories of securities: (i) New York Stock ExchangeListed Securities, (ii) Nasdaq-Listed Securities, (iii) American Stock Exchange-listed and Regional Exchange-listed securities, and (iv) Exchange-Listed Options. The actual disclosures are the bPercentage of Customer Orders Having a Market Value Less Than $200,000Q for securities, and for listed options, the bPercentage of Customer Orders Having a Market Value Less Than $50,000.Q Material aspects of the order-routing relationship between the broker and the market center. That is, indications of ownership in trading firms or trading systems, and payment for order flow arrangements. The percentage of orders in the following four categories: (i) all orders, (ii) market orders, (iii) limit orders, and (iv) other orders (stop orders, short sales, not held/discretionary orders, etc.).

Since the average options trade in 2003 was for 19.3 contracts with a value of $5887, the upper limit

B.W. Weber / Decision Support Systems 41 (2006) 728–746

of $50,000 indicated in #1 above screens out few transactions from the Rule 6 disclosures. Below are several examples of the disclosures’ bmaterial aspects of the order-routing relationshipQ (#2 above) sections: !

!

!

!

Credit Suisse First Boston (CSFB)—CSFB is a Competitive Market Maker in certain options traded on the International Securities Exchange (ISE). Consequently, CSFB may incur any gains or losses that are generated by acting in a market making capacity. T.D. Waterhouse—NISC (T.D. Waterhouse’s clearing affiliate) directs customer orders to designated Primary Market Makers (PMM) that trade listed option classes on the ISE. Not all PMM firms pay NISC for directing option orders. NISC receives payment on approximately 30% of all option trades routed to the ISE. Rates range from 40 cents to 1 dollar per contract. The average payment per contract received by NISC for this period was 26 cents. Goldman, Sachs—SLK-Hull Derivatives LLC (bSHDQ), an affiliate of Goldman, Sachs is a specialist, primary market maker or market maker on AMEX, CBOE, ISE, and PHLX. As an affiliate, Goldman Sachs stands to share indirectly in any profits that SHD or SLK generates from the execution of customer orders. Scottrade—Scottrade may receive payment for order flow ranging from $0.00 to $0.60 per option contract for orders routed to Knight for execution on the CBOE, AMEX, PHLX, ISE, and PSE.

735

Scottrade’s principal shareholder and President is a director and shareholder of Knight Trading Group, Scottrade may receive payment for order flow of $0.75 per eligible options contract for orders routed to ABN AMRO for execution on the CBOE, AMEX, PHLX, ISE, and PSE. Scottrade may direct orders to Adirondack Trading Partners for execution on the International Securities Exchange. Scottrade may receive payment for order flow of $0.75 per options contract. Scottrade maintains an approximate 2.8% ownership interest in Adirondack Trading Partners. Scottrade maintains an inactive seat on the International Securities Exchange. A plot of the SEC Rule 6 data for the sample firms along with the ISE’s overall market share (Graph 1) indicates that online brokers were initially quicker to adopt the ISE market. However, both groups increased their routing to the ISE at roughly the same rate as the ISE share grew. Also notable is that the sample firms have slightly lower average ISE use relative to ISE overall market share. I was not able to get an explanation for the shortfall. Three categories of nondirected orders are reported in Rule 6 disclosures: Market Orders—Any order in which a customer does not specify an execution price. Under normal market conditions, the order is filled immediately at the consolidated best bid or offer at the time of receipt by the market center.

Graph 1. A comparison of the ISE’s overall market share of options trading volume with the percentage of all orders routed to the ISE by the two groups. For comparability over the 11 quarters, only the seven Online Brokers and seven Full-Service Brokers with all 11 quarters of Rule 6 data are included. Sources: rule 11Ac1-6 reports, Securities Industry News, "Quarterly Statistical Reports", 2001 – 2003.

736

B.W. Weber / Decision Support Systems 41 (2006) 728–746

Limit Orders—Any order in which a customer specifies a limit to the price that they are willing to buy or sell a security. The order is usually filled if the consolidated best bid or offer btouchesQ the limit price of the order under normal market conditions. Other Orders—Odd lots, market opening and closing orders, orders submitted with stop prices, all-or-none orders, orders that must be executed on a particular tick or bid (such as nonexempt short sale orders), and bnot heldQ orders. 4.2. ISE use by broker type and order type For the full sample, limit orders were 37% of all customer options orders, while market and other orders accounted for the remaining 63% (Table 5). Customer orders from Online Brokers were far more likely to be limit orders than market or other orders. Section 5 examines whether the ISE attracts limit orders disproportionately. Tables A3 and A4 in Appendix A show how the two samples of firms increased their use of the ISE market over the 11 quarters examined. The tables include percentages of limit order, market and other orders, and all orders routed to the ISE. Since Rule 11Ac1-6 exempts broker/dealers from identifying execution venues that received less than 5% of a firm’s nondirected orders provided that 90% of the nondirected orders are identified, individual broker’s overall market share totals across the five options exchange did not necessarily sum to 100%. I do not, however, adjust the reported ISE data to reflect the possibility of ISE shares that were positive but less than 5%. The next section describes four research hypotheses, and estimates several multivariate regression models of firms’ use of the ISE in a given quarter. The models are developed from the quarterly disclosures Table 5 Online brokers in the sample reported routing more limit orders to options exchanges than full service brokers (n=188)

Limit orders Market/other orders

Online brokers (9)

Full service brokers (11)

All (20)

77.8% 22.1% 100%

50.7% 49.3% 100%

62.9% 37.1% 100%

from 20 major brokerage firms in the sample. The independent variables used to predict ISE use are: the firm’s ISE membership categories, if any, the type of firm (online or not), and the lagged (prior quarter) overall market share of the ISE, which rose from 8.1% in 2Q 2001 to 31.1% in 4Q 2003.

5. Hypotheses and multivariate analyses of ISE use The electronic markets literature identifies many broad factors that can contribute to a new market’s success. These include the need for mutual benefits for participants including users and market makers [14], for a critical mass of trading activity to develop [5], for adequate incentives to exist for traders to realize the cost-savings from online markets [8], and for characteristics of the traded instrument to be suited to screen-based trading [13]. We examine the specific factors that explain an individual brokerage firm’s level of use of a new market, such as the ISE. The factors that contribute to ISE use fall into two categories. First, the network effects from the prior period’s ISE market share among all options exchanges. Secondly, we include a number of measured, firm-specific factors including its membership categories in a particular quarter and whether it is an online broker or not. Finally, a firm fixed-effects model is estimated to account for unmeasured factors at the firm level that influence ISE use. The first hypothesis concerns the order types that the ISE attracts. Before developing an explanatory model of ISE use, it is important to examine whether the limit and market orders routed to the ISE need to be treated separately. Due to its market systems, it is possible that the ISE will attract proportionately more limit orders. The ability of the ISE’s electronic market to hold limit orders in price and time priority could attract proportionately more limit orders to the ISE. Hypothesis 1. The ISE will attract a higher proportion of brokerage firms’ customer limit orders than market orders. For the sample of online brokers, an average of 16.2% of limit orders were routed to the ISE, while market orders sent to the ISE averaged 15.6% over the 11 quarters. For the full service brokers, limit orders to

B.W. Weber / Decision Support Systems 41 (2006) 728–746

the ISE averaged 11.2%, and market orders to the ISE averaged 11.4% of the total number of option orders. t-Tests conducted on the individual brokers’ 11 quarterly disclosures indicated no significant differences in 18 of the 20 sample firms between the percentage of a firm’s market orders and the percentage of limit orders routed to the ISE in a quarter. Only Goldman Sachs routed a significantly larger proportion of limit orders to the ISE than market orders (34.0% and 22.0%, t=2.88, p=1.6%). The Lehman Brothers’ disclosures show it has routed no limit orders to the ISE, but on average has routed 6.6% of its market and other orders to the ISE (0.0% and 6.6%, t=2.85, p=1.7%). Since only one of the 20 firms routed a significantly greater percentage of limit orders to the ISE, Hypothesis 1 is not supported. While the electronic market system of the ISE maintains limit orders in price and time priority in a way that does not occur on exchange floors, this feature does not affect order routing to the ISE by order type. The analyses from hereon will only consider the percentage of sample firms’ total options orders routed to the ISE. Hypothesis 2a,b, 3, and 4 will be tested by developing three different model specifications. Each model has as its dependent variable the individual brokerage firms’ quarterly use of the ISE as a percentage of its total customer options order flow. The first model is an OLS regression with robust standard errors to account for the heteroskedasticity caused by the limited dependent variable (lower limit of 0.0 and upper limit of 1.0). The second model is a Tobit model that specifically ensures the model’s predicted values fall between 0 and 1. The third model is a fixed effect model that tests for unmeasured, firm-specific factors influencing their use of the ISE. Hypothesis 2a. The ISE will attract a higher proportion of brokerage firms’ customer orders when they have a membership affiliation in the ISE. Hypothesis 2b. The type of membership affiliation a broker has in the ISE (PMM, CMM, and EAM) will affect the broker’s level of ISE use. There are three membership types in the ISE: PMM, CMM, and EAM. Firms can have no ISE membership, or participate in one, two, or all three

737

types of membership. Even if a broker is not an ISE member, its orders can still be routed to the ISE if it directs customer orders to a clearing broker that is an ISE member. The ISE membership types for brokers in the sample are detailed in Tables A3 and A4 in Appendix A. Membership is coded as three indicator variables, one for each of the three membership categories. The indicator is set to one if the firm is that type of member in the quarter, or zero if it is not. In the cases where a firm becomes an ISE member during a quarter, the indicator variable is set to the approximate fraction of the quarter remaining in which it will operate on the ISE in that capacity. For instance, Ameritrade became an EAM on December 13, 2001 with only about a sixth (2 of 13 weeks) of 4Q 2001 remaining. Thus, the indicator is set to 0.167 for 4Q 2001, and 1 thereafter. Hypothesis 3. Online discount brokers will be more active participants in the ISE market and have greater ISE adoption rates (as evidenced by their Rule 11Ac16 order routing practices submissions) than fullservice brokers. An indicator variable (OLB) is set to 1 for online discount brokers in the sample and 0 for the full service brokers. Support for Hypothesis 3 is evident if the OLB coefficient in the regression model is positive and significant. Its value will reflect the incremental percentage of orders routed to the ISE by an online broker. It is possible that online firms’ technology, flexibility, and drive to lower cost result in greater adoption by online brokers of the ISE. In addition, the sample of full-service brokers includes firms with dedicated floor brokers at the major options exchanges, which could make them more likely to use in-house resources to trade options rather than the ISE. Hypothesis 4. Network externalities will draw more order flow to the ISE market as its volumes and liquidity grow. A trader adage states that bliquidity begets liquidity.Q At some point, it becomes disadvantageous to ignore a market that has attracted a significantly volume of trading. With the exception of E-Trade, all firms in the sample increased the fraction of options order sent to the ISE over the 11 quarter period 3Q

738

B.W. Weber / Decision Support Systems 41 (2006) 728–746

2001–1Q 2004. Tables A1 and A2 in Appendix A indicate that while only two of the sample firms routed orders to the ISE in 3Q 2001, all but one of 20 had routed some of its orders to the ISE by 2Q 2003, the eighth quarter of the sample. Another factor determining a broker’s use of the ISE may simply be the growing level of activity make its market more liquid and attractive to participate in. Of the firms in the sample, only A.G. Edwards, a St. Louis-based full-service firm, did not route orders to the ISE in any of the 11 quarters. E-Trade and Goldman Sachs were active, lead users, and could have been responsible for the ISE growing its market share and attracting additional user firms. A significant positive coefficient on the prior quarter’s ISE market share (PrQtr) indicates that the ISE’s lagged market share is related to individual brokers’ ISE use. Alternatively, if the coefficient is not significant, then the ISE’s growth comes from new users, rather than more volumes from existing ISE users. The OLS and Tobit models are developed with five independent variables. The first OLS model estimated is: ISE SHAREBroker

i; Quarter j

¼ Constant þ b1 ðOnline Broker IndicatorÞi þ b2 ðPMM indicatorÞi; j þ b3 ðCMM IndicatorÞi; j þ b4 ðEAM IndicatorÞi; j þ b5 ðISE Overall Market ShareÞQuarter

j1

þ eij

Additional specifications are considered for the OLS and Tobit models that leave out the ISE’s lagged market share. This is done to isolate the effect of the liquidity externality created by the ISE’s growing share of trading in the sample period. Finally, a firm fixed-effects specification is estimated that controls for unobserved, but salient features of the individual brokers in the sample (see Table 6). The estimated models that follow are the result of regressing the brokers’ use of the ISE in a quarter against the indicated independent variables (Table 7). The coefficients in the OLS and Tobit models are fairly consistent. The first OLS model explains 57% of the variance in ISE use over the brokerage firms, and has four significant, positive coefficients at the

Table 6 Descriptive statistics for the variables in the model Variable

Mean (n=188)

ISE share of broker i in quarter j (dependent) OLB indicator PMM indicator CMM indicator EAM indicator Prior quarter ISE share of options transactions

14.7% 46.8% 29.8% 45.1% 81.2% 21.8%

0.05 level. Being an online broker, a PMM, or an EAM are significantly related to the percent of orders routed to the ISE. Only the CMM indicator fails (slightly) to be significant at the 0.05 level in a twosided test. It would be significant in a one-sided test. In the first Tobit model, all five explanatory variables are positive and significant at the 0.05 level. The second OLS and Tobit models leave out the lagged ISE market share as an independent variable, which reduces the explanatory power of the models. The reduction in the models’ significance indicates that firms’ use of the ISE is related to the exchange’s prior quarter market share. That is, liquidity externalities are evident at the broker level as the ISE has grown over the nearly 3-year study period. The third model specification is a firm fixed-effects model with the 20 broker identities as the categorical factor. Including fixed effects for each firm, and dropping the OLB indicator (redundant with the firm categorical variable), gives the fixed effects model greater explanatory power. Based on the R 2, unobserved factors at the firm level account for an additional 26–27% of the variance in ISE use compared to the first OLS model. This is evidence of systematic differences in order routing to the ISE beyond the five measured variables. Notably in the firm fixed-effects model, CMM drops out of significance, and the PMM coefficient is now only marginally significant at the 0.10 level, and its value, 0.097, is about half what it was in the OLS and Tobit specifications. The EAM coefficient remains significant at the 0.05 level and is a similar size, 0.076, to that in the OLS (0.046) and Tobit (0.081) models. Apparently, fixed-effects pick up much of the influence of the PMM and CMM variables, but not the EAM indicator. The individual firm coefficients in the fixed-effects specification show that E-Trade has the largest positive coefficient, perhaps reflecting the fact that its founder, Bill Porter, is also the founder of the ISE. As of March

B.W. Weber / Decision Support Systems 41 (2006) 728–746

739

Table 7 Estimation results Variable

Online broker (0–1) PMM CMM EAM Prior Qtr ISE share ETRD BSC CSFB JBOX BOFA AMTD DATK GS MS AGE SCH FID SCOT BRWN TDW MER PRU LEHM SSB DBAB Constant Model ( F-stat) p-value R2 Adjusted R 2 (Likelihood ratio v 2) p-value

OLS with robust standard errors

Tobit model

Coefficient (t-statistic) p-value

Coefficient (t-statistic) p-value

Coefficient (t-statistic) p-value

Coefficient (t-statistic) p-value

Coefficient

0.0620218 (4.14) 0.000 0.1832548 (6.38) 0.000 0.0498074 (1.97) 0.051 0.0460201 (2.80) 0.006 0.4778999 (3.86) 0.000

0.0580603 (3.72) 0.000 0.160528 (5.38) 0.000 0.0780087 (3.17) 0.002 0.0610617 (3.18) 0.002 –

0.111404 (5.11) 0.000 0.1968406 (6.29) 0.000 0.08797 (2.80) 0.006 0.081016 (2.47) 0.014 0.7691195 (4.97) 0.000

0.1088379 (4.70) 0.000 0.1605806 (4.94) 0.000 0.1301359 (4.00) 0.000 0.1100216 (3.20) 0.002 –

–

0.1010812 (4.16) 0.000 (60.25) p=0.0000

0.0132062 (0.86) 0.393 (51.82) p=0.0000

0.2760581 (6.05) 0.000

0.1380609 (3.90) 0.000

0.5667 0.5548

0.5217 0.5113

2004, E-Trade Group owned approximately 4.6% of the ISE. The five most negative firm coefficients (DBAB, SSB, LEHM, PRU, MER) are all for fullservice brokers that have had trading operations on the floor option exchanges or long-standing relationships with the other four U.S. options exchanges. Hypothesis Tests. Overall, the results show support for Hypothesis 2a,b, 3, and 4. Hypothesis 2a,b is supported since members of the ISE routed more

Firm fixed-effects model

0.0974612 (1.75) 0.082 0.0265343 (0.93) 0.356 0.0761969 (2.36) 0.020 0.5616605 (7.22) 0.000 0.252577 0.137863 0.098843 0.097463 0.081644 0.079567 0.073466 0.067213 0.029577 0.008299 0.01 0.04142 0.04799 0.06586 0.08704 0.08746 0.11421 0.11488 0.13573 0.15729 0.0548137 (1.86) 0.065 (21.32) p=0.000 0.8422 0.8200

(156.90) p=0.0000

(132.47) p=0.0000

orders to the ISE. The coefficients of the PMM, CMM, and EAM indicator variables are positive and at least marginally significant in all but the case of the CMM variable in the fixed-effects model. The PMM indicator variable has the largest coefficient in all three models. A broker’s choice to become an ISE PMM is associated with greater activity level on the ISE. The EAM coefficient is smaller than the PMM coefficient, but is significant in all of the three models.

740

B.W. Weber / Decision Support Systems 41 (2006) 728–746

The use of 0–1 indicator variables allows interpretations to be drawn from the ISE membership coefficient values. From the first OLS and Tobit models, for instance, a firm that is a PMM is expected to route an incremental 18.3% and 19.7% of its customer options orders to the ISE in any given quarter. Being an EAM leads to 4.6% and 8.1% more orders being routed to the ISE. Hypothesis 3, which proposes that online brokers will use the ISE more, is strongly supported. The OLB indicator variable’s coefficient is significantly positive in all of the models it is included in. Its value of 0.062 in the OLS model and 0.111 in the Tobit model indicates that all else the same, an online broker can be expected to route an added 6–11% of its order flow to the ISE. Hypothesis 4 argues that more ISE market share in the prior quarter attracts greater ISE use, and is also supported. The positive and significant coefficients demonstrate that the prior quarter’s overall market share for the ISE is related to the sample brokers use of the ISE in a quarter. The OLS, Tobit, and fixedeffects coefficients of 0.478, 0.769, and 0.562 indicate that each additional percent of ISE market share in a quarter leads individual brokers to route an additional half to three-quarter percent of its orders to the ISE in the subsequent quarter. Additional insights can be drawn from the models by examining the relative contribution of the explanatory factors to the average ISE market share of the broker sample to be examined. Taking the Tobit and OLS model coefficients and multiplying by the mean values of the independent variables gives insight into the relative contribution to the models’ estimated ISE share of firms’ options orders. Table 8 shows that brokers’ use of the ISE in the model is about 60% apportionable to broker-specific

characteristics (PMM, EAM, OLB), and 40% accounted for by the network effects represented by the prior quarter’s ISE overall market share. Of the firm-specific factors, membership types (PMM and EAM) are about three times as influential as whether or not a firm is an online broker: summing to 42.8% vs. 15.1% in OLS, and 39.2% vs. 17.1% in Tobit.

6. Discussion and conclusions Understanding the mixed record of success of electronic financial markets is a challenge for I.S. researchers [6,13,18]. Unlike many new computerized markets, the ISE has succeeded in attracting a critical mass of volume and liquidity. A requirement of any new market is to attract use broadly or from a narrow set of active participants, and the newly available SEC Rule 6 disclosures provide a way to assess market adoption at the level of individual brokerage firms. The order routing patterns studied showed that online brokers overall and ISE member brokers in particular were rapid early adopters of the ISE. Even in electronic markets, exchange memberships are important, and were shown to influence brokers’ use of the ISE. The analysis shows that ISE members, in particular PMMs, direct substantially more order flow to the exchange than nonmembers. While technology can improve the functioning of a market and reduce costs, membership structures remain an important element in attracting order flow to an exchange. In addition to broker-specific factors, the network effects generated by the ISE’s growth served to draw additional orders from brokers. The liquidity externality is evident as individual brokers’ use of the ISE is

Table 8 Relative influence of the explanatory variables on the models’ estimated level of ISE use for the 20 sample firms

OLB PMM CMM EAM LagISE%

Mean value across 169 observations

Coefficient in OLS model

Proportional influence on estimated ISE use (%)

Coefficient in Tobit model

Proportional influence on estimated ISE use (%)

0.468 0.298 0.451 0.812 21.8%

0.0620 0.1833 0.0498 0.0460 0.4779

11.7 22.0 9.1 15.1 42.1

0.1114 0.1968 0.0880 0.0810 0.7691

13.6 15.3 10.3 17.1 43.7

B.W. Weber / Decision Support Systems 41 (2006) 728–746

significantly related to the lagged ISE market share of options trading in the prior quarter. Once the market began to grow, its adopters raise their usage. Liquidity does attract liquidity. The implications are that a well-designed membership structure leads to committed early users that contribute to the positive network effect from market liquidity. We conclude that the ISE benefited from its early efforts to attract members, who in turn contributed order flow to the market and attracted additional usage and new adopters. In particular, the activity in late 2001 from early lead users, such as ETrade and Goldman Sachs, established a liquidity base that encouraged other firms to use the ISE in competition with established open outcry markets. For market developers and exchange officials, the implications of this study of the ISE’s success factors are: (a)

early adoption of new markets across brokers is likely to be uneven, but the sustained use of early adopters can generate network externalities that draw in more participation (b) brokers and traders with membership positions in an electronic exchange use its screen-based market more actively, and membership advantages are helpful in attracting order flow that might otherwise be directed to the incumbent exchanges (c) segments of brokers, in the ISE’s case online brokers, are likely to adopt electronic trading more rapidly than other segments such as full service brokers (d) electronic market making and access can attract the order flow from securities firms that would not join a floor-based exchange that required a large staff.2

2 bMorgan Stanley was not in the [options trading] business pre-ISE because we didn’t deem it financially appropriate to maintain a staff of many brokers and floor traders in various exchanges.QUQuote in Business Week from Thomas R. Cardello, a managing director at Morgan Stanley. The article continued: bUsing ISE’s automatic trading system and algorithms derived from the stock prices, volatilities, and interest rates, Cardello says he can quote 50,000 options at any given time.Q From: bBest Little Options Exchange in America?—The arrival of the ISE broke up a cozy cartel,Q Business Week, September 2, 2002.

741

Using SEC Rule 11Ac1-6 disclosures from 20 major brokerage firms, a number of significant influences on firms’ use of a new electronic options market were identified: the firm’s ISE membership categories, if any, whether it is an online broker or not, lagged ISE market share, and firm fixed-effects are important determinants of ISE adoption at the brokerage firm level. The work illustrates the many nontechnological aspects of new trading systems that influence their adoption. It shows how online exchanges’ governance structures can create incentives for user firms and order providers to benefit as the liquidity and activity levels of the market increase. Creating business value from electronic markets and exchange is an important area of I.S. and economics research, and the work here shows that governance structures and membership affiliations can catalyze usage and liquidity in a new market. Although online markets do not have physical space constraints, ISE PMM and CMM members receive trading privileges in return for buying the membership and adhering to certain obligations to post quotes and trade. Evidence from the ISE suggests these member firms helped build initial liquidity in the critical early stages of the market. Other e-markets however are based on more open structures without costly memberships. Valuable future work on the adoption of new electronic markets will come from further study of membership privileges and obligations, new e-markets examples, and comparison of emarket order types and functionality. The new Boston Options Exchange (BOX) launched its electronic trading platform and became the sixth U.S. options exchange on February 6, 2004. In contrast to the ISE, BOX provides open access to all brokerage firms without expensive memberships, up-front costs, or formal market maker designations. The ISE’s member firms formed a committed user group. BOX’s ability to attract to volume will be a test whether electronic markets can succeed with nonexclusive memberless market structures. Complex orders, such as spreads and straddles, have traditionally been more suited to floor trading, but are now being introduced into electronic markets. The ability to accommodate more sophisticated orders may determine the future growth of the ISE. Further insight into the intermarket competition could come

742

B.W. Weber / Decision Support Systems 41 (2006) 728–746

from more detailed analysis of the bmaterial aspects of the order-routing relationshipQ disclosures from the Rule 6 filings. These could show how ownership in trading firms or trading systems, and payment for order flow arrangements influence brokers in choos-

ing among competing securities and options market centers. While it faced significant competitive obstacles, the ISE caught on and can now demonstrate what factors contribute to brokerage firms’ use of a new market and the market’s eventual success.

Appendix A Table A1. Online brokers in the sample routed a greater percentage of customer orders than the sample of fullservice brokers. Online brokers’ use of the ISE did not exceed the ISE’s overall market share among the five U.S. options exchanges.

AMTD (%) Limit orders 3Q01 0 4Q01 4 1Q02 33 2Q02 40 3Q02 36 4Q02 38 1Q03 34 2Q03 40 3Q03 40 4Q03 44 1Q04 38 Market/other orders 3Q01 0 4Q01 5 1Q02 18 2Q02 25 3Q02 24 4Q02 37 1Q03 27 2Q03 32 3Q03 32 4Q03 28 1Q04 42 Total nondirected 3Q01 0 4Q01 5 1Q02 29 2Q02 35 3Q02 30 4Q02 38 1Q03 33 2Q03 40 3Q03 40 4Q03 36 1Q04 39

BRWN (%)

DATK (%)

ETRD (%)

FID (%)

JBOX (%)

SCH (%)

SCOT (%)

TDW (%)

0 5 6 5 6 16

60 60 50 62 47 41 46 52 44 40 43

0 0 0 0 0 4 4 6 9 11 14

0 0 0 6 9 9 6 25 14 17 21

0 0 0 8 16 16 18 18 20 19 23

0 1 4 4 4 15 4 5 9 21 16

0 0 0 0 1 2 3 17 15 13 12

0 5 5 5 4 38

58 58 53 60 43 39 42 42 38 47 50

0 0 0 0 0 6 6 9 11 12 15

0 0 0 4 3 4 3 19 9 14 20

0 0 0 10 17 18 24 24 22 20 24

0 21 6 8 8 17 7 6 9 20 17

0 0 0 0 0 1 2 15 14 13 12

0 5 6 5 6 17

59 59 51 61 47 41 45 51 43 41 44

0 0 0 0 0 5 4 6 9 11 14

0 0 0 6 7 8 5 24 12 16 21

0 0 0 8 16 16 19 19 20 19 23

0 11 5 5 5 16 5 5 9 20 17

0 0 0 0 1 1 3 17 15 13 12

7 8 8 12 22

5 6 7 9 22

6 8 8 11 22

AMTD acquired DATK in September 2002 for $1.29 billion. Combined data first reported in 1Q03. NA: not available.

B.W. Weber / Decision Support Systems 41 (2006) 728–746

743

Table A2. ISE market share of full service brokers’ limit orders, and market and other orders, and total nondirected orders.

AGE (%) Limit orders 3Q01 0 4Q01 0 1Q02 0 2Q02 0 3Q02 0 4Q02 0 1Q03 0 2Q03 0 3Q03 0 4Q03 0 1Q04 0

BOFA (%)

34 26 31 8

Market/other orders 3Q01 0 4Q01 0 1Q02 0 2Q02 0 3Q02 0 4Q02 0 1Q03 0 2Q03 0 14 3Q03 0 14 4Q03 0 10 1Q04 0 35 Total nondirected 3Q01 0 4Q01 0 1Q02 0 2Q02 0 3Q02 0 4Q02 0 1Q03 0 2Q03 0 3Q03 0 4Q03 0 1Q04 0 NA: Not available.

29 23 25 26

BSC (%)

41 37 32 35 37

30 28 25 36 38

37 34 30 35 37

CSFB (%)

0 0 25 0 0 0 0 0 0

26 26 23 23 28 32 30 25 23

26 26 24 23 28 32 30 25 23

DBAB (%)

GS (%)

LEHM (%)

MER (%)

MS (%)

22 23 13 0 0 0 6

6 15 30 41 46 41 40 42 40 38 35

14 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 11 6 7 15 15 23

0 0 5 23 32 32 36 39 43 46 41

14 16 9 0 0 0 9

6 7 18 22 31 26 25 27 24 27 28

0 0 0 0 0 0 12 16 13 16 16

0 0 0 0 0 5 6 6 13 15 24

0 0 7 29 33 34 44 39 41 39 45

19 20 11 0 0 0 8

7 13 27 36 42 37 35 38 35 35 33

0 0 0 0 0 0 7 10 9 10 10

0 0 0 0 0 5 6 7 14 15 24

0 0 7 23 32 32 37 40 42 44 42

PRU (%)

SSB (%)

0 0 0 6 9 11 0 0 0

0 1 0 0 0 0 0 0 0 0 0

0 0 0 3 3 5 0 0 0

0 2 0 0 0 0 0 0 0 0 0

0 0 0 5 7 9 0 0 0

0 2 0 0 0 0 0 0 0 0 0

744

B.W. Weber / Decision Support Systems 41 (2006) 728–746

Table A3. Online brokers and their ISE membership and date.

Broker

Membership type

Date

Notes

AMTD

EAM CMM PMM EAM CMM None EAM CMM PMM EAM

13-Dec-01 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 20-Aug-01 – As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 18-Jun-02

Indicator variable for EAM set to 0 for 3Q01, 1/6=0.167 for 4Q01, and 1 thereafter. PMM and CMM via ownership interest in Adirondack trading partners wholly owned by JP Morgan-Chase, whose JP Morgan Securities unit was an EAM and later became a CMM.

None EAM EAM EAM CMM

– As of 31-Dec-00 As of 31-Dec-00 28-Nov-01 01-Apr-02

BRWN DATK ETRD

FID JBOX SCH SCO TDW

E*Trade is an EAM PMM and CMM via ownership interest in Adirondack Electronic Markets/KAP Group Indicator variable for EAM set to 0 for 1Q02 and before, 1/6=0.167 for 2Q02, and 1 thereafter.

Indicator variable for EAM set to 0 for 1Q02 and before, 1/6=0.167 for 2Q02, and 1 thereafter.

Table A4. Full-service brokers and their ISE membership and date.

Broker

Membership type

Date

AGE BOFA BSC

None EAM EAM CMM PMM EAM CMM EAM CMM PMM EAM CMM PMM EAM CMM EAM CMM PMM EAM CMM PMM EAM EAM

– As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 29-Aug-02 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 15-Nov-02 As of 31-Dec-00 10-Nov-03 10-Nov-03 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00 As of 31-Dec-00

CSFB DBAB

GS

LEHM MER

MS

PRU SSB

Notes

Sold its CMM and PMM memberships Nov 10, 2003. Indicator variable for PMM and CMM set to 0 for 3Q03, 4/9=0.44 for 4Q03, and 0 thereafter.

PMM, CMM, and EAM via Deutsche Bank Securities

PMM and via Hull Trading Co. subsidiary, EAM via Goldman Sachs and Co. and Spear, Leeds and Kellogg subsidiary Indicator variable for CMM set to 0 for 3Q02, 1/2=0.5 for 4Q02, and 1 thereafter. Indicator variable for PMM and CMM set to 0 for 3Q03, 5/9=0.56 for 4Q03, and 1 thereafter.

B.W. Weber / Decision Support Systems 41 (2006) 728–746

Table A5. ISE market makers.

References

Primary Market Makers, as of May 2004 (1) Knight Financial Products LLC (2) SLK-Hull Derivatives LLC (3) Adirondack Electronic Markets LLC (4) Citadel Derivatives Group LLC (5) UBS Securities LLC (6) Timber Hill LLC (7) Deutsche Bank Securities Inc. (8) Morgan Stanley & Co., Incorporated

Bin 1 Bin 2 Bins 3 and 4

Competitive market makers (1) Adirondack Electronic Markets LLC (2) Archelon LLC (3) Bear Wagner Specialists LLC (4) BNP Paribas Securitie

As of May 2004 Bins 1, 2, 5, 6, 10

(5) Citadel Derivatives Group LLC (6) Credit Suisse First Boston LLC (7) Cutler Group, LP (8) Deutsche Bank Securities (9) Geneva Trading LLC (10) Group One Trading, L.P. (11) J.P. Morgan Securities (12) Knight Financial Products LLC (13) Lehman Brothers (14) MAKO Global Derivatives LLC (15) Merrill Lynch Professional Clearing Corporation (16) Morgan Stanley & Co. Incorporated (17) Optiver US, LLC (18) PEAK6 Capital Management LLC (19) SLK-Hull Derivatives LLC (20) TD Options LLC (21) Timber Hill LLC (22) UBS Securities LLC (23) Wolverine Trading, LLC Source: http://www.iseoptions.com.

745

Bins 5 and 8 Bin 6 Bin 7 Bin 9 Bin 10

Bins 2, 3, 4, 5, Bins 2, 6, 7, 9 Bins 1, 2, 3, 4, 6, 7, 8, 9, 10 Bins 1, 2, 3, 4, 7, 9, 10 Bins 1, 2, 3, 4, 6, 7, 8, 9, 10 Bin 1 Bins 1, 2, 3, 4, 6, 7, 8, 10 Bin 10 Bins 8, 10 Bin 9 Bins 2, 3, 4, 5, 7, 8, 9, 10 Bins 1, 2, 3, 4, 6, 7, 8, 9, 10 Bin 9

7 5, 6, 5,

5,

6, 5,

Bins 1, 2, 3, 4, 7, 8, 9, 10 Bins 1, 2, 3, 4, 5, 6, 7, 8, 9 Bin 3 Bin 5 Bins 7, 8, Bins 7, 8 Bins 6, 8, Bins 7, 8, Bins 6, 7,

1, 3, 4, 5, 6, 9, 10 1, 3, 4, 5, 6, 1, 9, 1, 9, 1, 8,

2, 3, 4, 5, 10 2, 3, 4, 5, 10 2, 3, 4, 5, 9, 10

[1] M. Barclay, T. Hendershott, T. McCormick, Competition among trading venues: information trading on electronic communications networks, Journal of Finance 58 (2003) 2637 – 2665. [2] R. Battalio, B. Hatch, R. Jennings, Toward a national market system for U.S. Exchange-listed equity options, Journal of Finance 59 (2004) 933 – 962. [3] E. Bridges, Y.C. Kin, R. Briesch, A high-tech product market share model with customer expectations, Marketing Science 14 (1) (1995 Winter) 61 – 81. [4] E. Clemons, B. Weber, London’s big bang: a case study of information technology, competitive impact, and organizational change, Journal of Management Information Systems 6 (4) (1990) 41 – 60. [5] E. Clemons, B. Weber, Alternative securities trading systems: tests and regulatory implications of the adoption of technology, Information Systems Research (1996 June) 163 – 188. [6] E. Clemons, B. Weber, The Optimark experience, in: R.A. Schwartz (Ed.), Building A Better Stock Market: The Call Market Alternative, Kluwer Academic Publishers, 2001, pp. 353 – 364. Chapter 22. [7] P. De Fontnouvelle, R.P.H. Fishe, J.H. Harris, The behavior of bid-ask spreads and volume in options markets during the competition for listings in 1999, Journal of Finance 58 (2003) 2437 – 2464. [8] I. Domowitz, B. Steil, Innovation in equity trading systems: the impact on transaction costs and the cost of capital, in: B. Steil, D. Victor, R. Nelson (Eds.), Technological Innovation and Economic Performance, Princeton University Press, 2002. [9] M. Fan, J. Stallaert, A.B. Whinston, The Internet and the future of financial markets, Communications of the ACM 43 (11) (2000) 82 – 88. [10] W.H. Greene, Econometric Analysis, 3rd Edition, PrenticeHall, Upper Saddle River, NJ, 1997. [11] A. Grunbichler, F. Longstaff, E. Schwartz, Electronic screen trading and the transmission of information: an empirical examination, Journal of Financial Intermediation 3 (1994) 166 – 187. [12] J. Hamilton, Electronic market linkages and the distribution of order flow: the case of off-board trading of NYSE-listed stocks, in: H. Lucas, R. Schwartz (Eds.), The Challenge of Information Technology for the Securities Markets: Liquidity, Volatility, and Global Trading, Dow Jones-Irwin, 1989. [13] T. Hendershott, Technological innovations and electronic trading systems in financial markets, IEEE-IT Professional (2003 July–August) 10 – 14. [14] A. Kambil, E. Van Heck, Making Markets: How Firms Can Design and Profit from Online Auctions and Exchanges, Harvard Business School Press, 2002. [15] M. Massimb, B. Phelps, Electronic trading, market structure and liquidity, Financial Analysts Journal (1994 January– February) 39 – 50. [16] R.A. Schwartz, Reshaping the Equities Markets: A Guide for the 1990s, Business One Irwin, Chicago, 1993.

746

B.W. Weber / Decision Support Systems 41 (2006) 728–746

[17] R.A. Schwartz, B.W. Weber, Next-generation securities market systems: an experimental investigation of quote-driven and order-driven trading, Journal of Management Information Systems 14 (2) (1997 Fall) 57 – 79. [18] B. Weber, Elements of market structure for on-line commerce, in: Chris F. Kemerer (Ed.), Information Technology and Industrial Competitiveness: How Information Technology Shapes Competition, Kluwer Academic Publishers, Boston, 1998, pp. 15 – 32. Chapter 2. [19] B. Weber, Next-generation trading in futures markets: a comparison of open outcry and order matching systems, Journal of Management Information Systems 16 (2) (1999 Fall) 29 – 45. [20] B. Weber, Growing market liquidity at the international securities exchange, IT Professional, IEEE Computer Society 5 (4) (2003 July/August) 22 – 29. Bruce W. Weber is Associate Professor of Information Management at the London Business School, where he teaches bInformation ManagementQ and bTrading and Market StructuresQ in the MBA programme. His research examines electronic market systems and, in particular, IT-driven competition in online financial services and securities markets. He has an A.B. in Applied Mathematics from Harvard University and a Ph.D. in Decision Sciences from the Wharton School of the University of Pennsylvania. Prior to joining the London Business School in 2003, he was on the faculty of the Stern School of Business, New York University,

and Baruch College of the City University of New York, where he was founding director of the Subotnick Financial Services Center. He is on the editorial boards of Information Systems Research, Journal of MIS and Decision Support Systems. His articles have appeared in Management Science, Information Systems Research, Journal of Management Information Systems, Journal of Organizational Computing, and The London Stock Exchange Quarterly. His work has been cited in the Financial Times, the Wall Street Journal, and the New York Times, and he has been an invited speaker at regulatory hearings and at industry conferences. He is co-developer with Robert A. Schwartz of the NASD HeadTrader simulation, which is available at http://www.nasd.com/HeadTrader/BYP-main.htm. He has consulted on e-finance issues for several major financial services firms and the Nasdaq Stock Market and London Stock Exchange and has presented executive training programs in decision analysis and technology strategy to groups from U.S. and European firms.