Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/dsw
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 Commissions (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 firms membership role in the ISE,
and the network externality effect of the ISE markets 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
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 Exchanges 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, Londons
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
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
Low
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
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
(for 12 firms)
(9)
(8)
(6)
(6)
(3)
(2)
(3)
(4)
(4)
(3)
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
others 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 days 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
731
Table 2
Volumes on five SEC-registered exchanges for listed options trading
(Options contracts in millions)
2000
2001
2002
2003
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
732
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
Fig. 1. The route of an order in floor options markets. A customers 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.
733
Fig. 2. ISE market flows: a customers order is entered online or phoned to a broker. The order (at the brokerage firms 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.
734
Table 4
Brokers in the sample fit into two categories
Discount online brokers (OLBs)
AMTD
BRWN
DATK
ETRD
FID
JBOX
SCH
SCO
TDW
(2)
Ameritrade
BrownCo.
(online unit of
J.P. Morgan Chase)
Dateka
E-Trade
Fidelity
J.B. Oxford
Charles Schwab
Scott Trade
T.D. Waterhouse
(3)
735
Graph 1. A comparison of the ISEs 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
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%
737
738
i; Quarter j
j1
eij
Table 6
Descriptive statistics for the variables in the model
Variable
Mean (n=188)
14.7%
46.8%
29.8%
45.1%
81.2%
21.8%
739
Table 7
Estimation results
Variable
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
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
740
The use of 01 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 variables 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 611% 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 quarters 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
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%
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
2
bMorgan Stanley was not in the [options trading] business
pre-ISE because we didnt 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 ISEs 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
742
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 ISEs 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.
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
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.
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.
References
Bin 1
Bin 2
Bins 3 and 4
As of May 2004
Bins 1, 2, 5, 6, 10
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
746