1. Impact of Multilateral Trading Facilities on corporate
Security Trading Value Chains
by
Ulrich Benedikt Staudinger
A DISSERTATION
Submitted to
The University of Liverpool
in partial ful
2. llment of the requirements for the degree of
Master of Business Administration
2012
3. A Dissertation entitled
Impact of Multilateral Trading Facilities on the Security Trading Value Chain
by
Ulrich Benedikt Staudinger
We hereby certify that this Dissertation submitted by Ulrich Benedikt Staudinger
conforms to acceptable standards, and as such is fully adequate in scope and
quality. It is therefore approved as the ful
4. llment of the Dissertation require-ments
for the degree of Master of Business Administration.
Approved:
Dissertation Advisor Norman Williams, Date 28 June 2012
The University of Liverpool, 2012
1
6. cation Statement
I hereby certify that this paper constitutes my own product, that where the lan-guage
of others is set forth, quotation marks so indicate, and that appropriate
credit is given where I have used the language, ideas, expressions or writings of
another.
Signed, Ulrich Staudinger
2
26. rm's trading activities become
... by trading on MTFs? . . . . . . . . . . . . . . . . . . . . . . . . 81
4.28 Question 4.4.5 - Have or would the
27. rm's trading activities become
... by trading on MTFs? . . . . . . . . . . . . . . . . . . . . . . . . 82
4.29 Question 4.4.6 - Have or would the
28. rm's trading activities become
... by trading on MTFs? . . . . . . . . . . . . . . . . . . . . . . . . 82
4.30 Question 4.5 - Amount of dark pools, the company connects to . . . 82
4.31 Question 4.7 - Key factors for deciding to trade at an MTF . . . . . 83
4.32 Question 4.9 - KPIs most relevant in the context of MTFs . . . . . 83
4.33 Question 4.10 - Possible acceptance of MTFs for trading futures and
other derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Abstract
The
29. nancial industry hosts the securities trading value chain. Corporations
from all over the globe cooperate due to increased
30. nancialization and globaliza-tion.
Contrary to this, regulators act on a regional level. The European ESMA
has crafted the MIFID directive in 2004 and as a result a new European
31. nan-cial
landscape shaped. Multilateral trading facilities (MTF) oer services not
unlike those of public exchanges. Current academic literature covers MTFs only
non-satisfyingly and leaves large gaps in understanding the impact of MTFs on
both
32. nancial industry and securities trading value chain. In addition to this
gap, little available literature deals with securities trading value chain speci
33. cs,
such as its breadth. This study deployed a triangulated methodology consist-
10
34. ing of qualitative and quantitative research legs and analysis of results obtained.
Verbal, free-form interviews with six industry representatives from MTFs, pro-prietary
trading houses, clearers and independent software vendors yielded broad
information. A web-based survey tool provided a possibility for 27 survey takers
to answer various value chain, MTF and industry related questions. The qualita-tive
and quantitative study legs showed, that MTFs are accepted market places
to trade equity shares. The sample further con
35. rmed the validity of a proposed
securities trading value chain model. The average value grid depth assessment
showed that
36. rms in the sample interact per value chain phase usually with
3.25 counterparties. The sample con
38. rms' value chains and the industry. This study found that
MTFs successfully enable trading large quantities of shares. By attracting large
block traders, they oer deeper liquidity than public exchanges. By draining liq-uidity
from public exchanges and because of the increasing number of MTFs, liq-uidity
fragmentation sets in. The study sample perceives liquidity fragmentation
as a negative eect. The security trading value chain is found to pursue a stan-dardization
and interchangeability of services oered, thus successfully enabling
competition. The study concludes by advising MTFs increase client binding and
change barriers by oering value adding services, such as services that simplify
their clients' value chains. Clients should continue standardizing their commu-nication
and interaction interfaces to reduce vendor lock-ins, thus being able to
switch seamlessly between interchangeable service providers.
11
40. cant changes in the securities trading ecosystem over the last years bring
new opportunities and challenges. These changes result from new regulations,
advances in technology, overall increased
41. nancialization of our world and of
course MTFs. Dark Pools and their European pendant, multilateral trading
facilities (MTF), oer services satisfying the demand of large-block traders to
minimize market impact of their orders. As MTFs are a major force for market
change, it becomes clear that securities trading corporations of all type have to
be able to understand and foresee the impact these new players have on their
interconnected business and the industry.
Some of the many dierent approaches to describe a
43. rms are more suited to describe value
ow than others. As
12
44. an example, whereas supply chains deliver a big picture of goods and services
owing from supplier to consumers, business process modeling focuses on details
about interaction patterns and information exchange. Diametrical to these two
approaches, the value chain approach concentrates on modeling and understand-ing
the
ow of value in a
46. rms.
Little academic literature deals solely with MTFs and even less deal with the
entire value chain surrounding these multilateral security trading facilities. This
work analyzes through qualitative and quantitative research the security trading
value chain and to what extent related value grids have already changed or will
likely change because of MTFs.
Chapter 1 sets the scene in which this work looks at MTFs and their impact on
the security trading value chain. It presents the current market size and its frag-mentation
and de
47. nes the goals of this study. Chapter 2 looks at current liter-ature
about macro and micro structure of value chains, about Dark Pools and
MTFs. Chapter 2 also investigates the current literature about external change.
Chapter 3 de
48. nes the research approach and describes the methodology. Chapter
4 presents the collected results in graphical and textual form. Chapter 5 ana-lyzes
the situation by comparing synthesized data with information collected at
former stages, drawing conclusions about how MTFs impact the securities trad-ing
value chain. Chapter 6 concludes this study, points out additional research
possibilities and gives some advice to MTFs and clients and deducts three hy-potheses.
13
49. 1.2 Introduction to the alternative liquidity mar-
ket
Over the past three decades, the trading volume of major stock indices and at
exchanges in general has increased to a formerly unseen level. Although recent
geopolitical events, such as the collapse of the U.S. housing market in the CDS
aair or the latest European debt crisis, have left a mark on trading volumes,
market breadth is still well above the niveau of the beginning 2000s. Figure 1.1
on page 15 shows the absolute number of trading volume for the SP 500 and
the percentage change in the number of transactions month-by-month between
1950 and 2011. As can be seen, trading volumes have increased exponentially
until about 20081.
Nonetheless, the 90s and the 00s have given birth to signi
50. cant technical changes
which also transformed the security trading business: pit traders vanished and
got replaced with machines, new trading trends, such as High Frequency Trad-ing
or Algorithmic Trading have gained signi
51. cant popularity and have led to
challenging discussions in media worldwide (Weber, 2000). Broker ran trading
platforms, labeled Dark Pools, Automated Trading Systems, Electronic Crossing
Networks or MTFs, depending on the geographical region, shift trading activity
from exchanges to individual platforms. All this was possible due to the digital
revolution that includes engineers, mathematicians and economists (Paulden,
2007).
1Just as a side note, the wide-spread myth, that the Internet has resulted in an extraordi-
nary increase of transactions must be ruled void, as the afore mentioned chart shows no signs
of deviation from the long term market growth.
14
52. Number of transactions in SP 500 since 1950 per month
Jan
1950
Jan
1960
Jan
1970
Jan
1980
Jan
1990
Jan
2000
Jan
2010
0e+00 6e+09
Number of transactions
Jan
1950
Jan
1960
Jan
1970
Jan
1980
Jan
1990
Jan
2000
Jan
2010
−0.4 0.2 0.6
Growth of #transactions, SP 500 since 1950 per month
% delta volume month−to−month
Figure 1.1: Trading volume SP500, 1950-2011 (Yahoo, 2012)
15
53. 1.3 About MTFs
MTFs are a result of the European MIFID regulation from 2004, which for the
54. rst time permitted the creation of regulated, public bodies for exchanging eq-uity
shares away from their primary exchange. In layman's terms, the main dif-ference
from other, ordinary 2nd markets are the exchange like nature where
supply and demand meet in exchange like auctions, contrary to OTC transac-tions.
MTFs either belong to an existing
55. nancial enterprise, such as a bank, or they
are created from scratch by talented individuals. Examples of bank owned MTFs
are UBS MTF or BATS Chi-X Europe. The MTF market is in constant change.
The reason why banks or other corporations create MTFs are quickly told. Whereas
in the commodities markets, hardware or software is treated as the product; the
banking and
56. nancial industry is largely a service industry where large corpora-tions
oer unique service to the market. MTFs are new service oers by
57. rms to
extend their reach in the trading sector.
But not all MTFs are equal. MTFs run order books, just like public security ex-changes,
but some MTFs run dark and lit order books in parallel, whereas others
run only fully dark books. A dark order book does not display the current sup-ply
and demand in an equity share to trading participants, lit order books show
the current buy and sell orders. Trading in dark order books can be compared to
playing poker with hidden hands, where a reference hand - the public exchange -
is shown to everyone.
16
58. Exchange venue Turnover % share
Order Book - Lit Total EUR 93,730,587,662 35.76%
Order Book - Hidden Total EUR 77,101,719 0.03%
Order Book - Auction Total EUR 14,423,974,936 5.50%
Dark Order Book Total EUR 3,732,831,639 1.42%
O Order Book Total EUR 5,288,531,554 2.02%
MiFID OTC Total EUR 144,884,986,028 55.27%
Grand Total EUR 262,138,013,538
Table 1.1: Market share for all XETRA DAX index equities; Turnover January
2012(Thomson-Reuters, 2012)
By the time of this writing, the European MTF landscape contains about 15 dif-ferent
MTFs with dierent market share. After a merger between Chi-X and
BATS, both international players, long time market share leader UBS MTF
is now ranked second behind Chi-X Europe. Of relevance are also Turquoise,
TradeGate, Sigma-X and POSIT.
At the moment, MTFs, both lit and dark, are responsible for a very signi
59. cant
amount of all equity trades in Europe. According to Thomson-Reuters' market
share report (Thomson-Reuters, 2012), exchanges suer under a continuously
diminishing role, this means in numbers that only 30% to 50% of all trades in
XETRA equities take place on the main exchange, the rest happens either on lit
or dark MTFs or as plain OTC transactions; for details see table 1.1 on page 17.
MTFs gain more market share every month.
Contrary to popular belief, MTFs, dark pools and dark liquidity are not the
same. Dark liquidity is a very fuzzy term that usually includes all forms of or-ders
that are already placed or will be placed under certain conditions but which
are not publicly revealed. Already placed dark liquidity can take the form of
iceberg orders, which indicate only a fraction or none of the entire order to the
17
60. market. Industry-wide, two terms are used to label two additional subtypes of
dark liquidity: latent liquidity or intrinsic liquidity.
Latent liquidity includes orders that are already placed at exchanges, but which
are due to market data feed restrictions visible only to a small fraction of profes-sional
market data feed consumers. How can that happen? Usually, only the ten
best buy and sell price levels are disseminated through market data feeds and
thus orders that are far above or below the current price niveau are not publicly
announced.
Intrinsic liquidity adds additional liquidity to the market in case of need. Con-sider
the case of an Exchange Traded Fund (ETF), that replicates an underlying
index, which is built out of a basket of equity shares. While one might see that a
particular ETF shows only little liquidity, placing a large order in such a thinly
traded ETF might result in hedge orders, sent by the ETF issuer. These result-ing
orders are going to take or add liquidity to the underlying basket of stocks
and are called intrinsic liquidity.
1.4 Exceptions and simpli
61. cations for this study
Unless explicitly stated, this study is restricted to European MTFs and uses the
terms Dark Pools and Automated Trading Systems as synonyms for dark MTFs.
Although automatic trading platforms and dark liquidity emerged in the U.S.,
this study does not refer to U.S. speci
62. cs unless absolutely necessary. We further
exclude all areas of dark liquidity that are not part of the core business concept
of a venue, for example Iceberg orders or ways to algorithmically
64. minimal market impact. OTC markets, despite being the major dark source of
liquidity, are also not covered by this study.
This study also ignores ownership structures and talks about MTFs as if they
were entirely standalone entities. It covers ownership structures where they af-fect
the study topic.
1.5 Goals of this study
As the former sections reveal, this study is based on a general industry compre-hension,
where environmental forces in
uence existing structures to transfor-mation.
With this general understanding, it is impossible to set forth working
hypotheses, but possible to state research goals. Therefore, the study analyzes
existing security trading value chains and assesses how these changed as MTFs
became and become more relevant market places.
Restated, the primary objectives of this study are:
• Create a con
65. rmed model of the securities trading value chain.
• Describe the impact of MTFs on the securities trading value chain.
• Describe the impact of MTFs on the
66. nancial industry.
• Deduct one or several hypotheses resulting from impact assessments.
The following sections provide more details about each of these goals.
19
68. rmed securities trading value chain model
In order to analyze the securities trading value chain model, it is absolute imper-ative
to de
69. ne a securities trading value chain model before progressing to the
impact assessment phase. This model needs to be con
70. rmed as a result of this
study, so that all further argumentation and work builds on stable, defendable
ground.
1.5.2 Goal: impact assessment of MTFs on the securities trad-
ing value chain
The securities trading value chain is central to this study and deserves special
attention. In order to derive a theoretical model that describes the impact of
MTFs on the value chain, it is mandatory to collect data about the extent by
which MTFs have altered the value chain of
82. nancial industry makes full impact assessment a super-human
task, this dissertation suces to capture the most relevant changes due
to MTFs. Similar to assessing the impact of MTFs on the value chain, this study
aims at putting a 'price tag' on MTF related activities for the entire industry.
1.5.4 Goal: stating further hypotheses
Based on careful analysis of the impact of MTFs on the securities trading value
chain, this work postulates one or several hypotheses about generic patterns, oc-curring
when market environments of the
83. nancial industry and the securities
trading value chain change. These hypotheses can serve as a fundament for addi-tional
research.
21
84. Chapter 2
Literature research
The sheer overwhelming amount of academic, economic literature represents a
great fundament where to do research on. But this large knowledge corpus is
also hard to navigate. This chapter presents a snapshot of security trading value
chain and MTF related literature. It is split into seven sections which collect lit-erature
about key players in the
85. nancial system, the securities trading value
chain and MTFs plus a round-up of change related literature. It closes with
de
86. ning the generic securities trading value chain that serves as a fundament
for the entire study.
2.1 Key players in the European
87. nancial system
The security trading domain is compound of a buy-side, a sell-side and a regula-tory
layer, where each layer contains a speci
88. c set of requirements and rendered
services. It is not always apparent what quali
90. side participant, therefore this section uses an industry-accepted common view.
The buy-side includes all consumers of trading and brokering services, such as
institutional and private investors, including hedge funds or proprietary trading
desks. Such service consumers operate dierent trading approaches. Whereas
many private investors follow a very discrete approach, for example a Kostolany-like
buy and hold approach, proprietary trading desks can employ very special-ized
trading algorithms. With increased technological sophistication increase re-quirements
in terms of professional service. Usually, institutional investors dier
from more agile trading entities through their sheer mass of equities under man-agement.
According to OECD reports, institutional investors in the OECD (17)
countries held in 2005 average assets worth 122:01% of their individual GDPs
(Gonnard, Ki and Ynesta, 2008, p.4), with equity shares constituting a major
investment vehicle (ibidem, p. 21f). Concrete examples of institutional investors
include pension funds or large hedge funds.
The sell-side is a very broad term, de
92. nancial institutions that are involved in servicing buy-side clients and by
Harris (2003) as the entirety of dealers and brokers who provide exchange ser-vices
to the buy side, thus excluding technology or data providers like indepen-dent
software vendors or market data disseminators. In contradiction to Harris,
this study will include technology providers in the sell side.
Although geographically and politically separated, international regulatory insti-tutions
pursue similar goals through market permeating supervision. Contrary
to the U.S., the European Union does not have a central market supervision or-gan.
Although the fundamental political and economical changes past the sign-
23
93. ing of the Treaty of Maastricht in 1993 have led (a) to a harmonization of
94. nan-cial
policies and (b) to the formation of the Financial Services Action Plan 1, no
European market supervision had been formed until beginning 2012 (Jackson,
2010). Whereas the U.S. has the SEC, the FINRA and the CFTC as highest
bodies of market supervision, European markets are supervised by national bod-ies,
such as the German BAFIN. More recently, the responsibilities of the Euro-pean
Securities and Market Authority (ESMA) have shifted from providing pure
technical advice for ministries and councils to the crafting of laws and regula-tions
to be put into place (ESMA, 2011). But still, the national implementation
and supervision remains the sole responsibility of national bodies. But the need
for regulations and regulators are not without dispute. Davis, Neal and White
(2003) present that exchanges are historically self-regulatory bodies with clear
obligations on its market participants. This self-regulatory aspect stands in stark
contrast to the current European approach.
The key players in the
95. nancial system are summarized in table 2.1 on page 45.
This leads to describing a network of the dierent key players and their inter-dependencies,
shown in
96. gure 2.1 on page 25. Framing the key player model is
necessary to avoid ambiguities later on.
2.2 Value chain and value grid characteristics
Since its introduction by Porter (1985), the appealing value chain concept has
received signi
97. cant attention by academics and practitioners. The original idea
1... a process including several committees that also created the all-changing MIFID direc-
tive
24
99. of chain links, which combine value generating work interaction patterns into
complex chains has created a formerly unprecedented level of understanding
across market participants worldwide. The concept of value chains helps iden-tifying
the factors that drive competitive advantage (ibidem). Company-internal
value chains link to industry value chains in which a processor generates high
value output from low value input. The
100. nancial industry is special. Whereas
the manufacturing industry deals with physical goods, the
101. nancial industry re-sembles
an information processing industry, where information transformation of
lower information to higher information constitutes the value generating process.
Value chains include supply chains and demand chains (Porter, 1985), with sup-ply
chain being push driven and demand chains being pull driven. Each step
in the value chain has a supply side and a demand side. The demand chain is
perceived as belonging to the typical domain of marketing, sales and strategic
supply-chain managers (Walters and Rainbird, 2004), whereas the supply chain
belongs to tactical manufacturing and logistics personnel (ibidem).
Due to possibly tight integration of value chains between companies and indus-tries,
adverse eects at one place in the value chain can threaten the value chain
in its entirety. One example of these adverse eects are the so called bull-whip
eects (Lee, Padmanabhan and Whang, 1997). Obviously it must be in each
corporation's self interest to minimize possible negative consequences of tight
integration and dependence along the value chain through such measures as con-tinuing
standardization in areas like standard cargo container sizes, quality or
quantity standards and more importantly in the digital world through standard-ized
communication and information exchange protocols.
26
102. More recent additions to the value chain literature distinguish global value chains
into a buyer driven and a producer driven type (Gere, 2011). These two main
types are further re
104. ve main
subtypes. These subtypes are a) market like, where standardized products
ow
along the chain; price and exchange ability are a main decision factor, b) modu-lar,
where semi-standardized products incur low supplier or customer switching
costs, c) relational, with a high degree of complexity in more specialized trans-actions,
d) captive, where one large buyer controls many small suppliers and e)
hierarchical, where strongly integrated value chains are dominant due to com-plex
requirements (ibidem). Certain environmental and industry speci
105. c factors
are correlated with a dominance of a certain value chain type (Mahutga, 2012,
p.3).
Changes in environmental variables, for example technical innovation, can trans-form
entire value chains from buyer driven to producer driven and vice versa
(Sturgeon, 2009). Gere (2011) picks up this insight by presenting the electronic
industry, where a commodization of technology led to more modular value chains
along with a transition from on-shoring to o-shoring.
In summary, value chains are generic images of how value creation happens in a
particular, linear network of value producing parties. This simplicity of the con-cept
is also its biggest weakness. More recent additions to the value chain idea
are so called value grids, which add horizontal and diagonal dimensionality to
the linear concept of value chains (Pil and Holweg, 2006). Successful applications
of the value grid concept show that the value grid view generates meaningful in-formation
for corporate key deciders (Soilen, Kovacevic and Jallouli, 2012).
27
106. But how can one quantify value chains? Performance metrics are key to monitor,
manage and improve the premium obtained from value chains. Only with right,
critical performance measures, it is possible to decide about measures to improve
performance. Barber (2008) points out, of the various performance measuring
approaches, balanced scorecards have become the most widely used performance
measurement approach to measure the key components of
107. rms and business
processes. Although the original balanced scorecard approach was not meant to
be applied to value chains, but rather to individual pieces within a value chain,
the idea is directly transferable due to its simplicity. Barber (2008) continues
and distinguishes between
108. rst, second and third generation performance mea-surement
models, where particularly third generation models apply to the in-formation
technology industry and should therefore also apply to the securities
trading value chain with its large capital of intangible information assets. This
leads directly to the idea of key performance indicators, about which section 2.5
on page 38 elaborates.
2.3 Security Trading Value Chains
2.3.1 Macro-perspective on security trading value chains
Free trade and the security trading industry must be understood as a given fact
and a central theorem of today's economies of developed countries for which no
alternative exists. The trading of rights and non-tangibles is as old as human
kind, already the Romans traded rights, such as city council seats.
28
109. Security trading value chains can be split into three key areas: pre-trading, trad-ing
and post-trading activities; information
ows in both directions along the
industry value chain. Oxera (2008, p.3) splits trading and post-trading activities
into
112. nes in their work the description of these steps. For the sake of simplicity,
this section deals with the securities trading value chain in its three-block form:
a) pretrading, b) trading and c) posttrading.
The pretrading phase, an enterprise speci
113. c process chain, includes activities
leading to investment decisions (Knieps, 2006, p. 4). It is in its speci
114. c compo-sition
the blossom of corporate core competencies and can include such steps as
research, operating trading engines, fund share creation and redemption. As an
example, derivative engineering units deploy totally dierent decision mechanism
than portfolio investment managers.
The trading phase is at the heart of the entire value chain and includes all activ-ities
that match supply and demand in an orderly fashion. This phase includes
the OTC market, exchanges and other trading venues, such as MTFs. It goes
without saying, these venues are system relevant and their dysfunction can result
in systemic failure of the entire securities trading ecosystem.
The posttrading phase includes all value generating activities that are result of a
trade. This includes clearing, settlement and ex-post,
122. cantly and includes, in
agreement with Porter (1998, p. 199), specialized infrastructure, service or con-sulting
providers. And of course, market mechanisms dysfunction in the
123. nancial
sector, too. As soon as creating new service providers requires large monetary
substance, entry barriers raise and hardly any new participants enter. The idea
of having a central counterparty clearing house in Europe has seen appraisal be-fore
2006 and rejection past 2006 (Norman, 2011, p. 251).
Similar to other industries, also the security trading industry has undergone a
digitalization of its processes. Standardization of processes is a result of this dig-italization.
As noted in previous sections, standardization leads to modular or
even market like structures. Examples of such changes exist along the security
trading value chain, such as clearing back-oce tasks that are being outsourced
to specialists like Ridge (Anon, n.d.).
According to Oxera (2008), a same-day trade armation between investment
managers and broker/dealers, initiated through large degrees of automation,
will evoke positive business eects. This need for technological standardization
in the
124. nancial industry is highlighted by EIU (2000). Their study conducted
interviews and found that around 50% of all interviewed persons of the
125. nan-cial
sector perceive the integration of IT systems as a key success factor without
which a migration of their traditional, client-facing activities into e-commerce
processes, and thus increased business bene
126. ts, fail.
The corporate security trading value chain is tightly linked to global value chains,
such as the global derivative trading value chain outlined at Mai (2008). This
strong link results in theories which are valid for both types.
30
130. nition further, Williams (2010,
pp. 5-21) places individual and institutional investors and retail brokers and
dealers on the buy-side, but ignores mix types, for example banks that run their
own proprietary trading desks as client-less pro
131. t centers and which have to be
counted as buy side entities, too. Although trading on
132. nancial market places
includes buying and selling equity shares and other instruments at market venues,
this activity is in the following sections not meant by buy side and sell side.
Typical entities on the buy side are of course hedge funds, proprietary trading
houses, trading boutiques and private investors. But also large investment bank-ing
departments of multinational corporations qualify for the buy side. Although
the motivation for trading diers, their interaction with the sell side is always
similar. Despite these similarities, complexity is widely acknowledged (Nybo,
2012).
Market makers are particularly interesting because of their hybrid role. They
have, at least according to the strict de
133. nitions used by exchanges such as EU-REX,
an obligation to provide quotes in certain markets (EUREX, 2011) and at
the same time, consume exchange and clearing services. Risk management is of
uppermost relevance for market makers, at least it should be, and therefore po-sitions
in markets where a market maker has an obligation to trade and which
31
134. are a result of his market making activities, have to be hedged. But essentially
market makers oer services and consume trading services, too. Although they
remain invisible on the markets, hidden behind orders, their presence is crucial
for properly functioning markets.
Firms that launch an IPO are also consuming services of the sell side and can be
put into the buy side. Their relation with an launching bank and an underwriter
is best described as delicate (Roosenboom, 2007, pp. 1217-1219) and outside of
the scope of this study.
Inside the sell-side
The sell side of the securities trading value chain includes a large variety of par-ties.
The most important group members are banks, brokers, exchanges, MTFs
and crossing network operators and clearers, but not all are equally important.
An essential role in the sell side securities trading value chain play banks, their
analysts, market makers and underwriters. The critical role of investment banks
is not not only due to due to their large cash pools, which will nourish any trad-ing
channel through which these streams are funneled, but also because of their
large size with skilled and specialized sta. These large resources allow banks to
exercise more power than other players.
The sell side interfaces with the buy side through two types of communication
channels:
135. rst, personal communication channels, such as instant messengers,
phone calls and emails and second, through standardized computer-based com-munication
protocols such as FiX, SWIFT's or some other EDI.
32
136. MTFs and other trading platforms oer exchange-like services, but by default
they are not dark pools, but ordinary market places. Dark MTFs emerged in
Europe as a result of the MiFID initiatives. They have gained signi
137. cant mar-ket
share from 2005 to 2011 and attracted in November 2011 in average 5.92%
of market share in key order books (Thomson-Reuters, 2012). The sell side
138. rms
have to follow clear rules. The Financial Services Action Plan and the ESMA
are responsible for the MIFID, which gave advice how to regulate the
139. nancial
markets and which included also the four key pre-trade transparency exemp-tion
clauses that resulted in the advent of European MTFs (CESR, 2010). Just
like public exchanges, these venues are subject to regulation, which dictates how
certain business processes have to be implemented and what their output must
be. Regulated key areas are order matching principles, match reporting, trade
clearing and risk management. MiFID requires that all executions are at least
as good as or better as the European Best Bid/Oer price, order matching has
to happen between in the spread (ibidem). Although MTFs are exempt from
pre-trade transparency, they are not exempt from post-trade transparency. All
executions must be reported to a reporting endpoint within three minutes after
matching (Raan et al., 2011, p. ); central tapes do not exist, so that many non-exchange
trading platforms rely on third-party platforms, like MarkIt's BOAT2,
where trade reports from many trading platforms form a cross-venue ticker tape.
According to Ende, Gomber and Wranik (2007, p.706), MIFID says that the
buy-side may request evidence of best execution.
Based on the de
140. nition given in the buy-side section 2.3.2 on page 31, this study
2Commercial platform manufactured by MarkIt
33
141. also includes all types of funds in the sell side. As these oer securities trading
related services to other investors, they belong both to the sell- and buy-side.
ISVs, although also selling and providing software and services, are not a direct
member of the securities trading value chain, but rather a value chain enabler.
Their crucial relevance as downstream integrators from service providers to ser-vice
consumers puts them at risk of being squeezed out of the market by the
service provider or
142. rm internal business units - for example IT departments in
banks (Huang et al., 2009). Increased standardization and quality of services
oered, as in the SAP ecosystem, leads to an increase in the amount of ISVs
(ibidem). Particularly the service void around MTFs is
143. lled by ISVs that also
bridge the connectivity gap between MTF operator and client. The MTF opera-tor's
core business and core competency is operating an MTF and not counseling
customers. The same is true for customers, whose core competencies aren't soft-ware
engineering, but rather research into and development of trading systems
atop some software framework plus its operation.
2.4 Value generating activities in detail
The former high level descriptions provided enough detail to see the big land-scape
through which the securities trading value chain
ows. Throughout the
value chain, a common set of activities generates value. The work done in these
activities is not necessarily speci
144. c for a value chain link, but the output is; for
example, sta has to communicate with speci
145. c people in a special way, sta has
to maintain a speci
148. elds.
2.4.1 Software Development
Software development includes a large variety of activities around the act of cre-ating
and maintaining software to achieve business goals. Software development
almost always includes the coding, quality assurance and client acceptance of
business relevant key and support tools. In most trading corporations, line func-tions
are separated into back-, mid- and front-oce areas, software engineering is
no exception.
It is usually within the area of software engineering, that connectivity to coun-terparties,
whether trading venue, clearers or others is established. Fitting this
purpose, the most notable communication protocol is FiX, a standard that has
been created by a group around Bob Lamoureux, in the beginning 90s (Sand-man,
2004). The bene
149. ts of standardized connectivity are clear: it allows trading
parties to switch between and connect to venues in less time, thus reducing the
time to market and bene
150. ting from supplier independence. As software develop-ment
time to develop venue connectivity is reduced and corporate
exibility is
increased, standard ways of communication are widely preferred when feasible.
2.4.2 Research
Research is a crucial part in every company, but it carries dierent colors in dif-ferent
environments. Trading houses need so called quants (Joshi, 2007) for
researching new trading models, clearers need lawyers and economists for opera-
35
151. tional research and mathematicians for researching new risk models (Nosal, 2011,
p.141) and investment banks need economists as analysts to value corporations.
For many hedge funds and proprietary trading houses, research resembles the
bread and butter business in their corporation. As much as research input, ex-pectations
and actual output dier between houses, as much do research amount
and research tools dier. Quantitative and qualitative research very often uses
tailored 3rd party tools such as Bloomberg- , Thomson-Reuters or FactSet ter-minals
and services. In fully automated houses, the borders between research
and software development become fuzzy, as researchers tend to develop custom
software to cater for their organizations' needs.
2.4.3 Operations
The operational side of the securities trading value chain includes all activities
which are directly linked to running, meaning operating, a business in the securi-ties
trading area. Depending on the type of business, this can include accepting
client orders and routing these to exchanges, sending automated orders to venues
for execution, accepting and executing orders, but can also mean to settle trades
and notify included counterparties (Knight, 2010).
This large diversity of operational aspects of the securities trading value chain
results in changing professions and changing business landscapes at regular inter-vals,
be best understood by an example:
152. fteen years ago, the New York Stock
Exchange (NYSE) hosted hundreds of specialists that were responsible to match
supply and demand. Today, NYSE's entire trading operation is digitalized and
does not require human interaction. Another example comes from automated
36
153. trading houses, where sta sits and watches the automated trading systems run-ning
(IMC, 2012). In case there is an abnormal event on TV or indicated in the
Bloomberg terminal, high-salary individuals press the o button on a trading
system, just to press the on button when the situation is over.
2.4.4 Clearing
Clearers guarantee that a security transaction can take place at the prices given.
A transaction on an exchange is
154. nal only when the trade has been cleared,
meaning transactions have been booked into ledgers of all involved counterpar-ties.
Clearers communicate with the exchanges on one side, the custodians on
another side and equity depots on a third. The clearer ensures that all trans-actions
can be done, by maintaining and managing the margin accounts of its
clients. In the case of equity trading, the clearer also instructs the depot(s) to
settle the transaction. Once a transaction has been cleared, the clearer noti
155. es
the involved counterparties of a cleared transaction. Typical means to commu-nicate
a transaction include the FIX protocol, but also more rustic approaches,
such as CSV
157. le transmission3. Euroclear (2007) provides additional
background about technical mechanics around clearing. HM Treasury (2011)
mentions that the OTC market should contain more central clearers.
3CSV and PDF are industry standard
159. 2.5 Value chain KPIs
As the goals of this study are about quantifying the impact of MTFs on securi-ties
trading value chains and the industry at large, it is necessary to crystallize
possibilities how to quantify this impact. In modern corporations, KPIs mon-itor
activity and process eciency. Between the approaches of small and large
corporations exist dierences. While large corporations formalize performance
measurement due to more resources, they both follow a similar approach, namely
viewing their activities through indicators. Although most corporations mistake
key result indicators for key performance indicators (Parmenter, 2010, pp. 5-27),
it are these indicators that enable management to a) objectively understand the
business and b) measure the eectiveness of processes and process improvements.
Literature research does not yield a precise, concrete list of indicators to use in
the
160. nancial industry. The usual KPIs, like time to market and return on eq-uity,
also apply to the securities trading value chain, but do not cover securities
trading value chain speci
162. c pa-rameters,
such as costs per order, revenue per order or machine utilization per
revenue unit are interesting measures, contrary to coverage in academic litera-ture.
Clearly, there is a lot of room for research.
38
163. 2.6 Environmental changes where to draw paral-
lels from
Value chains have to either withstand or adapt to the elemental forces of change
that steer world evolution. Over the last decades, the world has been trans-formed
by digital technology in breadth only comparable to changes that re-sulted
from the invention of combustion engines. This section investigates how
industries in general react to fundamental changes, followed by drilling into how
speci
164. c value chains react to innovation.
Japan's country image awakens in one's mind associations of distinct culture,
social unity, loyalty and eciency, but also of hard facts, such as high-tech facto-ries
or megalopolitic cities. Japan's economy is also tightly associated with lean
production or just-in-time production; a result of scarce resources after the sec-ond
world war (Sturgeon, 2007, p. 5). As Japan was by the 1980s the leader in
living a lean-production concept, the lean idea in
165. ltrated many areas of national
industry value chains and resulted in value chain modularity (ibidem, p. 10),
but also over the years in strong vertical value chain integration. When other
countries, such as the United States, picked up the idea of modular value chains
and honed the idea of geographically scattered value chains by outsourcing large
parts of work to cheaper countries through an increased use of communication
technology, the original inventors of lean production suered increased compet-itive
pressure. Sturgeon (2007) concludes that the large vertical integration of
Japanese
166. rms has made these corporations in
exible, a state which forced these
39
167.
168. rms to react to increasingly global value chains through creating larger, uni-functional
production units, for example in strategic alliances. It is possible to
compare the situation of Japan in the 1980s with the current situation in the
169. -
nancial industry. In the latter, stable and established structures produced a well
functioning environment with moderately frequent market crashes. New regula-tions,
technology and other environmental forces result in increased competition.
The decline of the communistic east-block created extreme social and economic
changes in former soviet countries. Studies suggest that this increase of compet-itive
pressure, combined with social factors such as an increase of possibilities
for personal development, led to fundamental changes within the predominantly
agricultural industry value chains, eminent in pre-perestroika states. Using the
example of a county in Poland, Dannenberg and Kuemmerle (2010) show that
afore mentioned fundamental changes resulted a) in a professionalization and
growth of a few farms and b) an increase in the amount of small farms and c) a
dying away of medium-sized farms. They conclude, the agricultural value chain
was altered in such a way that the value grid width shrinked and condensed as
the number of industrial-scale agricultural goods producers decreased. The rele-vance
of this example comes from its similarity with the existing securities trad-ing
market: a market opened up to more suppliers and reacted with fundamental
transformation.
Returning to the
172. ned by the computerization of exchanges worldwide in the 1990s.
Continuous alteration of existing security trading value chains, by upgrading
and improving each individual link, increased pressure on
174. fore the turn of the century, digitalization had been perceived as a major trad-ing
enabling factor (Clemons and Weber, 1998). As trading technology contin-ues
to improve, companies have to move with their competitors just in order to
stay competitive, for example by using systems like the OptiMark system (ibi-dem).
Interestingly, some markets, such as the
175. xed income market, seem to
withstand computerization and the continuing anonymousness of trading in gen-eral.
Montazemi, Siam and Esfahanipour (2008) say that this is due to market
speci
176. cs: a) very strong inter-human relations, b) relatively large transaction
volumes involved in
177. xed income markets and c) the relatively short list of possi-ble
large scale market participants. Clearly, market volume in
178. xed income mar-kets
is very high and typically, large volume transactions involve more human
trust than small-scale transactions in high frequency trading (Montazemi, Siam
and Esfahanipour, 2008, p.266). So, obviously the other markets should exhibit
opposite characteristics. This computerization was not exclusive to the devel-oped
world, as Kanasro and Chandio (2011) show in their case study about the
Karachi, Pakistan stock exchange, and it is still showing in markets that develop
slower.
2.7 Dark liquidity, MTF and dark pool speci
179. cs
The term dark liquidity is a very broad term, for which Banks (2010) lists through-out
his comprehensive book various types of dark liquidity and dark pools. He
de
180. nes dark pools as a venue or mechanism containing anonymous, non-displayed
trading liquidity that is available for execution (ibidem, p. 3). Bank's de
182. includes dark-by-design venues, such as MTFs, but also liquidity not everybody
would readily include, such as iceberg orders, specialist orders and latent liquid-ity.
Although Banks does not include OTC transactions in his book, according
to Price (2011) public perceives OTC transactions as a form of dark liquidity.
Europe's major dark venues, several MTFs and broker crossing networks, ac-count
by the time of this writing for 5.92% of the total order book in equities
(Thomson-Reuters, 2011b), not including OTC transactions. Although six per-cent
does not sound much, the growth rates in this sector are impressive with an
increase from 4.14% to 5.92% year-to-date, which corresponds to around 43%.
The picture in the U.S. shows that dark pools account there for around 12% of
all transactions and that the total market share decreases as market growth at
public exchanges outpaces the trading volume increase on dark venues (Demos
and Makan, 2011).
Clearly, the dark pool venue market is fragmented, ten venues account for about
90% of all trading volume (Thomson-Reuters, 2011a). Based on these numbers,
the Her
183. ndahl index can be calculated to be 25.42%, which suggest a healthy
competitive market with a mediocre fragmentation. If Benston and Hagerman
(1974) are right, then this market fragmentation should result in tighter spreads
for individual securities and less unsystematic risk. Intuitively, market fragmen-tation
should indeed reduce risk of a total market failure stemming from failures
of individual market providers - compared to the highly monopolistic situation
within the security clearing industry.
Over the years, the topics covered in academic studies stayed mostly the same:
regulations, market eciency, optimal executions and strategies. More recent lit-
42
184. erature investigates how dark pools aect market eciency, market quality and
market fragmentation. So do, based on the de
185. nition of a weak form of mar-ket
eciency, O'Hara and Ye (2011)
186. nd that market fragmentation for indi-vidual
stocks improves market eciency. Contrary to this, Dick (2009) argues
that market fragmentation results in self-amplifying market movements as circuit
breakers at exchanges cease to function. Literature does not draw a concise con-clusion
and thus, it is not possible to estimate the precise impact of dark pools
on risk management of trading venues. But all criticism aside, as Clemons and
Weber (1998) remark, large orders create signi
187. cant market impact, which ex-plains
why these large block trades are preferably placed on special platforms.
2.8 The generic securities trading value chain
Like many other industries, the
188. nancial industry is assembled from plenty of le-gal
entities and business units, where activities of one convey value to another.
These value generating activities between entities from the
189. nancial sector form
the securities trading value chain. Whereas a large research body deals with
191. nancial mechanisms, a much smaller body covers the
securities trading value chain and its value generation mechanisms. Schaper
(2012) de
192. nes a relatively short and coarse securities trading value chain which
is valid but does not show the dierent value creating activities in enough de-tail
for this study to create comparable statistics between individual trading re-lated
193. rms and MTFs. Schaper's trading chain link seems to be too general,
although the observable similarities across trading corporations allow a
196. cation without loosing the validity of a generic value chain.
Based on the former elaborations, it is possible to derive an industry speci
197. c
value grid that will serve as a common ground for this study. The actual depth
of the grid remains unknown for the time being and is partly subject of the later
research. The constructed generic value chain is shown in
199. Participant role Description Examples
Client Initiates a transaction
through a broker. The
client is not always
needed, for example
prop-shops and market
makers do not have
clients in the classical
sense.
Hedge- and other funds,
private clients and
individuals
Broker/Trader The broker or trader
sends orders to the ex-change
and maintains
relations to custodian,
clearer and exchange.
Brokers like NewEdge,
banks like UBS or
Credit Suisse
Venue/MTF The place where orders
are matched. The MTF
instructs the clearer and
reports order
200. lls back
to the order origin.
Exchanges and MTFs
like Deutsche Boerse,
UBS MTF, TradeGate,
LSE
Custodian Holds and manages the
cash accounts.
Large banks and dedi-cated
custodians, such
as Penson
Clearer Guarantees that trades
can be done, instructs
the custodian to settle
transactions.
Institutions, such as Eu-roCCP,
Eurex Clearing
Regulator Regulators analyse and
monitor the market.
They craft regulations
to ensure working mar-ket
mechanisms.
Governmental bodies,
such as the FSMA
Market supervision Market supervision
monitors the adherence
of market participants
to regulations.
BAFIN, FINMA
ISVs Provide various services,
like software, to all
market participants.
Actant, T-Bricks, Orc,
Bloomberg, FactSet,
Reuters, etc.
Table 2.1: Key players in the
202. Chapter 3
Methodology
This chapter describes the research process undertaken for this study. First, the
methodology and goals are presented. Next, the implementation is presented and
approaches and tools used to
203. nd interviewees are described. This is followed by
how the responses were collected. This chapter closes by outlining how collected
data is analyzed.
It is a common known fact that many socio-economic studies deploy a dual leg
research approach to investigate and understand a phenomenon from two dier-ent
angles. These studies compensate with one research method for the weak-nesses
of the other. This approach, labeled triangulation, leads to more reliable
and more profound study results, delivering an increased level of con
204. dence and
trust in research results. Like in many academic areas, research methodology is
a domain that evolves constantly. Only few academic articles stand out of the
crowd of literature as much as Jick (1979, p. 606) does, with his article on ap-plied
triangulation, describing the practical application of a theory strongly en-
46
205. dorsed by Smith (1975, cited in Jick, 1979).
This work at hand followed Smith and deployed a triangulation through quali-tative
and quantitative analysis of practices, knowledge and assumptions of rep-resentatives
of key participants in the global
207. gure
2.1 on page 25. These two study approaches yielded results of varying breadth as
presented next.
3.1 Data collection
3.1.1 Qualitative study
Qualitative studies contribute to the understanding of a topic in areas where
quantitative works cannot reach (Burton, 2007, p. 6). In the qualitative part,
this study interviewed a list of persons having a good understanding of the dif-ferent
areas to cover.
Qualitative studies are not focused on numbers and quanti
208. able measures, but
more on crisp observations and experiences. Albarran, Chan-Olmsted and Wirth
(2005, p. 580) suggest that qualitative study design strives to achieve three ide-als:
(a) simplicity, (b) speci
209. city and (c) generalization. The qualitative section
of this work came close to these three ideals, by asking simple questions that
aimed straight at the heart of the topic. Similar interview questions were posed
to all participants of this study; the resulting data enabled deducting a general
model of how security trading value chains are altered by MTFs.
47
210. 3.1.2 Quantitative study
Contrary to soft, qualitative studies, quantitative studies focus on numbers and
hard, quanti
211. able measures. Whereas qualitative studies analyze small popula-tion
samples, quantitative studies aim to sample as broad as possible, using sta-tistical
tools and methods to identify and deduct generic patterns. King (1985,
p. 56) explains that quantitative studies should pursue and support a model
building process, whose outcome is a model that can be applied to other data
with symmetric empirical identi
212. cation (ibidem, p. 57). In order to deduct
a generic model, this study produced a survey and analyzed the responses of a
sample group and put the resulting numbers into context. As quantitative sur-veys
yield numbers, number-crunching analytical techniques
213. nd deeper patterns
which are up to interpretation by the researcher.
Although it is a noble goal to deduct a general theory about the impact of in-dustrial
innovations, such as MTFs, on the security trading value chain, it is also
very dicult to achieve. In a phenomenological study like this, any theory or
model can be only descriptive at best, but never all explaining.
3.1.3 Qualitative and quantitative survey process
Qualitative and quantitative part were undertaken in consecutive order. First,
the qualitative part investigated the topic in general from a wide angle, w here-after
the quantitative part provided statistical ground to key observations of the
qualitative part. The bene
214. ts of conducting the quantitative survey after the
qualitative survey include, but are not limited to, topics discussed by Chung
48
215. (2000, pp. 43-44), such as the ability to identify hypotheses through qualitative
work and to put them to test in quantitative surveys.
3.2 Goals and study design
3.2.1 Goals and design of the qualitative study
The qualitative part of this study aims at analyzing areas that are unique to and
which dier between corporations. The qualitative part oers
exibility where
a quantitative study requires strictness due to method inherent aspects, such as
the necessity to code the answers into hard facts.
Some topics evolving around security trading are better suited for analysis through
a qualitative study. The areas to be documented by a qualitative study are re-lated
to:
• the security trading chain, for example
{ alteration of the chain by trading at an MTF
{ changes in pro
217. c aspects
{ the requirements that had to be met by a company
{ the diculties encountered while migrating trading
ow to an MTF
The goals of the qualitative part are to provide answers to:
49
219. c value chains look like before and after migrating
trading to MTFs?
• Which procedures are critical to actually trade on MTFs?
• How do
220. rms change because of trading on MTFs?
Qualitative interviewing
Qualitative interviewing is a dicult task which includes a direct interaction be-tween
interviewer and interviewee. Depending on the communication media used
to conduct the interview, inter-human communication diculties, dierent be-liefs
or assumptions can add varying amounts of noise to their communication.
Nielsen (2007) distinguishes between three main scienti
221. c interview methods: a)
a phenomenological/hermeneutic position, b) a social constructionist position
and c) an action-oriented position. Although Nielsen's examples try to clearly
dierentiate themselves from each other, it is possible to imagine mixed proce-dures,
where the three methods are used to varying degree. Following Nielsen's
classi
222. cation, it becomes obvious that this work will be a mix between a phe-nomenological
and a constructionist approach.
This study pursued a phenomenological interview approach as far as possible,
leaving the interpretation and explanation of experiences, situations and obser-vations
to the interviewee whenever possible. In cases, where interviewees get
stuck, the interviewer induced conversation smoothing constructionist interview
approaches to generate an atmosphere of trust. To minimize the disturbance of
interviewees and to minimize a feeling of anxiousness but to maximize a feel-
50
223. ing of comfort and superiority, aligning reality between interviewee and other
sources, such as the interviewer, was minimized.
Success factors
Several factors are critical for the success of a qualitative study. Above all, the
interviewer has to connect to the interviewed human being to create a comfort-able
interview atmosphere; but to what extent is a face-to-face approach neces-sary?
Sedgwick and Spiers (2009) provided evidence about the success of lever-aging
technical communication channels to reduce interview costs and increase
interview comfort for remote interviewees at the same time. Modern communi-cation
channels include video conferencing over Skype, email or phone. Their
three step interview approach,
224. rst state the interview goal, then proceed with
questions and associators and
226. c word use, is comparable to recommendations by May (2002). May (2002,
p. 230) suggests to check the generalities and existing abstractions of intervie-wees
through case studies, before conducting the actual interview, thus getting
to know the coloring of a interviewee's mind. Particularly interviews about
topics that contain moral or ethical components should go through this sort of
interviewee background checking (ibidem). Although May and Sedgwick and
Spiers place the assumption checking at dierent locations in the interview pro-cess,
their inclusion of such a step justi
229. 3.2.2 Goals and design of the quantitative study
This study's quantitative part samples a large amount of members of the
230. nan-cial
caste through direct and targeted surveys. Coded and unambiguous ques-tions
and answers enabled statistical post-processing of collected responses and
strengthened the results obtained from the qualitative study leg. Directly de-ducted
from the study goals set forth in section 1.5.1 on page 20, the goals of the
quantitative section were to quantify:
• the value chain structure,
• the value chain complexity, before and after migrating trading
ow to MTFs,
• the costs to maintain and develop the value chain.
In order to reach quanti
232. able
are to be understood. Of help is a comparison with corporate environments,
where KPIs have become the popular approach to determine most relevant, quan-ti
233. able parameters to measure process eectiveness. But, as KPIs focus on com-pany
speci
234. c performance measures and rarely on value chain generics, these
KPI values are (a) probably sensitive, (b) measurable and (c) very hard facts.
In addition, according to the literature review in section 2.5 on page 38, the
newness of MTF trading has not produced accepted and profound MTF speci
235. c
KPIs, yet.
In an ideal world, all companies track their processes at a perfect level of de-tail
and all interviewed persons are aware of all and are willing to share all KPI
measures. But as reality is not ideal and perfect, this study had to rely on the
52
237. g-ures,
hoping for the best. In order to remedy this situation and also to avoid
a breach of nondisclosure and secrecy agreements between sta and
238. rms, the
quantitative survey did not ask for precise numbers, but rather for relative val-ues.
This intentional coarseness should have resulted in an increased intervie-wee
motivation to participate, as it requests no speci
239. c and secret information.
Coarse questions also speed up the survey processing, so that possible barriers to
participation are further removed.
3.3 Implementation
The following sections detail the actual implementation of the qualitative and
quantitative study parts. First, it describes the qualitative study, thereafter the
quantitative study.
3.3.1 Qualitative study
The qualitative part of the study was implemented as interviews, split over one
or two session of 60 minutes each. The interviews took place over whatever real-time
media possible. Preferably a face to face conversation takes place. All inter-viewees
were asked for permission to record the interview. The obtained record-ings
are stored by the interviewer in case recording had been possible and per-mitted.
The criteria used to determine a list of possible interviewees and
240. rms
are listed in table 3.1 on page 58. Appendix B.1 on 116 lists questions that were
53
241. posed to interviewees. This list was subsequently approached.
The target list of interviewees targeted an inclusion of at least (a) one MTF op-erator,
(b) one proprietary trading unit that trades at an MTF and (c) one bro-ker
that routes trac to MTF. Ideally more people and
242. rms participate during
qualitative interviews.
MTF operators
Applying the criteria list mentioned above, the following MTF operators were on
the long list.
• BATS
• CHI-X
• Credit Suisse Cross-
243. nder
• UBS MTF
Propshop units
Applying the criteria list mentioned above, the following propshops were on the
long list.
• GETCO
• IMC
• SIG
54
244. Brokers
Applying the criteria list mentioned above, the following brokers were on the
long list.
• Credit Suisse
• Deutsche Bank
• SwissQuote
• UBS
3.3.2 Quantitative study
Survey tool
The quantitative part of this study realized an anonymous, web-based online
survey. The web based tool LimeSurvey (Cleeland, J. et al, n.d.) was used to
create a survey with
245. ve main sections and 29 questions. Section B.2 on page
121 presents the entire survey. The entire survey was taken in an unsupervised
way over the Internet, where survey takers received a link to the survey.
The reasons why an online tool was chosen are manifold. First of all, web based
tools oer a convenient way of
246. lling out an online form and ease processing col-lected
data pro grammatically. High usability is further guaranteed by the lack
of a user registration or any other sort of process obstacles which could inhibit
high response rates. As most users in the target group use PCs and have Inter-net
access, a browser based approach was feasible.
55
247. Filling out a survey several times is impossible, because LimeSurvey uses cookies
as identi
249. es browser instances. As
long as interviewees use the same browser instance to surf the web and as long
as they do not clear their cookie cache, it is impossible for them to answer the
survey multiple times.
Finding survey participants
Quantitative studies bene
250. t from a large amount of survey participants. In order
to attain large reach, survey participants were sourced over the Internet. Sev-eral
existing social media tools, such as LinkedIn1 or Xing2, provide discussion
group features where interested hobbyists and professionals discuss topics of all
kind in special interest groups. Tables 3.2 on page 59 and 3.3 on page 60 list rel-evant
discussion groups. A call for participation in the online survey was sent
to all these groups. In addition to to posting survey participation invitations in
news groups, the researcher sent invitations to selected individuals in his address
book. Section B.3 on page 134 shows an example of such an invitation. About
150 personal invitations were sent out.
3.4 Analysis
Once all data was collected, the results were
251. rst visualized and then analyzed.
Most of the visualized results are included in this study. The analysis answered
1http://www.linkedin.com
2http://www.xing.com
56
252. the main research questions. The entire visualization part deserves a special
mentioning. LimeSurvey was used to conduct the survey. LimeSurvey comes
with built-in support to export all survey data into an R3-compatible format.
As this entire thesis was written in LaTeX, all charts are included when this doc-ument
is built. A wrapper script around LimeSurvey's R script did some data
preprocessing and generated all graphs into the same folder as this document. A
shell script around the LaTeX build step runs at
253. rst the R-wrapper and then
the LaTeX commands to build this dissertation in the spirit of repeatable re-search.
Once all data was visualized, a manual thought process analyzed the col-lected
information.
3R is a software and language similar to S for statistical data processing.
57
254. Parameter Description/Comments Relevance
Automation niveau of
trading
Many MTFs provide
only machine-usable
access points (thus
ooading GUI develop-ment)
High
Competency The competency of
interview partners is
crucial to receive ade-quate
information
High
Professional exposure exposure of a poten-tial
target
257. c
entity within the dark
pool domain
High
Trading volume The higher the trading
volume, particularly in
an MTF, the more rel-evant
should the
258. rm's
answers be
High
Table 3.1: Inclusion criteria for qualitative study
58
259. Group name Group size Relevance
Hedge Fund Group (HFG) 67547 High
Automated Trading Strategies 28370 High
Electronic Trading Group 26244 High
Algorithmic Trading 25256 High
High Frequency Trading 12291 High
Low Latency Infrastructure
11185 High
for Trading
Ultra-Low Latency Direct
Market Access (ULLDMA)
6605 High
Systematic Trading and Sta-tistical
Arbitrage Network
6238 High
Financial Engineering Group 4383 High
High Frequency Trading Tech-nologists
3438 High
Exchange Connectivity 3418 High
AlgoTrade 3077 High
The QUANTS Network 2315 Low
Quant Masters - Quantita-tive
Professionals - Predictive
Modeling, Statistics, ML,
Matlab R SPSS SAS
2053 High
Volatility Arbitrage Traders 1154
MTF - Multilateral Trading
Facilities Group
838 Very high
Smart Order Routing 578 High
Quant Entrepreneur 277 Low
IMC Financial Markets
239 High
Asset Management (Invitation
only)
Dark Pools 127 Very high
Swiss Quant Group 108 High
Table 3.2: Target LinkedIn groups (as of 20 January 2012)
59
260. Group name Group size Relevance
Aktien, Finanzdienstleistun-gen
und Versicherungen
7005 Low
Financial Risk Management 4492 Low
Quantitative Research in
Finance
1140 High
Daily Trading 842 Low
Algorithmic Trading 497 High
Table 3.3: Target Xing groups (as of 20 January 2012)
60
261. Chapter 4
Study results
4.1 Qualitative study
4.1.1 About the sample group
Due to severe diculties to
263. ve
interview partners were sampled in free-form verbal interviews.
Of the various MTF operators in Europe, this study contacted the two major
Swiss operators on the 20 December 2011. The
264. rst opportunity to get in touch
with these two entities posed a job fair in Z~Arich mid December 2011. In
266. nding the right party to talk to.
Both entities, independent from each other, stated that the corporate commu-nications
unit (CCU) is the right partner to talk to. As a follow-up to this very
267. rst contact, the researcher contacted the CCUs by phone in order to
269. vant interview partners within their organization. As a follow-up to this phone
call, an email with a summarizing presentation about the dissertation goals and
contact details was sent along. Of the two entities, only UBS MTF provided an
interview partner. This interview partner contacted the researcher on the 13
January 2012 and after an initial conversation, they agreed on realizing an in-terview
in the fourth week of January 2012 which subsequently took place. The
interview took place over the phone, was not recorded and took one hour.
The market maker Timber Hill was contacted on the 20 January 2011, but de-clared
they do not trade on dark pools, after the HR person on the phone had
to check at
270. rst whether they are a market maker, and thereafter the person had
check to whether they trade on dark pools. As the researcher reminded them of
their MTF trading activities and provided evidence thereof, they con
271. rmed their
trading activities. Due to time constraints, an interview did not take place.
The MTF operator BATS Trading was contacted on the 20 January 2011. BATS
Trading did not respond, but a representative was subsequently caught at Trade-
Tech 2012, where the representative agreed to have an informal conversation.
A proprietary trading
272. rm has been contacted by email on the 20 January 2011.
The proprietary trading
273. rm provided a head of trading who was interviewed
face-to-face in a one hour long session. SIG and Getco were also contacted, but
did not respond.
The brokers SwissQuote, Deutsche, CS and UBS were contacted, but did not
respond. Unfortunately, this ruled out the ability to interview someone from the
buy side with clients.
62
274. At the TradeTech 2012 conference in London on 25th and 26th April 2012 an
opportunity arose to remedy the shortcomings due possible participants rejecting
an interview. Three other people from industry relevant companies were inter-viewed.
Particularly interesting was the interview with a member of EuroCCP,
a pan-European global clearer. The character of interviewing at a conference is
very dierent to a formal interview situation, but fruitful nonetheless. An inter-esting
addition to the qualitative interviews proved to be panel conversations at
TradeTech 2012.
4.1.2 Findings
The qualitative expert conversations revealed that MTF providers aim to deliver
several features to clients, most notably a deeper market where large orders re-sult
in less market impact and a price improvement over the primary markets of
equities. They are bound by market regulations to provide at least the same or
a better price than the national best bid and oer. Usually, they undertake mid-point
pricing. At least UBS MTF looks into oering futures and other deriva-tives
over their market place. By the time of this interview, UBS MTF was the
fastest growing MTF in Europe.
The MTF landscape is rapidly changing, market shares are not written in stone.
The successful market introduction of the UBS MTF has resulted in an over-all
trading volume growth, but also in an attraction of capital from competing
MTFs.
One interviewee, belonging to a major central counterparty, raised the point that
63
276. rst quarter 2013, which then gives
aected parties half a year time to implement these new regulations. Asked
whether it is already possible to oer futures and derivatives on an MTF, he
declared that he assumes that the current MIFID regulations govern this suf-
278. rm is the Dodd-Frank act in the States
and the similarities and dierences to its European pendant 1 and the opportu-nities
a possible harmonization between these regions would yield. During the
conversation it became clear that the current regulatory actions are perceived as
political responses to former deregulation of the
279. nancial sector. It was agreed
that the counterparty business is extremely regulator dependent. Regulations
have the power to redesign the entire clearing landscape with a drop of an ink.
At this conference, in several shorter interviews, responses echoed that MTFs
are a result of the markets and that they are rather a reaction to a regulatory
action. Interestingly, one interviewee declared that regulations rather create
than stipulate innovation. Broker crossing networks and MTFs are just two op-tions
that many clients of the market access providers can choose from. More
and more often do broker crossing networks and MTFs become border crossing
networks. In a topic relevant panel conversation, the industry representatives
agreed: there seems to be a trend imminent to provide more and more client
demand speci
280. c trading venues, reducing the volumes in the OTC market and
increasing standardization of formerly non-standardized products, a motion that
is said to be further stimulated through MIFID2.
Data fragmentation and a consolidated tape are very important topics, said
1http://tinyurl.com/c3mvk95
64
281. keynote speaker David Lawton, FSA at TradeTech 2. As the market does not
provide a consolidated tape itself, the ESMA might have to mandate tape con-solidation
to one speci
282. c entity. David Lawton also made clear that the goal of
the FSA is to increase and protect public liquidity, but also liquidity in general.
The FSA tries to achieve a balance between liquidity and transparency. He con-tinued,
on a very international track, the G20 want to continue standardizing
OTC transactions, preferably migrating large fractions of OTC transactions into
Organized Trading Facilities. MTFs are a subtype of OTFs. Another aspect is
the declared goal to introduce more central clearers, particularly in the OTC
market, also noted on page 2.4.4 on 37.
4.2 Quantitative results
The quantitative part of this study bene
283. ted from undertaking a web based,
anonymous interview approach. An invitation to participate in an survey about
MTFs and the securities trading value chain was sent out to the author's address
book and also posted to various groups on LinkedIn and Xing as described in
section 3.3.2 on page 56. A copy of the sent invitation is shown at appendix B.3
on page 134. Unfortunately, the response rate was very low, particularly when
compared to the theoretical reach of the survey.
2http://www.wbresearch.com/tradetecheurope/daytwo.aspx
65
284. Figure 4.1: Question 1.1 - Do you feel well?
4.2.1 About the sample group
In total, 27 individuals participated in the survey. The survey's
291. gure 4.5 on 68). About 26% of all
participants did not continue to answer questions beyond the
292. rst survey section.
4.2.2 Knowledge and belief check of participants
Next, the domain speci
293. c knowledge of every participant was assessed. Of 20 in-terviewees,
19 knew the term Dark Pool. Of all MTFs that were presented as a
non-representative list of MTFs in question 2.2 (see
295. Figure 4.2: Question 1.2 - Gender distribution of interviewees
Figure 4.3: Question 1.3 - Hierarchical position of interviewees in their company
Figure 4.4: Question 1.4 - Does the interviewee belong to front, mid- or back-o
ce?
67
296. Figure 4.5: Question 1.5 - Seniority of interviewee
itSuisse's CrossFinder was best known, followed by UBS' MTF and Turquoise.
Asked for the positive, negative and irrelevant eects of MTFs, the results show
a fuzzy picture about the positive eects and the parameters which are not af-fected
by MTFs, but a very clear agreement on negative eects of MTFs. The
most concern of all participants was liquidity fragmentation (13 responses), fol-lowed
by market fragmentation (12 responses) and a negative impact on market
depth and breadth of individual products (6 responses, see
297. gure 4.9 on page
71). Looking at all possible answers, it becomes apparent that people tend to
uniformly see negative eects but cannot agree on the positive eects. The main
positive argument of MTF operators for MTFs (price improvement, see page 63)
was not perceived as equally important by the interviewees, where only 7 de-clared
that they think that MTFs have a positive eect on. Question 2.4 was
intentionally asked in a fuzzy way, so that participants interpret the question in
their own individual and environmental context.
The much clearer formulated question 2.5 revealed that most participants agreed
on the advantages and disadvantages of MTFs. Of all participants, 70% agreed
68
298. Figure 4.6: Question 2.1 - Does interviewee know what a Dark Pool is?
that the ability to execute large block trades in the dark is an advantage of MTFs,
followed by the ability to trade international assets from one market place, a fa-vorable
fee structure and favorable market prices (see
299. gure 4.11 on page 72).
Despite their pre-trade transparency exemption (see page 33), survey takers
perceived transparency as a disadvantage of MTFs, followed by elitism and the
tight regulations of MTFs (70%, 45% and 45%, see
303. Figure 4.11: Question 2.5a - Perceived advantages of MTFs
4.2.3 Value chain analysis
In order to align all interviewees and to make their responses comparable, the
generic value chain model from section 2.2 on page 44, which was veri
304. ed with
data collected in the qualitative session, was used to interview participants. The
used generic securities trading value chain model is presented in
305. gure 2.2 on
page 44. The large majority of participants agreed that the given model matches
very well, none of the participants rejected the model as not matching. Because
of this large agreement, the next questions, which target assessing the complex-ity
of each interviewee's value chain, standing on a stable ground.
Data about value chain depth at each step allows deducing a value grid dur-ing
result interpretation in the next chapter. Interestingly, only in the trad-
72
306. Figure 4.12: Question 2.5b - Perceived disadvantages of MTFs
ing decision making phase, most participants do not rely on external parties
to provide services (see