Competitive Strategies and Performance of Private Equity Firms

D R A F T Work In Progress
Competitive Strategy of Private Equity

Daniel A. Kukla
First version: July 2010
This version: August 2010
Working Paper
Abstract
Is Blackstone a PE firm or an investment bank? What are the most distinct competitive
strategies PE firms pursue today? I investigate two thirds of the PE universe and discover that
distinct Strategic Groups exist in the PE industry, with highly diametric strategic poles, with
sector focused investment firms on one end and multi-business investment firms on the
other end. The evidence suggests that inter-group performance differences are significant, and
that PE firms gravitate towards strategic business model centroids over time. Moreover I
discover that one of the industry-leading Strategic Groups, the multi-business investment
firm centroid, has traversed the confines of the strategic space of the traditional PE industry.
I speculate that this will shape the boundaries of the PE industry to the extent that substantial
parts of the PE universe will gradually converge towards the new gravity center.
Keywords: private equity, venture capital, competitive strategy, business model, strategic
groups, cluster analysis, sector focus, multi-business, investment firm, performance, return,
IRR, heterogeneity view.
Conceptual Framework and Hypotheses

IO Research
Industrial Organization (IO) researchers traditionally investigate industries at the macro
level in order to assess the impact and characteristics of an industry on consumer welfare. For
a long time the Structure, Conduct, Performance (SCP) paradigm, which was conceptualized
by Mason and Bain, has been the main theoretical framework for IO research. The SCP
paradigm investigates why price does not equal average cost in the long run and hence why
industries have different averages of profitability, assuming that the reason is in the existence
of barriers to entry. The concept of barriers to entry assumes that a potential entrant faces
additional cost with respect to the established firm when entering a market (Mason 1939; Bain
1956; Bain 1968).
The theory assumes that the existence and height of barriers to entry and the scale of
plants are the primary determinants of market structure, and that the industry structure is the
only element determining the performance of firms. The SCP paradigm builds on these and
on a number of other assumptions which are enlisted in the following:
1) Technology is given and is characterized by economies of scale
2) Plants differ in their scale and each firm only possesses one plant
3) Price is set by firms large plants and it is unique in the market
4) There are no differences between firms within the same market except for the scale of
operations
5) There is no uncertainty in the market: potential entrants and incumbent firms know the
demand and the cost curves
6) All the actors have the same set of information (i.e. there is no asymmetry of
information)
7) Actors rationality is not bounded
8) There is no opportunism in the market
9) There is no form of asset specificity
10) All firms operating within the industry are in direct competition among themselves
Traditional IO research assumes that firm level characteristics, besides of scale, have a
negligible effect on firm performance. In IO view, key decisions are determined by the
characteristics of the market structure and firm differences, such as organizational structure,
core competencies, or management, are unimportant. Traditional IO view was established
during an era, where mass manufacturing, large scale production and economies of scale were
-2-
cutting edge buzzwords in the field of business policy, which was later to be called
strategic management. Although not explicitly mentioned in the SCP assumptions, two other
parameters, which were typical of the post WWII period of the 1950s and 1960s, influenced
traditional IO thinking. First, markets were fairly confined regionally. Second, it was rather
straight-forward to draw lines between markets. With world societys rapidly accelerating
technological progress post WWII, growing intraregional and interregional
interconnectedness, these overly simplistic assumptions started to be challenged quickly.
From Entry Barriers to Mobility Barriers

Guided by Caves and Spence, in the early 1970s, a number of researchers, including Hunt,
Newman and Porter, were engaged in doctoral research studying the existence of structural
asymmetries within industries, their implications for market equilibrium, and modeling the
competitive behavior of firms on the basis of these asymmetries. This group originated the
concept of strategic groups and the initial theorizing about strategic groups (Faulkner and
Campbell 2003). It was written many times, that Hunt coined the term strategic groups in his
doctoral dissertation, so I will not make an exception: Hunt coined the term strategic groups
in 1972, whereas Caves and Porter framed the idea further with two seminal papers, published
in 1977 and in 1979 (Caves and Porter 1977; Porter 1979).
We can define strategic groups based on Porter as a set of firms within an industry that
are similar to one another and different from firms outside the group on one or more key
dimensions of their strategy (Porter 1979; Porter 1980). The theory of IO has by and large
viewed the industry as a homogeneous unit. In IO view, firms in an industry are assumed to
be alike in all economically important dimensions, except for their size (Porter 1979). Porters
concept of generic strategies provided one of the first typologies, and more generally, the new
paradigm laid the foundation for an economic understanding of the heterogeneity between
firms.
It has been recognized that firms established in other markets are often the least
disadvantaged entrants to an industry. There are two complementary ways to look at the
influence of a going firms assets on its behavior and strategic position as an entrant. First,
considering excess capacity, the going firm always lives in some short run and is apt to
possess some tangible assets that are lumpy and underutilized. Second, considering price
discrimination: the established firm possesses some assets or capacities that can be put to use
in various markets (Caves and Porter 1977; Porter 1979). These considerations lead to an
extension of Masons and Bains concept of barriers to new competition, and a new
theoretical model was formalized. It assumed that entry barriers are partly structural and
partly endogenous, and suggested that Bains theory becomes much richer when set forth as a
general theory of the mobility of firms among segments in the industry, thus encompassing
exit and intergroup shifts as well as entry. In a broadened framework of barriers of mobility,
diversification and the entry of established firms can be integrated to provide a basis for
Bains general condition of entry (Caves and Porter 1977).
-3-
The generalization of entry barriers into mobility barriers provides the theoretical
anchor of mobility barriers theory, which explains why some strategic groups are able to
achieve persistent performance advantages over other strategic groups (Bain 1968; Caves and
Porter 1977). Entry barriers are typically structural elements, whereas mobility barriers are
partly structural but also endogenous. This means that tacit collusion by strategic group
members to establish group protection can if the claim holds contribute to the future
superior performance of the group as a whole, as well as to that of the individual firm
(Fiegenbaum and Thomas 1995). It is possible to argue that mobility barriers could explain
persistent differences in profit rates between groups in an industry. Porter (Porter 1979)
suggested that the intensity of inter-group competition in the industry depend on three factors:
1) Number and size distribution of groups
2) Strategic distance between groups
3) Market interdependence between groups
Porter argued that these factors are linked to firm profitability, and suggested that
variances in performance may be affected across strategic groups for three reasons:
1) Differences in bargaining power
2) Differences in the degree of exposure of strategic groups to substitute products
3) Differences in the degree to which firms within the group compete with each other
While in the early 1970s, research initially focused on providing the existence of
differences in firms structural characteristics and the existence of stable group structures in
an industry, from the latter 1970s onward, a new paradigm placed the strategic discretion at
the heart of resource allocation thinking (Caves and Porter 1977; Caves 1980), which
effectively gave the emerging field of strategic management (earlier called business
policy) an economic rationale.
Strategic Groups
The research conducted by Hunt, Newman and Porter introduced an important extension
of the SCP paradigm, in suggesting that firms within the same industry are likely to differ on
traits other than size. It added new perspectives to the controversy in IO literature about
whether the firm or the industry, or some intra-industry group stratification, is the appropriate
unite for analysis (McGee and Thomas 1986; Faulkner and Campbell 2003)
McGee and Thomas proposed a refined strategic groups definition: a firm within a group
makes strategic decisions that cannot readily be imitated by firm outside the group without
substantial cost, significant elapsed time, or uncertainty about the outcome of those decisions
(McGee and Thomas 1986). Within the refined definition is the notion of mobility barriers,
-4-
expressing barriers either as absolute cost of movement from one strategic position (i.e.
group) to another, or as the operating cost penalty relative to the group incumbents that the
entrants must face. See Figure 1 for an overview of the mobility barriers framework
developed by McGee and Thomas, which corresponds broadly to differentiation and cost
strategies at the business unit level and characteristics of strategy at the corporate level
(McGee and Thomas 1986).
Market-related strategies
strategies
Market-related
Industry
Industry supply
supply characteristics
characteristics
Characteristics
Characteristics of
of firms
firms
Product line
Economies of scale
Production
Marketing
Administration
Ownership
User technologies
Market segmentation
Organization structure
Control systems
Manufacturing processes
Distribution channels
Management skills
R&D capability
Brand names
Geographic coverage
Boundaries of firms
Diversification
Vertical integration
Selling systems
Firm size
Marketing and distribution systems
Relationships with influence groups
Figure 1: Sources of Mobility Barriers

Strategic group theory and strategic mapping analyses can enrich strategic discussions
(McGee and Thomas 1986; Faulkner and Campbell 2003). Provided strategic groups really
exist, their formation can reflect strategy innovations and risky decisions. The nature of
mobility barriers forces us to think about the investments that underpin market position and
competitive advantage. Related considerations naturally point towards the nature of resources
and also to the idea of core capabilities. Moreover, it may also lead from history to predictive
ability, by giving an indication of new strategic dimensions and therefore the nature, pattern,
and intensity of future rivalry. Methodological experts suggest that strategic grouping
analyses provide three major benefits. First, a richer interpretation of current industry
structures and the interaction of firm asset structures with intra-industry competition, second,
a conceptual framework for analyzing change over time and across industries, and third, a
language for interpreting change in terms of asset structures of firms and the ensuing effects
on competition in the long run.
One example of a prominent strategic group framework is Porters two by two matrix,
mapping cost advantages accruing to larger competitors due to effects of scale or experience
on one axis, and advantages from specializing on customer groups or by developing unique
specialist skills on the other axis. Another prominent example of the essence of strategic
grouping is the BCG matrix, a model allowing for the investigation of the boundary of the
firm. The BCG matrix maps relative market share on one axis and industry growth rate on the
other axis. Although the BCG matrix prescribes the optimal intra-firm business portfolio mix
-5-
of a multi-business firm, it is based on variables which apply the industry as unit of analysis,
in other words it is conceptually framed by traditional IO view.
Many strategic group studies are concerned to link performance to group membership,
some are concerned with issues of industry and firm evolution and the way in which the
dynamics of group membership can contribute to understanding of these, whereas more recent
studies investigate dynamic strategic groups and hybrid strategic groups (McGee and Thomas
1986; Daems and Thomas 1994; DeSarbo and Grewal 2008; DeSarbo, Grewal et al. 2009).
Empirical evidence supporting a link between group membership and group profitability is
weak, and the predominant focus on performance has limited the insights to be gained from
strategic grouping research (Daems and Thomas 1994).
The drive for quantification does seem to have overshadowed the pressing, prior need to
adequately specify the model and the variables being addressed. A minority of strategic
grouping studies emphasizes detailed knowledge and understanding of the industry context in
specifying variables. Some studies rely on very broad indicators and few researchers have
chosen to build their own industry expertise from which variable identification and
specifications can proceed (McGee and Thomas 1986).
Hypotheses
Most research on PE to date has by and large investigated the PE industry and its
participants in a homogenous way. Quite often, academics and practitioners use the term PE
as a synonym for VC or buyout activity. Firms who had or still have a footprint in traditional
PE are generally being perceived as VC or buyout associations, even if traditional PE activity
makes only a small share of their current business. The pressure to publish statistically
quantifiable results in peer reviewed journals has been driving many research endeavors and
substantial intellectual capacity into a questionable direction. Although many studies on PE
are statistically significant and scientifically robust, at the same time, they offer rather
meaningless and shallow insights.
Typically researchers take a sample of firms marked as PE firms on the surface, from
recognized databases, without testing whether the operators of these databases have a deep
understanding of the PE industry. As a consequence, for example firms who may have less
than 20% in assets under management in PE, get clubbed with pure PE firms, just to have
large PE datasets on hand, for statistical tests and other academic exercise. In other occasions,
where databases offer a richer picture of contemporary PE firms, usually researchers exclude
everything from their samples which is non PE related. Although this procedure may simplify
research efforts, the downside is that it falsely reinforces the homogeneity paradigm of the PE
industry. In the words of an interview participant, this academic mental masturbation is
adding very little value, if any.
A thorough analysis of the PE industry from three perspectives (including historicist
approach, empiricist approach, and essentialist approach), and a thorough synthesis of
contemporary PE related academic literature leads to the conclusion that the prevailing
-6-
homogeneity view of the PE industry is severely myopic. This claim will be tested below, by
investigating whether the hypothesis A1 can be accepted by rejecting the null hypothesis A0.
A1: The PE industry is heterogeneous and Strategic Groups exist
A0: The PE industry is homogeneous and firms do not differ on specific traits
Based on the prevailing belief in A0, researchers have cultivated an ever growing appetite
to identify the golden formula for what makes PE firms successful, success typically being
measured by the IRR of PE investments, as the dependent variable. Nevertheless, considering
the fact base, it is highly questionable that all PE firms are supposed to be alike and that the
return on investment to LPs is the only success criterion. It seems more appropriate to assume
that different PE firms can successfully pursue various competitive strategies, depending on
their strategic group affiliation. Also the success of PE firms can be measured by other
variables, for example growth of assets under management, capital supply, reputation,
operational efficiency of the PE firm or operating margin of the PE firm. Not all these are
correlated to the IRR of the PE funds. In other words, if the heterogeneity view holds, it
seems very plausible that several successful competitive strategies exist for firms in the PE
industry. Respectively, this research also investigates whether the hypothesis B0 can be
rejected and B1 accepted.
B1: Several successful competitive strategies exist in the PE industry
B0: There is one best strategy in the PE industry
Empirical evidence from other strategic spaces suggests that in contemplating strategic
change, firms typically monitor the behavior of similar reference organizations (i.e. strategic
groups) in the same competitive environment in their search for new strategic options. Thus,
firms examine the recipes of competitive groups and focus initially on their own strategic
group as a reference point (Huff 1982; McGee and Thomas 1986; Huff, Huff et al. 1992;
Fiegenbaum and Thomas 1995). PE investment professionals are often based in major
financial centers, i.e. concentrated regionally, and many PE investment professionals belong
to relatively tight relationship networks. So in spite of the industrys secretiveness, the
proximity between PE investment professionals, across firms boundaries, possibly allows for
the diffusion of information, also regarding business models and competitive strategies.
Moreover, in recent years, PE firms also started to become increasingly transparent and to
disclose more information about strategic thrusts. Therefore, it may also be the case in the PE
industry, that successful PE firms monitor other firms in their strategic group (consciously
or unconsciously) as reference points, converging to these reference points over time. To
investigate whether this claim is justified for the PE industry, I test whether C0 can be
rejected and C1 accepted.
-7-
C1: Intra-group variances of successful Strategic Groups converge to the mean over time
C0: Intra-group variances of successful Strategic Groups stay stable or increase over time
The whole notion of industry definitions and boundaries of strategic spaces is fuzzy. Like
arbitrary set borders between countries, there is, quite often, no natural separating element
between strategic spaces. Strategic spaces under investigation by traditional IO research by
and large possessed some natural separating element. In the analogy of countries, sometimes a
high mountain or an ocean can provide a natural separating element. In the financial services
space or investment management space, it is virtually impossible to find such naturally
separating elements. Regulation may offer separating elements, still in reality very often
regulatory frontiers can be legally circumvented. Without such natural frontiers, it appears
reasonable to assume, that firms within the investment management or PE space gradually
expand into other strategic spaces. Empirical evidence possibly can show that D0 can be
rejected and D1 accepted, which would prove the claim that some firms in the strategic space
of PE have traversed out of their initially pre-confined space.
D1: One or more strategic groups traverse the boundaries of the PE industry
D0: All PE firms are within the confines of the PE industry
Provided D0 can be rejected, I see two possible interpretations. Interpretation a) a
strategic group is in the process of traversing the pre-confined space towards another strategic
space. Once it has traversed a certain threshold, the SIC staff, databases operators and
regulators will put firms of the strategic group into a new category. End of story.
Interpretation b) a strategic group is shaping the boundaries of a strategic space, i.e. it is either
expanding the pre-confined space or it is in the process of discovering a completely new
strategic space, which no one has ever systematically penetrated before. If interpretation b) is
correct, than hypothesis E0 can be rejected and E1 accepted.
E1: Traversing strategic groups shape the boundary of a strategic space
E0: Traversing strategic groups move on to other strategic spaces
A general problem of research is that studies often do not describe how crucial issues
were addressed. In any study, the rationale underlying methodological decisions should be
presented in sufficient detail to allow readers to make informed judgments about findings
(Cummings and Frost 1995). This is even more vital for studies using cluster analysis,
because of the role played by researcher judgment (Ketchen Jr and Shook 1996). In the
following chapter I will present the methodology used, as well as the underlying data which
-8-
was investigated during empirical testing. The subsequent chapter will interpret the results
and provide an outlook for future research.
-9-
Methodology and Data

Strategic Space and Strategic Variables
To empirically investigate the research questions, a new strategic groping approach was
designed. It builds on the strategic grouping framework developed by Fiegenbaum and
Thomas (Fiegenbaum and Thomas 1990; Fiegenbaum and Thomas 1995). The commonly
recognized framework was substantially expanded in three areas, in order to account for
specific circumstance of this research effort and to increase its robustness. First, the
identification of variables which capture competitive strategies of firms, was derived from
triangulating the essence from thorough industry analyses based on historicist approach,
empiricist approach and essentialist approach. Second, strategic grouping was carried out by
conducting a concurrent mixed-methods research, where qualitative case studies where
conducted in parallel to quantitative clustering methods, iterating several times up to the point
of statistically significant consistency in typology between strategic groups. Third, upon
consistency in typology has had been reached, a decision juncture tested whether at least one
of the strategic groups has substantially traversed the pre-confined strategic space. In the
following I will present this study in more detail.
Lets go to start. First step is to choose the strategic space (industry). The strategic
space under investigation is the PE industry. The main issue concerning the choice of the
strategic space relates to the identification of the boundaries of the industry. Practitioners and
researchers often rely on SIC codes, although the whole concept of strategic spaces
categorization is rather fuzzy. Besides, no explicit SIC code exists for the PE industry. So,
instead a recognized annual ranking, conducted by Private Equity International, of the top PE
firms by capital supply was used to identify the firms which belong to the strategic space
under investigation. An initial screening of these firms showed, that many PE firms are active
in both early investment stages (i.e. VC) and late investment stages. In many non-US PE
markets the difference between VC and late stage PE (e.g. buyouts) does not exist anyways.
The somehow artificial separation may be rooted in the lobbying strength of traditional VC
industry participants in the US, who (I am speculating now) have been trying to maintain
advantageous regulations for the clean early stage hands-on business development
enhancing activity of noble VCs, as opposed to the passionless late stage financially
engineered quick flips of greedy PEs. The data shows that most PE firms invest across a
variety of investment stages, so to pre-confine the strategic space by investment stage appears
not only myopic but simply false.
The unit of analysis is the corporate level of firms within the PE industry. Almost all PE
studies use the PE fund business as the unit of analysis, or the portfolio companies within the
PE fund business. My study goes up one level, to the corporate level. It is often suggested in
strategic grouping literature that one should not simply focus on business level characteristics
such as product and geographical scope or market segment. Corporate parenting can also
- 10 -
provide essentials for creating competitive advantage and it would be shortsighted to exclude
corporate effects. For example, functional strategies such as advertising intensity, reputation
development 1, and sales force characteristics can be critical investments in support of the
business strategy.
Choosing the variables along which to group observations, is the most fundamental step
in the application of cluster analysis, and thus, perhaps the most important (Ketchen Jr and
Shook 1996). Choosing the variables involves several critical issues and there are three basic
approaches (Dutton, Fahey et al. 1983; Porac and Thomas 1990; Thomas, Clark et al. 1993):
1) Inductive (focuses on exploratory classification of observations)
2) Deductive (number and nature of groups in a cluster solution are strongly tied to
theory, methodological research suggests that using deductive theory to guide variable
choice is often wise)
3) Cognitive (relies on the perceptions of expert informants such as industry executives,
has roots in research on interpretation in organizations, which posits that it is the
meaning that top managers attach to phenomena, not objective characteristics, which
directs subsequent organizational action and performance)
Strategy research is often exploratory, with a focus on theory building rather than testing.
It is recommended that variables should be chosen in a way that fosters rich description of a
samples characteristics. Both the inductive and cognitive approaches fit this requirement.
The latter may often be preferred, because its use of experts enhances the confidence that the
variables are important in a particular data set (Meyer, Tsui et al. 1993; Ketchen Jr and Shook
1996). It has also often been argued that there are a small number of dimensions that capture
the essence of strategic thrusts. This suggests that it is entirely possible to adopt a pragmatic
approach. Following these recommendations, I focused on the inductive and cognitive
approaches to choose key strategic variables in the PE industry. In addition to a thorough
industry analysis, I conducted in-depth case studies to reflect the view of industry participants
on competitive strategy. An initial cluster analysis was conducted to identify initial strategic
groups of firms within the PE industry. Subsequently in-depth interviews with senior PE
investment professionals from firms representing each strategic group were conducted. Each
interview was transcribed and the transcripts were approved by the interview participants. The
essence from the in-depth interviews, the essence from an overlaying and extensive historicist
approach, empiricist approach, essentialist approach, and from a thorough synthesis of
literature on contemporary PE, was triangulated to choose the key strategic variables. Insights
from this extensive development of deep understanding of the industry, moreover, were used
1
A recent example shows that these effects are real and that regulators sometimes even put up constraints against these
effects. After the widespread and substantial corrections of valuations of credit risks post the collapse of Lehman
Brothers in 2008, regulators in the US discussed whether PE activities should be banned from the range of services banks
should be allowed to offer. At the time of writing the new rules are still unclear. However, if banks will be allowed to
keep some PE activities, it is possible, that they might have to rebrand their PE businesses, so e.g. Goldman Sachs Capital
Partners would have to change its brand into something else.
- 11 -
to weight the strategic variables in order to reflect the importance of each variable relative to
the other variables. Early efforts in the search of strategic configurations, which had been
centered in the IO economics literature, were typically defined across narrow sets of variables,
often only one or two (e.g. Hunt and Porter in their doctoral dissertations). My study
incorporates seventeen variables, thirteen endogenous strategic variables and four external (i.e.
dependent) variables, used to test for statistical significance (see Figure 2).
Strategic dimension
Variable definition
Firm dependency
Organizational footprint in the US
Organizational centralization
Institutionalized experience
(control rights held by 3rd party strategic owners) / (total control rights)
sqrt (number of offices in US)
ln ((assets under management) / (number of offices))
full years since firm inception
Strategic space proximity

Strategic space fuzziness
(assets under management in PE) / (assets under management)

rank order of segment cluster to which the firm belongs
Capital supply
Deal flow
Investment size
Investment sector focus
Investment region focus
Investment stage focus
Buy to build focus
neg ( sqrt (1 / ((capital raised by firm over last five years) / (capital raised by all firms in samples over last five years
sqrt ((number of relationships to 3rd parties) / (average number of relationships to 3rd parties of all firms))
sqrt ((average PE deal size) / (average PE deal size of all firms))
sqrt ((investment value of top two sectors) / (total investment value))
(investment value of top two regions) / (total investment value)
(investment value of top two stages) / (total investment value)
(average investment holding period) / (average investment holding period of all firms)
External variables for significance tests (allows for significances tests)

Operational efficiency
Investment performance
Market share
Reputation
Combined external variables
ln ((assets under management in PE) / (investment professionals in PE))

sumproduct (net IRR; value of PE fund) / (sum of values of all PE funds of firm)
sqrt ((assets under management in PE) / (assets under management of all PE firms))
neg ((capital supply rank 2010) - (average capital supply rank 2010 of PE firms in sample))
CEV
Figure 2: Strategic Variables

Whether PE firms are independent or affiliated to e.g. banks, sovereign wealth funds, or
pension funds can have a considerable influence on strategic decisions, wealth transfers, and
the allocation and repatriation of generated wealth. The variable firm dependency
(FIRMDEP) measures the share of control rights held by 3rd parties. No comprehensive
database provides this information for all firms in the sample. Several sources had to be
triangulated and missing data points had to be estimated based to the best extent possible. I
hesitate to exclude this variable completely considering its strategic importance. Nevertheless,
it was weighted very low compared to the other variables, to minimize the noise from possible
errors based on not perfectly accurate data. The insight generated by the variable
organizational footprint in the US (ORGFOOT) is two-layered. The US market is the largest
and most mature PE market. Therefore ORGFOOT measures both the regional concentration
in the US, and the footprint in the worlds most mature PE market. If firms in the PE industry
undergo a natural evolution path, it is possible that those with a strong footprint in the US are
more likely to show signs of more advanced PE business models. Another truly central theme
in discussions of PE investment professionals, when it comes to key strategic decisions, is the
question of organizational centralization vs. decentralization. Some argue that centralization is
important, because the PE investment professionals are physically close to each other which
allows for exchange of market knowledge and for best practices sharing. Others argue that it
- 12 -
is virtually impossible to originate deals in all European countries out of London, and that you
need to have feet on the ground to make sure that you do not have to invest in companies
which other potential buyers considered unattractive investment opportunities. The variable
organizational centralization (ORGCENT) approximates the centralization degree of each
firm, by looking at the ratio of assets under management over number of offices. Small
financial services firms or investment management firms typically have to prove themselves
in one niche investment area. Over time the successful firms develop a good reputation and a
good investment track record, which reflects that he firm has deep experience in whatever its
focus is. In order to receive an indication of the magnitude of the firms experience, I
introduce the variable institutionalized experience (INSTEXP), which measures the full years
since firm inception. The variable strategic space proximity (SSPROX) has already been used
in other studies to measure the proximity of a firm towards the pre-confined strategic space. I
measure SSPROX by measuring the share of assets under management in PE of total assets
under management across all businesses. Although SSPROX tells us how large the part of a
firm is, which overlaps with the pre-confined strategic space, it does no tell us how fuzzy the
remaining part is. Therefore, I also introduce the variable strategic space fuzziness (SSFUZZ)
which ranks the order of the segment cluster to which the firm belongs. A cluster analysis was
carried out, specifically to identify groups of firms within the PE industry, which show
different degrees of strategic fuzziness. The alternative might have been to measure the
degrees of related and unrelated diversification. However, this would have created a strong
bias based on the beliefs of SIC code administrators or database operators. To avoid this bias,
SSFUZZ follows a more objective approach, by looking at the complexity and fuzziness of
each firm. Many industry experts argue that the supply of capital (CAPSUP) is the most
important strategic topic for a PE firm. Many would argue that compared to capital supply,
and other topic become almost irrelevant. There is no need to go that far to the extreme, still,
supply of capital does deserve a higher weighting relative to the other variables. Deal flow
(DLFLOW) is the other variable which is weighted as importantly as capital supply. It is
difficult to measure deal flow, especially the quality of deal flow. One would ideally need to
see what the Investment Committee of each firm sees, which obviously is not possible. It is
commonly recommended that the centrality in networks is an acceptable indicator for deal
flow. Therefore I measure DLFLOW by the number of firm relationships to third parties,
compared to the average number of relationship to third parties. The following five strategic
variables are commonly viewed as the most important criteria for investment decisions. First,
the variable investment sizes focus (INVSIZE) measures the average deal size of firm deals
compared to the average deal size of all firms in the strategic space. Second, the variable
investment sector focus (INVSECTOR) measures the concentration of firm investments in the
firms top two sectors. Third, the variable investment region focus (INVREGION) measures
the concentration of firm investments in the firms top two regions. Fourth, the variable
investment stage focus (INVSTAGE) measures the concentration of the firm investments in
the firms top two investment stages. And fifth, the variable buy to build focus (BYTOBILD)
compares the average investment holding period of the firm to the average investment holding
- 13 -
period of all firms in the sample, as an indicator for longer term business development
investment approach as opposed to a quick flip oriented investment approach. To test for
statistical validity of the results, four external variables have been used. First, the variable
operational efficiency (OPEREFF) measures the economics of each firm, by using the ratio of
assets under management in PE to investment professionals in PE as an indicator. Second, the
variable investment performance (INVPERF) measures the weighted Net IRR of firm PE
funds. The variable was tested for various vintages (default: 1994) thresholds and for various
payout ratios to investors (default: 60%). Third, the variable market share (MKTSHARE)
measures the share of firm assets under management in PE of assets under management in PE
of all firms. Fourth, the variable reputation (BRAND) measures the brand image of the firm,
by using the development firm of capital supply rank as an indicator. I also introduce a fifth
external variable, which is synthetic index, which comprises the weighted four other external
variables which I just presented.
Data
For each of the seventeen strategic variables, data was gathered for the largest 130 PE
firms worldwide. Samples of comparable strategic grouping studies are often much smaller,
e.g. 18 firms in Reger and Huffs study (Reger and Huff 1993), 19 firms in the study of Dess
and Davis (Dess and Davis 1984), or 16 firms in the study of Lewis and Thomas (Lewis and
Thomas 1990). The 130 firms analyzed in my study, were selected based on the PEI 300
2010, an annual ranking of the top 300 PE firms worldwide by capital supply, published by
Private Equity International. Private Equity International is a recognized publication for the
PE industry. The PEI 300 firms are ranked based on the amount of PE direct investment
capital each have raised or formed over a roughly five year period between January 2005 and
April 2010. The amount of capital supplied to the top 130 PE firms in this period totals
$1.08tr. Capital supplied to all PE firms worldwide in the same period totals $1.39tr (sourced
from VentureXpert, which provides data on PE deals from 1970 onwards). So the top 130 PE
firms by capital supply represent 78% of the total PE universe of PE firms. 37 firms had to be
excluded due to not applicable, missing or inconsistent data. With supplied capital of $0.90tr
over the last five years, the final sample of 93 firms represents 65% of the total PE universe.
For each variable for each firm bits and pieces from a set of different sources and
databases had to be hand collected and integrated. Frequent inconsistencies across various
sources had to be validated, and if gaps appeared too broad, the whole dataset had to be
excluded. FIRMDEP was sourced from desktop research and Thomson ONE Banker.
ORGFOOT was sourced from Merger Market, and validated through count of offices which
PE firms disclose on their websites. ORGCENT has two components, which were sourced
from both Merger Market and Thomson ONE Banker. INSTEXP was also sourced from
Thomson ONE Banker. The components for SSPROX were sourced from Merger Market and
Thomson ONE Banker. The data components needed for SSFUZZ were sourced from Merger
Market, Thomson ONE Banker, and Private Equity Intelligence, however, due to
- 14 -
inconsistencies, many data gap had to be manually complemented by a thorough desktop
research which included public filings, press releases, and firm websites. To provide a sense
of magnitude for the research effort conducted here, it should be highlighted, that just for
SSFUZZ, the net research time to complete the dataset for one PE firm averaged somewhere
between three and four hours. Centrality in network positions as a proxy for DEALFLOW
was sourced from Merger Market. The components for INVSIZE were sourced from Merger
Market and Thomson ONE Banker. Data for the three variables INVSECTOR, INVREGION
and INVSTAGE was sourced from Thomson ONE Banker. Average holding periods of PE
investments for BYTOBILD were sourced from Merger Market. The components for
OPEREFF were sourced from Merger Market, Thomson ONE Banker, and a thorough
desktop research of PE firms public filings and company website, especially on the number
of investment professionals. INVPERF was sourced from Private Equity International
MKTSHARE was sourced from Merger Market and Thomson ONE Banker, and BRAND
was sourced from Private Equity International.
The data was collected between July 2009 and December 2009, and updated between
April and June 2010. 35 sub-variables for almost each of the initial 130 PE firms had to be
hand collected, bringing the total number of hand collected data points to roughly 4,5002.
Two methods of data collection exist. The first is to manipulate the independent variables
by using different participants, i.e. different groups take part in each experimental condition
(called a between-group, between-subjects, or independent design). The second method is to
manipulate the independent variables using the same participants, e.g. by giving one group
access to something while restricting the group from access to something (called a withinsubject or repeated-measures design). I used the first method, i.e. independent design, to
collect the necessary data for this study.
Upon completion of the data gathering process, a thorough data exploration process was
carried out. First, the data was graphed and screened, especially to make sure that parametric
tests can be applied. In order to spot the obvious mistakes of abnormal distributions,
histograms and descriptive statistics were analyzed3. Data problems were corrected through
removals of outliers and through data transformation4. To test whether the distributions are
normal the Kolmogorov-Smirnov test was applied, which compares the scores in the sample
to a normally distributed set of sores with the same mean and standard deviation. If the test is
non-significant at p > .05 it suggests that the distribution of the sample is not significantly
different from a normal distribution (i.e. it is probably normal). If the test is significant at p
< .05 the distribution is significantly different from a normal distribution (i.e. it is probably
not normal). The applicability of the test is limited to small sample sizes, because with large
2
I wish to express my gratitude to Dinah Isabelle Winter, who kindly helped me to hand collect and assemble the data.
In particular skewness and kurtosis. Positive values of skewness indicate a pile-up of scores on the left of the distribution,
whereas negative values indicated a pile-up on the on the right. Positive values of kurtosis indicate a pointy distribution
whereas negative values indicate a flat distribution. The values of skewness and kurtosis should be zero in a normal
distribution.
4
For example, log transformation squashes the right tail of the distribution, square root transformation brings large scores
closer to the center, reciprocal transformation reduced the impact of large scores and increases the impact of small scores.
3
- 15 -
sample sizes it is very easy to get significant results from small deviations from normality. It
is commonly suggested to triangulate the test with the plotting of the data e.g. with Q-Q plots.
Variables with large ranges would be given more weight in defining a cluster solution that
those with small ranges, so that a subset of variables could easily dominate the definition of
clusters. The remedy is standardization, which transforms the distribution of elements along
variables so that each has a mean of zero and a standard deviation of one. This process allows
variables to contribute equally to the definition of clusters but may also eliminate meaningful
differences among elements (Edelbrock 1979; Ketchen Jr and Shook 1996; Hair 2006).
Also multicollinearity needs to be addressed. High correlation among clustering variables
can be problematic because it may also overweight one or more underlying constructs. There
are two types of correlations: bivariate and partial. A bivariate correlation is a correlation
between two variables whereas a partial correlation looks at the relationship between two
variables while controlling the effect of one or more additional variables5. In this study
multicollinearity was addressed through an extensive battery of correlation coefficients tests.
Clustering Firms into Strategic Groups

Rooted in IO research and often building on the SCP paradigm, strategic management
research generally focuses on the relationships among strategy, environment, organization,
and performance. The multidimensionality of these constructs creates a conceptual challenge
in that a vast array of specific combinations could be developed along these dimensions to
describe organizations (Ketchen Jr and Shook 1996).
The conceptual idea of cluster analysis is to take a sample of elements and group them
such that the statistical variance among elements grouped together is minimized while
between-group variance is maximized. Cluster analysis is a statistical technique that sorts
observations into similar sets or groups. Distance between groups can be considered as
approximating the height of mobility barriers while the distance between firms can be used as
a basis to analyze the differences between them. The use of cluster analysis presents a
complex challenge because it requires several methodological choices that determine the
quality of a cluster solution. Often the implementation of cluster analysis has been less than
ideal and the use of cluster analysis in strategic management has come under frequent attack.
One cause for concern is the extensive reliance on researcher judgment that is inherent in
cluster analysis. Also troubling is the fact that, unlike techniques such as regression and
variance analyses, cluster analysis does not offer a test statistic that provides a clear answer
5
Pearsons product-movement correlation coefficient is an example of a bivariate correlation coefficient. A one tailed test
should be selected for a directional hypothesis, whereas a two-tailed test should be used when the nature and direction of
the relationship is unknown. Considerable caution must be taken when interpreting correlation coefficients because they
give no indication of the causalitys direction. Correlation coefficients say nothing about which variable causes the other
to change. Moreover, in any bivariate correlation causality between two variables cannot be assumed because there may
be other measured or unmeasured variables affecting the results (this is known as the third variable problem or the
tertium quid). Spearmans correlation coefficient is an alternative test, and a non-parametric statistic. It can be used
when the data have violated parametric assumptions such as non-normally distributed data.
- 16 -
regarding the support or lack of support of a set of results for a hypothesis of interest. Instead,
to a large extent it is the researcher who is the arbiter of the meaning of results acquired
through cluster analysis (Barney and Hoskisson 1990; Meyer 1991). Another major issue is
the perception, that most applications of cluster analysis in strategy have lacked an underlying
theoretical rationale. Cluster analysis sorting ability is powerful enough that it will provide
clusters even if no meaningful groups are embedded in a sample. So, overall, cluster analysis
has the potential not only to offer inaccurate depictions of the groupings in a sample but also
to impose groupings where none exist (Barney and Hoskisson 1990; Reger and Huff 1993;
Ketchen Jr and Shook 1996). Some researchers in the 1990s even referred to the use of cluster
analysis as an embarrassment to strategic management and some prominent researchers
singled out cluster analysis as a methodological stigma (Ketchen Jr and Shook 1996).
Recognizing these views, measures suggested by methodological experts for improving
the application of cluster analysis were carefully studied and considered in this study.
Choosing the appropriate clustering algorithms, in other words the rules or procedures
followed to sort observations, is critical to the effective use of cluster analysis (Punj and
Stewart 1983). There are two basic types of algorithms: hierarchical algorithms and nonhierarchical algorithms.
Hierarchical algorithms process through a series of steps that build a tree-like structure by
either adding individual elements to (i.e. agglomerative) or deleting from (i.e. divisive)
clusters. Whereas agglomerative methods initially view each observation as a separate cluster
and then compile them into successively smaller number of groups, eventually putting all into
one group, divisive methods follow the opposite approach, by putting all observations into
one group initially, and the observations are divided into smaller groups until eventually each
observations becomes a separate cluster. The researcher must decide what level of division is
appropriate. Although the methods start at the opposite ends of the clustering process, the
number of groups identified should be the same regardless oh which one is used. The five
most popular agglomerative algorithms are single linkage, complete linkage, average linkage,
centroid method, and Wards method (Hair 2006). The differences among them lie in the
mathematical procedures used to calculate the distance between clusters. Each has different
systematic tendencies in the way it groups observations. Wards method was used in this
study, because it is best suited for studies where the number of observations in each cluster
are expected to be approximately equal, and where there are no outliers (Ketchen Jr and
Shook 1996). There are two types of divisive techniques: monothetic and polythetic
(Aldenderfer and Blashfield 1984; Everitt, Landau et al. 2001). Monothetic techniques are
used with binary variables, and these procedures group observations through successive rather
than simultaneous application of variables, which makes them less useful for configurational
research as it has traditionally been conducted in strategic management. Often the following
three problems of hierarchical algorithms have been highlighted. First, researchers often do
not know the underlying structure of a sample in advance, making it difficult to select the
appropriate algorithm. Second, hierarchical algorithms make only one pass through a data set,
consequently, poor cluster assignments can not be modified. Third, solutions are often
- 17 -
unstable when cases are dropped, especially when a sample is small. This is particularly
troublesome for research in strategic management, where sample sizes are often small
(Jardine and Sibson 1971). Because of these matters, confidence in the validity of a solution
obtained using only hierarchical clustering algorithms is limited.
Nonhierarchical algorithms, also known as K-means or iterative methods, partition a data
set into a pre-specified number of clusters. After initial cluster centroids (the center points
of clusters along input variables) are selected, each observation is assigned to the group with
the nearest centroid. As each new observation is allocated, the cluster centroids are
recomputed. Multiple passes are made through a data set to allow observations to change
cluster membership based on their distance from the recomputed centroids. To arrive at an
optimal solution, passes through a data set continue until no observations change in clusters
(Anderberg 1973). Nonhierarchical algorithms offer two advantages over hierarchical
methods. First, by allowing observations to switch cluster membership, nonhierarchical
methods are less impacted by outlier elements. Although outliers can initially distort clusters,
this is often corrected in subsequent passes as the observations switch cluster membership
(Aldenderfer and Blashfield 1984; Hair 2006). Second, by making the multiple passes through
the data, the final solution optimizes within-cluster homogeneity and between-cluster
heterogeneity. Obtaining this improvement, however, requires that the number of clusters be
specified a priori. In the field of strategic management this is problematic because cluster
analyses are often exploratory.
A solution advocated by methodological experts is to use a two-stage procedure where a
hierarchical algorithm is used to define the number of clusters and cluster centroids. These
results then serve as the starting points for subsequent nonhierarchical clustering (Punj and
Stewart 1983; Hair 2006). This procedure increases validity of solutions and the only cost is
the extra time and effort required on the researchers part (Punj and Stewart 1983; Ketchen Jr
and Shook 1996). Thus, experts of clustering methodologies suggest, that the best solutions
may be those obtained by using hierarchical and nonhierarchical methods in tandem.
However, many strategic grouping studies have ignored this guidance.
To increase the robustness and validity of my research design, I did follow this guidance.
First, I applied a hierarchical clustering algorithm (i.e. Wards method, as mentioned above)
to determine the numbers of clusters in the data sample. The most basic procedure is to
visually inspect a dendogram, a graph of the order that observations join clusters and the
similarity of observations joined (Ketchen Jr and Shook 1996). I screened the dendogram for
natural clusters of the data that are indicated by relatively dense branches, and by inspecting
the agglomeration coefficient, more specifically the incremental changes in the coefficient6.
This methods reliance on interpretation requires that it be used cautiously (Aldenderfer and
Blashfield 1984). A priori theory can serve as a non-statistical tool for determining the
number of clusters (Hair 2006). Comparison of emergent clusters with a theory-based
typology can provide evidence regarding the typologys descriptive validity (Ketchen,
6
Another option is to calculated and analyze the Cubic Clustering Criterion (CCC), which is a measure of within-cluster
homogeneity relative to between-cluster heterogeneity.
- 18 -
Thomas et al. 1993). Therefore I compared the findings with existing typologies and
frameworks which were reviewed during the extensive synthesis of literature on
contemporary PE. Second, upon the discovery of the number of clusters and of distinct
centroids, a nonhierarchical clustering algorithm (i.e. K-means) was applied to discover stable
strategic groups within the PE industry. Still, K-means is a heuristic algorithm, so although it
seeks for solutions among all possible ones there is no guarantee that the best will be found.
Due to the speed of the algorithm it is therefore commonly suggested to run it multiple times
with different starting conditions. These different starting conditions were derived from
multiple runs of the hierarchical clustering algorithms, in order to obtain a set of
constellations, which possibly are reasonably approximate to find the constellation which
comes closer and closer to the accurate solution. Statistical tests with external variables were
carried out for each run, to validate the degree of accurateness. The validation procedures will
be presented in the following chapter.
Validation of Strategic Groups

The goals of validation are to ensure that a cluster solution has external validity, i.e. is
representative of the general population of interest, and criterion-related validity, i.e. is useful
for the prediction of important outcomes (Cook and Campbell 1979; Kerlinger 1986).
Extreme care in validation is warranted because, despite the rigor used in previous, without
validation one is not assured of having arrived at a meaningful and useful set of clusters (Punj
and Stewart 1983).
Reliability, i.e. consistency, is a necessary but not sufficient condition of validity
(Kerlinger 1986). Therefore, the reliability of a cluster solution must be established before
validity is tested. There are two ways to evaluate reliability. First, as laid out above (and as
carried out in my research), researchers may perform a cluster analysis multiple times,
changing algorithms and methods for addressing multicollinearity. The degree of consistency
in solutions indicates reliability (Hair 2006). Second, researchers may split a sample and
analyze the two independently (Hambrick 1983). However, there is no standard for assessing
a satisfactory level of consistency, leaving this determination largely to researcher judgment
(Ketchen Jr and Shook 1996). Once reliability has been demonstrated, attention can turn to
external validity.
While cluster analysis always remains subjective, several statistical tests may be used to
evaluate the results (Daems and Thomas 1994). Validity can be assessed through significance
tests, such as analysis of variance, with external variables (Anderberg 1973; Aldenderfer and
Blashfield 1984). Such external variables should be theoretically related to clusters, but not
used in defining clusters. The variable determining the strategic group is the independent
variable. The external variable is the dependent variable because I assume that it will depend
on the type of the strategic group. I have used four external (i.e. dependent) variables, in
addition to the thirteen endogenous variables. Significance test with external variables offer a
powerful tool to establish validity of a cluster solution because the technique uses a test static
- 19 -
(often F-statistic) thereby avoiding having the researcher provide the meaning of results. It is
commonly advocated to use this technique whenever possible (Ketchen Jr and Shook 1996).
Differences in external variables, e.g. performance differences, created by unknown
factors are referred to as unsystematic variation. Differences in dependent variables created
by a specific experimental manipulation, e.g. strategic grouping, are referred to as systematic
variation. In between-group designs, or independent designs, there are differences between
the objects under investigation which contribute to the unsystematic variation. Therefore, the
error variation will typically be larger than if the same participants (i.e. repeated-measures
design) had been used. Effects of experimental manipulation are always against a background
of noise caused by random, uncontrollable differences between conditions. In a repeatedmeasures design this noise is kept to a minimum and so the effect of the experiment is more
likely to show up. If experimental manipulation, e.g. strategic grouping, is successful, then the
confidence intervals of the experimental groups should not overlap. In terms of the error bar
graph, if the bars on the error bar graph do not overlap this is indicative of a significant
difference between groups.
The t-test is a parametric test, and one of the simplest forms of statistically testing for
significant differences between means. In our study the independent means t-test suits better,
because it is used when there are two experimental conditions (or experimental manipulations,
such as strategic grouping), and different objects (which is the case in between-group design)
were assigned to each condition (or manipulation) 7 . The t-test has two limitations, which
make it less appropriate for this study. First, it assumes homogeneity of variance (which by
nature should not exist between strategic groups). Second, the t-test is limited to situations
where there are only two levels of independent variables, and it is most unlikely that there are
only two strategic groups in the PE industry. The alternative to the t-test is ANOVA.
ANOVA is a parametric test. ANOVA shows whether three or more means are
statistically the same. However, ANOVA is an omnibus test, which means that it tests for an
overall experimental effect. So although ANOVA tells us whether the experimental
manipulation was generally successful, it does not provide specific information about which
groups were affected. The F-statistic simply shows that the means of all groups are not equal.
It shows only whether the model fitted to the data accounts for more variation than extraneous
factors, but it doesnt show where the difference between groups lie. It is therefore
recommended, after conducting an ANOVA to carry out further analysis to find out which
groups differ. It is suggested to carry out so called post hoc tests.
Post hoc tests carry out pair-wise comparisons that are designed to compare all different
combinations of observations. There are several ways in which family-wise error rate8 can be
controlled. The easiest way is to divide by the number of comparisons, thus ensuring that
In contrast, a dependent means t-test is used when there are two experimental conditions and the same participants took part
in both conditions of the experiment, which, however, is not the case in my study.
8
In multiple comparison procedures, family-wise Type I error is the probability, that even if all observations come from the
same population, one will wrongly conclude that at least one pair of populations differ.
- 20 -
the cumulative Type I error is below .059. The trade-off for controlling the family-wise error
rate is loss of statistical power, which means that the probability of rejecting an effect that
does actually exist is increased (i.e. possibility of Type II error increases). Post hoc tests have
a limitation in that they perform poorly when group sizes are unequal and when population
variances are different.
A complementary and promising validation tool is experts opinions of clusters validity.
Although, like cluster analysis itself, the use of expert opinions relies heavily on subjective
perceptions. Still, if the experts are practitioners, between-methods triangulation is possible
because the perspectives and assumptions of researchers and managers are generally quite
different. There is evidence that while managers perceive groups of firms, these groups are
often based on different variables than those used by researchers (Reger and Huff 1993).
Therefore, as laid out above, overall a concurrent mixed-methods research design was chosen,
to apply between-methods triangulation between quantitative clustering methods and
qualitative in-depth case studies and expert interviews.
Besides, the discipline of strategic management claims to be an applied field (Rumelt,
Schendel et al. 1994). Expert opinions can also establish the practical value of a study. When
the experts are relevant practitioners like executives in a focal industry, their views can
establish the real world value of a set of results. So experts opinions can help maximize
validity and establish practical value, thus they should be sought whenever possible (Ketchen
Jr and Shook 1996). Most of the experts interviewed for this study were senior executives, or
quasi executives as members of executive or operating committees10.
Methodological experts highlight that the ability of research using cluster analysis to
generate knowledge has been hindered by the techniques implementation. It is commonly
suggested that for cluster analysis to be of maximum value, researchers must take steps to
overcome its weaknesses. As laid out above, I followed all steps suggested by methodological
experts to increase the robustness of the research design and to improve the validity of the
empirical results. The empirical results will be presented in the following chapter.
For example, if one conducts 10 tests, one uses .005 as criterion for significance. This method is known as the Beonferroni
correction.
10
I wish to express my gratitude to Prof. Laura B. Resnikoff, who helped me to obtain access to some of the most senior
executives in the PE industry during my stay as a Visiting Scholar at Columbia Business School in New York. Also I am
grateful to my doctoral supervisor Prof. Dodo zu Knyphausen-Aufse, who granted me access to a senior executive, who
is responsible for a whole region of one of the largest PE firms worldwide.
- 21 -
Empirical Results
As a first step, data for all variables was standardized and transformed into normal
distributions to increase the validity of parametric tests. To increase robustness, tests were
carried out with both non-normalized and normalized data. The results were compared and
they showed only small differences at the margins.
Correlation Analysis
The analysis of correlations between the strategic (independent) variables is shown in
Figure 3. The Pearson r shows the size of the effect, and p shows the significance level (two
tailed). 78 correlation relationships were investigated between the 13 independent variables.
44 of the 78 cases show no causality, and 8 cases show weak causalities with a statistical
probability of less than 5% (p < .05) that theses causalities occurred by chance. So 2/3 of all
cases show no or statistically weak signs of causality between each of the 13 independent
variables, which in turn shows that a large majority of the relationships between each of the
independent variables are not correlated. 14 cases (i.e. 18% of all cases) show moderate
causality (.2 < r < .5) significant at p < .01, and 7 cases (i.e. 9% of all cases) show moderate
causality, significant at p < .001. Finally, 6 cases show strong causality (r > . 5), significant at
p < .001.
Correlations
FIRMDEP
FIRMDEP
Pearson
C
l ti
Sig. (2-tailed)
ORGFOOT
Pearson
C
l ti
Sig. (2-tailed)
Pearson
C
l ti
Sig. (2-tailed)
N
INSTEXP
Pearson
C
l ti
Sig. (2-tailed)
N
SSPROX
Pearson
C
l ti
Sig. (2-tailed)
N
SSFUZZ
Pearson
C
l ti
Sig. (2-tailed)
N
CAPSUP
Pearson
C
l ti
Sig. (2-tailed)
DLFLOW
Pearson
C
l ti
Sig. (2-tailed)
N
INVSIZE
Pearson
C
l ti
Sig. (2-tailed)
N
INVSECTOR
Pearson
C
l ti
Sig. (2-tailed)
N
INVREGION
Pearson
C
l ti
Sig. (2-tailed)
N
INVSTAGE
Pearson
C
l ti
Sig. (2-tailed)
N
BYTOBILD
Pearson
C
l ti
Sig. (2-tailed)
N
ORGCENT
INSTEXP
SSPROX
SSFUZZ
CAPSUP
DLFLOW
INVSIZE
INVSECTOR
INVREGION
INVSTAGE
BYTOBILD
1
93
N
ORGCENT
ORGFOOT
.059
.572
93
93
.112
.166
.285
.111
93
93
93
.118
.088
.137
.259
.404
.190
93
93
93
93
-.313
.002
**
-.291
.005
**
-.326
.001
**
.066
93
93
93
93
93
.150
**
.471
.000
.257
.013
.062
-.102
.151
.552
.329
93
93
93
93
93
93
**
.309
.003
**
.433
.000
**
.318
.002
**
-.298
.004
**
.327
.001
.377
.000
.528
1
**
93
93
93
93
93
93
93
.400
.000
**
.123
.185
**
-.238
.021
**
.300
.004
**
.614
.000
.075
.335
.001
.242
93
93
93
93
93
93
93
93
.280
.007
**
.102
**
.503
.000
.105
-.134
**
.552
.000
.199
.618
.000
**
.315
.310
.003
**
.332
93
93
93
93
93
93
93
93
93
-.131
-.205
.048
-.028
**
.341
-.248
.017
-.197
.791
-.509
.000
.100
.212
.058
-.347
.001
.959
93
93
93
93
93
93
93
93
93
93
-.253
.015
.121
.254
.014
-.143
.100
.096
**
.359
.601
.371
.000
.338
-.391
.000
-.055
.171
-.263
.011
**
.246
93
93
93
93
93
93
93
93
93
93
93
-.087
-.309
.003
**
.080
-.263
.011
.139
-.105
.116
.040
**
.185
.318
.268
.704
.389
.000
.505
.000
.407
.448
**
.005
**
.078
.457
93
93
93
93
93
93
93
93
93
93
93
-.031
.061
.005
.052
-.160
.067
.139
.198
.053
-.040
.771
.564
.960
.619
.126
.521
.185
.057
.617
.705
-.279
.007
.436
93
93
93
93
93
93
93
93
93
93
93
93
**
**. Correlation is significant at the 0.01 level (2-tailed).

*. Correlation is significant at the 0.05 level (2-tailed).
Figure 3: Analysis of Correlations between Independent Strategic Variables
- 22 -
93
.082
1
93
The effect size and significance values suggest, that the deal flow is positively related to
capital supply (r = .614, p < .001). Considerable caution must be taken here, because these
results do not indicate the direction of the causality, i.e. it is unclear whether capital supply
originates from deal flow, or whether deal flow originates from capital supply, or whether
other variables have an effect on this causality relationship. All that can be said with
confidence is, that a strong positive relationship exists between deal flow and capital supply.
This strong effect is in line with literature and at least the causality directed from capital
supply to deal flow has been coined by Gompers and Lerner as the money chasing deals
phenomenon (Gompers and Lerner 2000). A better deal flow, which can be measured by
centrality in network positions (Brettel, Breuer et al. 2007), also implies, that better quantity
and quality of deals will result in better IRRs, in turn attracting more capital. So the pendulum
regarding the direction of the causality possibly moves depending on the position in the cycle.
Either way, this strong and significant effect is very much in line with some of the most
fundamental underlying dynamics of the PE business model.
Moving on to the strong and highly significant causalities, the analysis shows a strong
rectified relationship between INVSIZE and the degree of organizational centralization (r
= .503, p < .001), between INVSIZE and capital supply (r = .618, p < .001), and also between
INVSIZE and deal flow (r = .552, p < .001). All three cases are showing a strong positive
causality, and suggest that PE firms that have substantial capital supply invest in larger deals
(or, provided if the causality runs in the opposite direction, that PE firms with access to larger
deals receive more capital). Moreover, the positive causality towards organizational
centralization suggests a natural link towards organizational economies of scale.
A strong and reverse causality can be observed between the firms investment sector
focus and a firms institutionalized experience (r = -.509, p < .001). By nature, the direction of
this causality must run from institutionalized experience towards sector focus, because a PE
firm will hardly get younger, or older, due to changes in investment sector concentration. In
other words, this causality relationship suggests, that, the older PE firms get the less focused
they are on specific sectors. One can not conclude from this, however, that more established
PE firms sector expertise is shallower compared to sector expertise of younger firms. If a
firm managed grew larger towards a good business model, it may simply be at scale to have
deep sector expertise in more than just one or two sectors, which mathematically brings down
the variable INVSECTOR, which measures the sector concentration.
Finally, investment stage concentration also has a positive relationship with investment
sector concentration, i.e. firms in the PE industry which are more (or less) concentrated sector
wise are also more (or less) concentrated investment stage wise. The effect is strong and
significant (r = .505, p < .001). However, the causality between investment stage focus and
institutionalized experience is much weaker (r = .263, p < .05), compared to the causality
between INVSECTOR and INSTEXP. This suggests that, with growing age, firms in the PE
industry decrease their focus both in terms of investment sector and in terms of investment
stage, yet the investment sector diversification appears to be more pronounced.
- 23 -
*
* *
A moderate to strong positive relationship exists between strategic space fuzziness and
organizational footprint in the US, the effect being highly significant (r = .471, p < .001). This
is an interesting new finding. Recognizing that the US is the most mature PE market
worldwide [DK: introduce measure which proves the maturity claim], the strategic variable
ORGFOOT also shows how much exposure a PE firm has to more mature (and possibly more
advanced) business models. As laid out above, the higher SSFUZZ, the more non PE products
and services are under a firms corporate umbrella, i.e. the fuzzier it is. The strongly
pronounced and highly significant positive relationship between SSFUZZ and ORGFOOT
suggests that the central phenomenon under investigation, which has had been observed at the
initiation of this dissertation, is in fact a herald of a natural evolutionary step for firms at a
later stage of the life cycle of the PE industry.
*
* *
The causalities between capital supply and firm dependency, and between deal flow and
firm dependency, are also moderately positive (r = .377 and r = .400, respectively), both
significant at p < .001. Again, as mentioned above, the results do not indicate the direction of
the causality. Nevertheless, it appears counterintuitive that firms with a lot of capital supply
and a lot deal flow voluntary give up control rights. I speculate that the direction of the
causality goes the other way, i.e. from FIRMDEP to CAPSUP and from FIRMDEP to
DLFLOW. This would suggest, that firms in the PE industry, which are more affiliated with
other institutions (e.g. banks, pension funds, family offices), have both higher capital supply
and higher deal flow.
Also interesting is the relationship between investment region focus and deal flow. The
negative correlation coefficient (r = -.391), significant at p < .001, suggesting that firms who
are more concentrated on a specific region have lower deal flow, and that firms with a higher
deal flow are less concentrated in terms of region. Although this finding probably can be
categorized as common sense, it is still reassuring to see the corresponding evidence. Within
this context, the positive and significant correlation between investment region focus and
investment sector focus (r = .371, p < .001) shows that once PE firms expand in terms of
investment region (or investment sector) they also do so in terms of investment sector (or
investment region).
The observations of the correlation analysis, which I just laid out above, had been
considered during the weighting of the strategic variables. Once this step was completed, the
PE firms in the sample were clustered into strategic groups was, which will be presented in
the following chapter.
- 24 -
Clustering Analysis
As advocated by methodology experts, a two-stage procedure was applied, i.e.
hierarchical clustering algorithms were used in tandem with non-hierarchical clustering
algorithms. This procedure requires extra time and effort on the researchers part, though it
increases the validity of solutions (Punj and Stewart 1983; Daems and Thomas 1994; Ketchen
Jr and Shook 1996).
First, a hierarchical clustering analysis (Wards method using Squared Euclidean distance
between the groups) on the standardized variables was used to identify strategic groups based
on the degree of strategic space fuzziness, and to rank firms with regard to their affiliation to
one of these strategic groups. The ranking of this hierarchical clustering algorithm had
provided the input for the strategic variable SSFUZZ.
Second, the data of each of the thirteen standardized strategic independent variables
(including SSFUZZ) for each firm in the PE industry was used to carry out a non-hierarchical
clustering algorithm (K-means using 100 iterations) to evaluate the incidence of Strategic
Groups. As presented above, non-hierarchical clustering algorithms have some advantages
compared to hierarchical algorithms, including higher homogeneity within groups and higher
heterogeneity between groups. Also outliers create less noise. The only prerequisite is that
non-hierarchical clustering algorithms require the number of clusters as input. A triangulation
of the extensive literature review, the hierarchical clustering algorithm analysis and the
interviews with industry experts, suggested that three to ten clusters of strategic groups may
exist in the PE industry. Some methodology experts suggest ideally to perform a cluster
analysis multiple times, ceteris paribus, consistent group assignments despite different
methods would be evidence of stability whereas inconsistent assignments would suggest a
tenuous cluster solution (Ketchen Jr and Shook 1996). So repeated runs of different cluster
analyses were conducted, investigating different numbers of strategic clusters to test the
whole expected range. A five cluster solution shows the largest and most significant effects of
the controlled experiment.
- 25 -
Figure 4: Frequency of Strategic Groups

The distribution of firms in the PE industry in the five cluster solution is relatively
balanced (see Figure 4). Of the 93 firms of my sample, 19 firms are in SG1, 21 firms in SG2,
22 firms in SG3, 14 firms in SG4, and 17 firms in SG5, which corresponds to 20% of firms
being affiliated with SG1, 23% with SG2, 24% with SG3, 15% with SG4, and 18% with SG5.
Table 1 suggests that the means for each of the four dependent variables are different
between the Strategic Groups, significant at p < .001. The table shows that groups score
differently on each dependent strategic dimension.
Table 1: ANOVA between Strategic Groups

As laid out above, caution is required, because ANOVA is an omnibus test, which
increases the possibility for error, i.e. that the results show an effect which in reality might
not exist. Therefore robust tests (Welch and Brown-Forsyth) were applied (see Table 2). For
the OPEREFF, MKTSHARE and BRAND, both robust test confirm that the means between
- 26 -
the Strategic Groups are different at p < .001, and for INVPERF both robust test suggest that
the effect is significant at p < .01.
Table 2: Robust Tests (Welch and Brown-Forsythe) of Equality of Means

Taken together these tests indicate that there are significant between group differences
along the dependent strategic dimensions considered. Hypothesis A0 can be rejected at p
< .001, and A1 accepted. In contrast to the common prevailing view, distinct Strategic Groups
within the PE industry exist, which confirms the heterogeneity view of the firm.
The centroids of each SG are presented in Table 3. The centroids are synthetic and a
product of the non-hierarchical clustering algorithm. Centroids represent the mathematically
derived centers of the SGs. Firms affiliated to a specific SG differ on one or more traits from
the SGs centroid. Looking at one individual firm as an example of a particular SG typically
whirls dust. To avoid this unnecessary noise, the ample explanatory power of the centroids
should be fully exploited first.
Strategic Group
3
FIRMDEP
-.3202
-.5516
.0199
.8111
ORGFOOT
.5666
-.7644
-.4262
.6865
.3457
.2973
ORGCENT
.2715
-1.6098
.9346
2.0085
-1.1784
INSTEXP
-.3988
-.3280
.0716
.3179
.4965
SSPROX
-.0240
.3848
-.0029
-.4049
-.1114
-.6659
SSFUZZ
1.3453
-.8868
-1.3776
2.4778
CAPSUP
-.6532
-1.4982
.7113
2.7940
-.6405
DLFLOW
-.3665
-.3422
.0055
1.2388
-.1950
INVSIZE
-.7812
-.7711
.8395
3.3199
-1.9948
INVSECTOR
-.0610
.5261
.2020
-.2849
-.6085
INVREGION
.5181
.1286
-.2319
-.0661
-.3833
INVSTAGE
-.2381
1.3738
1.1341
.5815
-3.3775
BYTOBILD
.0874
.1371
.0236
-.1381
-.1839
Table 3: Centroids of Strategic Groups (based on Standardized Independent Variables)
- 27 -
The SGs institutionalized experience builds up from SG1 to SG5. SG1 and SG2 are
relatively young firms, SG3 is somewhere in the center in terms of age, and SG4 and SG5
comprise the oldest firm within the PE industry. This comes in handy, as SG1 and SG2 offer
the conceptual blueprint of two groups of entrants, whereas SG3, SG4 and SG5 offer three
more established centroids.
1
4
3
2
1
0
-1
-2
-3
-4
13
12
11
10
6
8
Figure 5: Centroids of Strategic Groups (Spider Graph)

A spider graph (see Figure 5) shows the pattern of each centroid relative to the position of
other SGs centroids. Each axis represents one of the thirteen dependent strategic variables,
and the sequence of the spider graph axes corresponds to the sequence of the dependent
variables as presented in Table 3. The graph indicates that SG1, SG2 and SG3 show a
comparably balanced pattern across all independent strategic dimensions.
SG4 is expanding on many dimensions, and showing extraordinary high scores on the
dimensions SSFUZZ, CAPSUP, DLFLOW and INVSIZE. In contrast, SG5s centroid pattern
is heading into the opposite direction, especially regarding scores on the dimensions
INVSIZE, INVSECTOR, INVREGION and mostly on the strategic dimension INVSTAGE.
The patterns of SG4 and SG5 suggest that from a certain maturity point onwards, firms in
the PE industry develop into two diametric models. One model, represented by the centroid
pattern of SG4, has industry highest organizational centralization (ORGCENT = 2.0085),
shows massive expansion into non PE products and services (SSFUZZ = 2.4778), has
industry leading capital supply (CAPSUP = 2.7940), industry leading deal flow (DLFLOW =
1.2388), and highest average investment size (INVSIZE = 3.3199), while, at the same time,
keeping investment sector concentration relatively focused (INVSECTOR = -.2849),
investment region concentration relatively focused (INVREGION = -.0661), and investment
stage concentration strongly focused (INVSTAGE = .5815). In contrast, the other seasoned
model, represented by the centroid pattern of SG5, is organizationally relatively decentralized
(ORGCENT = -1.1784), has remained relatively focused on traditional PE (SSFUZZ = .6659), has relatively smaller supply of capital (CAPSUP = -.6405), relatively smaller deal
- 28 -
flow (DLFLOW = -.1905), industry smallest average investment size (INVSIZE = -1.9948),
and, at the same time, has strongly lost focus in terms of investment sector concentration
(INVSECTOR = -.6085), investment region concentration (INVREGION = -.3833), and
mostly on investment stage concentration (INVSTAGE = -3.3775). The two models are
strongly diametric. Whereas SG4 has remains relatively focused on sector, region and stage,
while diversifying into non PE financial products and services, SG5 remains relatively
focused on traditional PE, while heavily diversifying in terms of sector, region and stage.
Figure 6: 95% Confidence Intervals of Investment Performance by Strategic Group

SG4 outperforms SG5 in terms of investment performance (see Figure 6). Concerning
market share, SG4 outperforms all other Strategic Groups, including SG5 (see Figure 7).
When looking at more mature firms in the PE industry, the evidence suggests, that in terms of
competitive strategy, the SG4 model is significantly more successful than the SG5 model.
A word of caution: the confidence intervals compare only firms within the PE industry to
each other. Even if SG5 shows a lower performance level compared other SGs, one should
not imply that the investments of SG5 are by and large bad. All the data shows is that PE
investments of firms affiliated with SG4 perform better than PE investments of firms
affiliated with SG5, on average.
- 29 -
Figure 7: 95% Confidence Intervals of Market Share by Strategic Group

The patterns of SG1 and SG2 also present two fairly diametric models (see Table 3). SG1
has a relatively strong footprint in the most mature market (ORGFOOT = .5666), the second
highest degree of strategic space fuzziness (SSFUZZ = 1.3453) after SG4, below industry
average capital supply (CAPSUP = -.6532), below average deal flow (DLFLOW = -.3665),
and below average investment size (INVSIZE = -.7812). Both investment sector
concentration (INVSECTOR = -.0610) and investment stage concentration (INVSTAGE = .2381) is below industry average, while SG1s regional concentration is highest in the
industry (INVREGION = .5181). In contrast, SG2 has the smallest footprint in the most
mature market (ORGFOOT = -.7644), the smallest degree of organizational centralization
(ORGCENT = -1.6098), the highest focus on traditional PE (SSPROX = .3848), and a
relatively low degree of strategic space fuzziness (SSFUZZ = -.8868). SG2s capital supply is
substantially lagging behind all other SGs (CAPSUP = -1.4982), deal flow is relatively low
(DLFLOW = -.3422), and also investment size is relatively low (INVSIZE = -.7711). SG2
cultivates industry leading focus in terms of sector concentration (INVSECTOR = .5261) and
also in terms of investment stage concentration (INVSTAGE = 1.3738). Again, the two
industry entrant models are also fairly diametric. SG1 has a strong regional focus with some
overlaying sector themes, while being diversified into non-PE products and services. In
contrast, SG2 is heavily focused on traditional PE and also has a strong focus in terms of
sectors, while being regionally more diversified than SG1, and with substantially less capital
supply compared to SG1.
The investment performance confidence interval of SG2 is broader than SG1s, offering
more upside, and also the investment performance mean of SG2 is slightly higher than SG1`s
(see Figure 6). However, in terms of market share, SG1 clearly outperforms SG2 (see Figure
7). SG1 also outperforms SG2 in terms of operational efficiency (see Figure 8). Taken
together, these results suggest that younger PE firms can successfully pursue both the SG1
model and the SG2 model. However, the significant outperformance of SG1 compared to SG2
- 30 -
concerning market share and operating efficiency possibly moves the pendulum of
comparative advantage in favor of SG1.
The firms with the best performing investments are affiliated with SG3 (see Figure 6).
The pattern of centroid of SG3 is quite distinct (see Table 3 and Figure 5). SG3s footprint is
more pronounced towards regions where the PE industry is not as mature as in the US
(ORGFOOT = -.4262). Age of firms affiliated with SG3 is close to worldwide industry
average (INSTEXP = .0716), and also diversification of SG3 firms into non PE businesses is
close to worldwide industry average (SSPROX = -.0029), while strategic fuzziness of SG3s
non PE businesses is lowest industry wide (SSFUZZ = -1.3776). SG3 has second largest
capital supply (CAPSUP = .7113), second largest deal flow (DLFLOW = .0055), and second
largest investment size (INVSIZE = .8395), all three after SG4, the industry leader concerning
these strategic dimensions. Investment sector concentration is slightly above industry average
(INVSECTOR = .2020) and investment region concentration is slightly below industry
average (INVREGION = -.3219).
Figure 8: 95% Confidence Intervals of Operational Efficiency by Strategic Group

SG3 clearly outperforms all other SGs in terms of investment performance (see Figure 6),
however SG3 clearly lags behind SG4 in terms of market share (see Figure 7) and also in term
of brand (see Figure 9). Concerning operational efficiency SG3 is comparable to SG4,
although SG4 has a broader confidence interval, with both upside and downside potential (see
Figure 8). Overall, it seems like SG3 comprises firms which are moderately old, heavily
concentrated on traditional PE with overlaying sector themes, more exposed towards less
mature PE markets, and who otherwise cultivate a fairly balanced business model.
Putting all this together, the inter-group performance differences, which I just laid out
above, suggest that several successful competitive strategies exist for firms in the PE industry.
Therefore hypothesis B0 can be rejected and B1 accepted.
- 31 -
Figure 9: 95% Confidence Intervals of Brand by Strategic Group

To investigate whether firms affiliated to a specific SG converge to the SGs centroid
over time, investment performance data was compared across two different time periods. One
time period included all vintages, while the other time period included only post 1996
vintages. 1997 as a threshold was chosen because in order to obtain most recent performance
data across the cycle for a full decade. Although the performance sample was drawn from
Preqin in 2010, no post 2007 vintages were distributed back to LPs by at least 60% of fund
value. Distribution levels below 60% make the performance data meaningless.
Figure 10 shows a comparison of the variances of investment performance by Strategic
Group between all vintages and post 1996 vintages. The variance of SG1, SG2, SG3 and SG4
has decreased over time, whereas the variance of SG5 has increased over time. As a second
measure, the development of the investment performance mean between the post 1996
vintages and all vintages was plotted on the right axis. The test shows, that the variance of
investment performance of firms affiliated with successful SGs converges to the mean over
time, and that the variance of investment performance of firms affiliated with the less
successful SG diverges over time. The likelihood of this effect to occur across all five SGs at
the same time is 3.125% (= .5 ^ 5).
Hypothesis C0 can be rejected, significant at p < .05, and hypothesis C1 can be accepted.
The evidence suggests that firms affiliated with a successful SG in the PE industry, converge
towards the SGs centroid over time.
- 32 -
1.2
0.5
0.9
0.6
0.3
Left axis: variance IRR all

vintages
0.0
Left axis: variance IRR post

1996 vintages
0.3
0.0
SG1
SG2
SG3
SG4
SG5
-0.3
-0.6
-0.3
-0.9
-1.2
Right axis: (mean IRR post

1996 vintages) - (mean IRR
all vintages)
-0.5
Figure 10: Comparison of IRRs between all Vintages and post 1996 Vintages by SG
Having presented evidence that Strategic Groups do exist in the PE industry, lets to turn
towards investigating whether one or more SGs traverse the confined strategic space of the
traditional PE industry. In spite of some exceptions, the phenomenon of contemporary PE
firms expanding into non PE products and services on the corporate level is by and large new.
At the time of writing, all practitioners and academic experts, with whom I discussed my
research, confirmed that the phenomenon is an important industry trend, and that some PE
firms have been and are undergoing massive transformations. The empirical results presented
here are the first which investigate this phenomenon in a controlled scientific environment.
The strategic variable which is mostly relevant here is SSFUZZ, and has already been
presented above. SSFUZZ measures how fuzziness of the boundary of each Strategic Group
and how far it may have traversed from the traditionally predefined strategic centroid of the
overall strategic space of the contemporary PE industry. Moreover, in order to put the most
possible richness of information to work, a fifth dependent variable was created: combined
external variable (CEV). CEV represents the average of the four other dependent variables, i.e.
operational efficiency, investment performance, market share and reputation. In other words,
CEV is an index which measures the performance of each Strategic Group across a range of
key performance indicators. Each SG was plotted in a matrix with SSFUZZ on the horizontal
and CEV on the vertical axis (see Figure 11). The color scheme of the bubbles in the matrix
corresponds to the color scheme of the Strategic Groups as presented above. The size of the
bubbles represents the aggregated PE assets under management of each SG. All strategic
variables, including SSFUZZ and CEV, are standardized. The dotted line in the matrix, which
cuts the SSFUZZ axis at 1.96, represents the upper bound of the 95% confidence interval, i.e.
the likelihood of a Strategic Group to traverse the confidence line by chance is less than 5%.
- 33 -
4
3
2
1
0
-4
-3
-2
-1
-1
-2
-3
-4
Figure 11: SSFUZZ and CEV Matrix

The SSFUZZ by CEV matrix shows that SG4 traverses the initially predefined strategic
space. SG4 includes some of the most distinguished and most respected firms in the industry,
such as The Blackstone Group, Bain Capital, or KKR. The effect is significant at p < .05. All
other Strategic Groups are within the predefined confined strategic space of the PE industry.
Hypothesis D0 can be rejected, and D1 accepted.
The SSFUZZ by CEV matrix shows that the least fuzzy group (SG3) and the most fuzzy
group (SG4) actually lead the industry, both in terms of performance (high CEV scores) and
in terms of PE assets under management. The two youngest Strategic Groups SG2 and SG1
have lower CEV scores and are positioned close to SG4 and SG3, respectively. SG5, the most
mature SG, remains relatively focused on traditional PE, has second lowest PE assets under
management, and underperforms all other SGs (lowest CEV scores).
With SG4, a large share of the PE universe is traversing the predefined strategic space of
traditional PE. Will this development reshape the boundaries of the PE industry? Is SG4 the
herald for a new and innovative financial services intermediary model? Or will SG4 firms
soon and SG1 firms later, simply traverse to adjacent strategic spaces such as traditional asset
management or merchant banking?
Some industry experts argue that capital supply is the most important strategic dimension
in the PE industry, making everything else close to irrelevant. The SSFUZZ by CAPSUP
matrix (see Figure 12), CAPSUP being represented by the vertical axis, shows that SG4 is
managing substantial PE assets under management, and has attracted more PE capital
commitments than any other Strategic Group in the PE industry. SG3 firms rank second in
terms of CAPSUP, on average.
If SG4 has managed to attract such a substantial share of PE capital commitments, of
which substantial shares yet have to be invested, one may carefully speculate, that this trend
will create a new center of gravity within the PE industry. This would imply that E0 can be
rejected and E1 accepted.
- 34 -
4
3
2
1
0
-4
-3
-2
-1
-1
-2
-3
-4
Figure 12: SSFUZZ and CAPSUP Matrix

Caution is required before making this conclusion. CAPSUP measures the amount of PE
capital commitments made to each firm in the sample over the last five years, i.e. between
2005 and 2010. As mentioned earlier, to date the common prevailing view of the PE industry
is highly homogeneous. Investors often seek PE investment opportunities in their quest for
specialty investments. At the time of writing, it is unclear to what extent the increasing
strategic fuzziness of SG4 will alienate investors away from SG4 and towards other or new
SG centroids. If this occurs, the center of gravity might also move back towards the more
traditional PE business model, and SG4 firms will either create their own niche or affiliate
with other existing strategic spaces. The evidence with regards to hypothesis E1 is too
inconclusive.
- 35 -
Interpretation and Discussion of Results

Summary
In contrast to the prevailing common view, the PE industry is not homogeneous, and
Strategic Groups do exist in the strategic space of PE. The evidence presented in this study
suggests that a substantial share of the PE universe is evolving towards partially diametric
gravity centers of distinct strategic patterns. Five Strategic Groups have been discovered,
significant at p < .001.
Firms affiliated with SG1 are product specialists. The product specialist centroid is
characterized by a pronounced footprint in the most mature market, younger tenure, higher
corporate diversification into non PE businesses, focus on mid cap transactions, and moderate
investment sector concentration. SG2 is the sector specialist, has the same young tenure as
the product specialist, yet otherwise the competitive position of the sector specialist is quite
diametric compared to the product specialist. The sector specialist centroid has a pronounced
footprint in less mature PE markets, is highly concentrated on traditional PE, and has industry
leading sector focus relative to the other centroids. In contrast to the sector specialist centroid,
rather than being overly focused on one or several particular sectors, the product specialists
unique selling proposition often includes one or several specialty financing products,
The centroid of SG3 (which I denominate sector focused investment firm), can be
interpreted as the more mature version of the sector specialist model. The business model
centroid of SG3 is characterized by moderate corporate diversification into non PE asset
classes and businesses which are fairly adjacent to the traditional strategic space of PE. The
sector focused investment firm has, on average, higher investment sector concentration, is
investing in larger deals, and is relatively more dispersed in terms of investment regions.
If SG3 is the more mature version of SG2, than SG4 can be seen as the more mature
version of SG1, therefore I call the centroid of SG4 the multi-business investment firm. The
multi-business investment firm is characterized by the highest degree of corporate bundling of
PE and non PE businesses in the industry. On average, firms affiliated with SG4 show
industry leading capital supply, industry leading deal flow, and industry leading investment
sizes. Both in terms of investment sector concentration and in terms of investment region
concentration the multi-business investment firm centroid remains fairly disciplined and only
slightly below industry average. The multi-business investment firms footprint is more
pronounced in the most mature market than any other centroid. The SG4 centroid is more
affiliated to other institutions and also more organizationally more centralized than all other
centroids. In terms of institutionalized experience, SG4 firms on average are the second most
mature firms in the PE industry.
Only SG5 firms are older, on average. The strategic pattern of SG5 is completely distinct
from all other SGs. I call the SG5 centroid the small cap generalist model. In addition to
being the most seasoned Strategic Group in the PE industry, small cap generalist firms are
- 36 -
characterized by a relatively high focus on traditional PE, small investment sizes, industry
lowest sector concentration, industry lowest regional concentration, and by far the industry
lowest investment stage concentration.
There is not one best strategy in the PE industry, depending on the performance measure,
both SG3 and SG4 are substantially outperforming the other SGs. The small cap generalist is
the worst performing centroid compared to the others. The relative average investment
performance of PE funds managed by SG1 firms, SG2 firms, SG3 firms and SG4 firms
improves over time, whereas the relative average investment performance of PE funds
managed by SG5 firms deteriorates over time. Intra-group homogeneity increases in
successful SGs over time. This finding suggests that better performing firms do converge
towards their strategic gravity centers more effectively than less successful firms.
By nature of the PE industry, the fuzziness of its boundaries is immense. I find evidence
that the multi-business investment firm centroid (i.e. SG4) is traversing the confines of the
strategic space of the traditional PE industry. This study is the first to shows this effect in a
scientifically controlled environment, significant at p < .05. SG4 has the largest share of
aggregated PE assets under management relative to all other SGs. The multi-business
investment firm model is setting the industrys upper benchmark in several additional
dimensions such as capital supply, deal flow, investment size, market share, reputation, and
ranks only second (after SG3) in terms of investment performance.
Almost all studies to date have focused only on the traditional PE part of the PE universe.
Recognizing the heavyweight of SG4 and SG1 within the PE universe, this appears myopic.
As simple as it sounds, still: PE in 2010 is not the same PE as in 1980. In the words of a
Senior Managing Director of a leading pure PE firm: the days of the one trick pony are
over.
Taken together, the evidence suggests that the hypotheses A1, B1, C1 and D1 can be
accepted. Strategic Groups do exist in the PE industry. There is no one best strategy as several
successful competitive strategies exist. Intra-group homogeneity converges to the mean in
successful Strategic Groups over time. And one Strategic Groups, actually the largest and
most recognized, is traversing the confines of the traditional PE industry.
Evidence concerning the fifth hypothesis E1 is inconclusive. I was not able to validate
whether the traversing group is shaping the boundaries of the PE industry, or whether it is
simply migrating to another strategic space. In can only speculate that the SG4 centroid SG4
(i.e. the multi-business investment firm) will shape the boundary of the PE industry to the
extent, that additional parts of the PE industry, predominantly more mature firms in more
mature PE markets, will gradually gravitate towards it. The second empirical part of this
dissertation will investigate this matter in greater detail.
- 37 -
Limitations
One limitation of this study is inherent in the data. First, although the final sample
represent two thirds of the PE universe by assets under management, one must not overlook
that there are some estimated 1,500 PE funds worldwide. Therefore the results are biased
towards the larger firms in the PE industry. Given that the concentration of smaller PE firms
is higher in SG1, SG2, and SG5, the relative magnitude of these three Strategic Groups is
possibly underrepresented. Second, the data used is from commercial databases, limited
partners, fund of funds, and from public filings. Often occurring inconsistencies had to be
aligned based on discretionary judgment. Particularly the performance data may be biased due
to window dressing efforts of PE firms. Third, the information which fed into SSFUZZ is
based on hand picked information from websites, public filings, press runs, equity analyst
reports, and expert interviews. There is no worldwide standardized taxonomy for products and
services in the investment management and financial services industry. Therefore I had to
decide on a case by case basis whether and how fuzzy products, services or asset classes fit
under the corporate umbrella of each PE firm.
Cluster analysis can be a valuable tool for strategy research because of the techniques
unparallel ability to classify a large number of observations along multiple variables. For
cluster analysis to be of maximum value, however, researchers must take steps to overcome
its weaknesses (Ketchen Jr and Shook 1996). The main problem is cluster analysis reliance
on researcher judgment. It makes the validity of results subject to serious doubts. The key to
surmounting this problem is vigorous pursuit of triangulation. Also all suggestions
recommended by methodology experts to increase validity have been considered. The results
can be statistically validated to the extent possible and also in terms of insights, the results are
coherent, in spite of the high degree of intertwined conclusions which could easily contradict
each other. Nevertheless, the methodology used does not allow the conclusion that this is the
only Strategic Grouping structure which exists in the PE industry. We can only conclude,
albeit with a significant degree of confidence is, that this is the best Strategic Groups structure
which my data shows.
Some methodology experts also suggest finding stable time periods prior to carrying out
cluster analyses. A commonly used approach is to think in term of punctuated equilibria, i.e.
periods of stability are punctuated by periods of change within which strategies are changed,
new positions taken up, and rivalry adjusts in response. Firms strategies and industry
structure are seen in equilibrium during each strategic time period. When the equilibrium ends,
because of exogenous shocks in the environment or alternatively triggered by autonomous
firm action, some firms change their strategies, new strategic groups are formed and others
disappear. Statistical techniques can be used to identify the relatively stable sub-periods
within which strategic groups are identifiable and respective transition points (Bogner,
Thomas et al. 1996). When studying Strategic Groups in the PE industry, or generally in the
financial services industry, over the last 20-30 years, it will be very difficult, if not impossible,
to find stable time periods. In his bestseller The Ascent of Money, Ferguson eloquently
- 38 -
explains the reason: In evolutionary terms, the financial services sector appears to have
passed through a twenty-year Cambrian explosion, with existing species flourishing and new
species increasing in number funds are the Galapagos islands of finance the rate of
innovation, evolution, competition, adaptation, births and deaths, the whole range of
evolutionary phenomena, occurs at an extraordinarily rapid clip (Ferguson 2008). Maybe
the only thing which does stay stable in the financial services industry is its pace of ongoing
rapid change. And if the financial services industry is an ever moving vessel, than the PE
industry is the speed boat. The limitation is that most academic studies about the PE industry
can only offer snapshots. This study is not an exception and its half-value period is probably
not much longer than one or two major economic cycles.
Implications
We come to the implications of this study for conceptual frames in the field of Strategic
Management and for the PE industry.
The established strategic grouping approach, conceptually framed by the seminal
contributions of Hunt, Newman and Porter, needs to be adjusted. These gentlemen have
developed a conceptual frame for us, which help to investigate companies traits within the
confines of a particular industry. These conceptual frames were adequately designed for the
industrial organization of the 1970s. For the rapidly evolving and fuzzy financial services
industry, which is at the center of capitalism and therefore at the center of economic activity,
the notion of confined industry spaces is misleading and imposes a major limitation. We
should upgrade the common industry analyses frameworks in a way, so that one can reiterate
to the very initial starting point, where the strategic space under investigation has been
defined and where it can be continuously redefined. This will ensure that the common
conceptual perception of an industry's boundaries and the real industry space will converge to
a higher degree. Often prevailing conceptual perceptions of strategic spaces, such as the PE
industry, have been defined at one specific point of time, then frozen, and then copy & pasted
for years and decades to come. In the words of a Senior Managing Director and Corporate
Officer of one of the largest PE firms: "For us it has been an effort [to become a multibusiness investment firm] ... for over two decades ... still people think of us as a PE firm, even
if it's only a fourth of our business".
The proposed adjustment of the strategic grouping approach might allow mitigate one of
the most criticized limitations of common frameworks, which simply do not allow for an
adjustment of the predefined industry space. Porter's background is rooted in the IO school of
thought, where the boundaries between industries were naturally clearer to be drawn. For
example, we can easily draw a conceptual line between the strategic space of a car
manufacturer in Detroit and the industry of a home appliance manufacturer in Birmingham.
Now, how do we draw the line between an active equity fund manager based in New York,
and a passive credit fund manager based in London, who both raise capital from the same
pension fund in California, and who both back the financing of the same underlying business
- 39 -
in the Middle East? At the time of writing, we simply do not have the appropriate conceptual
frames which would help academics and practitioners to investigate the most basic strategic
matters of financial services and investment management firms.
The direct implications from this study and from this new school of thought, which I am
just proposing, are multifaceted. Overall the evidence suggests that the life cycle of the one
trick pony LBO associations of the 1980s is coming to its decline phase. Two large Strategic
Groups of established firms in the PE firms exist, with fairly diametric models. One SG is
centered by the model of a sector focused investment firm, whereas the other is centered by
the multi-business investment firm model. Younger firms in the PE industry also gravitate
towards two fairly diametric strategic centroids, one towards the sector specialist model and
the other towards the product specialist model. The sector focused investment firm model
is a more mature blueprint of the sector specialist, and the multi-business investment firm
is a natural extension of the product specialist. The unresolved question is, whether the next
generation models which this study outlines, will shape the industrys boundaries, or whether
they will migrate towards other strategic spaces.
For PE firms, this finding leads to specific strategic imperatives. Younger PE firms have
to decide whether they want to prioritize developing industry leading sector expertise or
whether they can offer and maybe bundle innovative niche financing. This decision is linked
to the firms strategic fit with either the sector focused investment firm model or with the
multi-business investment firm model at a later development stage. One key finding during
the interviews was that PE firms who rush too fast towards the multi-business investment firm
model will likely fail in their efforts, given that this transformation has to be implemented
over a long time horizon with rigorous discipline to stay focused on core capabilities. More
established PE firms face the same situation. Especially those established PE firms which
today are positioned somewhere half-way between the industrys strategic gravity centers
might be well advised to decide one way or the other.
Finally, PE firms which decide to migrate towards the multi-business investment firm
centroid, need to be very careful going down that road. The path offers great potential, and
firms who succeed will most likely constitute the next generation of stars. Yet, many matters
are still highly uncertain. At the time of writing, it is still unclear whether the model will
remain within the confines of the PE industry by shaping its boundaries, or whether it will
migrate towards another space. The first would be advantageous in terms of competitive
position for any firm who manages to establish its SG4 footprint early. The latter would imply,
that yesterdays small and secretive niche financing PE firms would enter tomorrows arena
of powerful financial services firms.
There is a third way, if the multi-business investment firm manages to carve out its own
strategic space. For the foreseeable future, commercial and investment banks will face
regulatory restrictions concerning their proprietary investing activities. They will have to, and
have already started, to abandon the highly juicy sweet spot of bundling financing activities
under the corporate umbrella with a strong reputation. The synergies from the bundling of
financial activities are real and substantial (Yasuda 2005; Ivashina, Nair et al. 2009; Ivashina
- 40 -
and Kovner 2010). The only natural obstacle, which can stop a well integrated investment
firm are regulatory restrictions or an even better integrated investment firm. Commercial and
investment banks being forced by regulators to abandon this juicy pasture creates an
opportunity for fast moving less regulated financial services firms to penetrate this space.
Fergusons view on the evolution of financial services firms enriches this point: In the
evolutionary process, animals eat one another, but that is not the driving force behind
evolutionary mutation and the emergence of new species and sub-species. The point is that
economies of scale and economies of scope are not always the driving force in financial
history. More often, the real drivers are the process of speciation whereby entirely new
types of firms are created (Ferguson 2008). Maybe the next generation of PE firms is
currently in the pole position to penetrate the widening attractive market gap. If Ferguson is
right than the PE firms, regulators, and academics, might be well advised to unlock the forces
of speciation.
Further Research
There is a subliminal critique that researchers often suggest topics in their further
research sections which are too complex or too vague to be researched instantly. Otherwise
why wouldnt they investigate them themselves?
You may rest assured that I will not be the exception. However, I do hope that I will be
able to work on some of these areas at a later point of my career. If you feel tempted to join in
on some of these, please do let me know.
Cluster analyses can enable the testing of multilayered phenomena. Yet they are limited
in the ability to capture complexity. To take a more comprehensive academic view, studies
might include Strategic Group membership as one of several factors in a structural equation or
path analysis model designed to predict performance (Ketchen Jr and Shook 1996). Cluster
analysis has a place in strategic managements methodological toolbox, yet the technique
must be applied prudently in order to ensure validity of insights that it provides.
Cluster analysis should be combined with other methods. Practitioners would benefit
from an integrated strategic toolbox, with Strategic Grouping approach playing a central part.
This could marvelously upgrade the value creation activities of active investors, management
consultants, and corporate development staff.
Moreover, I hope that my research contribution will inspire others to investigate the
evolution of the Strategic Groups (especially the multi-business investment firm) less as an
exception, and more as a real-life experiment, from which we can derive some dos and donts
on how to design the next generation of effective financial intermediaries.
The recent economic downturn has demonstrated that the business models of many of the
most renowned commercial and investment banks are poorly designed. Low Tier 1 capital
ratios are not the cause but the effect of an inherent problem which goes much deeper. The
real underlying cause of our current difficult situation is that financial institutions, and their
employees, have become too detached from the underlying businesses. It is not necessary for
- 41 -
every banker to become an entrepreneur. However, in the words of a PE investment
professional, you can simply not fairly value and financially sponsor businesses while sitting
behind a Bloomberg terminal.
Many people have argued that the PE is a scam. The evidence shows that this may be true
in some situations but, on average, PE seems to create value. We have the opportunity to use
parts of the PE universe as an incubator to design the architecture for the future generations of
effective financial intermediaries, which will be capable to integrate financing functions with
deep, responsible and long-term thinking about the underlying businesses. This is my major
plea in terms of where future research energy concerning PE should be invested.
I end this chapter by linking my investigation of the roots of PE, which shows that
contemporary PE can be traced back to merchant banks in medieval Europe, to one of
Fergusons insightful observations: It also seems possible that wholly new forms of financial
institutions will spring up in the aftermath of the crisis this might be just the perfect
opportunity to set up an old-fashioned kind of merchant bank, and aiming to build the trust
that so many established banks have forfeited (Ferguson 2008 p.359)
- 42 -
Bibliography
Alchian, A. A. and H. Demsetz (1972). "Production, Information Costs, and Economic

Organization." The American Economic Review 62(5): 777-795.
Aldenderfer, M. S. and R. K. Blashfield (1984). Cluster analysis. Beverly Hills, Sage
Publications.
Anderberg, M. R. (1973). Cluster analysis for applications. New York,, Academic Press.
Ansoff, H. I. (1965). Corporate strategy; an analytic approach to business policy for growth
and expansion. New York,, McGraw-Hill.
Bain, J. S. (1968). Industrial organization. New York,, Wiley.
Barney, J. B. and R. E. Hoskisson (1990). "Strategic Groups: Untested Assertions and
Research Proposals." Managerial and Decision Economics 11(3): 187-198.
Berger, P. L. and T. Luckmann (1990). The social construction of reality : a treatise in the
sociology of knowledge. New York, Anchor Books.
Brettel, M., W. Breuer, et al. (2007). "The Syndication Network and Deal Flow of Venture
Capital Firms." SSRN eLibrary.
Burns, T. and G. M. Stalker (1966). The management of innovation. London, Tavistock
Publications.
Burrell, G. and G. Morgan (1979). Sociological paradigms and organisational analysis :
elements of the sociology of corporate life. London, Heinemann.
Campbell, A. and K. S. Luchs (1992). Strategic synergy. Oxford ; Boston, ButterworthHeinemann.
Caves, R. E. (1980). "Industrial Organization, Corporate Strategy and Structure." Journal of
Economic Literature 18(1): 64-92.
Caves, R. E. and M. E. Porter (1977). "From Entry Barriers to Mobility Barriers: Conjectural
Decisions and Contrived Deterrence to New Competition*." The Quarterly Journal of
Economics 91(2): 241-261.
Chandler, A. D. (1962). Strategy and structure: chapters in the history of the industrial
enterprise. Cambridge,, M.I.T. Press.
Chandler, A. D. (1977). The visible hand : the managerial revolution in American business.
Cambridge, Mass., Belknap Press.
- 43 -
Clark, J. B. (1899). The distribution of wealth; a theory of wages, interest and profits. New
York,
London,, The Macmillan company;
Macmillan & co., ltd.
Coase, R. H. (1937). "The Nature of the Firm." Economica 4(16): 386-405.
Cook, T. D. and D. T. Campbell (1979). Quasi-experimentation : design & analysis issues for
field settings. Dallas, Houghton Mifflin.
Creswell, J. W. (2009). Research design : qualitative, quantitative, and mixed methods
approaches. Thousand Oaks, Calif., Sage Publications.
Crotty, M. (1998). The foundations of social research : meaning and perspective in the
research process. London ; Thousand Oaks, Calif., Sage Publications.
Cummings, L. L. and P. J. Frost (1995). Publishing in the organizational sciences. Thousand
Oaks, Calif., Sage Publications.
Daems, H. and H. Thomas (1994). Strategic groups, strategic moves and performance. New
York, Pergamon.
Denzin, N. K. (2009). The research act : a theoretical introduction to sociological methods.
New Brunswick, NJ, AldineTransaction.
DeSarbo, W. S. and R. Grewal (2008). "Hybrid strategic groups." Strategic Management
Journal 29(3): 293-317.
DeSarbo, W. S., R. Grewal, et al. (2009). "Dynamic strategic groups: deriving spatial
evolutionary paths." Strategic Management Journal 30(13): 1420-1439.
Dess, G. G. and P. S. Davis (1984). "Porter's (1980) Generic Strategies as Determinants of
Strategic Group Membership and Organizational Performance." The Academy of
Management Journal 27(3): 467-488.
Dutton, J. E., L. Fahey, et al. (1983). "Toward Understanding Strategic Issue Diagnosis."
Strategic Management Journal 4(4): 307-323.
Edelbrock, C. (1979). "MIXTURE MODEL TESTS OF HIERARCHICAL CLUSTERING
ALGORITHMS: THE PROBLEM OF CLASSIFYING EVERYBODY." Multivariate
Behavioral Research 14(3): 367.
Everitt, B., S. Landau, et al. (2001). Cluster analysis. London
New York, Arnold ;
Oxford University Press.
Faulkner, D. and A. Campbell (2003). The Oxford handbook of strategy. Oxford ; New York,
- 44 -
Ferguson, N. (2008). The ascent of money : a financial history of the world. New York,
Penguin Press.
Fiegenbaum, A. and H. Thomas (1990). "STRATEGIC GROUPS AND PERFORMANCE:
THE U.S. INSURANCE INDUSTRY, 1970-84." Strategic Management Journal 11(3): 197215.
Fiegenbaum, A. and H. Thomas (1995). "STRATEGIC GROUPS AS REFERENCE
GROUPS: THEORY, MODELING AND EMPIRICAL EXAMINATION OF INDUSTRY
AND COMPETITIVE STRATEGY." Strategic Management Journal 16(6): 461-476.
Gompers, P. and J. Lerner (2000). "Money chasing deals? The impact of fund inflows on
private equity valuation." Journal of Financial Economics 55(2): 281-325.
Goold, M., A. Campbell, et al. (1994). Corporate-level strategy : creating value in the
multibusiness company. New York, J. Wiley.
Grossman, S. J. and O. D. Hart (1986). "The Costs and Benefits of Ownership: A Theory of
Vertical and Lateral Integration." The Journal of Political Economy 94(4): 691-719.
Guba, E. G. (1990). The Paradigm dialog. Newbury Park, CA, Sage Publications.
Hair, J. F. (2006). Multivariate data analysis. Upper Saddle River, N.J., Pearson Prentice Hall.
Hambrick, D. C. (1983). "An Empirical Typology of Mature Industrial-Product
Environments." The Academy of Management Journal 26(2): 213-230.
Hart, O. and J. Moore (1990). "Property Rights and the Nature of the Firm." The Journal of
Political Economy 98(6): 1119-1158.
Huff, A. S. (1982). "Industry Influences on Strategy Reformulation." Strategic Management
Journal 3(2): 119-131.
Huff, A. S. (2009). Designing research for publication. Thousand Oaks, SAGE Publications.
Huff, J. O., A. S. Huff, et al. (1992). "STRATEGIC RENEWAL AND THE INTERACTION
OF CUMULATIVE STRESS AND INERTIA." Strategic Management Journal 13: 55-75.
Ivashina, V. and A. Kovner (2010). "The Private Equity Advantage: Leveraged Buyout Firms
and Relationship Banking." SSRN eLibrary.
Ivashina, V., V. B. Nair, et al. (2009). "Bank Debt and Corporate Governance." Rev. Financ.
Stud. 22(1): 41-77.
Jardine, N. and R. Sibson (1971). Mathematical taxonomy. London, New York,, Wiley.
Kerlinger, F. N. (1986). Foundations of behavioral research. New York, Holt, Rinehart and
Winston.
- 45 -
Ketchen, D. J., Jr., J. B. Thomas, et al. (1993). "Organizational Configurations and
Performance: A Comparison of Theoretical Approaches." The Academy of Management
Journal 36(6): 1278-1313.
Ketchen Jr, D. J. and C. L. Shook (1996). "THE APPLICATION OF CLUSTER ANALYSIS
IN STRATEGIC MANAGEMENT RESEARCH: AN ANALYSIS AND CRITIQUE."
Khandwalla, P. N. (1977). The design of organizations. New York, Harcourt Brace
Jovanovich.
Klier, D. O. (2009). Managing diversified portfolios what multi-business firms can learn from
private equity. Contributions to management science. Heidelberg, Physica-Verlag.
Knight, F. H. (1921). Risk, uncertainty and profit. Boston, New York,, Houghton Mifflin
Company.
Lewis, P. and H. Thomas (1990). "THE LINKAGE BETWEEN STRATEGY, STRATEGIC
GROUPS, AND PERFORMANCE IN THE U.K. RETAIL GROCERY INDUSTRY."
Lincoln, Y. S. and E. G. Guba (1985). Naturalistic inquiry. Beverly Hills, Calif., Sage
Publications.
Mace, M. L. (1971). Directors: myth and reality. Boston,, Division of Research, Graduate
School of Business Administration, Harvard University.
Marshall, A. (1890). Principles of economics. London ; New York, Macmillan and Co.
McGee, J. and H. Thomas (1986). "Strategic Groups: Theory, Research and Taxonomy."
Mertens, D. M. (1998). Research methods in education and psychology : integrating diversity
with quantitative & qualitative approaches. Thousand Oaks, Calif., Sage Publications.
Meyer, A. D. (1991). "What Is Strategy's Distinctive Competence?" Journal of Management
17(4): 821.
Meyer, A. D., A. S. Tsui, et al. (1993). "Configurational Approaches to Organizational
Analysis." The Academy of Management Journal 36(6): 1175-1195.
Miller, D., P. H. Friesen, et al. (1984). Organizations : a quantum view. Englewood Cliffs,
N.J., Prentice-Hall.
Mintzberg, H. (1979). The structuring of organizations : a synthesis of the research.
Englewood Cliffs, N.J., Prentice-Hall.
- 46 -
Newman, H. H. (1978). "Strategic Groups and the Structure-Performance Relationship." The
Review of Economics and Statistics 60(3): 417-427.
Patton, M. Q. (2002). Qualitative research and evaluation methods. Thousand Oaks, Calif.,
Sage Publications.
Penrose, E. T. (1959). The theory of the growth of the firm. New York,, Wiley.
Porac, J. F. and H. Thomas (1990). "Taxonomic Mental Models in Competitor Definition."
Academy of Management Review 15(2): 224-240.
Porter, M. E. (1979). "The Structure within Industries and Companies' Performance." The
Review of Economics and Statistics 61(2): 214-227.
Porter, M. E. (1980). Competitive strategy : techniques for analyzing industries and
competitors. New York, Free Press.
Punj, G. and D. W. Stewart (1983). "Cluster Analysis in Marketing Research: Review and
Suggestions for Application." Journal of Marketing Research 20(2): 134-148.
Reger, R. K. and A. S. Huff (1993). "STRATEGIC GROUPS: A COGNITIVE
PERSPECTIVE." Strategic Management Journal 14(2): 103-123.
Richardson, G. B. (1972). "The Organisation of Industry." The Economic Journal 82(327):
883-896.
Robinson, E. A. G. (1931). The structure of competitive industry. London,, Nisbet & Co. ltd.;.
Rorty, R. (1982). Consequences of pragmatism : essays, 1972-1980. Minneapolis, University
of Minnesota Press.
Rumelt, R. P. (1986). Strategy, structure, and economic performance. Boston, Mass., Harvard
Business School Press.
Rumelt, R. P., D. Schendel, et al. (1994). Fundamental issues in strategy : a research agenda.
Boston, Mass., Harvard Business School Press.
Schwandt, T. A. (2007). The SAGE dictionary of qualitative inquiry. Thousand Oaks, Calif.,
Sage Publications.
Slife, B. D. and R. N. Williams (1995). What's behind the research? : discovering hidden
assumptions in the behavioral sciences. Thousand Oaks, Calif., Sage Publications.
Tashakkori, A. and C. Teddlie (2010). Sage handbook of mixed methods in social &
behavioral research. Thousand Oaks, SAGE Publications.
Thomas, J. B., S. M. Clark, et al. (1993). "STRATEGIC SENSEMAKING AND
ORGANIZATIONAL
PERFORMANCE:
LINKAGES
AMONG
SCANNING,
- 47 -
INTERPRETATION, ACTION, AND OUTCOMES." Academy of Management Journal
36(2): 239-270.
Usher, A. P. (1920). The industrial history of England. Boston,, Houghton Mifflin.
Williamson, O. E. (1985). The economic institutions of capitalism : firms, markets, relational
contracting. New York
London, Free Press ;
Collier Macmillan.
Woodward, J. (1965). Industrial organization: theory and practice. London ; New York,
Yasuda, A. (2005). "Do Bank Relationships Affect the Firm's Underwriter Choice in the
Corporate-Bond Underwriting Market?" The Journal of Finance 60(3): 1259-1292.
Yin, R. K. (2009). Case study research : design and methods. Los Angeles, Calif., Sage
Publications.
- 48 -

Competitive Strategies and Performance of Private Equity Firms

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Competitive Strategies and Performance of Private Equity Firms

Transféré par

Droits d'auteur :

Formats disponibles

D R A F T Work In Progress

Competitive Strategy of Private Equity

Conceptual Framework and Hypotheses

From Entry Barriers to Mobility Barriers

Marketing and distribution systems

Relationships with influence groups

Figure 1: Sources of Mobility Barriers

Methodology and Data

Strategic space proximity

(assets under management in PE) / (assets under management)

External variables for significance tests (allows for significances tests)

ln ((assets under management in PE) / (investment professionals in PE))

Figure 2: Strategic Variables

Clustering Firms into Strategic Groups

Validation of Strategic Groups

**. Correlation is significant at the 0.01 level (2-tailed).

Figure 3: Analysis of Correlations between Independent Strategic Variables

Figure 4: Frequency of Strategic Groups

Table 1: ANOVA between Strategic Groups

Table 2: Robust Tests (Welch and Brown-Forsythe) of Equality of Means

Table 3: Centroids of Strategic Groups (based on Standardized Independent Variables)

Figure 5: Centroids of Strategic Groups (Spider Graph)

Figure 6: 95% Confidence Intervals of Investment Performance by Strategic Group

Figure 7: 95% Confidence Intervals of Market Share by Strategic Group

Figure 8: 95% Confidence Intervals of Operational Efficiency by Strategic Group

Figure 9: 95% Confidence Intervals of Brand by Strategic Group

Left axis: variance IRR all

Left axis: variance IRR post

Right axis: (mean IRR post

Figure 11: SSFUZZ and CEV Matrix

Figure 12: SSFUZZ and CAPSUP Matrix

Interpretation and Discussion of Results

Alchian, A. A. and H. Demsetz (1972). "Production, Information Costs, and Economic

Vous aimerez peut-être aussi