Weitere Àhnliche Inhalte
Ăhnlich wie Ferguson, Deephouse & Ferguson 2000 Do Strategic Groups Differ In Reputation (20)
KĂŒrzlich hochgeladen (20)
Ferguson, Deephouse & Ferguson 2000 Do Strategic Groups Differ In Reputation
- 1. Strategic Management Journal
Strat. Mgmt. J., 21: 1195â1214 (2000)
DO STRATEGIC GROUPS DIFFER IN REPUTATION?
TAMELA D. FERGUSON1*, DAVID L. DEEPHOUSE2 and
WILLIAM L. FERGUSON3
1
College of Business Administration, University of Louisiana at Lafayette, Lafayette,
Louisiana, U.S.A.
2
E. J. Ourso College of Business, Louisiana State University, Baton Rouge,
Louisiana, U.S.A.
3
College of Business Administration, University of Louisiana at Lafayette, Lafayette,
Louisiana, U.S.A.
While most strategic group research has focused on performance implications, we consider the
relationship between strategic group membership and reputation. Using strategic group identity
and domain consensus concepts, we propose strategic groups have different reputations. We
ïŹnd signiïŹcant differences exist in reputation across three identiïŹed strategic groups of U.S.
property/casualty insurers, supporting our contention that reputation is a multilevel concept.
Post hoc analyses suggest strategic groups with higher reputation also have higher performance
on some critical measures, indicating reputation may be a mobility barrier beneïŹcial to members
of certain groups. Practitioner implications include challenges of within-group differentiation
in ïŹrm reputation. Copyright © 2000 John Wiley & Sons, Ltd.
Strategic groups represent collections of ïŹrms that relationship between strategic groups and repu-
are similar on key strategic dimensions (Hunt, tation. SpeciïŹcally, the discussion is theoretically
1972; Porter, 1979). The primary goal of most expanded by using not only strategic group iden-
prior strategic groups research has been to deter- tity (Peteraf and Shanley, 1997), but also domain
mine if signiïŹcant performance differences consensus (Thompson, 1967) at the strategic
existed across strategic groups (e.g., Cool and group level as a way to explain the process
Schendel, 1987; Mehra, 1996). Yet there may be linking strategic groups and reputation. We use
other implications of strategic groups. For a broader deïŹnition of reputation which includes
instance, Cool and Dierickx (1993) found that not only the favorableness component highlighted
group structure affects rivalry, which then affects by Peteraf and Shanley (1997) but also the true
performance. Peteraf and Shanley (1997) pro- characteristics component important to economists
posed that strategic groups with strong identities (Weigelt and Camerer, 1988). The relationship
would have more positive reputations. Dranove, between strategic groups and reputation is empiri-
Peteraf, and Shanley (1998) also suggested some cally investigated in the property/casualty sector
strategic groups develop reputations that serve as of the U.S. insurance industry. In addition to
mobility barriers that may affect performance, furthering knowledge of nuances in both repu-
such as strategic groups that specialize in high- tation and strategic groups, a demonstrated
quality products. empirical relation would provide more evidence
In this paper, we consider in more detail the to deïŹect criticism that strategic groups are arti-
ïŹcial, artifactual collections of ïŹrms (e.g., Barney
and Hoskisson, 1990).
Key words: reputation; strategic group identity; domain Our paper is structured as follows. First, past
consensus; mobility barriers strategic groups research is brieïŹy reviewed. This
*Correspondence to: Tamela D. Ferguson, College of Business
Adminstration, University of Louisiana at Lafayette, PO Box is followed by the development of relationships
45370, Lafayette, LA 70504-3570, U.S.A. between strategic groups and reputation using the
Copyright © 2000 John Wiley & Sons, Ltd. Received 22 March 1999
Final revision received 7 June 2000
- 2. 1196 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
concepts of strategic group identity and strategic ïŹrm performance. According to Dranove et al.
group domain consensus. Next, a description of (1998), the Cool and Dierickx study is noteworthy
how our hypothesis was tested using a sample of as it is one of the few to test the impact of
84 insurers is presented, followed by discussion group-level structures on ïŹrm-level performance.
of results and implications. Finally, limitations of To evaluate the body of research testing perfor-
our study and suggestions for future research mance implications of strategic groups, Ketchen
conclude the paper. et al. (1997) meta-analyzed 40 original tests and
found signiïŹcant performance differences across
strategic groups. Collectively, these studies pro-
STRATEGIC GROUPS AND vide better evidence that strategic groups are a
REPUTATION useful tool in the strategic management toolbox.
Recently, Peteraf and Shanley (1997) proposed
The strategic groups concept appeared in the that strategic groups may have identities, much
1970s as industrial organization economists like organizations. Organizational identity has
sought to ïŹnd ways to understand differences been deïŹned as the central, distinctive and endur-
within industries (Hunt, 1972). A strategic group ing feature of an organization (Albert and
represents a collection of ïŹrms within an industry Whetten, 1985). Peteraf and Shanley (1997: 166)
that differs systematically from ïŹrms outside the extended organizational identity to the strategic
group along certain strategic dimensions (Hunt, group level and deïŹned strategic group identity as
1972; Porter, 1979). Caves and Porter (1977) the set of mutual understandings among managers
applied the industry-level concept of entry bar- regarding the central, distinctive, and enduring
riers to the strategic group level. They argued characteristics of a cognitive intra-industry group.
strategic groups were subsets of an industry sepa- By deïŹnition, a key distinctive factor of each
rated by mobility barriers that limit movement strategic group is its strategic recipe. In their
across groups. An important implication of theory, strategic group identity is based both on
mobility barriers is that strategic groups should micro-level social learning and social iden-
differ in performance. However, empirical tests tiïŹcation processes, and on macro-level economic,
of this proposition were mixed. Moreover, most historical and institutional processes (Peteraf and
research formed strategic groups by cluster ana- Shanley, 1997). They also suggest these processes
lyzing archival strategy variables (McGee and may lead groups with stronger identities to have
Thomas, 1986; Ketchen, Thomas, and Snow, more positive reputations.
1993). Given the goal of cluster analysis is to Reputation has been used in related ways in
create groups and the mixed results in past strategic research, as reviewed recently by Dol-
research regarding the relationship between stra- linger, Golden, and Saxton (1997). Most research
tegic groups and performance, some researchers focused at the ïŹrm level of analysis, where repu-
suggested strategic groups may be mere method- tation has been deïŹned as the knowledge about
ological artifacts (Hatten and Hatten, 1987; Bar- a ïŹrmâs true characteristics and the emotions
ney and Hoskisson, 1990). towards the ïŹrm held by stakeholders of the
Researchers responded to these issues in sev- ïŹrm (Weigelt and Camerer, 1988; Hall, 1992;
eral ways. Some developed cognitive strategic Fombrun, 1996). In essence, reputation reïŹects
groups, formed from the groupings used by man- what stakeholders think and feel about a ïŹrm.
agers themselves (e.g., Reger and Huff, 1993). Different types of reputation have been studied,
Others improved conceptualizations of archival such as for being a tough competitor (Milgrom
variables used to form strategic groups, focusing and Roberts, 1982), for being a good place to
on ïŹrm scope and resource commitments (e.g., work (Gatewood, Gowan, and Lautenschlager,
Cool and Schendel, 1987). Strategic groups 1993), and for having quality products (Shapiro,
formed by cognitive methods were found to be 1983). Reputations also provide information about
similar to those formed through cluster analyzing expected future behavior (Alchian and Demsetz,
archival variables (Nath and Gruca, 1997). Cool 1972; Weigelt and Camerer, 1988). Thus, a ïŹrm
and Dierickx (1993) examined the implications might be expected in the future to be a tough
of groups on rivalry and found that intergroup competitor, a good place to work, and/or offer
and intragroup rivalry had differential effects on quality products.
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 3. Do Strategic Groups Differ in Reputation? 1197
Interest in reputation has grown in the past as a differentiation signal. We agree a strong
decade. At the ïŹrm-level, research in the identity may increase group visibility, but we are
resource-based view of the ïŹrm proposed that uncertain if a strong identity necessarily increases
reputation may be a resource leading to superior group reputation based on research on both ïŹrms
performance (Dierickx and Cool, 1989; Barney, and industries. Consider the tobacco industry.
1991). For instance, U.K. managers rated com- Given major threats to its legitimacy and few
pany reputation as the most important of 13 ïŹrms in the tobacco industry (Peteraf and Shan-
intangible resources (Hall, 1992). Additionally, ley, 1997; Dranove et al., 1998), a strong identity
Rao (1994) showed how the winners of legit- would be expected. Compared to other industries,
imation contests in the embryonic U.S. auto however, the tobacco industry is commonly per-
industry developed reputations that increased their ceived to have a bad reputation (e.g., Miles,
survival chances. Researchers have begun to con- 1982). Similarly, at the ïŹrm level, Dutton and
sider differences in reputation across industries, Dukerich (1991) noted the Port Authority of New
as well. For instance, Bennett (1998, 1999) found York and New Jersey had a strong engineering
U.K. residents viewed mutual building societies identity, but a poor reputation with the public.
as friendlier than stockholder-owned banks. In Thus, while we question the extent to which
addition to ïŹrm and industry reputations, Peteraf identity strength increases reputation, we agree
and Shanley (1997: 179) proposed that a strategic identity and reputation may be related at the
group with a strong identity will increase its strategic group level. We extend Peteraf and
reputation, which they deïŹne as âfavorable and Shanleyâs (1997) discussion by applying past
publicly recognized standing.â We expand this research on organizational identity and reputation.
deïŹnition to include knowledge of true character- At the ïŹrm level, identity, strategy, and repu-
istics of strategic groups, recognizing the sup- tation have been connected theoretically and
plementary perspective of economic theory empirically. Identity and strategy are reciprocally
(Alchian and Demsetz, 1972; Weigelt and Cam- related, in that strategic choices are concrete
erer, 1988). examples of ïŹrm identity (Ashforth and Mael,
Given this overview of strategic groups and 1996; Whetten and Godfrey, 1998). When Sara-
reputation, we proceed to develop a proposition son (1995) asked managers what was central,
connecting these concepts using two different distinct, and enduring about their ïŹrm, many
theoretical logics. The ïŹrst expands upon strategic responded by describing their ïŹrmâs strategy. The
group identityâreputation propositions initially centrality of strategy is certainly consistent with
made by Peteraf and Shanley (1997). The second most strategic management research. Identity and
applies the domain consensus concept of Thomp- strategy are related to reputation in the following
son (1967) to the strategic group level. In contrast way: A ïŹrm projects images that reïŹect its iden-
to the former, the latter perspective does not tity to its stakeholders (Whetten, Lewis, and Mis-
require the existence of a strategic group identity chel, 1992). These images include not only adver-
for there to be a reputation, nor does it assume tising and public relations, but also strategic
that outsiders who assess reputations perceive actions and verbal statements of strategy, such as
groups. Underlying both logics is the theoretical those communicated through annual reports or
strategy of applying ïŹrm-level theory to the stra- speeches by CEOs. In turn, stakeholders view
tegic group level, analogous to the multi-level these images, interpret them and form reputations
theorizing on threat rigidity by Staw, Sandelands, based on them (Dutton, Dukerich, and Harquail,
and Dutton, (1981). 1994; Whetten, 1997). Strategy has also been
connected directly to reputation. Most notably,
Fombrun and Shanley (1990) found ïŹrm diversi-
Strategic group identity and reputation
ïŹcation strategy was related to reputation.
An initial proposition connecting strategic groups We propose similar reasoning may connect
and reputation was presented by Peteraf and identity, strategy, and reputation in strategic
Shanley (1997: 179). They assert: â[a] stronger groups. Each strategic group has its recipe or
strategic group identity will increase a groupâs core strategy (Porac, Thomas, and Baden-Fuller,
positive reputation,â reasoning that a strong iden- 1989; Reger and Huff, 1993). As such, core
tity is more visible to outsiders and would serve strategy is one of the embodiments of strategic
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 4. 1198 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
group identity that is projected to the external those formed from archival strategy variables.
environment. External stakeholders view this Given both archival and cognitive group research
image of each strategic groupâs identity and form has shown strategic interactions among group
reputations based on it. The reputation of each members in a variety of industries, and likely
group may differ because the identity and strategy similarities between cognitive and archival
of each group differ. Moreover, this relationship groups, it is possible there are interactions among
may be inïŹuenced by the numerous factors that strategic groups in other industries constructed
affect strategic group identity strength as outlined from archival strategy variables. Thus, the
by Peteraf and Shanley (1997). These include relationship between strategic groups and repu-
social learning and identiïŹcation, economic, his- tation may hold in both types of groups.
torical and institutional forces, network properties,
and many other factors.
The domain consensus of strategic groups
The strategic identity logic is based on cogni-
tive strategic groups. The relationship between A second theoretical logic for linking strategic
strategic groups and reputation may hold not just groups and reputation has its basis in Thompsonâs
for cognitive strategic groups but also for archival (1967) concept of domain consensus, which does
strategic groups, if there are sufïŹcient mobility not require the existence of a strategic group
barriers and differences in strategic interactions identity. Instead, it assumes external observers
among groups. Cognitive strategic group research who assess reputations face cognitive limitations
highlights the fact that ïŹrms within a group may and use categorization schemes. Thompson (1967)
affect each other even if they do not directly deïŹned ïŹrm domain as the markets a ïŹrm serves
recognize all the other ïŹrms as competitors. Dif- and the technologies (i.e., resources) it uses to
ferent groups of ïŹrms were shown to exist in serve them. His deïŹnition of domain is quite
the Scottish knitwear (Porac et al., 1989, 1995), similar to the deïŹnition of strategy as a ïŹrmâs
banking (Reger and Huff, 1993) and hotel indus- realized allocation of resources to its product
tries (Lant and Baum, 1994). Managers within market choices (Chandler, 1962; Mintzberg, 1978;
each group did not always recognize every mem- Wernerfelt, 1984). A ïŹrm is embedded in a task
ber of the group as a competitor. Nevertheless, environment, consisting of various stakeholders
Porac et al. (1995) found that the density of including customers, suppliers, competitors, and
named rivalries was far greater within a group regulatory groups. This latter category includes
than across groups. The implication is that the both governmental regulatory agencies and other
actions of particular ïŹrms in a group have greater quasi-regulatory bodies such as trade associations,
inïŹuence on other ïŹrms in the group because the professional organizations, and rating agencies. A
network property of structural equivalence (i.e., ïŹrm negotiates a domain consensus, representing
links to the same ïŹrm without direct ties) sup- a set of expectations about what the organization
plements cohesion (Burt, 1987; Galaskiewicz and will do with respect to its stakeholders
Burt, 1991). This is especially important for larger (Thompson, 1967: 29). As such, this set of expec-
groups, such as those with âtens of ïŹrmsâ found tations can be subdivided into individual compo-
by Porac et al. (1995), which is comparable to nents that are related to the economic perspective
the insurer groups we identify. on reputation (Alchian and Demsetz, 1972; Wei-
Some archival strategic group research also gelt and Camerer, 1988). For instance, customers
indicates there are group-level strategic inter- may expect a company to produce high-quality
actions. Tremblay (1985) found that advertising products (Shapiro, 1983).
expenditures by regional and national brewing According to the embeddedness perspective,
groups in the United States inïŹuenced demand the negotiation of a domain consensus involves
asymmetries. Cool and Dierickx (1993) found not only economic but also social exchanges
that group rivalry was related to ïŹrm performance (Granovetter, 1985). Economic exchanges include
in archival strategic groups in the pharmaceutical resource, product, and monetary transactions,
industry. Furthermore, Nath and Gruca (1997) whereas social exchanges provide information
connected cognitive and archival strategic group about ïŹrm characteristics and trustworthiness.
research. They found convergence between Such reputational information spreads through
groups formed from managerial cognitions and individual and interorganizational networks and
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 5. Do Strategic Groups Differ in Reputation? 1199
coalesces into ïŹrm reputation (Fombrun, 1996). ity. The insurance industry has many character-
For example, Shrum and Wuthnow (1988) found istics enumerated by Peteraf and Shanley (1997)
that network interactions affected the reputation that may increase the strength of strategic group
of research institutes. This phenomenon also has identities. These characteristics include social
been observed in boundary-spanning inter- learning, social identiïŹcation, historical and insti-
relationships (Galaskiewicz and Wasserman, tutional development, network linkages, exoge-
1989; Galaskiewicz and Burt, 1991). nous shocks, managerial movement, corporate
Both the domain consensus and network argu- diversiïŹcation, and ïŹrm entry and exit. Moreover,
ments may also apply at the strategic group level. they also facilitate strategic interactions and the
Firms in a strategic group have similar domains, maintenance of mobility barriers. Although Pet-
that is, they use similar resources to serve similar eraf and Shanley (1997) presented theoretical
markets. The associated social interactions within characteristics, the empirical features of the indus-
the task environment may lead to similar percep- try often embody a combination of characteristics.
tions of ïŹrms in the group by those in the There are dozens of insurance industry trade
environment. Social cognition theory implies that and professional associations, such as the Alliance
people categorize their environment to make of American Insurers (AAI), the American
sense of it (Mervis and Rosch, 1981; Fiske and Association of Insurance Services (AAIS), the
Taylor, 1991; Weick, 1995). In the context of American Insurance Association (AIA), the
strategic groups, research has shown that man- Insurance Services OfïŹce (ISO), the Insurance
agers in an industry categorize ïŹrms into groups Information Institute (III), and the National
(Porac et al., 1989; Reger and Huff, 1993). Mem- Association of Insurance Commissioners (NAIC).
bers of the task environment also face cognitive These organizations have many different effects.
limitations and therefore may also categorize First, they are evidence of the institutional devel-
ïŹrms in the focal industry into groups. Seeing opment of the industry. With this development,
one ïŹrm as similar to another may evoke a there are extensive network ties, an important
schema for interpreting both ïŹrms in terms of component in the development of a strategic
their true characteristics and their emotional group and industry macroculture (Abrahamson
appeal (Ashforth and Mael, 1996). As in the and Fombrun, 1994; Peteraf and Shanley, 1997).
case of an individual ïŹrm, members of the task In the case of the aforementioned groups, they
environment exchange reputational information facilitate exchange of information among different
through social networks, such as through trade companies, product lines, and professionals. In
publications or professional organizations. Over other words, the trade associations contribute to
time, this information may coalesce into a repu- social learning and competitive dynamics (Peteraf
tation for the strategic group as a whole. Because and Shanley, 1997; Kraatz, 1998). To the extent
strategic groups have different domains, their that certain ïŹrms focus on different product lines
reputations may differ. Given the connections and classes of business, this increases the density
between archival and cognitive groups presented and structural equivalence of certain segments
previously, the logic of domain consensus of (e.g., American Land Title Association, Crop
strategic groups may hold for both cognitive and Insurance Research Bureau).
archival groups. Second, trade and professional organizations
In sum, both the strategic group identity and can strengthen identiïŹcation within certain indus-
domain consensus perspectives suggest the fol- try groups. For instance, the National Association
lowing: of Mutual Insurance Companies (NAMIC) lobbies
various levels of government and develops public
Proposition: Different strategic groups may relations campaigns on behalf of mutual insurers.
have different reputations. Additionally, professional organizations such as
the Independent Insurance Agents of America
We develop a testable hypothesis to investigate (IIAA) and professional education/designation
the proposition that strategic groups may differ in programs such as the American Institute for Char-
reputation in the context of the property/casualty tered Property Casualty Underwriters (AICPCU)
segment of the insurance industry, focusing on encourage various levels of what Galaskiewicz
reputation for ïŹnancial stability and product qual- and Wasserman (1989) refer to as network ties
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 6. 1200 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
via boundary-spanning personnel. Given that the industry have different reputations for
purpose of many professional organizations is the product/ïŹnancial quality.
dissemination of knowledge and the socialization
of members into the profession (e.g., the Code
of Ethics of the CPCU Society), it is conceivable METHODS
that members may come to view their environ-
Sample
ment in similar ways. Having many parallel
effects to these trade and professional associations Recent research recommends that strategic group
are well-developed trade media. These range from studies focus on a single industry in order to
general industry news (e.g., Business Insurance, develop a richer understanding of the key prod-
National Underwriter) to professional journals ucts and resources of the industry (Peteraf and
(e.g., CPCU Journal, Risk Management) to prod- Shanley, 1997; Mehra and Floyd, 1998). We
uct or segment speciïŹc publications (e.g., Rough extend this logic by testing our hypothesis using
Notes, Surplus Line Reporter & Insurance News). a single sector within an industry, namely the
Historical factors also may be important in the property/casualty segment of the U.S. insurance
strength of strategic group identity as well (see industry. Our use of individual ïŹrms as the level
Birkmaier and Laster, 1999, for a recent history of analysis also differs from, and improves upon,
of insurance organizations). For instance, prior insurer strategic group research (e.g., Fie-
insurance in general and mutual insurers in parti- genbaum, 1987; Fiegenbaum and Thomas, 1990).
cular evolved out of afïŹliations of individuals These early works analyzed entire insurer âïŹeetsâ
who faced similar loss/risk exposures (e.g., farm- (or holding companies) that spanned both the
ers, tradesmen, wooden home owners in cities, life/health and property/casualty sectors. SigniïŹ-
ocean marine shippers). The ïŹrst successful large- cant differences exist, both across and within
scale stock insurers did not emerge until the mid- each sector, along myriad strategic dimensions
1800s, when the overall economic system had including operating strategies, product offerings,
developed sufïŹciently to better support such for- regulatory oversight, scope of operation, and
profit ventures. As a result of their early com- resource deployment. Our more ïŹne-grained
monality of purpose, mutual insurers are still approach recognizes that insurers, including those
viewed today as being more in touch with their within ïŹeets, tend to operate as unique entities
customer-owners in contrast to the investor- and may focus their strategic efforts in one or
owners of stock insurers. relatively few geographic areas, lines of business,
In recent years, there have been several or even individual products. Thus, we consider
exogenous shocks that may increase strategic the appropriate level of analysis to be individual
group identity, such as major hurricane and earth- ïŹrms in a single industry sector. Furthermore,
quake losses, and increased pressure for ïŹnancial our choice of the ïŹrm as unit of analysis parallels
services deregulation and integration among the choice of the ratings agencies, who prefer to
insurers, banks, and investment brokerages. There rate individual ïŹrms whenever possible.
has been net growth in new industry entrants, Due to time lags in the collection and dissemi-
both in terms of new U.S. start-ups and inter- nation of data, 1996 is the last year for which
national insurers seeking growth opportunities in complete data had been reported at the time of
our domestic markets (Insurance Information initial research in this study, and will therefore
Institute, 1999). Most ïŹrms, even the well estab- be the year of analysis. Approximately 3350 com-
lished, may not compete in every major line of panies sold some form of property/casualty
business but instead focus on a limited number insurance as of year end 1995 (Insurance Infor-
of products or markets. In sum, the insurance mation Institute, 1998). The top 100 companies
industry has many characteristics that may by 1996 sales for which complete strategic proïŹle
heighten strategic group identity and cause stra- data were available were included in the original
tegic interactions to vary among groups. Thus, sample. Eleven ïŹrms were dismissed from the
we propose: analysis due to their status as reinsurers, whose
strategic focus is on other insurers rather than
Hypothesis: Different strategic groups in the end-user insurance consumers. Five ïŹrms were
property/casualty segment of the insurance dismissed as being extreme outliers and not rep-
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 7. Do Strategic Groups Differ in Reputation? 1201
resentative of the sample (see later discussion), Although the value of objective assessment of
resulting in a ïŹnal sample of 84 ïŹrms who insurer insolvency risk or failure cannot be
accounted for approximately 60 percent of total ignored or downplayed (Ferguson, Barrese, and
property/casualty industry premium volume in Levy, 1998), the actual occurrence rate has been
1996. The statistical power of this sample is such relatively low (e.g., less than 20 failures per
that we can detect all large and medium effects thousand per year throughout the decade preced-
at α = 0.01, but we cannot be sure of detecting ing our analysis). Yet policyholders, who are
all small effects (Ferguson and Ketchen, 1999). somewhat insulated from the full consequences
of insurer insolvency due to the existence of
limited state guaranty funds, may still be expected
Data sources
to willingly pay a premium for products offered
The primary source of data was the statutory by insurers with a better reputation for quality
annual ïŹnancial statements of individual and safety (Sommer, 1996). Thus, reputation rat-
insurance ïŹrms. These are submitted in uniform ings are important to insurer owners and man-
format governed by the National Association of agers because ratings are widely touted in stra-
Insurance Commissioners (NAIC) to the regula- tegic marketing, promotion, and placement
tory agency in states where ïŹrms are licensed activities.
or domiciled. These statements are collectively Rating agency opinions carry signiïŹcant weight
available from OneSource, a private ïŹrm licensed beyond a simple threshold ability to meet future
by NAIC to distribute these data. claims. For example, ïŹrms with lower relative
ratings may be excluded from competing for cer-
tain customers or classes of business (Ferguson
Measures
et al., 1998). Many corporate risk managers have
institutional policies that preclude placing busi-
Insurer reputation
ness with lower rated insurers without stringent
There are many types of reputation (Dollinger et justiïŹcation. Insurance producers (e.g., agents,
al., 1997). One critical type of reputation in the brokers, solicitors) also have a signiïŹcant personal
insurance industry is the ability to meet future interest in insurer ïŹnancial reputation ratings as
claims. This type of reputation is fundamentally a producer may be held ïŹnancially liable to their
meaningful for insurance customers because of policyholders by statute for placing business with
the intangible, contingent, and future-oriented na- an insurer that later fails (Hardigree and Howe,
ture of insurance products. The quality of the 1990). The use, or misuse, of rating information
insurance product to the consumer depends in is also important to state insurance regulators in
large part upon the insurer being in business and their role as protectors of the public interest
having sufïŹcient reserves to pay claims, should through the surrogate monitoring of market con-
a loss occur, at some point in the future. How- duct activities. Further, a large body of literature
ever, most consumers ïŹnd it too difïŹcult and/or exists that explores adverse ïŹnancial effects (e.g.,
time consuming to accurately assess the ïŹnancial in ïŹrm bond or stock price) that may result from
strength and stability of insurers. Instead, they rating downgrades (e.g., Hand, Holthausen, and
rely on rating agencies having comparative advan- Leftwich, 1992).
tage in the collection, analysis, and dissemination A number of rating agency intermediaries cur-
of such information (Wakeman, 1981). Reliance rently compete in the market to provide infor-
on specialized intermediaries is common in for- mation regarding insurer ïŹnancial strength and
mation of reputation (Fombrun, 1996).1 claims-paying ability, each having signiïŹcant ïŹ-
nancial incentives to provide accurate and timely
rating opinions (Ferguson et al., 1998). We
1
There may be other measures of insurance company repu- employ the rating opinions issued by three major,
tation, most notably ratings of customer satisfaction by con-
sumer groups. Consumer Reports ratings would perhaps be well-respected rating agencies (the A. M. Best
the most highly regarded of these, although severely limited Company, Standard & Poorâs Corporation (S&P)
in terms of âratingâ frequency, sampling method, and other
factors (Lichtenstein and Burton, 1989). Most critically for
our study, CR has rated just a handful of ïŹrms in two products rated. Further, ïŹrms themselves are not rated, only
personal lines (auto and home owners), with no commercial speciïŹc policy claim satisfaction.
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 8. 1202 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
and Weiss Ratings, Inc.) as our reputation meas- group identity to cognitive strategic groups, Nath
ure. Each agency has developed a proprietary and Grucaâs (1997) demonstration of the conver-
rating philosophy and method to realize its indi- gence between cognitive and archival groups
vidual competitive advantage (Klein, 1992), and implies that archival groups develop identities as
their opinions are widely disseminated and pub- well and thus should be appropriate. Moreover,
licized to the insurance markets. Though each we follow the precautions for researchers using
agency does not rate every insurer, the three cluster analysis outlined by Ketchen and Shook
agencies we employed regularly rate the largest (1996).
cross-section of ïŹrms (Ferguson et al., 1998). Strategic group membership traditionally has
Weiss rates ïŹrms about a normally distributed, been deïŹned along proïŹles and characteristics
academic-based scale, using alphabetic ranks of that inïŹuence competitive advantage (McGee and
A, B, C, D, and E with a âplusâ or âminusâ Thomas, 1986). Later research extended this
available for each, along with a simple âFâ (i.e., approach by acknowledging that strategic groups
no plus or minus). Thus, 16 levels of ratings are are produced by two types of traits critical to
available to label ïŹrms (i.e., A+ = 16, A = 15, competition, notably scope of operations and
⊠E = 2, F = 1). Ratings of D+ or below resource deployment methods (Cool and Schen-
indicate potential vulnerability or substantial del, 1987; Mehra, 1996). The choice of particular
weakness that may cause the ïŹrm to experience strategic variables was based on both prior stra-
signiïŹcant ïŹnancial difïŹculties, especially in an tegic group studies of the insurance industry
unfavorable economic environment. A. M. Best (Fiegenbaum, 1987; Fiegenbaum and Thomas,
rates ïŹrms following a 15-level modiïŹed AâF 1990) and discussions with an expert panel con-
scale (i.e., A++, A+, A, A , B++, B+, B, B , sisting of seven consultants and researchers in
C++, C+, C, C , D, E, and F). However, Best both strategic management and insurance (Mehra
ratings are essentially normally distributed around and Floyd, 1998). Variables capturing scope of
the A rating, and rating categories D, E, and F operations include product scope and diversity,
are reserved for ïŹrms below minimum standards, ïŹrm size, age, and ownership form. Variables
under state supervision or in liquidation, respec- capturing resource deployment include distri-
tively. Ratings of B+ and above are considered bution methods, production methods, and ïŹnan-
secure, while ratings of B and below are con- cial and investment strategies. Table 1 lists the
sidered vulnerable. S&P ratings follow an 18- speciïŹc measures. A more complete description
level basic AâC pattern (i.e., AAA, AA+, AA, of logic behind their usage follows, beginning
AA , A+, A, A , BBB+, BBB, BBB , BB+, with scope of operations variables, then method
BB, BB , B+, B, B , CCC and R). Firms with of developing competitive advantage variables.
ratings of BBB and above are considered
secure, while ratings of BB+ and below are con-
Scope of operations variables
sidered vulnerable. The âRâ rating is reserved for
ïŹrms undergoing regulatory action. Scope of operations is the degree to which an
A composite reputation measure for each ïŹrm organization sells products offered by the indus-
in the sample was generated by standardizing the try, or the number of niches in which the ïŹrm
numeric equivalents of the letter opinion level operates. The insurance industry is commonly
assigned the ïŹrm by each rating agency and then characterized as having two important scope of
averaging the standardized values. Overall, the operations dimensions: (1) type of customer (i.e.,
reliability coefïŹcient for this three-item set of personal or commercial lines), and (2) type of
observations indicates an alpha of 0.768, which product (e.g., life/health or property/casualty). In
is generally acceptable for exploratory work such addition, the diversity in business lines sold by
as the current research (Hair et al., 1995). the ïŹrm, as well as ownership structure and
organizational age and size, are expected to be
indicators of relative scope of operations.
Strategic variables
Personal vs. commercial lines (PPERS) rep-
We form strategic groups by cluster analyzing resents the division of products primarily sold to
archival strategic data. Although Peteraf and individuals (e.g., home owners, private passenger
Shanley (1997) limited their theory of strategic auto) and those sold to businesses (e.g., ocean
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 9. Do Strategic Groups Differ in Reputation? 1203
Table 1. Corporate strategy variables
Strategic component Strategic DeïŹnition
Variable
Scope of operations
Product Scope Personal vs. PPERS Personal Net Premiums Written (NPW)
Commercial Lines Personal NPW + Commercial NPW
Product Scope Property Lines PPROP Property NPW
Total NPW (All Lines Written)
Product Scope Financial Lines PFNAN (Fidelity + Surety + Guaranty, etc.) NPW
Total NPW (All lines Written)
Product Diversity DIVER n
H=1â P2
i
i=1
where: Pi is relative size of ith line in ïŹrm
portfolio (i = 1 ⊠n lines)
Size LSIZE Ln (Total Admitted Assets)
Ownership Form OWNER Stock = 1; Nonstock = 0 (i.e., Mutuals,
Mutual-owned stocks, Reciprocals, Lloyds)
Age AGE 1996 Year of Incorporation
Resource Deployment
Distribution AGENCY Agency = 1; Nonagency = 0
Production REIN Direct Premiums Written -NPW
NPW
Finance LEVER Net Earned Premium
Policyholder Surplus
Investment INVEST Mortgages + Commâl Real Estate + Junk + etc.
U.S. Governments + Investment Grade Corporates
marine, ïŹdelity, workers compensation, commer- Proportion of property lines (PPROP): Three
cial multi-peril). Both personal and commercial major types of ïŹnancial loss exposures are gener-
lines sectors exhibit unique demand and market ally covered by property/casualty insurers: prop-
characteristics that may induce insurers to com- erty exposures (e.g., direct and associated indirect
pete in providing products to satisfy the individual losses suffered by the primary policyholder), lia-
needs of consumers in each market (e.g., Mayers bility exposures (e.g., to indemnify others for acts
and Smith, 1990). The PPERS variable measures that are the responsibility of the primary insured),
the proportion of personal lines business relative and other miscellaneous ïŹnancial exposures that
to total net premiums written. may adversely affect the primary policyholder
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 10. 1204 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
due to the failure of a third party to perform ownership characterizes the overwhelming
(e.g., ïŹdelity, surety bonding, credit). The PPROP majority of insurers. Mutual insurers, though far
variable measures the proportion of property lines fewer in number (comprising less than 10% of
business relative to total net premiums written. ïŹrms), control approximately 40 percent of total
Proportion of ïŹnancial products (PFNAN): The industry assets and premium volume (Insurance
proportion of operations derived from products Information Institute, 1998).
dealing with the third major product area above Differences in ownership structure have very
(ïŹnancial exposures resulting from the failure of important implications for managerial incentives
others) such as ïŹdelity coverage, surety and per- and the relative discretion of managers to act
formance bonds, mortgage guaranty, and credit upon those incentives, including breadth and
insurance, allows a better distinction among com- scope of insurer operations (e.g., Jensen and
panies specializing in these important niche areas. Meckling, 1976; Mayers and Smith, 1981, 1988,
To avoid multicollinearity, the second major 1994, et seq). For example, managers of nonstock
product segment (the relative proportion of lia- ïŹrms (e.g., mutuals) have considerably more
bility insurance) is not included. The PFNAN discretion and incentives to act in their own best
variable measures the proportion of business interest, rather than the policyholders, because of
derived from ïŹnancial lines relative to total net their relative insulation from the market for
premiums written. corporate control (Mayers and Smith, 1994).
DiversiïŹcation (DIVER): The insurance indus- Because of differences inherent in policyholder
try as a whole offers over 30 different major and stockholder goals, nonstock ïŹrms are charac-
product lines, with a multitude of different poli- terized by their cost-centered orientation, whereas
cies sold within each line. The number of lines stock ïŹrms are considered more profit oriented
an organization chooses to offer, as well as the (Mayers and Smith, 1981). A binary variable is
relative emphasis placed on each line, reïŹects employed to proxy strategic differences in
strategic choices having many implications (e.g., owner/managerial incentives between stock and
reduced portfolio risk). A HerïŹndal index is used nonstock ownership forms. Also, following May-
to measure the extent of ïŹrm diversiïŹcation ers and Smith (1981), stock insurers whose ulti-
across lines of business (Pitts and Hopkins, 1982). mate parent is in fact a mutual or other nonstock
Higher index values indicate greater diversiïŹ- form are coded as nonstock ïŹrms since they can
cation. be expected to behave more like their parent than
Size (LSIZE): Size can inïŹuence organizational a traditional widely held stock insurer.
market power, ïŹexibility and strategic response Age (AGE): Age is an important factor in
to environmental concerns. For instance, larger determining scope of operations in that insurers
ïŹrms have greater market power that tends to acquire customers, credibility and capacity to sell
increase the sustainability of competitive actions multiple product lines over time, and is strongly
and outcomes, yet larger ïŹrms may also have correlated with reputation and ïŹrm survival
greater bureaucratic structure or rules which tend (Anderson and Formisano, 1988). Reputational
to decrease ïŹexibility (Hitt, Ireland, and Hoskis- capital is a vitally important asset in the business
son, 1995). Insurer size is expected to inïŹuence of insurance or any ïŹnancial service industry in
scope of operations due to potential economies which fundamental success necessarily requires
of scale and scope (e.g., Doherty, 1981; Johnson, public trust and conïŹdence. Further, rating agen-
Flanigan, and Weisbart, 1981). The natural log cies will not issue a rating opinion until an insurer
of total admitted assets is employed in order to has been sufïŹciently âseasonedâ over a number
mitigate adverse effects of skewness in insurer of years of operation. Age is calculated as 1996
size that exist across the industry (Hair et al., (the year of analysis) less the year of incorpo-
1995). ration.
Ownership form (OWNER): The insurance
industry is dominated by two major forms of
Resource deployment variables
ownership: mutual and stock. Mutual insurers
combine the owner and policyholder roles, while In general, organizations use their resources to
stock insurers clearly demarcate owner/investor enhance organizational value, particularly through
and policyholder functions. The stock form of efïŹcient operations and ïŹnancial management.
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 11. Do Strategic Groups Differ in Reputation? 1205
The level of resource commitment to various between total direct premiums written and net
ïŹrm functions can be indicative of organizational premiums written (i.e., after reinsurance ceded),
commitment to production efïŹciency. Strategic divided by net premiums written.
choices regarding capital resources and invest- Financial leverage (LEVER): Insurance com-
ment also inïŹuence opportunities to create value panies rarely issue traditional debt instruments
by attentiveness to ïŹnancial management issues. due to the contingent liability nature of the pri-
Distribution system (AGENCY): The method of mary obligations they assume issuing contracts
product dissemination into their target market(s) of insurance. However, strategic use of ïŹnancial
represents a crucial competitive strategic decision leverage by insurers can magnify potential returns
for any ïŹrm (see Anderson and Schmittlein, 1984; obtained through underwriting operations and
Anderson, 1985). Much property-liability commensurate investment activities (Anderson
insurance in the United States is sold through the and Formisano, 1988). The LEVER variable is
âAmerican agencyâ (or âindirectâ) system, wherein calculated as the commonly used net earned pre-
insurance products are channeled to customers miums (i.e., that portion of total policy premiums
through either independent agents (i.e., inde- written where the contracted obligation for cover-
pendent contractors who may represent a number age already has been provided as amortized over
of otherwise unrelated insurers) and/or brokers the life of the policy) divided by policyholder
(i.e., contractors who have no advance commit- surplus (i.e., the accounting difference between
ment to any insurer and legally represent insureds available asset base and total ïŹrm liability
as clients). Other insurers, known as âdirect writ- obligations).
ers,â utilize either exclusive agents (i.e., contrac- Investment strategy (INVEST): Investment strat-
tors who represent one insurer or group of com- egies represent organizational choices of accept-
monly owned insurers only), salaried employees, able risk/return relationships, with investment
and/or mass merchandising techniques (e.g., mail, earnings offsetting (supplementing) underwriting
Internet) to market and distribute their products. losses (profits). While ïŹrms in most other indus-
The choice whether to use agency (indirect) or tries are virtually free to invest in stocks, bonds,
direct distribution channels has signiïŹcant stra- derivatives, etc. as they wish, insurers operate in
tegic implications regarding relative managerial a highly regulated environment where investment
control over product marketing and degree of choices are constrained by statute. For instance,
potential market penetration, as well as overall property/casualty companies are prohibited from
cost effectiveness, among other competitive fac- investing more than 10 percent of their assets in
tors (Barrese and Nelson, 1992). A binary vari- stock of any one nonclosely related corporation,
able (agency = 1; nonagency = 0) is used to and real estate holdings cannot exceed more than
depict whether an agency or nonagency distri- 10 percent of total assets (Huebner, Black, and
bution system is utilized. Webb, 1996: 653). Such regulatory constraints
Reinsurance (REIN): Reinsurance is the transfer induce the typical insurer to invest heavily in
of all or a portion of a particular risk by a relatively lower-risk government and higher-
primary insurer to another insurer or insurers, quality commercial securities, with equity invest-
which further spreads the risk and reduces ments generally being predominantly preferred
exposure to extraordinary losses. Reinsurance pro- stock (Insurance Information Institute, 1998).
vides several beneïŹts, including increased ïŹnan- However, insurers can and do invest in other
cial capacity, stabilization of profits, reduction of than low-risk government and high-grade corpo-
unearned premium reserves, and surplus relief. rate securities. The proportion of more risky
One of the most important beneïŹts of reinsurance investments (e.g., mortgages, junk bonds, com-
may be facilitation of entry or exit from a parti- mercial real estate) to relatively safe investments
cular line of business, thereby potentially dimin- (e.g., government securities, municipal bonds,
ishing a mobility barrier created by regulatory high-grade corporate stocks) for a given insurer
obligation to offer continuity of coverage is an appropriate measure that can differentiate
(Trieschmann and Gustavson, 1995: 607). Use of strategic investment choices and resultant com-
reinsurance can thus expedite capitalization of petitive position. The INVEST variable represents
new competitive opportunities (Mayers and total investments other than government and
Smith, 1990). REIN is captured by the difference investment grade corporate securities divided by
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 12. 1206 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
aggregate government and investment grade agency system. Group 2 consisted mainly of non-
corporate securities. stock insurers that exhibited low overall product
line diversity. These ïŹrms tended to be more
narrowly focused, with signiïŹcant personal lines
Statistical analysis
operations. Greater relative conservatism with
respect to investment portfolio and strategy also
Strategic groups
was evident. Group 3 represented the middle
Strategic groups were formed using a two-step ground between the other two groups along most
clustering approach (hierarchical and k-means), measures. Moderately diverse product lines, often
which reduces potential biases introduced by sold on a direct basis to individuals and small
employing a single method (Ketchen and Shook, businesses, characterize this group. A large num-
1996). The hierarchical method used was visual ber of mutual insurers also populate the group.
inspection of tree-plots, a common method of Our hypothesis proposed that strategic groups
determining the appropriate number of clusters differ in their average reputation. ANOVA results
(see Miles, Snow, and Sharfman, 1993; Ketchen as presented in Table 5 indicate signiïŹcant repu-
et al., 1993). Consistent with prior strategic tation differences across the three identiïŹed
groups research, ïŹve outliers also were eliminated groups (F = 7.506; d.f. = 2,81; p < 0.001),
and the clustering procedure was repeated on the providing support for our hypothesis. We also
ïŹnal sample. Initial cluster centers taken from the examined differences between pairs of strategic
ïŹrst step were used in the k-means step (i.e., groups. Bonferroni post hoc tests indicate the
Wards clustering method), eliminating problems overall signiïŹcant F-statistic is being driven by
associated with random seed setting (Hair et al., relationships between strategic Groups 1 and 3
1995). (p < 0.007) and Groups 2 and 3 (p < 0.006).
There were no statistically signiïŹcant differences
in reputation between Groups 1 and 2.2
Hypothesis testing
At the strategic group level, Dranove et al.
Analysis of variance (ANOVA) was used to test (1998) stated that investments in high-quality
the hypothesis of differences in reputation across reputations by group members could be a
strategic groups. Pairwise differences in repu- mobility barrier that separates strategic groups
tation across the strategic groups were assessed and contributes to performance differences among
with Bonferroni post hoc tests. them (Caves and Porter, 1977). To investigate this
possibility, we examined post hoc performance
differences among groups on two measures of
RESULTS importance in the insurance industry: the loss
ratio and the expense ratio.
Pearson zero-ordered correlations among the vari- The loss ratio provides a measure of the rela-
ables used are presented in Table 2. Statistical tive success of an insurer in attaining an overall
analysis revealed three strategic groups with sig- profitable distribution among all exposures
niïŹcantly different strategic competitive proïŹles accepted, which is not only important in current
based on scope of operations and resource performance assessment, but also is crucial to
deployment emerged from the two-step clustering long-run ïŹrm survival (Huebner, Black, and
procedure, with a Wilksâ lambda F = 17.417 (d.f. Cline, 1982). The loss ratio is calculated as the
= 22, 142; p < 0.001). Nine of the 11 strategic
variables were signiïŹcantly different at α = 0.05 2
We also examined limited Consumer Reports claims satisfac-
across the three identiïŹed clusters, as presented tion ratings available for 29 ïŹrms in our sample. At the ïŹrm
in Table 3. Companies within each strategic group level, this rating was signiïŹcantly correlated (0.606, p < 0.01)
are listed in Table 4. with our reputation measure. ANOVA revealed signiïŹcant
differences across groups (F > 4.145, p < 0.027), being driven
Group 1 consisted mainly of larger, older, stock by the difference between Groups 3 and 2 (p < 0.029).
insurers offering very diverse product lines, Overall, Group 3 (N = 17 ïŹrms), had the highest satisfaction;
particularly to commercial rather than personal Group 1 (N = 3), had the second highest; and Group 2 (N =
9), the lowest. These ïŹndings are consistent with our hypoth-
lines clients. Product distribution was esis test results and provide some supporting evidence that
accomplished primarily through the independent claims satisfaction may be associated with reputation.
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 13. Table 2. Correlations
Copyright © 2000 John Wiley & Sons, Ltd.
Variables 1 2 3 4 5 6 7 8 9 10 11 12
Scope of operations
1. Personal Lines PPERS
2. Property Products PPROP 0.10
3. Financial Products PFNAN 0.50 0.04
4. Line Diversity DIVER 0.41 0.21 0.44
5. Size LSIZE 0.23 0.18 0.22 0.17
6. Ownership Form OWNER 0.37 0.11 0.37 0.39 0.03
7. Age AGE 0.42 0.11 0.40 0.39 0.39 0.10
Resource deployment
8. Agency AGENCY 0.40 0.23 0.43 0.51 0.01 0.58 0.30
9. Reinsurance REIN 0.07 0.02 0.19 0.20 0.09 0.18 0.15 0.20
10. Leverage LEVER 0.41 0.13 0.08 0.20 0.29 0.06 0.30 0.11 0.05
11. Investment Mix INVEST 0.27 0.18 0.24 0.08 0.18 0.05 0.28 0.12 0.02 0.33
Performance
12. Loss Ratio LR 0.18 0.21 0.18 0.12 0.22 0.04 0.21 0.15 0.10 0.11 0.04
13. Expense Ratio ER 0.32 0.41 0.37 0.56 0.01 0.22 0.35 0.61 0.17 0.04 0.06 0.14
N = 84. All correlations 0.22 are signiïŹcant at p < 0.05; all correlations 0.28 are signiïŹcant at p < 0.01.
Do Strategic Groups Differ in Reputation?
Strat. Mgmt. J., 21: 1195â1214 (2000)
1207
- 14. 1208 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
Table 3. Results of two-step cluster analysis
Wilksâ lambda F = 17.417; d.f. = 22, 142; p < 0.001
Group 1 (N = 26) Group 2 (N = 17) Group 3 (N = 41)
Strategy Mean (S.D.) Mean (S.D.) Mean (S.D.) F ratio Sig. of
variable F
PPERS 0.6752 (0.4641) 1.3112 (0.2794) 0.2350 (0.8820) 43.868 0.000
PPROP 0.0054 (0.7505) 0.1089 (1.0546) 0.2395 (1.0150) 0.495 0.611
PFNAN 0.0873 (0.1603) 0.2388 (0.0064) 0.1841 (0.0805) 11.882 0.000
DIVER 0.6664 (0.2154) 0.8515 (0.6853) 0.0326 (0.9249) 22.388 0.000
SIZE 0.6065 (0.8384) 0.6219 (0.9150) 0.1585 (0.9639) 10.194 0.000
OWNER 0.8800 (0.3300) 0.4700 (0.5100) 0.5600 (0.5000) 5.481 0.006
AGE 1.2020 (0.8883) 0.7195 (0.5906) 0.2287 (0.5519) 51.208 0.000
AGENCY 0.8500 (0.3700) 0.2400(0.4400) 0.4400 (0.5000) 10.801 0.000
REIN 0.1216 (0.5032) 0.1633 (0.4486) 0.0985 (0.6391) 1.685 0.192
LEVER 0.2422 (0.5355) 1.5449 (0.7245) 0.3200 (0.6569) 56.429 0.000
INVEST 0.0860 (0.0444) 0.1202 (0.0221) 0.0949 (0.0526) 3.013 0.055
proportion of losses plus associated loss variables and the two performance measures. The
adjustment expenses to premiums earned. The second model added the dummies for two of the
expense ratio is a widely accepted measure that groups; including the third would create a per-
reïŹects an organizationâs ability to adequately fectly collinear model. Table 6 presents the
manage operational expenses (e.g., administrative results. When adding the group dummies, the R2
expenses, commissions, contingency for profit) increased from 0.430 to 0.480, an increase that
which generally must be recognized when a pol- was signiïŹcant at the p < 0.01 level (F = 3.270;
icy is ïŹrst written or renewed, according to NAIC d.f. = 2,68). Dummy variables for both Groups
statutory accounting rules. The expense ratio is 2 and 3 were signiïŹcant (t = 2.075, p < 0.042
calculated as the ratio of underwriting expenses and t = 2.54, p < 0.013, respectively). These
to net premiums written. results suggest that knowledge of the strategic
Using MANOVA, we found signiïŹcant overall group structure helps us better explain reputation
differences across the three groups in loss and over and above when only the ïŹrm-level strategy
expense ratio performance measures (Wilksâ and performance variables are used.
lambda = 5.430; d.f. = 4, 160; p < 0.000). This
relationship was driven by signiïŹcant differences
across groups in the expense ratio (exact F = DISCUSSION AND CONCLUSION
7.482; d.f. = 2; p <0.001), whereas the loss ratio
was not signiïŹcantly different (F = 1.908; d.f. = This paper used the concepts of strategic group
2; p < 0.155). Bonferroni tests revealed that identity and domain consensus to propose that
Group 3, the group with the highest average strategic groups have different reputations. Analy-
reputation, had signiïŹcantly better overall per- sis in the property/casualty sector of the U.S.
formance than the other two groups. These results insurance industry provided support for this
provide preliminary support for the idea that stra- hypothesis. Our study has several implications for
tegic group reputation is a mobility barrier. future research in both the strategic group and
To further evaluate the credibility of the stra- reputation realms.
tegic groups, we examined whether knowledge of Our study contributes to strategic group
the strategic group structure adds to our ability research in at least two ways. First, ïŹnding differ-
to predict ïŹrm reputation, as Tremblay (1985) ences in reputation among groups provides further
and Cool and Dierickx (1993) did for perfor- evidence of the usefulness of strategic groups,
mance. We did this through hierarchical regression. contrary to past criticism. Second, we suggest
The ïŹrst model included all-11 ïŹrm-level strategy that strategic group reputation may be a mobility
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 15. Table 4. Strategic group membership
Group 1 Group 2 Group 3
NAIC# NAIC# NAIC#
19038 Aetna Casualty & Surety Co 34754 Commerce Ins Co 19690 American Economy Insurance Company
19380 American Home Assurance Company 21636 Farmers Ins Co of OR 19275 American Family Mutual Insurance Co
20443 Continental Casualty Co 21652 Farmers Ins Exchange 19704 American States Insurance Company
13935 Federated Mutual Ins Co 21660 Fire Insurance Exchange 19976 Amica Mutual Insurance Company
22292 Hanover Insurance Co 22063 Government Employees Ins Co 21202 Auto Club Insurance Association
22357 Hartford Accident & Indemnity Co 26298 Metropolitan Property & Cas Ins Co 18988 Auto Owners Ins Co
19682 Hartford Fire Ins Co 24260 Progressive Casualty Ins Co 20788 Buckeye Union Ins Co
Copyright © 2000 John Wiley & Sons, Ltd.
22713 Insurance Co of North America 32352 Prudential Property & Cas Ins Co 15539 California State Auto Asn Inter-Ins
23043 Liberty Mutual Ins Co 28959 Prudential Property & Cas Ins Co NJ 31534 Citizens Ins Co of America
22977 Lumbermens Mutual Casualty Co 43796 State Farm Indemnity Co 19410 Commerce & Industry Ins Co
19356 Maryland Casualty Co 21709 Truck Ins Exchange 20621 Commercial Union Ins Co
20478 National Fire Ins Co of Hartford 12963 Twentieth Century Ins Co 20990 Country Mutual Insurance Co
25623 Phoenix Insurance Co 25968 USAA Casualty Ins Co 26271 Erie Ins Exchange
24457 Reliance Insurance Co 21458 Employers Ins of Wausau a Mutual Co 24732 General Ins Co of America
26980 Royal Ins Co of America 27553 Mercury Ins Co of CA 38288 Hartford Ins Co of Illinois
24767 St Paul Fire & Marine Ins Co 25143 State Farm Fire and Casualty Co 15598 Interins Exch-Auto Club of So CA
25658 Travelers Indemnity Co 43419 State Farm Lloyds 19437 Lexington Ins Co
25887 United States Fidelity&Guaranty Co 21687 Mid-Century Ins Co
21113 United States Fire Ins Co 22012 Motors Ins Corp
16535 Zurich Ins Co US Branch 23779 Nationwide Mutual Fire Ins Co
20281 Federal Ins Co 23787 Nationwide Mutual Ins Co
21873 Firemans Fund Ins Co 12122 New Jersey Manufacturers Ins Co
20850 Firemens Ins Co of Newark NJ 24074 Ohio Casualty Ins Co
16691 Great American Ins Co 20346 PaciïŹc Indemnity Co
25534 TIG Insurance Company 24988 Sentry Ins a Mutual Co
19445 National Union Fire Ins Co of 23388 Shelter Mutual Ins Co
Pittsburgh 18325 Southern Farm Bureau Cas Ins Co
35076 State Compensation Insurance Fund
25941 United Services Automobile Assoc
41181 Universal Underwriters Ins Co
25976 Utica Mutual Ins Co
20397 Vigilant Ins Co
19232 Allstate Insurance Company
10677 Cincinnati Ins Co
Do Strategic Groups Differ in Reputation?
35289 Continental Ins Co
21970 General Accident Ins Co of America
24740 Safeco Ins Co of America
11762 Vesta Fire Ins Corp
Strat. Mgmt. J., 21: 1195â1214 (2000)
1209
28207 Anthem Insurance Companies Inc
25178 State Farm Mutual Automobile Ins Co
40827 Virginia Surety Co Inc
- 16. 1210 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
Table 5. Reputational differences barrier leading to increased performance. The
resource-based view of the ïŹrm proposed that
F = 7.506; d.f. = 2, 81; p < 0.001 reputation takes time to develop and can be hard
Number of Standardized
ïŹrms mean reputation to imitate, and thus should affect performance
(S.D.) (Dierickx and Cool, 1989; Barney, 1991; Hall,
1992).
Group One 26 0.2951 Our study also contributes to research on repu-
(0.7398) tation in that we show reputation applies not just
Group Two 17 0.3911 to individual ïŹrms and industries but also to
(0.9699)
Group Three 41 0.3253 strategic groups. If reputation is a mobility barrier
(0.7221) at the strategic group level, as well as a barrier
Bonferroni comparisons to imitation at the ïŹrm level, then managers may
Groups SigniïŹcance 95% conïŹdence need to consider the impact of the actions of
interval their ïŹrm on the collective reputation of the
1 and 2 1.000 ( 0.50, 0.69)
1 and 3 0.007 ( 1.10, 0.14) group. Caves and Porter (1977) noted that ïŹrms
2 and 3 0.006 ( 1.27, 0.16) might invest in mobility barriers that defend the
group. Thus, group members face collective
action issues in deciding how much to invest in
reputation building and maintenance at the group
Table 6. InïŹuence of strategic group membership on ability to predict reputationa
Cluster/performance model Model with control variables
Independent variables ÎČ t ÎČ t
Scope of operations
Personal Lines 0.166 1.269 0.003 0.022
Property Products 0.063 0.576 0.049 0.461
Financial Products 0.155 1.232 0.146 1.193
Line Diversity 0.127 1.007 0.206 1.544
Size 0.227â 2.163 0.303ââ 2.860
Ownership Form 0.159 1.291 0.118 0.974
Age 0.183 1.507 0.039 0.264
Resource deployment
Agency 0.117 0.815 0.153 1.093
Reinsurance 0.037 0.392 0.063 0.673
Leverage 0.462âââ 4.106 0.537âââ 3.366
Investment Mix 0.090 0.890 0.132 1.321
Performance
Loss Ratio 0.215â 2.015 0.202â 1.943
Expense Ratio 0.390ââ 2.676 0.378ââ 2.680
Control variables
Group2 0.557ââ 2.075
Group3 0.456ââ 2.540
R2 0.430 0.480
F 4.054âââ 4.178âââ
R2 0.430 0.050
F 13, 70 / 2, 68 4.054âââ 3.270ââ
a
N = 84; â p < 0.10; âp < 0.05; ââ
p < 0.01; âââ
p < 0.001
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 17. Do Strategic Groups Differ in Reputation? 1211
level. However, the managers of an individual tation may also require collective action by group
ïŹrm must also ïŹnd a way to have their reputation members in order to maintain this reputation. In
stand out from their group so their ïŹrm can this context, managers of individual ïŹrms face
develop competitive advantage over other group the challenge of differentiating the reputation of
members. their ïŹrm from that of their peers, particularly on
There are limitations to our research design the strategic group level.
which provide opportunities for future research.
Our study concentrates on one sector of a single
industry in a single year, thus limiting generaliz- ACKNOWLEDGEMENTS
ability. Future research should assess if differ-
ences in reputation across strategic groups exist The authors wish to thank Dr. James Barrese and
in other sectors of the insurance industry (i.e., Barbie Keiser of the College of Insurance for
life/health insurance), in other industries, and their assistance in data collection. We also wish
across time. Second, we assume that constituents to thank our expert panel, two anonymous ref-
categorize ïŹrms, consistent with cognition theory. erees, and Dr. Karel Cool for their valuable com-
Future research should examine the extent to ments in the preparation of this paper. An earlier
which such categorization occurs. Third, determi- version was presented at the Conference on Cor-
nation of whether our knowledge of the group porate Reputation, Image and Competitiveness in
structure helps explain ïŹrm outcomes might be San Juan, PR, January 1999. All errors remain
accentuated with the use of a group-level variable. our responsibility.
Future research should seek to develop group-
level variables that may affect reputation.
Another limitation of this paper is that we REFERENCES
did not directly assess the extent of strategic
interactions within and across strategic groups. Abrahamson E, Fombrun CJ. 1994. Macrocultures:
Rather, we assumed there was a greater density determinants and consequence. Journal of Manage-
of interactions within a group based on insurance ment Review 19(4): 728â755.
industry characteristics and past research (e.g., Albert S, Whetten D. 1985. Organizational identity. In
Cool and Dierickx, 1993; Reger and Huff, 1993; Research in Organizational Behavior, Cummings
LL, Staw BM (eds). Vol. 7: JAI Press: Greenwich,
Porac et al., 1995). Future research should exam- CT; 263â295.
ine this assumption by asking managers to iden- Alchian AA, Demsetz H. 1972. Production, information
tify who their competitors are (Lant and Baum, costs, and economic organization. American Eco-
1994; Porac et al., 1995) or by examining com- nomic Review 62: 777â795.
petitive attack and response dynamics in publicly Anderson DR, Formisano RA. 1988. Casual factors in
P-L insolvency. Journal of Insurance Regulation
available documents (e.g., Young, Smith, and 6(4): 449â461.
Grimm, 1996). Moreover, given our interest in Anderson E. 1985. The salesperson as outside agent or
reputation, it is important to consider the types employee: a transaction cost analysis. Marketing
of interactions that would affect this variable in Science 4: 234â254.
addition to those that affect performance. Finally, Anderson E, Schmittlein DC. 1984. Integration of the
sales force: an empirical examination. Rand Journal
future research should investigate the relative of Economics 15: 385â395.
importance of strategic group identity and domain Ashforth BE, Mael FA. 1996. Organizational identity
consensus in the strategic group-reputation con- and strategy as a context for the individual. In
text. Advances in Strategic Management, Baum JAC,
In sum, we theoretically connected strategic Dutton JE (eds). Vol. 13: JAI Press: Greenwich,
CT; 19â64.
groups and reputation using the principles of Barney JB. 1991. Firm resources and sustained competi-
domain consensus and identity applied at strategic tive advantage. Journal of Management 17: 99â120.
group level. Our empirical analysis indicated that Barney JB, Hoskisson RE. 1990. Strategic groups:
strategic groups differ in reputation. Thus, repu- untested assertions and research position. Mana-
tation appears to be a multilevel concept. Post gerial and Decision Economics 11: 187â198.
Barrese J, Nelson JM. 1992. Independent and exclusive
hoc analysis of performance differences indicate agency insurers: a reexamination of the cost differen-
strategic group reputation may be a mobility bar- tial. Journal of Risk and Insurance 59(3): 375â397.
rier that beneïŹts group members. Finally, repu- Bennett R. 1998. Ethnicity, the representativeness heu-
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)
- 18. 1212 T. D. Ferguson, D. L. Deephouse and W. L. Ferguson
ristic, and customer perceptions of banking organis- School Press: Boston, MA.
ations. Paper presented at the Conference on Corpor- Fombrun CJ, Shanley M. 1990. Whatâs in a name?
ate Reputation, Identity and Competitiveness, Reputation building and corporate strategy. Academy
Amsterdam. of Management Journal 33: 233â258.
Bennett R. 1999. Corporate reputation of UK banks Galaskiewicz J, Burt RS. 1991. Interorganization con-
and building societies among ethnic minorities. Cor- tagion and corporate philanthropy. Administrative
porate Reputation Review 2: 104â114. Science Quarterly 36: 88â105.
Birkmaier U, Laster D. 1999. Are mutual insurers an Galaskiewicz J, Wasserman S. 1989. Mimetic processes
endangered species? SwissRe SIGMA 4/99. within an interorganizational ïŹeld: an empirical test.
Burt RS. 1987. Social contagion and innovation: Administrative Science Quarterly 34(3): 454â480.
cohesion versus structural equivalence. American Gatewood RD, Gowan MA, Lautenschlager GJ. 1993.
Journal of Sociology 92: 1287â1335. Corporate image, recruitment image and initial job
Caves RE, Porter ME. 1977. From entry barriers to choice decisions. Academy of Management Journal
mobility barriers: conjectural decisions and contrived 36(2): 414â427.
deterrence to new competition. Quarterly Journal of Granovetter M. 1985. Economic action and social struc-
Economics 91: 241â261. ture: the problem of embeddedness. American Jour-
Chandler AD. 1962. Strategy and Structure. MIT Press: nal of Sociology 91: 481â510.
Cambridge, MA. Hair JF, Anderson RE, Tatham RL, Black WC. 1995.
Cool K, Dierickx I. 1993. Rivalry, strategic groups, Multivariate Data Analysis. 4th edn, Prentice-Hall:
and ïŹrm profitability. Strategic Management Journal Englewood Cliffs, NJ.
14(1): 47â59. Hall R. 1992. The strategic analysis of intangible
Cool K, Schendel D. 1987. Strategic group formation resources. Strategic Management Journal 13(2):
and performance: the case of the U.S. pharmaceu- 135â144.
tical industry, 1963â1982. Management Science 33: Hand JR, Holthausen RW, Leftwich RW. 1992. The
1102â1124. effect of bond rating agency announcements on bond
Dierickx I, Cool K. 1989. Asset stock accumulation and stock prices. Journal of Finance 47: 733â753.
and the sustainability of competitive advantage. Hardigree DW, Howe V. 1990. The insurer selection
Management Science 35: 1504â1513. process by independent insurance agents. Journal of
Doherty NA. 1981. The measurement of output and Insurance Issues 13: 27â46.
economies of scale in property liability insurance. Hatten KJ, Hatten ML. 1987. Strategic groups, asym-
Journal of Risk and Insurance 48(3): 390â402. metrical mobility barriers, and contestability. Stra-
Dollinger MJ, Golden PA, Saxton T. 1997. The effect tegic Management Journal 8(4): 329â342.
of reputation on the decision to joint venture. Stra- Hitt MA, Ireland RD, Hoskisson RE. 1995. Strategic
tegic Management Journal 18(2): 127â140. Management: Competitiveness and Globalization.
Dranove D, Peteraf M, Shanley M. 1998. Do strategic West Publishing: St Paul, MN.
groups exist? An economic framework for analysis. Huebner SS, Black K, Cline RS. 1982. Property and
Strategic Management Journal 19(11): 1029â1044. Liability Insurance. 3rd edn, Prentice-Hall: Engle-
Dutton JE, Dukerich JM. 1991. Keeping an eye on wood Cliffs, NJ.
the mirror: image and identity in organizational Huebner SS, Black K, Webb BL. 1996. Property and
adaptation. Academy of Management Journal 34: Liability Insurance. 4th edn, Prentice-Hall: Upper
517â554. Saddle River, NJ.
Dutton JE, Dukerich JM, Harquail CV. 1994. Organiza- Hunt M. 1972. Competition in the major home
tional images and member identiïŹcation. Adminis- appliance industry, 1960â1970. Doctoral dissertation:
trative Science Quarterly 39: 239â263. Harvard University.
Ferguson TD, Ketchen DJ. 1999. Organizational con- Insurance Information Institute. 1998. Property/
ïŹgurations and performance: the role of statistical Casualty Insurance Facts. Insurance Information
power in extant research. Strategic Management Institute Publications: New York.
Journal 20(4): 385â395. Insurance Information Institute. 1999. Fact Book 2000.
Ferguson WL, Barrese J, Levy DT. 1998. The cost of Insurance Information Institute Publications: New
biased insurer ratings. Journal of Insurance Issues York.
21(2): 138â150. Jensen MC, Meckling WH. 1976. Theory of the ïŹrm:
Fiegenbaum A. 1987. Dynamic aspects of strategic managerial behavior, agency costs and ownership
groups and competitive strategy: concepts and structure. Journal of Financial Economics 3(4):
empirical examination in the insurance industry. 305â360.
Doctoral dissertation, University of Illinois at Johnson JE, Flanigan GB, Weisbart SN. 1981. Returns
UrbanaâChampaign. to scale in the property-liability insurance industry.
Fiegenbaum A, Thomas H. 1990. Strategic groups and Journal of Risk and Insurance 48(1): 18â45.
performance: the U.S. insurance industry, 1970â Ketchen DJ, Shook CL. 1996. The application of cluster
1984. Strategic Management Journal 11(3): 197â analysis in strategic management research: an analy-
215. sis and critique. Strategic Management Journal
Fiske ST, Taylor SE. 1991. Social Cognition. 2nd edn, 17(6): 441â458.
McGraw-Hill: New York. Ketchen DJ, Thomas JB, Snow CC. 1993. Organiza-
Fombrun CJ. 1996. Reputation. Harvard Business tional conïŹgurations and performance: a comparison
Copyright © 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 1195â1214 (2000)