r Academy of Management Journal

2015, Vol. 58, No. 6, 1614–1636.

http://dx.doi.org/10.5465/amj.2013.1204

YOU DON’T FORGET YOUR ROOTS: THE INFLUENCE OF

CEO SOCIAL CLASS BACKGROUND ON
STRATEGIC RISK TAKING
JENNIFER J. KISH-GEPHART
University of Arkansas

JOANNA TOCHMAN CAMPBELL

University of Cincinnati

Social class is increasingly recognized as a powerful force in people’s lives. Yet despite
the long and extensive stream of research on the upper echelons of organizations, we
know little about how executives’ formative childhood experiences with social class influ-
ence their strategic choices. In this study, we investigate the influence of chief executive
officers’ (CEOs’) perceived social class origins on firm risk taking. We also explore the
moderating influences of other important career experiences, in the form of elite education
and diverse functional backgrounds. Our theory and findings highlight that an executive’s
social class origins have a lasting and varying impact on his or her preferences, affecting his
or her tendency to take risks. By examining this novel managerial characteristic, we offer
important implications for theorizing about social class and upper echelons.

Americans have a really hard time talking about the a profound influence on individuals’ perspectives
class system, because they really don’t want to admit and decision making (Côté, 2011; Fiske & Markus,
that the class system exists. But the reality is it does. 2012; Liu, Ali, Soleck, Hopps, Dunston, & Pickett,
. . . No matter how you define it, coming to terms with 2004). Moreover, anecdotal evidence implies that
class in America means understanding how it shapes the effects of childhood social class are not easily
most everything about you. (PBS, 2001) discarded—even after an individual experiences
The terms “corporate elite” and “upper echelons” objective success and movement into a higher social
often invoke images of strategic leaders at the apex class (see, e.g., Lubrano, 2004). As one chief executive
of the societal hierarchy. Yet the attendant wealth officer (CEO) from a lower social class background
and privilege of such positions may obscure the fact explained to the authors:
that not all executives share the same social class You grow up in that kind of a world where you have to
origins, or early experiences with, and access to, work for what you have. You don’t have a lot of extra
valuable resources while growing up (Domhoff, things. You know you’ve got to work together. How
2010; Freeland, 2011). Recent empirical and theo- can that not help but have an impact on you?1
retical work in the fields of psychology, education,
and sociology suggests that social class can have Despite such evidence that social class matters,
organizational researchers know little about how
The authors contributed equally to the preparation of
childhood experiences with social class—and sub-
this manuscript. Authorship order was determined by coin sequent upward mobility—influence individuals in
toss. We extend our many thanks to Barbara Gray, Don the workplace, including those who have ascended
Hambrick, Eden King, and Chris Tuggle for providing to the top of the corporate hierarchy.
feedback on earlier drafts of the manuscript. Thank you to In this study, we take an initial step toward
Aimee Hamilton for assistance with coding and to Nina addressing this conspicuous oversight and exam-
Gupta for officiating the coin toss. We are grateful for ine the influence of CEOs’ perceived social class
generous financial support from the Sam M. Walton Office
of Diversity and Inclusion, and the Office of the Vice Pro-
1
vost for Research and Economic Development at the Uni- As part of a larger project aimed at understanding
versity of Arkansas. Finally, thank you to Scott Graffin and the role of social class background in executive careers,
our three anonymous reviewers for encouragement and the authors conducted interviews with CEOs from the
insights throughout the review process. S&P1500.

1614
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express
written permission. Users may print, download, or email articles for individual use only.
2015 Kish-Gephart and Campbell 1615

background on strategic risk taking at the firm capital, which will result in higher levels of strategic
level. We define social class as “a person’s per- risk taking.
ceived place in an economic hierarchy” (Liu et al., In this paper, we build on and integrate disparate
2004: 9). This economic hierarchy reflects differ- literatures on upper echelons, social class, and im-
ences in access to material resources (more or less printing to theorize and test a model that highlights
abundant) and societal rank (where one stands in the complex ways in which social class origins affect
comparison to others in society) (Côté, 2011: 47; risk taking. In so doing, we contribute to the research
Kraus, Piff, Mendoza-Denton, Rheinschmidt, & on the upper echelons of organizations, and extend
Keltner, 2012). Drawing on imprinting theory our understanding of why and how managerial
(Marquis & Tilcsik, 2013), we argue that individ- characteristics influence strategic choices. Thirty
uals develop a social class imprint that is highly years ago, Hambrick and Mason (1984: 201) observed
resistant to decay over time. We apply this per- that “there has been almost no attempt in the orga-
spective to the corporate suite and, building on nizational literature to relate socioeconomic back-
upper echelons theory (Hambrick & Mason, 1984), ground to organizational strategy.” To our knowledge,
suggest that top executives’ social class origins will our study represents the first direct empirical test of
shape how they view and respond to strategic sit- the effects of perceived CEO social class background
uations. Following prior research (for example, on strategic decision making. We also contribute to
Fiske, Moya, Russell, & Bearns, 2012; Gray & Kish- the emergent stream of social class research in orga-
Gephart, 2013), we distinguish between three levels nizational studies and extend researchers’ limited
of social class—lower, middle, and upper—and understanding of the effects of upward mobility in
compare those executives who have experienced organizational settings. Our results suggest that an
the least and the most resource constraints (the individual’s social class origins are not easily shed as
upper and lower social classes, respectively) to the he or she rises to the corporate elite; rather, social class
“normative” middle class (Skeggs, 2004) with origins, and the experiences inherent in them, have
regard to strategic risk taking. We focus on risk a lasting and differential impact on the individual’s
taking as the firm-level outcome because of its risk-taking preferences.
relevance to a number of influential organizational
theories beyond upper echelons, including agency
theory (Eisenhardt, 1989) and prospect theory THEORETICAL BACKGROUND
(Kahneman & Tversky, 1979).
Upper Echelons
Consistent with the idea that the “effects and ex-
ternal manifestations” of imprints can strengthen or Hambrick and Mason’s (1984) seminal work forms
decay over time (Marquis & Tilcsik, 2013: 195), we the basis of the empirical stream of research on the
also consider how certain salient characteristics of “upper echelon” of organizations. In contrast to
an executive’s career experiences influence the re- organizational demography research (see Pfeffer,
lationship between his or her social class imprint 1983), which highlighted the importance of the de-
and his or her strategic risk taking. Starting with mographic make-up of entire organizations (that is,
college education, we suggest that an elite college spanning multiple organizational levels), Hambrick
education will have unique implications for strate- and Mason (1984) brought the focus to the organi-
gic risk taking, because social class origins influ- zation’s top level, arguing that characteristics of the
ence how individuals perceive and respond to that upper echelon‒‒executive values and preferences‒‒
success—as something to be leveraged or something are reflected in organizational outcomes. According
to be protected. Moving beyond prior research that to this theorizing, “the givens that a manager brings
has used elite education as a proxy for social class to an administrative situation” (Hambrick & Mason,
background (see, e.g., Westphal & Khanna, 2003), 1984: 196) form the manager’s mental model, which
this allows us to separate theoretically and empiri- has a bearing on managerial perceptions, interpreta-
cally the influences of perceived social class back- tions, and subsequent strategic choices.
ground and elite education. We also consider the Since the publication of that formative article,
moderating influence of the executive’s functional a stream of studies on executive characteristics and
background. To the extent that executives have ex- their impact on organizational outcomes has fol-
perience in diverse and varied functional areas of lowed. For instance, researchers have examined the
organizations, we suggest that they will be able to effects of executive tenure (see, e.g., Finkelstein &
build on their existing mental models and social Hambrick, 1990) and various executive traits, such
1616 Academy of Management Journal December

as hubris (Hayward & Hambrick, 1997), narcissism addition to material conditions (reflecting eco-
(Chatterjee & Hambrick, 2007), and charisma nomic capital), Bourdieu (1984, 1994) argued that
(Waldman, Javidan, & Varella, 2004). Upper echelons individuals in different social classes share similarities
researchers have also analyzed a score of relation- in social capital (e.g., connections and networks) and
ships between top management team (TMT) compo- cultural capital (e.g., tastes and practices that are cul-
sition and/or demographic characteristics and firm turally valued). This is because, through socialization
outcomes (for reviews, see Carpenter, Geletkanycz, and life experiences, individuals develop a “habitus,”
& Sanders, 2004; Finkelstein, Hambrick, & Canella, or a set of dispositions, expectations, and behaviors
2009). that are considered normative for members of a partic-
Nevertheless, among the plethora of executive ular social class. An individual’s habitus operates pri-
characteristics examined, one construct mentioned marily beneath consciousness, and manifests in tastes,
by Hambrick and Mason (1984) has not been given styles, and preferences (such as leisure time and ways
much research consideration—namely, executive of interacting with community institutions). It is also
social class background. Some studies have exam- reproduced through everyday actions and interactions
ined the relationship between elite education and (Gray & Kish-Gephart, 2013). As Stephens, Markus,
executive behavior (see, e.g., Haunschild, Henderson, and Townsend (2007: 814, emphasis original) noted:
& Davis-Blake, 1998; Palmer, Jennings, & Zhou, 1993). “Although the material conditions of the sociocultural
Tihanyi, Ellstrand, Daily, and Dalton (2000), for ex- context do not determine people’s actions, they do
ample, found that the average level of elite college promote certain kinds of actions and increase the
education of the TMT was positively related to in- likelihood that these actions will become normative
ternational diversification. Although attendance at an and preferred. . .”
elite college institution is often used as a proxy for Within organizational studies, direct consider-
an upper social class background (Domhoff, 1970; ation of social class is limited (for exceptions, see,
Westphal & Khanna, 2003), this practice is problem- e.g., Gray & Kish-Gephart, 2013; Leana, Mittal, &
atic because it assumes that those who graduated Stiehl, 2012; Scully & Blake-Beard, 2006; Smith,
from an elite college institution share such a back- Menon, & Thompson, 2012). As Côté (2011: 44)
ground. In addition, in an examination of acquisition noted, “organization science typically ignores social
behavior in the 1960s, Palmer and Barber (2001) used class,” and when it is considered, its indicators (such
elite secondary education (and/or family membership as occupational position) tend to be “‘treated as
in a social register) as a proxy for upper social class nuisance variables whose influence must be ex-
background, finding that elite secondary education cluded’ (Christie & Barling, 2009: 1474–1475).” Yet
was negatively related to acquisition behavior. While researchers in education, psychology, and sociology
informative, this work did not account for meaningful increasingly recognize social class as “one of the
differences in the early experiences of executives from most meaningful cultural dimensions in people’s
the lower and middle social classes. Consequently, lives” (Liu et al., 2004: 3), influencing such outcomes
much remains to be understood about the potentially as long-term health, well-being, and judgments and
highly influential nature of childhood social class decision making (Côté, 2011; Fiske & Markus, 2012;
origins—a topic to which we turn now. Scott, 2005).
Furthermore, whereas much of the aforemen-
tioned work related to social class in organizations
(Lingering) Effects of Social Class Background
has focused on current social class standing, less is
As defined earlier, social class represents “a per- known about situations in which individuals have
son’s perceived place in an economic hierarchy” experienced movements into a higher social class
(Liu et al., 2004: 9). This economic hierarchy reflects rank—as is the case for the CEOs from lower and
differences in individuals’ access to material re- middle social class backgrounds in the current
sources and societal rank (Côté, 2011). According to study. According to Kniffin (2007: 56), a common
social class theorizing, the objective and subjective assumption, even among some researchers, is that
aspects of social class are highly interrelated upwardly mobile adults experience a “molting pro-
(Bourdieu, 1984; Kraus, Tan, & Tannenbaum, 2013; cess” and discard “their old cultural skins” when
Markus & Fiske, 2012). Material conditions, through they experience movement into a higher social class.
their influence on what a person owns and experi- However, qualitative research (see, e.g., Ostrove &
ences, shape how individuals perceive where they Stewart, 1998) and writings by successful upwardly
stand in comparison to others (Kraus et al., 2012). In mobile adults (see, e.g., Lubrano, 2004) hint at the
2015 Kish-Gephart and Campbell 1617

enduring nature of social class origins. As Mahony During these periods, individuals look to and adopt
and Zmroczek (1997: 4) have explained, “class is cues from their environment regarding “appropri-
deeply rooted, retained, and carried through life ate” cognitive and behavioral patterns. This per-
rather than left behind (or below). In this sense, it is spective is consistent with social class theorizing
more like a foot which carries us forward than suggesting that childhood serves an especially im-
a footprint which marks a past presence.” Moreover, portant role in socializing “different ways of being”
research in biology suggests that childhood social for members of different social classes (Stephens,
class continues to have pervasive effects on the hu- Markus, & Phillips, 2014: 626; Bourdieu, 1984). For
man body, even after accounting for movement to example, activities as simple as storytelling by fam-
a higher social class in adulthood (Kittleson, Meoni, ily members communicate distinctive ways of per-
Wang, Chu, Ford, & Klag, 2006; Marin, Chen, & ceiving and responding to the world (e.g., a focus on
Miller, 2008). Miller and colleagues (2009) argued opportunity and self-expression versus a focus on
that childhood social class influences the body’s bi- dangers and risk; Miller, Cho, & Bracey, 2005). These
ological programming such that “repeated social messages are further reinforced by teachers in grade
adversity in early life can program a ‘defensive’ schools, who socialize children to the conditions of
phenotype . . . that persists across decades and the context (Stephens et al., 2014). In addition, in-
thereby accentuates vulnerability to disease later in dividuals tend to be surrounded by, and spend the
life” (Miller et al., 2009: 14719). From this perspec- majority of their time with, others of similar social
tive, childhood social class has the potential to leave classes to themselves (for example, via neighbor-
a “biological residue” (Miller et al., 2009). hoods, schools, and work institutions), rarely cross-
ing social class boundaries (Hout, 2008; Kraus et al.,
2012; Liu et al., 2004).
Social Class “Imprint”
Imprinting theory thus provides support for the
Similar to a “biological residue,” social class ori- longevity of the effects of social class origins even
gins may leave a “cognitive residue that explains after movement from one’s childhood social class
variations in the decisions of organizational mem- into a new social class standing. To the extent that
bers” (Côté, 2011: 59). Drawing on organizational people adopt the attendant norms and cognitions
theorizing, one explanation for the long-term influ- learned through their childhood social class back-
ence of perceived childhood social class despite ground, “their subsequent behaviors bear the stamp
upward mobility is imprinting. “Imprinting” refers [or imprint] of the environment they experienced”
to the process through which, during one or more (Marquis & Tilcsik, 2013: 201). This social imprint is
sensitive time periods in a person’s life, that person likely to persist, such that it influences behavior over
“develops characteristics that reflect prominent time and across contexts (Bianchi, 2013; Higgins,
features of the environment, and these charac- 2005). Indeed, some related empirical research sup-
teristics persist despite significant environmental ports the influence of early macroeconomic cir-
changes in subsequent periods” (Marquis & Tilcsik, cumstances on beliefs and behavior years later (see,
2013: 199). Originally inspired by Stinchcombe’s e.g., Bianchi, 2013; Malmendier & Nagel, 2011). As
(1965) theorizing about organizations, the concept of one example, Giuliano and Spilimbergo (2009)
imprinting has also been applied to individual-level found that experiencing a recession in early adult-
phenomena. Researchers have investigated, for in- hood influenced individual beliefs about success.
stance, how imprints from early career experiences In the following section, we integrate the litera-
(such as employment at a particular company) per- tures on upper echelons and social class to argue
sist throughout careers and across organizations that, because of childhood experiences with material
(Bianchi, 2013; Higgins, 2005). Others have sug- resources and societal rank, senior executives will
gested that imprinting shapes managers’ mental differ with regard to firm-level risk taking. Using the
models, “which they then use to interpret the envi- normative middle class as the baseline, we begin by
ronment and guide their actions” (Holburn & Zelner, contrasting executives from the upper social classes
2010: 1293). with those from the middle social classes. We then
As the definition implies, social imprints are most compare the risk-taking preferences of executives
likely to occur during sensitive or vulnerable periods from the lower social classes with the preferences of
in an individual’s life, including childhood and those from middle social class backgrounds. Next,
other significant role transitions (such as entry into building on the idea that the influence of imprints
college or job transition: Marquis & Tilcsik, 2013). may vary within certain boundaries (Marquis &
1618 Academy of Management Journal December

Tilcsik, 2013), we consider the interactive effects of continued employment, creating vulnerability to
other important experiences during an executive’s ebbs in the economy (Gray & Kish-Gephart, 2013).
career—that is, the level of elite education and varied Further, the middle rank position involves both the
career functional background—on the relationship potential for upward movement (into the upper so-
between perceived social class origins and risk cial classes) and the potential to fall into the lower
taking. social classes. For this reason, the psychology of the
middle class has been described as “fear of falling”
(Ehrenreich, 1989) and “status panic” (Mills, 1951),
SOCIAL CLASS ORIGINS AND RISK TAKING
such that those from the middle classes are espe-
Individuals’ risk preferences are influenced by cially motivated to maintain their current position
their motivation toward avoidance or opportunity, and minimize the likelihood of status loss (Schwalbe,
as well as by relevant situational factors (Lopes, 2008; Smith, 1982).
1987; March & Shapira, 1987). While some in- Based on the aforementioned research, we argue
dividuals may pay more or less attention to the worst that executives from the upper social classes will be
(or best) possible outcomes, situational factors (such more willing than their middle class counterparts to
as the risk of a highly undesirable outcome or his or engage in strategic risk taking. First, because of dif-
her current reference point) can also direct an in- ferences in the size of their childhood safety nets
dividual’s focus toward either avoidance or oppor- (social and economic resources), individuals with an
tunity (March & Shapira, 1987). Building on this, upper social class imprint are less likely than their
we argue that an executive’s experience with child- middle class counterparts to perceive the same ac-
hood social class—and its attendant resources and tion as risky. According to March and Shapira (1987:
rank—directs his or her attention toward avoidance 1407), managers perceive high risk as “the threat of
(or threat) or opportunity. This emphasis on avoid- a very poor outcome.” An individual’s definition of
ance or opportunity shapes an individual’s social a very poor outcome, however, depends on the level
class imprint, persisting across time and influencing of resources available to him or her (Lopes, 1987). All
strategic decisions years later. else being equal, those with more abundant re-
sources (as is the case for those from the upper social
classes) will have a higher tolerance for potential
Upper Social Class Background and Risk Taking
loss. As Williams (2012: 42) noted: “Because money
Having grown up with an abundance of resources, can buy second chances, those who have it have
individuals from the upper social classes have ex- a different attitude toward novelty and risk.”
perienced the benefit of a substantial safety net that Second, given their superior societal rank, in-
provides not only economic security, but also—and dividuals from the upper social classes are likely to
perhaps more importantly—psychological security. have developed an approach motivation, character-
As Williams (2012: 42) pointed out, individuals from ized by a focus on rewards and disinhibited behavior
this type of background “are encouraged to experi- (Keltner, Gruenfeld, & Anderson, 2003). This is be-
ment, secure in the knowledge that if they get into cause social class standing reflects differences in
scrapes, parents usually can get them out.” As such, power, or control over valuable resources (Bullock &
individuals from the upper social classes tend to Lott, 2010; Smith et al., 2012). According to Keltner
perceive the world as safe, welcoming, and full of and colleagues (2003: 269), “acting within reward-
opportunity (Kusserow, 2012). Furthermore, upper rich environments and being unconstrained by
social class standing is characterized by superior other’s evaluations or the consequences of one’s ac-
societal rank. This position yields such benefits as tions” is related to elevated perceptions of power.
higher levels of optimism, self-esteem, and sense of Such elevated power increases approach-related
control (Kraus et al., 2012; Marmot, 2004; Twenge & behavior (including risk taking), because indi-
Campbell, 2002). viduals with elevated power focus more on rewards
In contrast, the middle social class position is and downplay the potential for loss in a given
characterized by adequate access to resources that situation—in other words, they are more optimistic
allow for basic necessities and limited extras (for about their chances for success (Keltner et al., 2003).
example, family vacations and college education; Anderson and Galinsky (2006) found empirical sup-
Williams, 2012). In comparison to the upper social port for this perspective, showing a positive relation-
classes, the middle class safety net is smaller and less ship between power and risk taking, as mediated by
secure, because access to resources is predicated on optimism. Consistent with existing research suggesting
2015 Kish-Gephart and Campbell 1619

that individuals from the upper social classes expe- class norms), but also rests precariously on contin-
rience elevated levels of optimism, self-esteem, and ued employment (Skeggs, 2004). Individuals from
control (Lamm, Schmidt, & Trommsdorff, 1976; the middle class are particularly focused on dangers
Marmot, 2004; Twenge & Campbell, 2002), percep- (March & Shapira, 1987) and are motivated to
tions of self-esteem and control have also been em- maintain the status quo (Schwalbe, 2008; Smith,
pirically linked to risk taking (Anderson & Galinsky, 1982). This perspective is supported by the limited
2006; March & Shapira, 1987). research on middle status positions. For example,
Consequently, in comparison with those from the according to Nagi (1963), in comparison to upper and
middle classes, we expect that upper social class lower status positions, individuals in the middle
upbringing will heighten managers’ sense of security status position experience the highest level of anxi-
and control, thus increasing their optimism about ety when facing risks, because risks in the middle
future prospects (Keltner et al., 2003; Malmendier & class position represent a “double threat”: They
Tate, 2005). In line with research suggesting that threaten the individual’s current middle class
CEOs who are optimistic or overconfident are more standing (with its many advantages) and the poten-
likely to undertake acquisitions, especially unrelated tial to increase that standing in the future (Nagi, 1963:
ones (signifying a willingness to engage in risky 442; Smith, 1982). In contrast, individuals in lower
strategies: Malmendier & Tate, 2008), we expect that social class positions face the lowest amount of po-
CEOs of upper social class origins will engage in tential loss (Nagi, 1963). They have less to lose in the
greater levels of strategic risk taking in comparison to event of failure and may be more willing to take risks
their middle class counterparts. Formally: because the negative consequences of failure are not
perceived to be as destructive (in comparison to the
Hypothesis 1. In comparison with their middle consequences perceived by those in a middle status
class counterparts, CEOs of upper social class position). Therefore, their disadvantaged position
origins engage in higher levels of strategic risk may focus their attention on opportunities for gain,
taking. rather than on the dangers of loss (March & Shapira,
1987).
Lower Social Class Background and Risk Taking In sum, this paper suggests that, in comparison
with those from the middle classes, CEOs from the
Having experienced the lowest level of resources lower social classes may perceive that they have
and societal rank, executives of lower social class “less to lose” and may be more willing to accept
origins are familiar with circumstances character- the potential downside of a risky decision. Execu-
ized by regular uncertainty and lack of control tives from the middle social classes, in contrast,
(Shipler, 2004). In such environments, “every small may be more concerned about avoiding any loss
error [has] large consequences” because individuals from their current position. Thus, consistent with
do not have adequate safety nets to cushion against Hambrick and Mason’s (1984) original proposition
unexpected negative events (Shipler, 2004: 45). that CEOs from lower social class backgrounds are
Parents in lower social class contexts teach their more likely to take strategic risks, we hypothesize
children to recognize the potential downsides of an that:
action and “to avoid errors” (Stephens et al., 2014:
618). Over time, individuals in lower social class Hypothesis 2. In comparison with their middle
contexts are likely to develop a “greater sensitivity to class counterparts, CEOs of lower social class
potential social and environmental threat” (Kraus origins engage in higher levels of strategic risk
et al., 2012: 550; Miller, Chen, & Cole, 2009a), or an taking.
avoidance motivation (Keltner et al., 2003; March &
Shapira, 1987).
Despite this heightened vigilance toward threat, SOCIAL CLASS ORIGINS, POST CHILDHOOD
EXPERIENCES, AND RISK TAKING
we argue that individuals from the lower social
classes will be more likely than their middle class According to imprinting theory, the “effects and
counterparts to engage in risk taking. As discussed external manifestations” of imprints can “vary over
already, middle class individuals are in a vulnerable time, reflecting an interplay of the past and the pres-
position: Their societal rank comes with many ad- ent” (Marquis & Tilcsik, 2013: 195). Consequently, it
vantages (including adequate access to resources is important to understand when and how an exec-
and access to institutions designed around middle utive’s social class origins may interact with other,
1620 Academy of Management Journal December

less distal, experiences. In this section, we consider their upper social class counterparts, and thus lead
the potential for an executive’s experience with elite us to hypothesize that rather than increasing risk
education and diverse career functional background taking, an elite education for those from the lower
to influence the relationship between his or her so- social classes will attenuate risk taking. First,
cial class origins and firm-level risk taking. whereas the upper social classes benefit from a non-
stigmatized identity and are familiar with elite in-
stitutions’ dominant norms, those with lower social
Elite Education
class origins carry with them the burden of a stig-
For those from the upper social classes, an elite matized identity (Bullock & Lott, 2010; Skeggs,
education is not uncommon (Domhoff, 2002) nor is 2004). Similar to the experiences of gender and ra-
the environment (characterized by upper social class cial minorities, a stigma does not have to be fully
norms) unusual (Johnson, Richeson, & Finkel, 2011). internalized to represent a threat to an individual’s
However, it does represent an opportunity for an positive sense of self (Gray & Kish-Gephart, 2013);
individual to improve his or her access to resources simply making one’s social class (minority) status
and to form valuable connections—the latter being salient is enough to trigger stereotype threat and its
easier for those from the upper social classes because attendant consequences (e.g., increased anxiety and
of their familiarity with navigating upper class cul- fear of being perceived according to the stereotype;
ture (Stephens et al., 2014). Indeed, prior research Hall, 2012; Steel & Aronson, 1995). Indeed, prior
suggests that people are able to easily and quickly research demonstrates that social class differences
discern the social class standing of others (Hout, are likely to be salient during the individual’s edu-
2008; Kraus & Keltner, 2009; Liu et al., 2004; Tudor, cational experience, especially for those from the
1971). This is especially likely in elite circles, in lower social classes in an elite environment (Bullock
which cultural capital plays a crucial role in in- & Lott, 2010; Lubrano, 2004; Stephens et al., 2007,
teractions (Fussell, 1983; Skeggs, 2004). In addition, 2014). Managing this stigmatized identity (for ex-
people in a state of elevated power tend to be socially ample, by concealing it from others or avoiding sit-
uninhibited (Keltner et al., 2003) and “feel less con- uations in which it may become salient) exacts
strained by their in-groups” (Smith et al., 2012: 71). a high emotional and cognitive toll (Clair, Beatty, &
The formation of networks should be easier for those MacLean, 2005), including psychological strain
from the upper social classes, leading to exposure to (Ragins & Cornwall, 2001) and self-regulatory de-
new ideas (for example via connections with people pletion (Johnson et al., 2011). As Smart and Wegner
across industries), as well as additional resources on (2000: 221) noted, “concealing a stigma leads to an
which to fall back in case of loss. Research in entre- inner turmoil that is remarkable for its intensity and
preneurship shows that informal industry networks its capacity for absorbing an individual’s mental
play an important role in opportunity recognition or life.” Based on the high level of cognitive effort in-
the individual’s level of alertness to new business volved, we expect that it will be more difficult for
opportunities (Ozgen & Baron, 2007). Moreover, Cao, individuals of lower social class origins to capitalize
Simsek, and Jansen (2012) argue that external social fully on the benefits of an elite education, such as
ties, through their ability to provide external feed- forming interpersonal relationships and expanding
back, give CEOs “a greater sense of mastery, control their networks (Clair et al., 2005). In line with this,
and preparedness” that encourages higher levels of recent empirical research suggests that when low
entrepreneurial risk. Additional resources should status individuals experience threat, they “protect
also lessen the perception of the threat of risk, be- themselves from the uncertainties implicit in net-
cause any negative outcomes can be more easily working among less familiar contacts” (Smith et al.,
mitigated as the individual’s safety net grows (March 2012: 70).
& Shapira, 1987). Consequently, for executives from A second difference is the degree to which grad-
the upper social classes, we expect that an elite ed- uation from an elite institution represents a nota-
ucation will increase the likelihood of taking strate- ble and rare accomplishment for those from the
gic risks. lower social classes. For those in the lower societal
For those from the lower social classes, an elite ranks, graduation from any institution of higher
education may potentially result in a similar boost of education—let alone an elite institution—is often
resources. However, two points make the experience out of reach (Resnick & Wolff, 2003; Stephens,
of an elite education for individuals from the lower Fryberg, Markus, Johnson, & Covarrubias, 2012). As
social classes fundamentally different from that of Bullock and Lott (2010: 417) noted, “for low
2015 Kish-Gephart and Campbell 1621

income families, [a four-year education] means Hitt and Tyler (1991: 334) pointed out: “Top execu-
tremendous, if not impossible sacrifice.” As such, tives often have experience in multiple functions, al-
we expect that graduation from an elite institution though they may have dominant experience in one.”
will change the reference point for evaluating future We thus consider how the influence of the social
risks (Lopes, 1987). Drawing on our earlier argument, class imprint varies based on whether the CEO is
rather than having “nothing to lose,” individuals of a broad generalist or a narrow functional specialist
lower social class origins who have graduated from (Bunderson & Sutcliffe, 2002; Cannella et al., 2008).
an elite institution now have a “positive balance” This moderating influence comes from two sour-
that is likely to reduce their motivation to engage in ces. First, executives with diverse functional expe-
subsequent high-risk efforts (Bateman & Zeithaml, rience are likely to have more extensive personal
1989; Kahneman & Tversky, 1979). This is consistent networks and to “have a good perception of where
with March and Shapira’s (1987) theorizing that the knowledge is and how to tap into it” (Bunderson,
moving to just over a performance target shifts an in- 2003: 460; Cannella et al., 2008: 769). In other words,
dividual’s focus from opportunities for gain to dan- functional job rotation expands executives’ net-
gers of loss. As one CEO from a lower social class works, and allows for a broader flow of information
background communicated to us, “you don’t need to and exchange of ideas (Geletkanycz & Black, 2001).
throw another Hail Mary pass if you’re ahead by 20 This in turn increases the potential set of investment
points.” Formally, we hypothesize that: opportunities and should positively affect strategic
risk taking.
Hypothesis 3a. Elite education moderates the
Second, because executives tend to view strategic
relationship between upper social class origins
decisions through the prism of their personal back-
and risk taking, such that it amplifies the ten-
grounds, those “with high intrapersonal functional
dency of CEOs of upper social class origins to
diversity are less likely to be narrowly parochial”
engage in risk taking.
(Cannella et al., 2008: 769; Raskas & Hambrick, 1992)
Hypothesis 3b. Elite education moderates the and more likely to have a broader, open-minded
relationship between lower social class origins perspective. Experience in different functional areas
and risk taking, such that it reduces the ten- broadens an executive’s knowledge of strategic ap-
dency of CEOs of lower social class origins to proaches and practices, and lessens his or her com-
engage in risk taking. mitment to the status quo (Geletkanycz & Black,
2001). Conversely, as tenure within a single func-
tional area increases, an executive’s perspective
Executive Functional Background successively narrows and more closely reflects the
Another potentially influential factor is a CEO’s dominant logic of the particular functional area
experience acquired via a varied career background. (Geletkanycz & Black, 2001).
In the upper echelons framework, managers’ percep- For those from both the upper and lower social
tions, problem-solving frameworks, and—ultimately— classes, this reasoning suggests that having a diverse
strategic decisions are a reflection of their backgrounds functional background—reflecting a broad, “general
(Wiersema & Bantel, 1993). Executive functional management,” career track—will enhance the like-
background in particular is one of the key character- lihood that these executives will perceive new in-
istics outlined in upper echelons theory (Hambrick & vestments that entail a level of risk as comparatively
Mason, 1984). And while multiple studies have ex- less risky (vis-à-vis those with narrow functional
amined TMT heterogeneity or diversity, including backgrounds). As one CEO explained to us:
functional background heterogeneity (Bunderson & [Y]ou can take more risks because you’ve been there
Sutcliffe, 2002; Carpenter et al., 2004), little research and you’ve done that and you have more leeway to do
examines diverse managerial experience (that is, the it and you have more confidence in doing it. And you
diversity of experience within an individual, as op- also realize that you know you are going to do the right
posed to within a team). We focus on intrapersonal thing because you just have that experience that’s
functional diversity, defined as “breadth of functional built up . . . I did about every marketing and sales job
experience within TMT members” (Cannella, Park, & in [name of company] when I was there, and general
Lee, 2008: 769, emphasis original); however, it should management job, and presidents’ jobs at three differ-
be noted that our focus is on the single most powerful ent major ones. I did worldwide marketing, I did op-
executive—the CEO—rather than the entire TMT. As erations in the U.S. and Canada, I did all of our
1622 Academy of Management Journal December

channels. So, you know, I think you are a victim of the CEOs to their respective firms and constructed a
building blocks of your experience. And the more panel dataset of 265 executives (308 unique CEO–
experience you get, the more diversity you have. . . firm combinations) over 10 years (that is, 2002–2011).
Consequently, we expect a diverse, general man- Our final sample used in the regressions (after miss-
agement functional background to positively affect ing data) consists of 1,042 observations.
risk taking by amplifying the effect of CEOs’ social
class origins. Formally, we hypothesize that: Measures
Hypothesis 4a. A general management func- Independent variable: CEO social class background.
tional background moderates the relationship We measured subjective social class using a cate-
between upper social class origins and risk gorical measure adapted from existing research (see,
taking, such that it amplifies the tendency of e.g., Davis & Robinson, 1988; Jackman & Jackman,
CEOs of upper social class origins to engage in 1973; Smith et al., 2012). To gauge childhood social
risk taking. class (versus current social class), CEOs were asked
Hypothesis 4b. A general management func- to answer the question, “Which of the following best
tional background moderates the relationship describes your family’s socioeconomic situation
between lower social class origins and risk tak- while you were growing up?” Response choices in-
ing, such that it amplifies the tendency of CEOs cluded five categories: lower, lower-middle, mid-
of lower social class origins to engage in risk dle, upper-middle, and upper (Jackman & Jackman,
taking. 1973). The distribution of the responses in our sample
was as follows: 3.40% lower, 38.49% lower-middle,
38.11% middle, 16.98% upper-middle, and 3.02%
METHODS upper class.4
To examine the validity of our measure, we coded
Sample and Data Sources interview data for two indicators of objective social
The sample frame for our study included the CEOs class: parents’ occupation and highest level of com-
of S&P1500 firms. We obtained data on CEO social pleted education. The interviews were conducted as
class background via a mail survey sent to all current part of a larger data collection effort and comprised
chief executives in August 2012. The packet in- 31 CEOs who responded to our initial paper sur-
cluded a solicitation letter featuring the school’s vey request (11.7% of all respondents). One of
logo, a one-page survey instrument, and a prestamped, the authors and an independent coder (unfamiliar
pre-addressed envelope. We received responses from with the hypotheses of this study) coded each of
272 CEOs,2 making the response rate 16.3%—a favor- the interviews separately. Following Adler, Epel,
able outcome, given our target group3 and the sensitive Castellazzo, and Ickovics (2000), each parent’s edu-
nature of our research topic. We addressed the poten- cation was coded into one of five categories (no de-
tial for self-selection bias among our responder group gree, high school degree, college degree, master’s
in our analyses (see “Additional Analyses and Ro- degree, and higher professional degree); occupation
bustness Checks”). was coded into one of three categories (blue collar or
We obtained demographic data and other back- service, clerical or self-employed, and professional
ground information on the CEOs from a variety of or managerial). Initial agreement between coders
sources, including the Execucomp database, com- was 100% for education and 97% for occupation; the
pany websites, Businessweek.com, Forbes.com, and two coders subsequently discussed and reached
company annual reports. Firm-level data were col- agreement on codes for the two discrepant cases for
lected from the Center for Research in Security Prices occupation. Consistent with the idea that “indi-
(CRSP) and Compustat databases. We matched the viduals identify with the ‘highest’ class possible” in
their household (Davis & Robinson, 1988: 109), we
2 calculated Spearman’s rank-order correlation using
Seven of the CEOs declined the invitation to
the occupation and education codes from the parent
participate.
3
As Cycyota and Harrison (2006) discussed, based on with the highest level of occupation or education,
their 12-year study, the survey response rates for top ex-
4
ecutives are significantly lower compared to other pop- We addressed the potential nonrepresentativeness of
ulations (including employees and lower-level managers), our sample and self-selection bias via a Heckman maxi-
and appear to be steadily declining. mum likelihood procedure, as discussed later.
2015 Kish-Gephart and Campbell 1623

respectively. The correlation between CEOs’ per- had an undergraduate degree from an elite university
ceived social class origins and parents’ occupation (5.6%), and “3” if both degrees were from elite uni-
was 0.68 (p , .001); and the correlation between the versities (8.7% of our sample). Because our study
CEOs’ perceived social class origins and parents’ focuses on the effects of early, formative life experi-
highest level of completed education was 0.63 ences, our index places a high importance on the
(p , .001). These results are consistent with exist- elite status of undergraduate education as a key pe-
ing research showing a moderate-to-strong corre- riod of growth and self-discovery in the life of young
lation between subjective and objective measures of adults.5 In line with prior research (see, e.g., Bigley &
social class (see, e.g., Côté, 2011; Hout, 2008; Smith Wiersema, 2002), we used the list of elite universities
et al., 2012). Subjective and objective measures of reported in Finkelstein (1992).
social class have also been found to similarly pre- Moderator: General management functional
dict a variety of outcomes, including those related background. Following Hambrick and Mason’s
to health, cognition, and behavior (see, e.g., Adler & (1984) classification scheme, each CEO’s career
Snibbe, 2003; Adler et al., 2000; Piff, Kraus, Côté, functional background was coded into one of the
Cheng, & Keltner, 2010; Smith et al., 2012). In the three following categories: throughput functions
final analyses, and in line with existing theorizing (R&D, operations, engineering); output functions
(Fiske et al., 2012; Gray & Kish-Gephart, 2013), the (marketing and sales); and peripheral functions (law,
lower and lower-middle categories were aggregated finance, accounting). If the CEO’s functional back-
into one category (lower social class origins), and ground did not fit into one of these categories (39%
the upper and upper-middle categories were ag- of the CEOs in our sample), it was coded as general
gregated into a second category (upper social class management functional background (Bednar, 2012).
origins). This coding scheme thus resulted in four functional
Dependent variable: Strategic risk taking. background indicators. In the analyses, we took
Following Martin, Gomez-Mejia, and Wiseman a conservative approach and used peripheral back-
(2013) and Devers, McNamara, Wiseman, and Arrfelt ground as the baseline because of its rather broad
(2008), we measured strategic risk taking using three focus in terms of organizational functions vis-à-vis
variables: research and development (R&D) expen- other options (that is, throughput or output).6
ditures, capital expenditures, and the value of the Control variables. We included a number of
firm’s long-term debt. This approach is consistent industry-level, firm-level, and individual-level control
with our focus on managerial risk taking, “defined as
management’s proactive strategic choices involving
5
the allocation of resources” (Palmer & Wiseman, This also accounts for the fact that many of the CEOs in
1999: 1038, emphasis original). Consistent with prior our sample received their MBAs from elite universities,
research, we use the unscaled versions of the vari- when they were already on a management career path. Our
ables and explicitly control for firm size in our results hold if we drop the last category and use an elite
undergraduate education index that ranges from 0 to 2.
models, which avoids the issue of spurious correla-
They also hold if we use a measure of elite education rep-
tions common to ratio measures (Wiseman, 2009).
resenting the years of education spent at elite institution(s).
Principal components factor analysis returned a sin- This particular measure is less precise because we know
gle factor, with a 1.58 eigenvalue, explaining 53% of only if a CEO has a degree from said institution, but not the
the variance. The factor loadings were 0.66 for R&D, number of years spent there; as such, we assume four years
0.81 for capital expenditures, and 0.71 for long-term for an undergraduate degree, two years for a graduate de-
debt, justifying the aggregation of the variables into gree, and six years if both degrees were from an elite school.
a single composite indicator of firm risk taking. We The years of elite education measure is correlated at 0.86
used a one-year lag between the measurement of the with that used for our primary analyses.
6
dependent variable and our predictors (strategic risk By using peripheral background as the baseline, we are
taking was measured at time t11). better able to isolate the effect of background diversity
versus background focus, because both output and through-
Moderator: Elite education. Degree of elite edu-
put functions are concentrated in certain parts of organi-
cation was measured using a modified version of the
zational process, while peripheral functions such as law
index used by Tihanyi and colleagues (2000). The and accounting can apply to all stages of organizational
variable was coded as “0” if the CEO did not com- production function. However, our results and conclu-
plete a formal four-year degree (15.4% of our sam- sions remain essentially unchanged if we use all three
ple), “1” if he or she had an undergraduate degree nondiverse categories combined (throughput/output/
from a nonelite university (70.3%), “2” if he or she peripheral) as the baseline.
1624 Academy of Management Journal December

variables to isolate the influence of CEO social class were included in all models, but are not reported due
background. First, we included a lagged value of the to space constraints.
dependent variable (strategic risk takingt), a control
for firm accounting performance (as return on as-
Analysis
sets, or ROA: Bromiley, 1991), and market perfor-
mance (as total stockholder return: Sanders & Because of the panel structure of our data and the
Hambrick, 2007). We also included a control for potentially endogenous nature of the elite education
firm size using the natural log of assets (Martin et al., variable (whereby the effects of social class back-
2013) and diversification via the entropy index ground and elite education are confounded as a result
(Hoskisson, Hitt, Johnson, & Moesel, 1993). We used of their interdependent relationship),7 we employed
CEO age to control for the fact that risk-taking pro- endogeneity-corrected models suited to panel data
pensity may vary with age (Devers et al., 2008; analysis. Specifically, we relied on two-stage least
Hambrick & Mason, 1984). We included a separate squared (2SLS) instrumental variable regressions (see
control for CEO tenure at the focal firm. We con- Bascle, 2008). While the matter of instrument appro-
trolled for the CEO’s compensation and incentives priateness is ultimately determined statistically, our
by including measures for his or her salary, bonus initial search was theoretically motivated. Two vari-
compensation, (the Black-Scholes value of) stock ables had good statistical properties and aligned well
option awards compensation, and stock owner- theoretically with elite education: (a) an indicator for
ship percentage (CEO ownership). Because research whether the CEO was awarded an honorary PhD, and
suggests that risk taking may vary by gender (b) the number of external board appointments (re-
(MacCrimmon & Wehrung, 1986) and founder status flective of the individual’s status and social network).
(Begley, 1995), we included female and founder in- The two requirements for a good instrument
dicator variables. We also included three additional (Murray, 2006) are that it is relevant (that is, a
educational controls: MBA, CPA/CFA, and professional “strong,” or “not weak,” predictor of the po-
degree (JD or PhD) dummies. Moreover, we controlled tentially endogenous variable) and valid (that is,
for the organization’s task environment via measures exogenous—not correlated with the error term of the
of industry munificence, dynamism, and complexity second-stage equation). We performed appropriate
(Keats & Hitt, 1988). statistical tests that confirmed both instrument rele-
Because risk perception is relative and may vary vance and exogeneity. First, the F-statistic for the in-
with exposure to various business environments, we strument relevance test was equal to or greater than
also included binary (dummy) controls for each ex- 17.23 (F . 10) for all models (p . F 5 0.000), providing
ecutive’s career background. Specifically, we col- strong evidence that our instruments were relevant
lected information from executive biographies about and not weak predictors (Staiger & Stock, 1997). In
his or her industry experience and coded for the support of this conclusion, both instrumental vari-
following industries: aerospace, commercial bank- ables were statistically significant in the first-stage
ing, finance and investment banking, biotechnol- models and had significant partial R2 contributions to
ogy or pharmaceuticals, business services, casino the first-stage model (p , .001 for all models). Further,
operations, construction, chemicals, consulting, the Anderson canonical correlations likelihood-ratio
consumer goods, industrial goods, health care and (LR) test rejected the null hypothesis in all cases, in-
medical equipment, insurance, information tech- dicating that the instruments were relevant and the
nology, law/legal, manufacturing, oil and gas, public model was identified (StataCorp, 2011).
accounting, publishing, retail, software, technology, Second, because our models included more in-
telecommunications, transportation/freight, and util- struments than potentially endogenous regressors,
ities. Industry indicators at the two-digit standard we were able to perform a direct test of instrument
industrial classification (SIC) level are also included exogeneity (Wooldridge, 2002). Both the Sargan and
in all models to control for the variance in risk taking the Basmann tests failed to reject the null hypothesis
resulting from (current) industry affiliation (Martin of exogeneity—that is, that the instruments are un-
et al., 2013). Finally, we included year dummy vari- correlated with the error term and correctly excluded
ables to remove the issue of contemporaneous corre- from the estimated equation. Finally, we investi-
lation in panel data (Certo & Semadeni, 2006) and to gated whether there was any evidence that the elite
control for general macroeconomic fluctuations.
These three sets of dummy variables (CEO industry 7
We thank a thoughtful reviewer for pointing this con-
background experience, the firm’s industry, and year) cern out to us.
2015 Kish-Gephart and Campbell 1625

education variable was in fact endogenous. The risk taking, while Hypothesis 4b predicted that gen-
Durbin x2 test, as well as the Wu-Hausman F-statistic eral management functional background positively
showed some evidence of variable endogeneity moderates the relationship between lower social
(p values often less than or around .10), suggesting class origins and risk taking (amplifying the ten-
that an endogeneity-corrected model was warranted.8 dency of both groups to engage in risk taking). Based
All nonindicator variables used in the creation on the positive coefficients on the interactions re-
of the interaction terms were mean-centered to re- ported in Model 4 of Table 2 (p , .01 and p , .001,
duce nonessential multicollinearity. We ran multi- respectively), both hypotheses receive statistical
collinearity diagnostics and the results indicated support. All significant interactions are depicted in
that all individual variance inflation factor (VIF) Figures 1 and 2.
values were below 10, with a mean VIF of approxi-
mately 2, suggesting that multicollinearity did not
Additional Analyses and Robustness Checks
affect our results.
Our findings indicate that CEOs from both the up-
per and lower social classes take higher levels of
RESULTS
strategic risk than their middle class counterparts.
Table 1 reports summary statistics and correla- Further tests indicate that while both groups take
tions. Table 2 reports the results of our hypotheses more risk than executives from the middle class,
testing. Hypothesis 1 predicted that, in comparison CEOs with upper social class backgrounds neverthe-
with their middle class counterparts, CEOs of upper less take greater levels of strategic risk than those of
social class origins engage in higher levels of strate- lower social class origins (p , .001). In other words,
gic risk taking. The coefficient on the upper social while both groups of CEOs differ from those from the
class background indicator variable in Model 2 of middle class, they also differ from one another.
Table 2 is positive and statistically significant (p , To help to ensure the robustness of our results, we
.001), supporting Hypothesis 1. collected additional data on the firm’s governance
Hypothesis 2 predicted that, in comparison with characteristics, including board size, average di-
their middle class counterparts, CEOs of lower rector age, average director tenure, average director
social class origins engage in higher levels of strate- ownership, outside director percentage, and outside
gic risk taking. The coefficient on the lower social director ownership. None of the governance controls
class background indicator variable in Model 2 were statistically significant or substantively changed
of Table 2 is positive and statistically significant our results; as such, they were excluded for parsi-
(p , .05), supporting Hypothesis 2. mony and to preserve degrees of freedom.
Hypothesis 3a proposed that elite education mod- Moreover, we fitted a model that accounts for self-
erates the relationship between upper social class selection into our sample (which thus accounts for
origins and risk taking, such that it strengthens (am- the potential nonrepresentativeness of our sample
plifies) the relationship between upper social class compared to the more general pool of CEOs). Spe-
origins and risk taking. The coefficient on the in- cifically, we relied on a Heckman self-selection
teraction between upper social class status and elite model using an indicator variable for the selection
education in Model 3 of Table 2 is not statistically equation whereby we coded CEO-firm observations
significant, thus failing to support Hypothesis 3a. In in our sample as “1” and coded the others (that is, the
Hypothesis 3b, we argued that elite education reduces remaining group of S&P1500 CEOs) as “0”. We sus-
the tendency of CEOs with lower social class origins to pected that the firm’s prominence or visibility may
engage in risk taking. The coefficient on the interaction have played a role in whether the CEO chose to re-
between lower social class origins and elite education spond to our survey, especially given the nature
in Model 3 of Table 2 is negative and statistically sig- of our topic. It is also probable that upper-middle
nificant (p , .05), supporting Hypothesis 3b. and upper class CEOs are more likely to be hired
Hypothesis 4a predicted that general management into more prominent firms (which would help to
functional background positively moderates the re- explain the perhaps lower-than-expected proportion
lationship between upper social class origins and of upper-middle class respondents). As such, as
a proxy for the firm’s prominence or visibility, we
8
We performed a similar analysis treating the social collected data from the I/B/E/S database on the
class indicators as endogenous and found no evidence of number of financial analysts following the firm
endogeneity. (that is, issuing recommendations). The higher the
 1626

TABLE 1
Descriptive Statistics and Correlations
Variable Mean SD 1 2 3 4 5 6 7 8 9 10 11 12

1 Strategic risk takingt11 0.00 1.00 1.00

2 Lower social class origins 0.42 0.49 20.03 1.00
3 Upper social class origins 0.20 0.40 0.08 20.42 1.00
4 Elite education 1.10 0.75 0.12 20.17 0.12 1.00
5 General management functional 0.39 0.49 0.03 0.04 0.01 20.14 1.00
background
6 Strategic risk takingt 0.00 1.00 0.82 0.02 0.01 0.11 0.10 1.00
7 Accounting performance (ROA) 0.04 0.14 20.04 0.09 20.06 0.03 0.03 20.03 1.00
8 Market performance 11.25 73.74 0.00 0.03 20.02 0.01 20.02 20.01 0.03 1.00
9 Firm size 7.64 1.67 0.15 0.01 20.03 0.13 0.08 0.37 20.03 20.01 1.00
10 Diversification 0.40 0.60 20.04 0.07 20.05 0.08 20.02 20.02 20.03 20.01 0.04 1.00
11 CEO age 53.83 6.82 20.03 0.19 20.09 20.03 0.01 20.01 0.04 0.01 20.02 0.04 1.00
12 CEO tenure 4.61 6.59 20.08 0.08 20.08 20.02 0.07 20.09 0.01 20.02 20.08 20.08 0.43 1.00
13 Salary 579.51 307.14 0.14 0.04 20.08 0.04 0.11 0.28 0.02 0.01 0.29 0.02 0.34 0.26
14 Bonus 304.48 682.46 0.03 0.02 20.03 0.07 0.01 0.13 0.11 20.01 0.11 20.04 0.08 0.11
15 Options awards 314.12 1120.24 20.01 20.03 0.03 0.07 20.01 20.01 20.02 20.01 0.09 0.02 20.10 20.06
16 CEO ownership 0.01 0.05 20.02 20.01 0.08 20.12 0.01 20.05 0.01 0.00 20.18 20.04 0.24 0.33
17 Female 0.03 0.18 20.02 20.03 0.02 20.02 20.11 20.03 20.07 0.00 20.08 20.03 20.08 20.01
Academy of Management Journal

18 Founder 0.10 0.30 20.03 0.03 20.08 20.09 0.06 20.06 20.03 20.01 20.15 20.04 0.13 0.46
19 MBA 0.40 0.49 0.05 20.05 20.08 0.11 0.22 0.08 0.00 20.02 0.00 20.04 0.06 0.12
20 CPA/CFA 0.10 0.30 0.00 0.03 0.01 20.03 20.17 20.02 20.05 20.01 0.01 0.01 0.00 0.01
21 Professional degree 0.11 0.31 20.02 0.02 0.04 0.11 20.12 20.02 0.02 20.01 20.06 0.02 20.04 0.06
22 Output background 0.09 0.29 0.00 20.08 20.10 20.06 20.26 20.03 20.01 0.09 20.03 0.03 20.04 0.00
23 Throughput background 0.21 0.41 20.02 20.01 0.04 0.09 20.41 20.07 0.03 20.01 20.06 20.04 0.02 20.03
24 Munificence 0.04 0.10 0.01 20.04 0.03 0.06 0.00 0.04 0.07 20.01 0.06 20.08 20.07 20.06
25 Dynamism 0.03 0.03 20.02 20.06 0.05 0.07 20.05 0.01 20.06 20.03 0.13 0.05 0.01 20.01
26 Complexity 1.05 0.17 0.00 20.02 0.02 0.04 20.01 0.01 0.12 0.00 0.02 0.00 20.04 20.02
December
2015 Kish-Gephart and Campbell 1627

number of analysts following the firm, the higher its

1.00
20.19
25 prominence or visibility to the public.
We used the newly constructed prominence
variable—the average number of analysts following

1.00
20.21
0.25
24

the firm in a given year—as our instrument in the

first-stage model. The variable was highly significant
in the model (p , .001) and the sign was negative, in

1.00
20.03
0.02
0.02
line with our expectations. We relied on the Heck-
23

man maximum likelihood procedure to draw our

conclusions, because the literature indicates that it
1.00
20.16
20.07
0.01
20.02
has better econometric properties than the tradi-
22

tional, “two-step” Heckman model (although our

results are very similar and our conclusions un-
changed if we use the two-step model). Our con-
1.00
20.11
0.00
0.08
0.13
0.03
21

clusions regarding our hypotheses tests remain

substantively unchanged from our primary analyses
(that is, there are only minor differences in sig-
1.00
20.12
20.08
20.15
0.06
20.01
20.01

nificance levels and/or magnitudes of individual

20

coefficients), with one important exception: Hy-

pothesis 3b, regarding the moderating effect of
1.00
20.02
20.12
20.02
20.12
0.06
20.08
0.04

elite education on the relationship between lower

19

social class origins and risk taking, is no longer

supported, while Hypothesis 3a, regarding the
moderating effect of elite education on the re-
1.00
0.01
20.03
0.04
0.01
20.07
20.03
20.05
20.02
18

lationship between upper social class origins and

TABLE 1

risk taking, is now supported (in other words, what

(cont’d)

changes is which half of Hypothesis 3 is sup-

1.00
0.03
20.04
0.03
0.03
0.16
20.03
20.01
20.03
0.00

ported). This is likely to be the result of the endo-

17

geneity of our elite education variable. As such,

since all other conclusions are essentially un-
1.00
0.01
0.33
20.06
0.00
0.04
0.01
20.04
20.05
20.01
0.02

Note: Correlations greater than 0.04 are statistically significant at p , .05.

changed, we use the models accounting for endo-

16

geneity resulting from unobserved variable bias

instead of self-selection for our primary analyses.
1.00
20.02
20.02
0.05
20.01
0.01
0.00
20.03
0.02
0.06
0.02
0.06
15

DISCUSSION
Social class is increasingly recognized as a pow-
1.00
0.19
0.14
20.06
0.08
20.02
20.03
0.03
20.02
20.03
0.06
0.02
0.05
14

erful force in people’s lives. Yet our understanding of

the influence of social class and upward mobility at
work, including how executives’ formative child-
1.00
0.18
20.02
20.03
20.09
0.09
0.01
20.03
20.03
20.06
20.06
20.13
20.01
20.06

hood experiences with social class influence their

13

strategic choices, has been limited. In this paper, we

integrated theorizing from several literatures—upper
echelons, social class, and imprinting theory—to
Throughput background

suggest that top executives are likely to vary system-

Output background
Professional degree

atically in terms of their strategic risk-taking prefer-

CEO ownership
Options awards

ences because of different experiences with childhood

Munificence

Complexity

societal rank and access to resources, or their social

Dynamism
CPA/CFA
Founder

class imprint. Using a direct measure of CEOs’ per-

Female
Salary
Bonus

MBA

ceived social class origins, our results suggest that

social class imprints do indeed endure and exert
a lasting influence on executive decision making. We
13
14
15
16
17
18
19
20
21
22
23
24
25
26

found that the perceived social class origins of top

1628 Academy of Management Journal December

TABLE 2
Results of 2SLS Instrumental Variables Regression Models for Strategic Risk Takingt 1 1

Model 1 Model 2 Model 3 Model 4 Model 5

Control Main Elite General

model effects education management Full model

Strategic risk takingt 0.15*** 0.15*** 0.14*** 0.14*** 0.14***

Accounting performance (ROA) 0.01 0.01 0.01 0.01 0.00
Market performance 0.00 20.00 20.00 20.00 20.00
Firm size 0.00 0.01 0.01 0.01 0.01
Diversification 20.01 20.01 20.00 20.00 20.00
CEO age 0.00 0.00 0.00 0.00 0.00
CEO tenure 0.001 0.001 0.00 0.00* 0.00
Salary 0.07** 0.07** 0.07* 0.07** 0.07**
Bonus 0.03** 0.03* 0.04*** 0.03* 0.04**
Option awards 0.04 0.01 0.01 0.03 0.03
CEO ownership 20.10 20.10 20.07 20.10 20.08
Female 20.08** 20.08** 20.09** 20.08** 20.09**
Founder 20.11*** 20.10*** 20.08** 20.13*** 20.09***
MBA 20.04* 20.03 20.04* 20.02 20.04*
CPA/CFA 0.09*** 0.09*** 0.09*** 0.08*** 0.08**
Professional degree 0.04* 0.06** 0.09*** 0.05* 0.08***
Output background 0.20*** 0.21*** 0.22*** 0.20*** 0.21***
Throughput background 0.07*** 0.07*** 0.09*** 0.07*** 0.10***
Elite education 0.06 0.06 0.131 0.06 0.121
General management background 0.13*** 0.13*** 0.14*** 0.07** 0.10**
Munificence 20.12 20.11 20.171 20.12 20.181
Dynamism 0.01 20.08 20.09 20.10 20.11
Complexity 0.02 0.01 0.01 0.01 0.01
Upper social class origins 0.09*** 0.08*** 0.06** 0.05**
Lower social class origins 0.03* 0.03* 20.01 20.01
Upper social class origins 3 Elite education 20.00 20.01
Lower social class origins 3 Elite education 20.14* 20.14*
Upper social class origins 3 General management 0.10** 0.07*
background
Lower social class origins 3 General management 0.10*** 0.06*
background
Constant 20.19 20.32 20.28* 20.27 20.25*
F df 98 100 102 102 104
F 36.89*** 37.89*** 36.89*** 37.97*** 36.26***
N 1032 1032 1032 1032 1032
R2 0.79 0.80 0.80 0.81 0.81

Note: Industry, year, and CEO industry experience indicator variables are included in all models.
1
p , .10
* p , .05
** p , .01
*** p , .001

executives help to explain their risk-taking prefer- Furthermore, consistent with the idea that the
ences at the firm level years later, such that CEOs of manifestations of imprints can change over time, we
lower and upper social class origins take greater found that experiences with elite education and ca-
strategic risks than their middle class counterparts. reer diversity influenced the relationship between
Post hoc analyses also revealed that while differing executives’ perceived social class origins and stra-
from individuals of middle social class origins on tegic risk taking. First, our results suggest that CEOs
risk taking, CEOs of upper and lower social class from lower social class backgrounds become more
origins also differed from each other. In particular, risk averse as the amount of elite education in-
CEOs from the lower social classes engage in lower creases. However, contrary to our expectations, the
levels of strategic risk taking than their upper social level of elite education did not significantly influ-
class counterparts. ence the relationship between upper social class
2015 Kish-Gephart and Campbell 1629

FIGURE 1
Influence of Elite Education on Risk Taking among CEOs of Lower Social Class Origins

origins and risk taking. This finding may be explained between upper echelons and imprinting theorizing,
by the fact that, for those from the upper social classes, and suggests the importance of going beyond tradi-
obtaining an elite education is not uncommon or un- tional demographic variables (such as age or gender)
expected (Domhoff, 2002). As such, individuals from to consider how (and when) early experiences in-
the upper social classes may not perceive (or objec- fluence executives’ perceptions and decision mak-
tively accrue) any significant increases in their eco- ing (see also Crossland, Zyung, Hiller, & Hambrick,
nomic or social capital as a result of graduation from an 2014).
elite institution. Second, our study sheds new light on the con-
Second, presumably through its influence on ex- structs of elite education and executive functional
panding executives’ networks and knowledge base, background. Importantly, we provide empirical ev-
we found that a diverse functional (general manage- idence that elite education may not be an adequate
ment) background serves to amplify the relationship proxy when studying social class origins. Although
between CEO social class origins and strategic risk prior research has used elite education as a proxy for
taking among executives from both the lower and social class background (e.g., Westphal & Khanna,
upper social classes. In contrast to the asymmetric 2003), graduation from an elite college institution
effects of elite education, this finding suggests that does not necessarily equate to an upper social class
executives of lower and upper social class origins background. In our sample, 25% of the CEOs of up-
may be able similarly to capitalize on the networking per social class origins attended an elite un-
and informational (that is, social and cognitive) ben- dergraduate institution, compared to 10% and 14%
efits of a diverse functional background. of those from the lower and middle social classes,
respectively. By showing that the effect of elite edu-
cation is, in fact, contingent on social class back-
Implications for Research
ground, we demonstrate the promise of using
Our study has a number of implications for stra- primary data to assess executive social class origins,
tegic management and organization science re- despite the challenges inherent in collecting such
search. First, by integrating social class theorizing data. We also introduce the concept of executive
and imprinting theory with upper echelons work, diverse functional background—which we label
our study contributes new insights to the question of “general management background”—as another key
why, how, and when managerial characteristics im- contextual variable and show its utility for upper
pact firm outcomes and, specifically, strategic de- echelons research. Whereas prior research has
cision making. In particular, we demonstrate that highlighted the role that functional background di-
childhood social class matters. Despite some exec- versity plays in a team context, our study brings the
utives taking the “long road” to the corporate elite, focus to the individual (intrapersonal) level.
executives appear to hold on to the lessons and un- Third, our work contributes to research on social
derstandings garnered from their formative years. class diversity in organizations more broadly. De-
Moreover, our work points to the synergy that exists spite increasing interest in the topic, organizational
1630 Academy of Management Journal December

FIGURE 2
Influence of General Management Functional Background on Risk Taking among CEOs of (a) Upper Social
Class Origins and (b) Lower Social Class Origins

researchers still have much to learn about how social Finally, our study extends researchers’ under-
class influences individuals in the workplace. In standing of how and why executive risk-taking
calling for research to address this oversight, one preferences vary, offering a contribution to multiple
scholar admonished organizational researchers for theories in management scholarship. For example, by
making “the assumption that employees doff their introducing a novel construct and documenting its
social class when they come to work each day and influence on executive risk aversion with respect to
put it on again when they leave work” (Cotton, 1994: firm investments, our theory and findings directly
412). Our work helps to answer this and other calls to contribute to the normative stream of agency theory,
consider social class in organizations (see, e.g., Côté, which seeks to understand what influences execu-
2011; Fiske & Markus, 2012; Gray & Kish-Gephart, tives’ willingness to bear firm-specific risk (Eisenhardt,
2013; Leana et al., 2012; Scully & Blake-Beard, 2006). 1989). Thus, our findings point to several potential
We demonstrate that the individual’s social class applications of the concept of social class origins to
background is not checked at the organization’s other theoretical domains.
door; rather, individuals appear to import their early
experiences with material resources and societal
Future Research Opportunities
rank into organizations. Moreover, the effects of so-
cial class origins persist, influencing decisions in the Our study points toward several avenues for future
workplace even among the most successful in our research. One fruitful area for future research—and the
society—the corporate elite. most natural extension of this study—is to examine
2015 Kish-Gephart and Campbell 1631

other firm-level outcomes that may be influenced by dominated by middle and upper social class cultural
social class origins. For example, prior research sug- norms (Domhoff, 2002; Johnson et al., 2011; Stephens
gests that people in the middle and upper social et al., 2012). Future research might consider how ex-
classes tend to value independence and influence, ecutives’ experiences with upward mobility influ-
while those in the lower social classes value in- ence their sense of identity and connection to the
terdependence, integrity, and care for others (Kraus organization, as well as their interactions with co-
et al., 2012; Lucas, 2011). This is consistent with workers, leaders, and subordinates (see also Fiske &
empirical evidence demonstrating that individuals Markus, 2012; Gray & Kish-Gephart, 2013).
in the lower social classes tend to be more other-
oriented, charitable, and generous than individuals
Limitations
in the upper social classes (see, e.g., Piff et al., 2010).
Future research might consider whether and how As with all research, our study has limitations,
social class origins influence CEOs’ decisions about which offer further opportunities for future research.
employment programs (such as benefits or work First, because our study is set in a North American
arrangements) and cooperative strategies, such as context, future research might explore how our
strategic alliances (for example, joint ventures). findings translate to other social and institutional
Strategic alliances, for instance, may be used more settings. For example, the traditional social struc-
frequently by CEOs of lower social class origins who ture of the Indian society, with well-defined and
value working with others and thus may be more entrenched social class boundaries, may provide an
likely to view alliances as a value-creating strategy. interesting setting in which to study the effects of
We also speculate that the types of alliance partners social class background.
may differ (for example, in terms of status) between A second limitation of our study is our focus on the
CEOs from different social class backgrounds. CEO alone as a result of data availability. While re-
Future research could also examine the moder- cent studies point to a comparatively large CEO effect
ating effect of social class origins on executive fi- on firm performance (see, e.g., Hambrick & Quigley,
nancial or compensation incentives. Surprisingly, 2013; Mackey, 2008), especially in the United States
individual differences between executives and the (Crossland & Hambrick, 2007), collecting data on the
resulting motivational effects of incentive compen- social class background of other TMT members—
sation have yet to receive significant research atten- from S&P1500 firms or other populations—would
tion (Gerhart, Rynes, & Fulmer, 2009; Hambrick, allow for richer theoretical and empirical treatment
2007). This avenue seems especially promising given of executive social class roots. Specifically, a focus
our findings regarding differences in risk-taking pre- on the entire team would allow researchers to assess
ferences, because incentive compensation is often two dimensions—average TMT social class back-
issued with the goal of discouraging risk aversion and ground and the variance in TMT social class
incentivizing managers to take more risk than they background—reflective of a novel and unexplored
would otherwise be inclined to accept. Future work dimension of TMT diversity.
might examine whether CEOs of upper social class Finally, while we obtained direct measures of
origins are more sensitive to risk-inducing compen- perceived social class background, we were unable
sation incentives, such as stock options, because of to assess executive decision making directly. We
their lifelong experience with a financial “safety net.” thus encourage researchers to attempt a more prox-
In addition to the effects of compensation, social class imal approach to measuring executive decision
origins may also influence the structure of compen- making, such as scenario analysis, computer simu-
sation packages that CEOs are willing to accept (as lation (see Hambrick, 2007), or policy capturing (Hitt
their own compensation) and approve (as pay for & Tyler, 1991). Although the population of S&P1500
other TMT members). CEOs may be beyond reach for such an approach,
Lastly, our research points to upward mobility as smaller or regional firms (or even executive MBA
a rich area for future research in organizational stud- programs) may be a good starting point.
ies. As Zweigenhaft and Domhoff (1998: 184) noted,
social class has become “the factor most affecting
CONCLUSION
the likelihood of reaching the highest levels of the
institutional structure for all Americans.” Those Despite the appeal of “rags to riches” stories in
who are able to achieve upward mobility, especially American discourse, the long-term influence of so-
into the corporate elite, must navigate environments cial class origins on the upwardly mobile in the
1632 Academy of Management Journal December

workplace—especially those who have reached the Bourdieu, P. 1994. Structures, habitus, power: Basis for
heights of the socioeconomic hierarchy—has been a theory of symbolic power. In N. B. Dirks, G. Eley &
unclear. By integrating work on social class, im- S. B. Ortner (Eds.), Culture/power/history: A reader in
printing, and upper echelons theorizing, we provide contemporary social theory: 155–199. Princeton, NJ:
an important initial step toward understanding the Princeton University Press.
effects of the social class background of strategic Bromiley, P. 1991. Testing a causal model of corporate risk
leaders on firm-level outcomes. Overall, our results taking and performance. Academy of Management
demonstrate that early formative experiences with Journal, 34: 37–59.
social class matter. We thus strongly encourage fu- Bullock, H. E., & Lott, B. 2010. Social class and power. In
ture research related to upward mobility and the role A. Guinote & T. K. Vescio (Eds.), The social psychol-
of social class in the workplace. ogy of power: 408–427. New York: Guilford Press.
Bunderson, J. S. 2003. Team member functional back-
ground and involvement in management teams:
Direct effects and the moderating role of power cen-
REFERENCES tralization. Academy of Management Journal, 46:
Adler, N. E., & Snibbe, A. C. 2003. The role of psychosocial 458–474.
processes in explaining the gradient between socio- Bunderson, J. S., & Sutcliffe, K. M. 2002. Comparing al-
economic status and health. Current Directions in ternative conceptualizations of functional diversity in
Psychological Science, 12: 119–123. management teams: Process and performance effects.
Adler, N. E., Epel, E. S., Castellazzo, G., & Ickovics, J. R. Academy of Management Journal, 45: 875–893.
2000. Relationship of subjective and objective social Cannella, A. A., Park, J. H., & Lee, H. U. 2008. Top man-
status with psychological and physiological func- agement team functional background diversity and
tioning: Preliminary data in healthy white women. firm performance: Examining the roles of team
Health Psychology, 19: 586–592. member colocation and environmental uncertainty.
Anderson, C., & Galinsky, A. D. 2006. Power, optimism and Academy of Management Journal, 51: 768–784.
risk-taking. European Journal of Social Psychology, Cao, Q., Simsek, Z., & Jansen, J. J. P. 2012. CEO social
36: 511–536. capital and entrepreneurial orientation of the firm:
Bascle, G. 2008. Controlling for endogeneity with in- Bonding and bridging effects. Journal of Management.
strumental variables in strategic management re- Published online ahead of print. [DOI: 10.1177/
search. Strategic Organization, 6: 285–327. 0149206312469666].
Bateman, T. S., & Zeithaml, C. P. 1989. The psychological Carpenter, M. A., Geletkanycz, M. A., & Sanders, W. G.
context of strategic decisions: A model and conver- 2004. Upper echelons research revisited: Anteced-
gent experimental findings. Strategic Management ents, elements, and consequences of top manage-
Journal, 10: 59–74. ment team composition. Journal of Management,
30: 749–778.
Bednar, M. K. 2012. Watchdog or lapdog? A behavioral
view of the media as a corporate governance mech- Certo, S. T., & Semadeni, M. 2006. Strategy research and
anism. Academy of Management Journal, 55: panel data: Evidence and implications. Journal of
131–150. Management, 32: 449–471.
Begley, T. M. 1995. Using founder status, age of firm, Chatterjee, A., & Hambrick, D. C. 2007. It’s all about me:
company growth rate as the basis for distinguishing Narcissistic chief executive officers and their effects
entrepreneurs from managers of smaller businesses. on company strategy and performance. Administra-
Journal of Business Venturing, 10: 249–263. tive Science Quarterly, 52: 351–386.
Bianchi, E. C. 2013. The bright side of bad times: The Christie, A., & Barling, J. 2009. Disentangling the indirect
affective advantages of entering the workforce in links between SES and health: The dynamic roles of
a recession. Administrative Science Quarterly, 58: work stressors and personal control. The Journal of
587–623. Applied Psychology, 94: 1466–1478.
Bigley, G. A., & Wiersema, M. F. 2002. New CEOs and Clair, J. A., Beatty, J. E., & MacLean, T. L. 2005. Out of sight
corporate strategic refocusing: How experience as heir but not out of mind: Managing invisible social iden-
apparent influences the use of power. Administrative tities in the workplace. Academy of Management
Science Quarterly, 47: 707–727. Review, 30: 78–95.
Bourdieu, P. 1984. Distinction: A social critique of the Côté, S. 2011. How social class shapes thoughts and actions
judgment of taste. Cambridge, MA: Harvard Univer- in organizations. Research in Organizational Be-
sity Press. havior, 31: 43–71.
2015 Kish-Gephart and Campbell 1633

Cotton, C. 1994. Social class as a neglected variable in or- How societal rank influences interaction: 234–251.
ganizational behavior. The Journal of Psychology, New York: Russell Sage Foundation.
128: 409–417. Freeland, C. 2011. The rise of the new global elite. The
Crossland, C., & Hambrick, D. C. 2007. How national sys- Atlantic (January/February). Available online at http://
tems differ in their constraints on corporate execu- www.theatlantic.com/magazine/archive/2011/01/the-
tives: A study of CEO effects in three countries. rise-of-the-new-global-elite/308343/.
Strategic Management Journal, 28: 767–789.
Fussell, P. 1983. A guide through the American status
Crossland, C., Zyung, J., Hiller, N., & Hambrick, D. C. 2014. system. New York: Touchstone.
CEO career variety: Effects on firm-level strategic and
Geletkanycz, M. A., & Black, S. S. 2001. Bound by the
social novelty. Academy of Management Journal,
past? Experience-based effects on commitment to the
57: 652–674.
strategic status quo. Journal of Management, 27:
Cycyota, C. S., & Harrison, D. A. 2006. What (not) to expect 3–21.
when surveying executives: A meta-analysis of top
manager response rates and techniques over time. Gerhart, B., Rynes, S. L., & Fulmer, I. S. 2009. Pay and
Organizational Research Methods, 9: 133–160. performance: Individuals, groups, and executives.
The Academy of Management Annals, 3: 251–315.
Davis, N., & Robinson, R. 1988. Class identification of men
and women in the 1970s and 1980s. American So- Giuliano, P., & Spilimbergo, A. 2009. Growing up in a re-
ciological Review, 53: 103–112. cession: Beliefs and the macroeconomy. National
Bureau of Economic Research Working Paper No.
Devers, C. E., McNamara, G., Wiseman, R. M., & Arrfelt, M.
15321. Available online at http://www.nber.org/
2008. Moving closer to the action: Examining com-
pensation design effects on firm risk. Organization papers/w15321.
Science, 19: 548–566. Gray, B., & Kish-Gephart, J. J. 2013. Encountering social
Domhoff, G. W. 1970. The Higher Circles. New York: class differences at work: How “class work” perpetu-
Random House. ates inequality. Academy of Management Review,
38: 670–699.
Domhoff, G. W. 2002. Who rules America? Power and
politics. Boston, MA: McGraw-Hill. Hall, C. C. 2012. Behavioral decision research, social class,
and implications for public policy. In S. T. Fiske &
Domhoff, G. W. 2010. Who rules America: Challenges to
H. R. Markus (Eds.), Facing social class: How societal
corporate and class dominance. New York: McGraw-
rank influences interaction: 175–194. New York:
Hill.
Russell Sage Foundation.
Ehrenreich, B. 1989. Fear of falling: The inner life of the
middle class. New York: Pantheon Books. Hambrick, D. C. 2007. Upper echelons theory: An update.
Academy of Management Review, 32: 334–343.
Eisenhardt, K. 1989. Agency theory: An assessment and
review. Academy of Management Review, 14: Hambrick, D. C., & Mason, P. A. 1984. Upper echelons:
57–74. The organization as a reflection of its top managers.
Academy of Management Review, 9: 193–206.
Finkelstein, S. 1992. Power in top management teams:
Dimensions, measurement, and validation. Academy Hambrick, D. C., & Quigley, T. J. 2013. Toward more ac-
of Management Journal, 35: 505–538. curate contextualization of the CEO effect on firm
performance. Strategic Management Journal, 35:
Finkelstein, S., & Hambrick, D. C. 1990. Top-management-
473–491.
team tenure and organizational outcomes: The mod-
erating role of managerial discretion. Administrative Haunschild, P. R., Henderson, A., & Davis-Blake, A. 1998.
Science Quarterly, 35: 484–503. CEO demographics and acquisitions: Network and
Finkelstein, S., Hambrick, D. C., & Canella, A. A. 2009. cognitive effects of educational and functional
Strategic leadership: Theory and research on ex- background. In A. J. Leenders & S. M. Gabbay (Eds.),
ecutives, top management teams, and boards. New Corporate social capital: 266–283. New York:
York: Oxford University Press. Addison-Wesley.
Fiske, S. T., & Markus, H. R. 2012. Facing social class: How Hayward, M., & Hambrick, D. C. 1997. Explaining the
societal rank influences interaction. New York: premiums paid for large acquisitions: Evidence of
Russell Sage Foundation. CEO hubris. Administrative Science Quarterly, 42:
Fiske, S. T., Moya, M., Russell, A. M., & Bearns, C. 2012. 103–127.
The secret handshake: Trust in cross-class encounters. Higgins, M. C. 2005. Career imprints: Creating leaders
In S. T. Fiske & H. R. Markus (Eds.), Facing social class: across an industry. San Francisco, CA: Jossey-Bass.
1634 Academy of Management Journal December

Hitt, M. A., & Tyler, B. B. 1991. Strategic decision models: Kusserow, A. 2012. When hard and soft class: Class-based
Integrating different perspectives. Strategic Man- individualism in Manhattan and Queens. In S. T. Fiske
agement Journal, 12: 327–351. & H. R. Markus (Eds.), Facing social class: How soci-
Holburn, G. L., & Zelner, B. A. 2010. Political capabilities, etal rank influences interaction: 195–215. New York:
policy risk, and international investment strategy: Russell Sage Foundation.
Evidence from the global electric power genera- Lamm, H., Schmidt, R. W., & Trommsdorff, G. 1976. Sex
tion industry. Strategic Management Journal, 31: and social class as determinants of future orientation
1290–1315. in adolescents. Journal of Personality and Social
Hoskisson, R. E., Hitt, M. A., Johnson, R. A., & Moesel, D. D. Psychology, 74: 763–773.
1993. Construct validity of an objective (entropy) cat- Leana, C. R., Mittal, V., & Stiehl, E. 2012. Organizational
egorical measure of diversification strategy. Strategic behavior and the working poor. Organization Sci-
Management Journal, 14: 215–235. ence, 23: 888–906.
Hout, M. 2008. How class works: Objective and subjective Liu, W., Ali, S., Soleck, G., Hopps, J., Dunston, K., & Pickett,
aspects of class since the 1970s. In A. Lareau & J. T. 2004. Using social class in counseling psychology
D. Conley (Eds.), Social class: How does it work?: research. Journal of Counseling Psychology, 51:
25–64. New York: Russell Sage Foundation. 3–18.
Jackman, M. R., & Jackman, R. W. 1973. An interpretation Lopes, L. L. 1987. Between hope and fear: The psychology
of the relation between objective and subjective of risk. Advances in Experimental Social Psychol-
social status. American Sociological Review, 38: ogy, 20: 255–295.
569–582.
Lubrano, A. 2004. Limbo: Blue-collar roots, white-collar
Johnson, S. E., Richeson, J. A., & Finkel, E. J. 2011. Middle dreams. Hoboken, NJ: Wiley.
class and marginal? Socioeconomic status, stigma,
Lucas, K. 2011. The working class promise: A commu-
and self-regulation at an elite university. Journal of
nicative account of mobility-based ambivalences.
Personality and Social Psychology, 100: 838–852.
Communication Monographs, 78: 347–369.
Kahneman, D., & Tversky, A. 1979. Prospect theory: An
MacCrimmon, K. R., & Wehrung, D. A. 1986. Taking risks:
analysis of decision under risk. Econometrica, 47:
The management of uncertainty. New York: Free
263–291.
Press.
Keats, B. W., & Hitt, M. A. 1988. A causal model of linkages
Mackey, A. 2008. The effect of CEOs on firm performance.
among environmental dimensions, macro organiza-
Strategic Management Journal, 29: 1357–1367.
tional characteristics, and performance. Academy of
Management Journal, 31: 570–598. Mahony, P., & Zmroczek, C. 1997. Why class matters. In
P. Mahoney & C. Zmroczek (Eds.), Class matters:
Keltner, D., Gruenfeld, D. H., & Anderson, C. 2003. Power,
“Working-class” women’s perspectives on social
approach and inhibition. Psychological Review, 110:
class: 1–7. London: Taylor & Francis.
265–284.
Malmendier, U., & Nagel, S. 2011. Depression babies: Do
Kittleson, M. M., Meoni, L. A., Wang, N., Chu, A. Y., Ford,
macroeconomic experiences affect risk taking? The
D. E., & Klag, M. J. 2006. Association of childhood
Quarterly Journal of Economics, 126: 373–416.
socioeconomic status with subsequent coronary heart
disease in physicians. Archives of Internal Medicine, Malmendier, U., & Tate, G. 2005. Does overconfidence af-
166: 2356–2361. fect corporate investment? CEO overconfidence mea-
sures revisited. European Financial Management,
Kniffin, K. M. 2007. Accessibility to the PhD and pro-
11: 649–659.
fessoriate for first-generation college graduates: Review
and implications for students, faculty, and campus Malmendier, U., & Tate, G. 2008. Who makes acquisitions?
policies. American Academic, 3: 49–70. CEO overconfidence and the market’s reaction. Jour-
Kraus, M. W., & Keltner, D. 2009. Signs of socioeconomic nal of Financial Economics, 89: 20–43.
status: A thin-slicing approach. Psychological Sci- March, J. G., & Shapira, Z. 1987. Managerial perspectives
ence, 20: 90–106. on risk and risk taking. Management Science, 33:
Kraus, M. W., Piff, P. K., Mendoza-Denton, R., Rheinschmidt, 1404–1418.
M. L., & Keltner, D. 2012. Social class, solipsism, and Marin, T. J., Chen, E., & Miller, G. E. 2008. What do tra-
contextualism: How the rich are different from the poor. jectories of childhood socioeconomic status tell us
Psychological Review, 119: 546–572. about markers of cardiovascular health in adoles-
Kraus, M. W., Tan, J. J. X., & Tannenbaum, M. B. 2013. The cence? Psychosomatic Medicine, 70: 152–159.
social ladder: A rank-based perspective on social Markus, H. R., & Fiske, S. T. 2012. Introduction: A wide-
class. Psychological Inquiry, 24: 91–96. angle lens on the psychology of social class. In
2015 Kish-Gephart and Campbell 1635

S. T. Fiske & H. R. Markus (Eds.), Facing social class: PBS. 2001. People like us: Social class in America. Docu-
How societal rank influences interaction: 1–12. New mentary transcript. Available online at http://www.
York: Russell Sage Foundation. pbs.org/peoplelikeus/resources/transcript.pdf.
Marmot, M. G. 2004. The status syndrome: How social Pfeffer, J. 1983. Organizational demography. Research in
standing affects our health and longevity. New York: Organizational Behavior, 5: 299–357.
Henry Holt & Co. Piff, P. K., Kraus, M. W., Côté, S., Cheng, B. H., & Keltner, D.
Marquis, C., & Tilcsik, A. 2013. Imprinting: Toward 2010. Having less, giving more: The influence of social
a multilevel theory. The Academy of Management class on prosocial behavior. Journal of Personality
Annals, 7: 193–243. and Social Psychology, 99: 771–784.

Martin, G. P., Gomez-Mejia, L. R., & Wiseman, R. 2013. Ragins, B. R., & Cornwall, J. M. 2001. Walking the line: Fear
Executive stock options as mixed gambles: Revisiting and disclosure of sexual orientation in the workplace.
Paper presented at the Annual Meeting of the Acad-
the behavioral agency model. Academy of Manage-
emy of Management, Washington, DC.
ment Journal, 56: 451–472.
Raskas, D. F., & Hambrick, D. C. 1992. Multifunctional
Miller, G. E., Chen, E., Fok, A. K., Walker, H., Lim, A.,
managerial development: A framework for evaluating
Nicholls, E. F., Cole, S., & Kobor, M. S. 2009. Low early
the options. Organizational Dynamics, 21: 5–17.
life social class leaves a biological residue manifested
by decreased glucocorticoid and increased proin- Resnick, S., & Wolff, R. 2003. The diversity of class anal-
flammatory signaling. Proceedings of the National ysis: A critique of Erik Olin Wright and beyond. Crit-
ical Sociology, 29: 7–28.
Academy of Sciences of the United States of Amer-
ica, 106: 14716–14721. Sanders, W. G., & Hambrick, D. 2007. Swinging for the
fences: The effects of CEO stock options on company
Miller, P. J., Cho, G. E., & Bracey, J. R. 2005. Working-class
risk taking and performance. Academy of Manage-
children’s experience through the prism of personal
ment Journal, 50: 1055–1078.
storytelling. Human Development, 48: 115–135.
Schwalbe, M. 2008. Rigging the game: How inequality is
Mills, C. W. 1951. White collar: The American middle reproduced in everyday life. New York: Oxford
classes. New York: Oxford University Press. University Press.
Murray, M. P. 2006. Avoiding invalid instruments and Scott, J. 2005. Life at the top in America isn’t just better, it’s
coping with weak instruments. The Journal of Eco- longer. New York Times, May 16. Available online at
nomic Perspectives, 20: 111–132. http://www.nytimes.com/2005/05/16/us/class/life-at-
Nagi, S. Z. 1963. Status profile and reactions to status the-top-in-america-isnt-just-better-its-longer.html?_r50.
threats. American Sociological Review, 28: 440–443. Scully, M. A., & Blake-Beard, S. 2006. Locating class in
Ostrove, J., & Stewart, A. 1998. Representing Radcliffe: organizational diversity work: Class as structure, style
Perceptions and consequences of social class. Journal and process. In A. M. Konrad, P. Prasad & J. Pringle
of Adult Development, 5: 183–193. (Eds.), Handbook of workplace diversity: 431–454.
London: Sage.
Ozgen, E., & Baron, R. A. 2007. Social sources of in-
formation in opportunity recognition: Effects of men- Shipler, D. K. 2004. The working poor: Invisible in
tors, industry networks, and professional forums. America. New York: Knopf.
Journal of Business Venturing, 22: 174–192. Skeggs, B. 2004. Class, self, culture. New York: Routledge.
Palmer, D., & Barber, B. M. 2001. Challengers, elites, and Smart, L., & Wegner, D. M. 2000. The hidden costs of hid-
owning families: A social class theory of corporate den stigma. In T. F. Heatherton, R. E. Kleck, M. R. Hebl
acquisitions in the 1960s. Administrative Science & J. G. Hull (Eds.), The social psychology of stigma:
Quarterly, 46: 87–120. 220–241. New York: Guilford Press.

Palmer, D. A., Jennings, P. D., & Zhou, X. 1993. Late Smith, E. B., Menon, T., & Thompson, L. 2012. Status dif-
adoption of the multidivisional form by large ferences in the cognitive activation of social networks.
U.S. corporations: Institutional, political, and eco- Organization Science, 23: 67–82.
nomic accounts. Administrative Science Quarterly, Smith, K. K. 1982. Conflict in organizations: Prisons in
38: 100–131. disguise. Dubuque, IO: Kendall-Hunt.
Palmer, T. B., & Wiseman, R. M. 1999. Decoupling risk Staiger, D., & Stock, J. H. 1997. Instrumental variables regres-
taking from income stream uncertainty: A holistic sion with weak instruments. Econometrica, 65: 557–586.
model of risk. Strategic Management Journal, 20: StataCorp. 2011. Stata 12 base reference manual. College
1036–1062. Station, TX: Stata Press.
1636 Academy of Management Journal December

Steel, C. M., & Aronson, J. 1995. Stereotype threat and the Wiersema, M. A., & Bantel, K. A. 1993. Top management
intellectual test performance of African Americans. team turnover as an adaptation mechanism: The role
Journal of Personality and Social Psychology, 69: of the environment. Strategic Management Journal,
797–811. 14: 485–504.
Stephens, N. M., Fryberg, S. A., Markus, H. R., Johnson, Williams, J. C. 2012. The class culture gap. In S. T. Fiske &
C. S., & Covarrubias, R. 2012. Unseen advantage: H. R. Markus (Eds.), Facing social class: How societal
How American universities’ focus on independence
rank influences interaction: 39–58. New York: Russell
undermines the academic performance of first-
Sage Foundation.
generation college students. Interpersonal Relations
& Group Processes, 102: 1178–1197. Wiseman, R. M. 2009. On the use and misuse of ratio
Stephens, N. M., Markus, H. R., & Phillips, L. T. 2014. So- measures. In D. Bergh & D. Ketchen (Eds.), Research
cial class culture cycles: How three gateway contexts methodology in strategic management: 75–110. Bing-
shape selves and inequality. Annual Review of Psy- ley: Emerald Group.
chology, 65: 611–634. Wooldridge, J. M. 2002. Econometric analysis of cross
Stephens, N. M., Markus, H. R., & Townsend, S. S. M. 2007. section and panel data. Cambridge, MA: MIT Press.
Choice as an act of meaning: The case of social class. Zweigenhaft, R. L., & Domhoff, G. W. 1998. Diversity in
Journal of Personality and Social Psychology, 93: the power elite: Have women and minorities
814–830.
reached the top? New Haven, CT: Yale University
Stinchcombe, A. L. 1965. Social structure and organiza- Press.
tions. In J. G. March (Ed.), Handbook of organizations:
142–193. Chicago, IL: Rand-McNally.
Tihanyi, L., Ellstrand, A. E., Daily, C. M., & Dalton, D. R.
2000. Composition of the top management team and
Jennifer J. Kish-Gephart ([email protected]) is an assis-
firm international diversification. Journal of Man-
tant professor of management at the Walton College of
agement, 26: 1157–1177.
Business, University of Arkansas. She received her PhD in
Tudor, J. F. 1971. The development of class awareness in organizational behavior from Pennsylvania State Univer-
children. Social Forces, 49: 470–476. sity. She is broadly interested in social issues in man-
Twenge, J. M., & Campbell, W. K. 2002. Self-esteem and agement, with a particular focus on social class diversity,
socioeconomic status: A meta-analytic review. Per- inequality, and behavioral ethics.
sonality and Social Psychology Review, 6: 59–71. Joanna Tochman Campbell ([email protected]) is
Waldman, D. A., Javidan, M., & Varella, P. 2004. Charismatic an assistant professor of management at the Lindner Col-
leadership at the strategic level: A new application lege of Business, University of Cincinnati. She holds a PhD
of upper echelons theory. The Leadership Quarterly, in strategic management from Texas A&M University. Her
15: 355–380. research interests include top executive characteristics
Westphal, J. D., & Khanna, P. 2003. Keeping directors in and their effects on firm outcomes, corporate governance
line: Social distancing as a control mechanism in the issues, stakeholder strategy, and configurational methods.
corporate elite. Administrative Science Quarterly,
48: 361–398.
Copyright of Academy of Management Journal is the property of Academy of Management
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.