ArticlePDF Available

The role of social proximity in professional CEO appointments: Evidence from caste/religion-based hiring of CEOs in India

The role of social proximity in professional CEO
appointments: Evidence from caste/religion-based
hiring of CEOs in India
Naga Lakshmi Damaraju
| Anil K. Makhija
Strategy Area, Indian School of Business,
Hyderabad, India
Department of Finance, Fisher College of
Business, The Ohio State University,
Columbus, Ohio
Naga Lakshmi Damaraju, 8127 AC8 Level 1,
Indian School of Business, Hyderabad 500 111,
Research Summary: The role of homophily in CEO
appointments at the largest corporations is an important
subject in corporate governance. This subject is particu-
larly important in a country like India where a multitude
of religions, castes, and communities form its social fab-
ric. We test for the role of homophily in professional CEO
appointments in India by empirically examining the pref-
erence for same caste/religion CEOs by the largest firms.
Using a unique dataset, assembled by detailed identifica-
tion of castes/religions from family names and counterfac-
tuals obtained through the Coarsened Exact Matching
technique, we find that caste/religion plays a crucial role
in CEO selection as a source of information (positive dis-
crimination). The evidence is not consistent with its use to
pursue taste-based preferences (negative discrimination).
Managerial Summary: We test for the role of homophily in
the appointments of CEOs in India by empirically examining
the preference for same caste/religion CEOs by the largest
firms. We find that caste/religion plays an important role in
CEO selection, i.e., as a form of information or positive dis-
crimination.The evidence is not consistent with its use to
pursue taste-based preferences or negative discrimination.
caste/religion in India, CEO appointments, corporate
governance, homophily, social proximity
CEOs are an important part of corporate governance in organizations. Consequently, there has been
substantial research interest in understanding factors affecting CEO selection, succession, and
Received: 15 September 2015 Revised: 17 February 2018 Accepted: 19 February 2018 Published on: 3 April 2018
DOI: 10.1002/smj.2787
Strat Mgmt J. 2018;39:20512074. Copyright © 2018 John Wiley & Sons, Ltd. 2051
performance (Hambrick & Mason, 1984; Quigley & Hambrick, 2015; Tian, Haleblian, & Rajagopa-
lan, 2011; Zajac, 1990). Most of this research has focused on understanding the effects of CEOs
demographic characteristics, whether a CEO is an insider or outsider (Zhang & Rajagopalan, 2010),
the influence of industry structure and firm-specific attributes (Datta & Rajagopalan, 1998; Rajago-
palan & Datta, 1996; Zhang & Rajagopalan, 2003), in addition to different network influences on
CEO selection (Ocasio, 1994; Tian et al., 2011). There has been considerable interest among finance
scholars as well in this regard (e.g., Parrino, 1997).
While the role of industry backgrounds, experience, education, and demographic characteristics
have been explored in this literature (e.g., Datta & Rajagopalan, 1998; Tian et al., 2011), there are
other interesting but unexamined aspects of homophily,i.e., the extent of social proximity by vir-
tue of belonging to particular caste/religion, that could play a role in CEO appointments (Guiso,
Sapienza, & Zingales, 2009; Lazarsfeld & Merton, 1954; McPherson, Smith-Lovin, & Cook, 2001).
Such homophily can be for positive performance-enhancing reasons, e.g., common value systems,
enhanced trust and information flows, etc., (Guiso et al., 2009), or for purely taste-basedreasons
that can adversely affect performance (Becker, 1957). Consequently, both the underlying reasons for
same caste/religion influences in CEO hiring and the performance implications of such influences
are unclear.
In this paper, we examine the influence of social proximity on CEO appointments using a novel
dataset from one of the largest developing economies in the world, India. India offers a natural set-
ting to test social proximity-based explanations in view of the multitude of religions, castes, and
communities that form its social fabric. We also test competing hypotheses regarding performance
consequences of CEO selections based on social proximity.
Our dataset covers the largest 1,000 publicly listed firms from 2001 to 2009. We focus on profes-
sional CEO appointments in the private sector to study the influence of Hindu castes and other reli-
We document a significant in-grouppreference in professional CEO hiring by firm owners/
chairpersons for the Indian firms. However, this preference could be due to informationreasons, or
purely taste-basedreasons, or both. We then empirically distinguish the informationreasons by
examining the same caste/religion preference between insidercandidates and outsidercandidates
to the firm. Results show, across a broad range of firm categories, that the preference for in-group
hiring is lesser in case of insidercandidates compared to outsidercandidates. These results sup-
port the information hypothesis,since informational and trust advantages are less likely for a group
of candidates who are already known to the firm. Also, there is no evidence for adverse effects of
in-groupCEO selections on firm performance, again across a broad variety of firm samples. Thus,
the evidence from this study supports the information hypothesisand not a taste-basedexplana-
tion for same caste/religion CEO selection among the largest publicly traded firms in India.
Homophily is the principle that similar people are socially proximate, interacting more frequently
among themselves than they do with dissimilar people.
Many different characteristics define simi-
larity, e.g., race, ethnicity, gender, age, education, occupation, and social class, and these character-
istics serve as bases for homophily (McPherson et al., 2001). The effects of homophily have been
Government-owned institutions are likely to be influenced by different decision-making criteria for CEO appointments than private
sector firms. Therefore, we focus only on private sector firms termed non-PSUs (non-Public Sector Undertakings) in this paper.
This idea is popularly expressed as birds of a feather flock together,people love those who are like themselves(Aristotle, 1934,
p. 1155), similarity begets friendship(Plato, 1968, p. 837), etc.
studied in a wide variety of contexts, e.g., the classic study of propinquity effect on selection of
spouses in Philadelphia (Bossard, 1932); friendship networks among children (Shrum, Cheek Jr, &
Hunter, 1988), college students (Lewis, Kaufman, Gonzalez, Wimmer, & Christakis, 2008), and
adults (Marsden, 1988); and the coadoption of technology services by friends (Aral, Muchnik, &
Sundararjan, 2009), to cite a few.
Organizational phenomena, such as venture capital investments (Gompers, Vladimir, & Yuhai,
2016; Hegde & Tumlinson, 2014), alliance formation (Ahuja, Polidoro Jr, & Mitchell, 2009), gender
differences and their impact in the workplace (Ibarra, 1992), CEO advice networks (McDonald &
Westphal, 2003), and entrepreneurial phenomenon such as the structure of founding teams (Ruef,
Aldrich, & Carter, 2003), have also seen influential research based on the principle of homophily
and provide evidence that birds of a feather flock together.Prior studies document both positive
and negativeeffects of homophily-based decisions in different contexts (e.g., Hegde and Tumlin-
son [2014] and Gompers et al. [2016] in the context of venture capital investments).
Given its pervasive nature, prescribing an individuals choices ranging from the mundane
(e.g., dietary practices) to the far-reaching (e.g., choice of marriage partner), caste too can likely
engender social closeness. In terms of the classic work on homophily by Lazarsfeld and Merton
(1954), caste may engender strong homophily because it entails similarity in both social status and
the value systems of individuals.
2.1 |The caste system in India and corporate decision-making
With its true beginning and purpose shrouded in ancient history, the caste system in India dates back
to at least 3000 years and is mentioned in the canonical sacred Hindu text, the Rig Veda, from circa
1500 B.C. The traditional Hindu religion comprises four castes (Varnas). They are Brahmin (priests
and scholars), Kshatriya (warriors), Vysya (merchants), and Shudra (artisans, farmers, and other
workers). Over time, the traditional castes have morphed into a complex pattern of many caste-based
communities, or Jatis. The caste system, which was perhaps originally a means for division of labor
classifying individuals by professions they pursue, came to imply a hierarchy with the Brahmins at
the top, followed by Kshatriya, Vysya, and Shudra, in that order. There was yet another group
which were termed the fifthvarna or the out of caste.These were the untouchables.
caste system has been allegedly the source of massive social and economic discrimination. A num-
ber of influential political parties derive their support by pursuing agendas that cater to certain castes
and have proposed extending the caste-based job quotas to the private sector.
There is a large literature on the caste system (and the religions) of India from sociological and
anthropological perspectives (e.g., Banerjee & Knight, 1985; Borooah & Iyer, 2005; Hnatkovska,
Lahiri, & Paul, 2012). A few studies also examine the economic consequences of caste (Anderson,
2011), and a handful of studies examine a corporate decision, e.g., recruitments, with mixed results
for the impact of caste/religion (e.g., Banerjee, Bertrand, Datta, & Mullainathan, 2009; Jodhka,
2008; Siddique, 2008). Nevertheless, many studies of caste deal with relatively small survey-based
Over six decades ago, soon after Indias independence, the Constitution of India explicitly produced schedules of the most disadvan-
taged castes (the fifthvarna or out of caste)asScheduled Castes and Scheduled Tribes,(SC/ST). This is the group commonly
referred to as Dalits (the oppressed) today and was called Harijans (children of God) by Mahatma Gandhi. The government also
generically noted other lower castes as Other Backward Classes (OBC)as targets of affirmative action through job reservation in
the public sector and seat reservation in government-aided educational institutions. Thus, the percentage of seats in higher educational
state-related institutes reserved for SC, ST, plus OBC, today stands at 49.5% in many states of India, prompting members of some
castes to seek official low caste status (De Zwart, 2000). Otherwise, in an attempt to engender a caste-less society, India has eschewed
any recognition of upper castes in its dealings.
samples or at an aggregate level of broadly distinguishing between forward/upperand backward/
Breakdowns of population, job, and other economic data are not readily available across all
castes, in particular on upper castes which are also called Forward Castes (i.e., Brahmin, Kshatriya,
and Vysya). Very little reliable data or analyses are available about the role of castes in the private
sector, which is allegedly dominated by the Forward Castes. Thus far, caste-based research has been
very challenging to undertake since one must develop primary data for caste identification.
Of course, Chen, Chittoor, and Vissa (2015), Fisman, Paravisini, and Vig (2017), and Vissa
(2011) use a dyadic approach and study caste-based homophily influences on lending by state-
owned banks, entrepreneurial tie-formation intentions, and accuracy of analystsearnings forecasts,
respectively. Our approach is closest to these studies which adopt a dyadic methodology. To our
knowledge, this paper is the first study of the role of caste in CEO selection.
While more than 80% of the roughly1.25 billion population of India is Hindu, there are large
numbers of adherents of other religions, including Muslims, Christians, Jains, Sikhs, and Parsees
(or Zorastrians). We do not study caste among non-Hindus of India, although it has been claimed
that they too practice caste discrimination, perhaps as lingering vestiges of their sometimes pre-
conversion Hindu ancestry.
2.2 |How caste/religion affects CEO hiring
2.2.1 |The case for positive discrimination
As outlined earlier, social proximity (or homophily) implies individuals similar in their personal
or social characteristics and sharing a common identity (e.g., race, gender, culture) tend to have
common interests and worldviews, which explains the formation of ties based on interpersonal
attraction (Lazarsfeld & Merton, 1954; McPherson & Smith-Lovin, 1987). Similarity engenders trust
and understanding and greater levels of social affiliation than would be expected among dissimilar
individuals. In-groupindividuals are likely to be perceived as cooperative, honest, and trustworthy
compared to out-groupindividuals (Brewer, 1979; Ruef et al., 2003). Also, similarity in interests,
shared values, and worldviews can facilitate communication leading to enhanced interactions than
would be the case when individuals differ widely on these dimensions.
For these reasons, caste (or religion)-based social proximity can also lead to formation of expres-
sive ties based on interpersonal attraction that foster trust and reciprocity. Extant studies
(e.g., Menning [1997] on textile industry in Surat in North India and Rudner [1994] on Chettiar
community in South India) provide empirical evidence for trust as the underlying mechanism for
inter-firm ties that are based on common caste affiliations. Guiso et al. (2009) argue and provide
evidence that cultural similarity, be it religious, genetic or somatic, can improve bilateral trust, lead-
ing to more trade.
Bird (1989) suggests homophily preferences should be more pronounced in settings which
require sizeable investments of time and resources, e.g., organizational founding teams. CEO
appointments are closest in this sense. Consequently, we hypothesize the following for CEO hires,
where same-caste/religion CEOs are those with same caste or religion as that of the owners/
Indeed, the last formal enumeration of the population by all castes, other than for the SC, ST, and OBC that are the targets of posi-
tive discrimination by constitutional mandate, goes back to the census of 1931 under British Raj. Finally, the government recently
acknowledged that the issue will not disappear by looking away and is undertaking the first ever caste-based census survey in inde-
pendent India. Even the census will only partially ease caste-related research, helping with estimates of caste populations but not with
the crucial problem of identifying castes of subjects in a study.
chairpersons of the hiring firm (Caste assignments to family and non-family firms are discussed later
in the Methodology section).
Positive discrimination hypothesis
Hypothesis 1 (H1) Information reasons lead to an abnormally high incidence of same
caste/religion CEO hires.
2.2.2 |The case for negative discrimination
Homophily or in-groupdiscrimination based on taste preferences also has a large literature, start-
ing with Becker (1957). Taste-based preferences too predict the selection of in-group CEOs. In the
process of exercising taste-based preferences, individuals may sacrifice quality, either consciously or
unconsciously, for the opportunity to work with similar others, which maximizes their personal util-
ity (Gompers et al., 2016).
In a rigid traditional sense, caste has a pervasive influence on the individual, giving rise to strong
taste preferences for forming personal associations between members of the same caste. Caste can
affect dietary practices, whose touch is polluting,marriage choices, limits on associations with
other castes, religious rituals, etc.
This pervasive influence of caste/religion can extend to the
choice of CEO candidates by owners of the firm for purely taste-based reasons. This is akin to the
argument by Khurana (2002) that the CEO labor market does not fit the traditional view of a market
in that CEOs are selected, not on the basis of supply and demand, but on non-economic
Negative discrimination hypothesis
Hypothesis 2 (H2) Taste-based preferences lead to an abnormally high incidence of
samecaste/religion CEO hires.
In contrast to H1 and H2, the performance outcomes for firms can be very different based on the
underlying drivers of caste/religion-based hiring of CEO candidates.
2.3 |How caste/religion-based CEO hiring effects performance
2.3.1 |The effect of positive discrimination
It has been argued that social proximity can lead to economically efficient outcomes. Since
social homogeneity is likely to facilitate communication due to shared knowledge, values, atti-
tudes, and vocabulary (Huston & Levinger, 1978; Rogers & Bhowmik, 1970), it can make
behavior more predictable in workplaces thus enhancing instrumental relationships (Lincoln &
Miller, 1979). Hegde and Tumlinson (2014) show positive performance effects for start-ups
when venture capitalists and the start-ups they invest in have ethnic proximity. Superior commu-
nication and coordination due to co-ethnicity are suggested as drivers of these positive perfor-
mance effects. Similar arguments can be made for high quality information and resource flows
among actors with common caste affiliations (Chen et al., 2015), improving quality of decision
It is an empirical issue regarding how industrialization, and its consequent urbanization, has thrown together individuals from across
castes and has thus weakened caste preferences in India, though researchers have long argued that caste has transformed and survived
(Niehoff, 1959). Srinivas (2003), an eminent caste scholar, has predicted the demise of the caste system due to the economic changes
underway in modern India.
making and leading to better performance outcomes. The economic benefits of these social net-
works are also emphasized in a number of other studies (e.g., Banerjee, Besley, & Guinnane,
1994; Kalnins & Chung, 2006).
Furthermore, social proximity can reduce moral hazard through social sanction, thereby lower-
ing information and monitoring costs, according to Stiglitz (1990). In particular, Freitas (2006)
models the Indian caste system as an information-sharing institution with the clout for contract
enforcement. Those who default on members of their caste are punished by being denied services
produced by the caste. Indeed, in an experiment, Hoff, Kshetramade, and Fehr (2011) show the
willingness of upper-caste Hindus to punish norm violations by their members. Caste thus
becomes an informal enforcement system particularly in the absence of efficient institutional
enforcement mechanisms. Jackson and Schneider (2011) find that taxi drivers in New York City
exhibit reduced effects of moral hazard when they lease from members of their own community
by country of birth. They note that these network effects operate through social sanctions.
Performance hypothesis for positive discrimination
Hypothesis 3a (H3a) If information and/or trust reasons drive the hiring of same
caste/religion CEOs, then firms with same caste/religion CEOs will perform better
than their peers.
2.3.2 |The effect of negative discrimination
On the other hand, it has also been argued that social proximity can lead to economically ineffi-
cient outcomes. In this view, homophily, based on characteristics other than ability, is likely to
lead to inefficient decision making because it could induce social conformity and groupthink
(e.g., Asch, 1951; Ishii & Xuan, 2014; Janis, 1982). This is likely the case since individuals who
are similar to each other are more likely to prefer unanimity and support each other, ignoring the
downsides of a favored decision and advice from experts outside the group. Furthermore, as men-
tioned earlier, individuals may sacrifice quality, either consciously or unconsciously, for the
opportunity to work with similar others because they derive personal utility from the collaboration
(Gompers et al., 2016).
Gompers et al. (2016) show this cost of friendshiphas adverse performance outcomes for
common investments where venture capitalists are similar based on affinity-related characteristics.
In the context of top management teams, higher level of advice-seeking from friends and similar
others in response to poor firm performance has been shown to exacerbate executive biases inhibit-
ing strategic change and leading to further deterioration in firm performance (McDonald & West-
phal, 2003). For all these reasons, hiring same caste/religion CEOs based on taste-based preferences
may lead to adverse performance consequences for firms. Therefore,
Performance hypothesis for negative discrimination
Hypothesis 3b (H3b) If taste-based preferences drive the hiring of same caste/religion
CEOs, then firms with same caste/religion CEOs will perform worse than their peers.
3.1 |Sample
For our sample, we draw the largest 1,000 listed firms by market capitalization from the Bombay
Stock Exchange (BSE) or the National Stock Exchange (NSE) over the period, 20012009. We
focus on these firms because their hiring decisions should have the greatest impact on the Indian
economy. We obtain accounting data from the Prowess database from CMIE (Center for Monitoring
Indian Economy). In order to conduct the study, we needed to also identify family-owned firms,
CEO turnover events, and the castes of both the firm and the CEO. Since we are interested in how
CEOs are selected, we focus on new CEO appointments, rather than all existing CEOs. Doing so
avoids survivorship bias and also eliminates the problem of identifying the rival candidates of some
existing CEOs who were hired long ago.
3.2 |Identifying family firms and CEO transitions
Similar to Perez-Gonzalez (2006), we categorize a firm as a family firmif (a) there is at least one
shareholder with a 5% or greater block holding, (b) two or more individuals related by blood or mar-
riage are reported as shareholders, officers, or directors, or (c) a founder/promoter is identified with
some continuing family ownership in the firm. During the sample period, we identify all CEO turn-
overs using Prowess.
We then categorize each transition as familyor non-family/professional.
Family successions are those in which the CEO is related by blood or marriage to (a) founder/pro-
moter CEO, or (b) a large shareholder. Other CEO transitions are classified as non-family/profes-
sional(i.e., unrelated). Next, we use the Directors Database managed by the Bombay Stock
Exchange to extract manager and director profiles. We use these profiles to categorize directors as
relatedor unrelated,”“independentor non-independent,and to categorize CEOs as founders/
promoters,family successors,or professional.This classification also helped in assigning fam-
ily firm status based on relationships among the officers/directors. In addition, the information about
directors helped in constructing the insidercandidates for CEO selection (as described later).
3.3 |Assigning surnames to family and non-family firms
For family firms, we assigned the surname of the CEO who was the founder/owner of the firm.
When information is not readily available regarding founders, we study the company incorporation
history to know the founder/promoter details and assign the surname accordingly. The Bombay
Stock Exchange, The Economic Times, and websites were the sources for this
information. If no such information is found, we use data from Prowess to identify the business
group to which the firm belongs. We then assign to the firm the surname of the family that owns
the business group.
Even non-family firms are assigned surnames, and subsequently castes/religions. We followed
the same procedure for the non-family firms as we did for the family firms, i.e., located the founder/
promoter and assigned their surname, or if the firm belonged to a business group, we assigned the
surname of the owning family of the business group. In instances where none of these pieces of
CEOs in India may assume a wide variety of designations such as chairman, president, managing director, director, or executive
director, etc. CEOs were identified by (a) identifying the highest paid official from Prowess, and (b) by manually verifying the top
position according to the annual report of the firm on the website
Any other religions such as Buddhists have been grouped in this group of Parsee and others, but simply referred to as Parsee.
information were available, we assigned the surname of the family, if the firm was previously family
owned, or finally, the chairpersons surname, since chairpersons are likely to have influence on
CEO appointments.
3.4 |Identifying religion and caste of CEO, firm owners, chairpersons, and board of
Individuals in this sample could be from different religions to begin with, i.e., Hindu, Muslim,
Christian, Sikh, Jain, or Parsee (Zorastrian).
Foreignwas also used as a religion,where foreign
ownership or foreign CEOs were involved. While there could be castes (sects) within other reli-
gions, we consider castes only for the Hindus and do not consider them for non-Hindus.
As mentioned earlier, over time, the four castes (Varnas) of the traditional Hindu religion,
i.e., Brahmin, Kshatriya, Vysya, and Shudra, have morphed into a complex pattern of many caste-
based communities, or Jatis. First, to capture the broad patterns, we extend the traditional four castes
to six, recognizing the emergence of at least two castes, Kayastha and Jats, which may fall between
the traditional ones but have acquired sufficient identity of their own. Second, we formulate an
Otherscaste, which includes the Shudras and OBCs,
the Scheduled Castes (SC), and the Sched-
uled Tribes (ST). The detailed steps taken for these caste assignments are described in Appendix
1 in File S1.
3.5 |Calibration against population of castes/religions for bias tests
A high percentage of CEOs from any caste/religion could simply be the result of a larger population
of members of that caste/religion. Since centuries of privilege might have led the Forward Castes to
be better educated and have more business experience than the other castes, they are likely to be
overrepresented in the pool of qualified CEO applicants relative to the general population. We use
the caste/religion percentages from the pool of qualified CEO applicants (explained in footnote 12)
for calibration in our bias tests rather than the percentages in general Indian population to take care
of this aspect.
3.6 |Selection of CEO from among potential rival candidates
For more rigorous regression analyses, since CEOs are hired from a group of rival candidates, we
formed a pool of insiderand outsidercandidates for comparison. We need both insidersand
outsidersin the pool to be able to parse out whether a bias in CEO hiring is due to information
reasons or due to taste-basedpreferences. A significant bias in favor of same caste/religion hiring
in the overall pool could be for either information reasons, or taste-based preferences, or both. How-
ever, if the bias in favor of same caste/religion is less in case of candidates who are insidersto the
firm, then it will be clean evidence in support of informationreasons (positive discrimination
hypothesis) since information advantages are unlikely through same caste/religion hiring in case of
We use Shudras and OBCsince not all Shudras are OBCs.
One could also think of why firms do not simply recruit insidersinstead of outsiders,if it was about information regarding the
individual. Firms may want outsidersat least for a few reasons. First, firms may want an outsiderif they want to bring some
changes in the company for which insiders may not be suitable because they are too ingrained with the practices within the firm. Sec-
ond, firms may want outside talent which is exposed to the strategies and ideas at other firms due to which such outsiders can have a
fresh eye to examine the ways of working and strategies within the recruiting firm. Nevertheless, in choosing from among
outsiders,firms may still use same caste/religion as a source of soft information about these candidates.
insiderswho are already known to the firm.
There is no equivalent variable to distinguish the
taste-basedpreferences in this study.
The detailed steps taken for building the pool of rival candidates for CEO selection (insiders
and outsiders), along with the data sources used, are in Appendix 2 in File S1. These insiders
and outsiderscould be from same caste/religionor not from same caste/religionas that of the
firm owners/chairpersons. We then use the Coarsened Exact Matchingtechnique for obtaining the
set of rival candidates for CEO selection (described later in the sub-section under the multivariate
3.7 |Variables and measures
3.7.1 |Dependent variable
The dependent variable for the multivariate analyses for CEO hiring (H1 and H2) was whether the
candidate was selected to be the CEO from the pool of potential candidates. This variable was coded
Selected= 1, for the hired candidates, and Selected= 0, for the non-hires in the pool.
For the multivariate analyses of the impact of same caste/religion CEOson firm value (H3a
and H3b), the dependent variable was ROA (return on assets calculated as EBITDA over total
assets) as a measure of firm value. We chose to use this measure since it is less likely to be endoge-
nous to same caste/religion hiring bias, i.e., ROA is less likely to adjust itself based on a same caste/
religion CEO, which is likely the case with a market based measure such as Tobins Q. The ROA of
the year following CEO change was used in these firm performance regressions.
3.7.2 |Independent variable
The independent variable of interest for both multivariate analyses, i.e., CEO hiringand firm
performance,was a dummy variable that indicated whether the CEO candidate belonged to the
same caste/religion as the firm owners/chairpersons. This dummy variable was coded 1if the
CEO candidate was from the same caste/religion as firm owners/chairpersons and 0otherwise.
This variable was used in the Coarsened Exact Matching procedure for defining the treatmentand
controlgroups. We also used an interaction term of CEO candidate being an insider to the firm
TABLE 1 Sample description for firm types and CEO appointments
Panel A. Firm type and
no. of CEO appointments
Panel B. Non-PSU
CEOs classified by
Firm type
No. of CEO
No. of professional
CEO appointments CEO type N
All firms 7,461 677 468
Non-PSU firms 7,093 557 348 Promoter+ 167
Indian non-PSU 6,489 455 255
Indian, family firms 6,175 400 201 Descendant 42
Indian, Hindu-family firms 4,987 306 149 Professional 348
Foreign firms (non-PSU) 604 102 93
Panel A displays the total CEO appointments and the professional CEO appointments by firm type included in the sample. While we
started with a total of 804 CEO appointments in the sample, after applying the filters mentioned, we were left with 677 CEO appoint-
ments. Out of these, 468 were professional CEO appointments. For the non-PSU sample, total CEO appointments were 557, which
included 348 professional CEO appointments. +There were firms with multiple promoters who succeeded each other, either by rota-
tion, or due to demise of one of the promoters, or due to other reasons.
and CEO candidate being from same caste/religion as firm owners/chairpersons.This interaction
term helps in parsing out informationreasons for preference of same caste/religion candidates
(as discussed later in the analysis and results section).
3.7.3 |Control variables
Several individual and firm level control variables were used following previous studies on CEO
selection. A few additional control variables adapted to the context were also included. The details
of the control variables used are in Appendix 3 in File S1.
4.1 |Descriptive statistics
Table 1 displays the sample characteristics. The sample contained only firm-years where the firm
could be classified as a familyor non-familyfirm, both the firm owners/chairpersonand
CEO caste/religionwere categorized, and where the CEO type could be identified as promoter
(or founder),”“descendant,or professional.Professional CEOs are individuals who are unrelated
to the firm owners. These filters left 7,461 observations out of the original 9,000 observations. The
private sector firms, i.e., non-public sector undertakings (non-PSUs), which are the most relevant for
this study accounted for 7,093 of these observations. All analyses that follow are based on the non-
PSU sample. Family firms are a predominant form of ownership in India and comprise about 87%
of these non-PSU observations. About 81% of these family firms in this sample belong to the Hindu
religion, the largest religion in India.
Panel A displays the total CEO appointments and the professional CEO appointments by firm
type included in the sample. Panel B displays the break-up of total CEO appointments by CEO type
in the non-PSU firms. A large majority of these were professional CEO appointments, with about
63% (i.e., 348 appointments) falling in this category.
4.2 |Distribution of firm owners/chairpersons and professional CEOs by caste/religion
Table 2 displays the distribution of firm owners/chairpersons by caste and religion and the caste and
religion of professional CEOs appointed by them. This distribution is only for the non-PSU firms
which, as mentioned earlier, are the focus of this study.
There are, at least, three important observations from this table. First, a majority of the firm-
years (Column 1) belong to Vysya (36.87%) and Kshatriya (24.37%) castes of the Hindu religion,
There were several instances in this dataset where promoters succeeded each other, either by rotation or due to the death of one of
the promoters, e.g., in the case of Asian Star, Dinesh Shah, Arvind Shah, and Vipul Shah were all promoters and became Managing
Directors one after the other.
We also coded OBC/Shudra and SC/ST castes separately where the classification was possible. There were no CEO transitions in
the SC/ST owned firms or professional CEOs appointed from SC/ST, to the extent our classification permitted. Therefore, we did not
proceed with further analyses with such classification.
This pool was formed by taking into consideration all firms that went through a professional CEO transition. This assumption was
considered reasonable, since these firms experienced instability at the top and were, therefore, likely to provide candidates into pool
for CEO selection. The pool included the non-independent directors/key executives of firms (including CEOs) in the year. PSUs can-
didates were included since private sector firms could potentially hire them as CEO candidates. Percentages of castes in this pool
adjust for potential over-representation of higher castes of the Hindu religion compared to their actual percentage in the Indian popu-
lation. Column 16 shows the lead of Forward Castes of Hindu religion in this pool although they account for less than 20% of the
total Indian population.
TABLE 2 Distribution of CEO and firm owner/chairperson caste/religion for professional CEO appointments (non-PSU firms)
Column 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Firm owner/chairperson
caste/rel. % firm-years Brahmin Kshatriya Kayastha Vysya Jat Others Muslim Christian Jain Sikh Parsee Foreign Total
caste %
in pool
Brahmin 9.19 13 3 1 0 0 1 0 1 2 0 0 2 23 56.52 20.45
Kshatriya 24.37 9 17 3 6 3 0 0 2 2 1 1 3 47 36.17 22.80
Kayastha 1.07 1 2 1 0 0 0 0 0 0 0 0 0 4 25 6.23
Vysya 36.87 23 29 10 39 0 2 0 2 6 3 1 5 120 32.5 21.99
Jat 0.18 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.43
Others 2.26 1 1 0 1 0 1 0 0 0 0 0 0 4 25 1.62
Total Hindu 73.94 47 52 15 46 3 4 0 5 10 4 2 10 198 73.52
Muslim 1.54 1 0 0 0 0 0 1 0 0 0 0 0 2 50 2.31
Christian 1.35 1 0 0 0 0 0 0 1 0 0 0 0 2 50 2.05
Jain 9.35 3 3 2 2 0 0 0 0 2 0 1 0 13 15.38 3.71
Sikh 1.55 1 2 0 1 0 0 0 0 0 3 0 0 7 42.86 2.10
Parsee 3.75 12 5 3 2 0 1 1 0 0 0 5 4 33 15.15 1.54
Foreign 8.52 15 23 4 4 0 0 3 2 2 0 0 40 93 43.01 14.77
Total other rel. 26.06 33 33 9 9 0 1 5 3 4 3 6 44 150 26.48
Total Professional CEO
100 80 85 24 55 3 5 5 8 14 7 8 54 348 100
followed by Jain religion (9.35%), and Brahmin caste of Hindu religion (9.19%), respectively. This
shows the lead of Forward Castes in business.
Second, firm owners/chairpersons seem to appoint a higher number of CEOs from their own
castes/religions. The joint distribution of castes/religions of firm owners/chairpersons and profes-
sional CEOs appointed by them (Columns 213), captures these patterns. The most striking pattern
is along the diagonal of the distribution where the numbers of CEO hires are greater when they are
from the same caste/religion as that of the firm owners/chairpersons.
Third, for all castes/religions, the percentages of same caste/religion professional CEOs
appointed by firm owners/chairpersons (Column 15) are greater than the percentages of same caste/
religion candidates in the potential pool of candidates for CEO selection (Column 16), pointing to
same-caste/religion bias in professional CEO hiring.
It is also interesting to observe that even
Foreignowners/chairpersons seem to appoint a majority of professional CEOs from Forward
Hindu castes.
4.3 |Bias towards same-caste/religion professional CEO hires
Table 3 offers a statistical test of the caste/religion hiring bias in the selection of professional CEOs
by the non-PSU firms. As mentioned earlier, these tests assume the castes/religions of the CEO can-
didates follow the distribution of the proportion of castes/religions as in the potential pool of CEO
In the full sample of non-PSU firms, same-caste/religion professional hires constitute 35.59% of
the total professional CEO hires. Based on the castes/religions represented in the potential pool for
CEO hires, the expected percentage of same caste/religion is only 8.33%. The bias, 27.26% is eco-
nomically and statistically significant. When the analysis is restricted to Only Indianfirms, the
bias is similar. In the sample limited to Only Hindufirms, the same caste/religion bias is slightly
TABLE 3 Tests of within caste/religion hiring of professional CEOs in non-PSU firms
Percentage same
caste/religion hires
Expected percentage
same caste/religion hires+ Bias
Signrank Test
Full sample
All Firms 35.59 8.33 27.26 0.003
Only Indian firms 34.86 7.75 27.11 0.005
Indian Hindu firms 35.04 12.25 22.79 0.043
Indian, Non-Hindu firms 34.68 2.34 32.34 0.043
Family firm sample
All firms 45.42 8.33 37.09 0.007
Only Indian firms 43.34 7.75 35.59 0.001
Indian Hindu firms 39.32 12.25 27.07 0.108
Indian, Non-Hindu firms 45.74 2.34 43.4 0.043
Non-family firm sample
All firms 47.77 8.33 39.44 0.028
Only Indian firms 50.00 7.75 42.25 0.046
Indian Hindu firms 53.34 12.25 41.09 0.079
Indian, Non-Hindu 33.33 2.34 30.99 0.317
+The caste/religion percentages used are from the pool of potential hires as in Table 2. This test is similar to the Yonker (2017) home
bias test.
lower at 22.79% and for non-Hindufirms the bias is slightly higher at 32.34%. All these biases
are statistically significant.
When the sample is further split by familyand non-familyfirms, in the familyfirm sam-
ple, the Indian Hindu familyfirms seem to exhibit much less bias (27.07%, weak significance at
slightly greater than 0.1) than non-Hindu Familyfirms (43.4% and significant). The non-family
firm sample shows evidence of greater bias than in familyfirm sample in all categories. This is an
intriguing result although the bias is weakly significant in the case of Hindu non-familyfirms and
is insignificant for Indian non-Hindu, non-familyfirms.
These tests, although they establish the existence of same caste/religion bias in professional
CEO hiring, do not help distinguish whether the bias is due to information or trustreasons (posi-
tive discrimination), or due to taste-basedpreferences (negative discrimination). Therefore, we
resort to more rigorous multivariate tests with different pools of potential candidates for CEO selec-
tion to be able to tease out the nature of the bias.
4.4 |Multivariate tests
To begin with, we summarize the characteristics of the full sample of candidates used for multivari-
ate regression for CEO selection (i.e., the selectedprofessional CEO candidates, and the compari-
songroup of rival candidates as described earlier under the sub-head Selection of professional
CEOs from among potential rival candidates). These summary statistics and the correlations among
the variables used in the selection regression are in Table 4. This table does not include firm level
variables since they are controlled for through hiring observation fixed effects. There are no high
correlations indicative of collinearity problems. The variables CEO candidate from same caste as
firm owners/chairpersonsand CEO candidate being an insider to the firmare positively corre-
lated with being selected as CEO, according to expectation.
We then used the Coarsened Exact Matching procedure (Iacus, King, & Porro, 2011) to obtain
the treatedand controlgroups of candidates for CEO selection. This matching technique helps
in pruning observations from the data so that the remaining data have better balance between the
treatedand controlgroups in terms of the distributions of the covariates in the groups being
more similar.
The treatmentin this set up was whether the candidate belongs to same caste/reli-
gion as that of firm owners/chairpersons. We coded the dummy variable treated= 1 for the group
of candidates from the same caste/religion as that of the firm owners/chairpersons.This formed
the treatment group. We coded treated= 0 for the candidates who were not same caste/religion
as that of the firm owners/chairpersons.This formed the control group. The variables used for
coarsening were all the covariates used in the CEO selection regression, i.e., CEO candidate age,
CEO candidate was an insiderto the firm,”“CEO candidate has masters or professional degree,
CEO candidate is from an elite Indian Institute,”“CEO candidate has a throughput background,
CEO candidate was also a chairperson of a firm,and the industry categoriesto obtain candi-
dates matched by industry category.
The technique produces an overall imbalance score and also the weights to be assigned to the
observations while running the regressions. The imbalance score for this sample was 0.429 which
meant the samples were decently matched (a score closer to 1would indicate high degree of
imbalance between the treatmentand controlgroups along the covariates). For the regressions,
The t-tests, in fact, showed significant differences along all covariates and therefore CEM achieves a better balance.
These regressions were also run for the public-sector firm subsamples. Results show that same caste/religion CEO candidates are
likely to be preferred and this holds in the case of insidercandidates as well. This result is not unexpected since public sector firms
are likely to be influenced by different decision-making criteria for CEO appointments than private sector firms.
TABLE 4 Summary statistics and correlation matrix for the sample of non-PSU observations
S. no. Variable n Mean s.d. Min Max 1 2 3 4 5 6 7 8
1 Selected CEO candidate 249 1
2 CEO candidate same caste as firm
7,563 0.249 0.433 0 1 0.043 1
3 Insider CEO candidate 7,563 0.098 0.298 0 1 0.265 0.199 1
4 CEO candidate age 7,563 52.19 10.57 20 94 0.026 0.019 0.012 1
5 CEO candidate has masters/professional degree 7,563 0.672 0.469 0 1 0.001 0.073 0.0002 0.057 1
6 CEO candidate attended an elite Indian Institute 7,563 0.089 0.285 0 1 0.083 0.082 0.019 0.048 0.115 1
7 CEO candidate has throughput background 7,563 0.114 0.318 0 1 0.008 0.021 0.011 0.034 0.08 0.002 1
8 CEO candidate also a chairperson 7,563 0.065 0.247 0 1 0.029 0.049 0.064 0.065 0.134 0.015 0.001 1
No high correlations.
the original covariates get used along with the weights produced through this matching process in
order to control for any remaining differences after the matching process (Blackwell, Iacus, King, &
Porro, 2009).
We test for the same caste/religion bias in professional CEO hiring with linear probability
models. These are Ordinary Least Squares (OLS) regressions with the dependent variable 1if the
candidate was a selected CEO, and 0otherwise.
We run these regressions with the complete
pool of insiderand outsiderCEO candidates obtained from the CEM procedure. Since each can-
didate is unique to the hiring observation (company-year) and the CEM procedure matches the can-
didates only on observable characteristics, i.e., CEO candidate age, education, etc., we use hiring
observation fixed effects to take care of unobservable characteristics. These hiring observation fixed
effects are more meaningful in taking care of differences in size of the group and share of same
caste, etc., with split samples of insidersand outsiders,since CEM procedure uses only
insidersor outsidersin the group if the selected CEO is an insider or an outsider. Therefore, we
run these regressions in the split-samples of pools of only insidersand pool of only outsiders,
as well. Differences in firm level variables such as firm profitability etc., also get controlled through
the hiring observation fixed effects.
Table 5 displays the results from these regressions for the key variables of interest using the
complete pool of insiderand outsidercandidates for CEO selection (Panel A), pool of only
insidercandidates (Panel B), and pool of only outsidercandidates (Panel C).
These regres-
sions include the covariates used for coarsening purposes and the complete results with all variables
are in Appendix 4 in File S1.
In Panel A, the coefficient of the key independent variable, CEO candidate from same caste as
firm owners/chairperson,is 0.014 at p= .029 for the Only Indianfirms subsample (column 2).
All other subsamples also show positive coefficients with varying p-values. This is evidence in sup-
port of H1 and H2, i.e., a preference for same caste/religion professional CEO hires.
However, the coefficients of the interaction term in Panel A, i.e., CEO candidate from the
same caste/religion as firm owners/chairpersons × CEO candidate insiderto the firmare nega-
tive at p= .00 for most subsamples. These results suggest that the preference for same caste/reli-
gion hiring is lesser in case of insidersthan in case of outsidersto the firm.
These results are further corroborated with the results from the split-samples of insidersand
outsiders,with controls for the hiring observation fixed effects.
The negative coefficients for the
CEO candidate from the same caste/religion as firm owners/chairpersonsvariable in all subsam-
ples in Panel B (i.e., the pool of only insidercandidates for CEO selection) suggest that, among a
pool of insidercandidates, being from same caste/religion as firm owners/chairpersons decreases
the probability of being selected as a CEO. On the other hand, the positive coefficients for this same
caste/religion variable in all subsamples in Panel C (i.e., the pool of only outsiderCEO candi-
dates) suggest that, among a pool of outsidercandidates, being from same caste/religion as firm
owners/chairpersons increases the probability of being selected as a CEO. In terms of effect sizes, in
The numbers for the overall sample in the regressions are different from the numbers in the summary statistics table, since CEM
procedure was done including PSU firms.
Hiring observation fixed effects enable estimating the average across all the effects within hiring decisions, i.e., after subtracting
from each variable the average of the variable within each hiring decision. In the regressions with the complete pool of insidersand
outsiders,which involve an interaction term (Panel A), the average of the interaction gets subtracted but not the product of each
variable minus its average within each hiring decision. Therefore, the results using split samples are more precise.
We also checked the cross-correlations of all the OLS errors of regressions with variables used for creating CEM groups as the
dependent variable and the hiring observation fixed effects as regressors. We largely observe zero correlations or very small correla-
tions less than 0.1 throughout the whole matrix. This is in line with the L-statistic of 0.429 for the effectiveness of the CEM
TABLE 5 Hiring professional CEOs within caste/religionLinear Probability Models (non-PSU firms)+
Model (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Sample category All non-PSU
family Kshatriya/Jat Vysya
DV: Selected = 1 Panel A: Complete
comparison pool of
insiderand outsider
CEO candidate same caste as
firm owners/chairpersons
0.003 0.014 0.013 0.016 0.009 0.005 0.015 0.108 0.110 0.013 0.005
[.524] [.029] [.116] [.068] [.183] [.578] [.081] [.000] [.000] [.127] [.544]
CEO candidate insider to the
firm (B)
0.189 0.191 0.188 0.199 0.211 0.219 0.188 0.136 0.124 0.128 0.191
[.000] [.000] [.000] [.000] [.000] [.000] [.000] [.000] [.000] [.000] [.000]
A×B 0.08 0.101 0.131 0.115 0.105 0.143 0.101 0.186 0.204 0.109 0.150
[.000] [.000] [.000] [.001] [.000] [.000] [.004] [.000] [.000] [.000] [.000]
DV: Selected = 1 Panel B: Comparison pool of
only insidercandidates
CEO candidate same caste as
firm owners/chairpersons
0.133 0.143 0.168 0.027 0.146 0.177 0.027 0.173 0.173 0.045 0.203
[.000] [.001] [.000] [.839] [.002] [.000] [.838] [.090] [.090] [.615] [.000]
DV: Selected = 1 Panel C: Comparison pool of
only outsidercandidates
CEO candidate same caste as
firm owners/chairpersons
0.001 0.01 0.010 0.011 0.003 0.001 0.011 0.106 0.108 0.014 0.002
[.770] [.045] [.148] [.091] [.517] [.923] [.096] [.000] [.000] [.038] [.804]
No. of hiring observations 234 172 133 39 127 91 36 45 43 48 81
Total observations 7,715 5,449 3,804 1,645 4,374 2,779 1,594 1,074 1,024 1,893 2,787
R-squared 0.119 0.123 0.121 0.167 0.129 0.131 0.152 0.129 0.126 0.135 0.123
+CEM procedure used; insiderand outsiderpools drawn after CEM; Variables used for coarsening were controlled for (Table 5A, B, and C of Appendix 4 in File S1 are the complete set of results
for Panel A, B, and C, respectively); p-values in brackets; Hiring observation fixed effects in all models and pools.
Panel A, being from same caste/religion as firm owners/chairpersons drops the odds of selection for
insiders by about 10% to 20% while it raises the odds of selection for outsiders by under 2%. These
effect sizes for insidersand outsidersare similar in Panel B and Panel C, respectively.
Therefore, taken together, these results from Panels A, B, and C in Table 5 support the informa-
tion hypothesis(H1) since information benefits based on caste/religion are less likely in case of
insidersto the firm.
Also, since taste preferencesbased on same/caste religion are likely to be
same for insidersor outsiders,the small positive net effect in case of outsiderscould be either
because there are unobserved negative reasons for hiring an outsider from the same caste/religion
which offset large positive information benefits or because there are no unobserved negative reasons
for hiring from same caste/religion and that the information benefits are small. While there is no
way to resolve which of these could be the likely scenario, the small net effect in case of outsiders
also is a support for the information hypothesis.
It is interesting to note that the Indian familyand Indian Hindu familysubsamples (column
3 and column 6 in Panel A of Table 5, respectively) do not exhibit an overall bias in favor of same
caste/religion CEO candidates for hiring. Further, in these subsamples, the coefficients of the inter-
action term are negative suggesting that insidersfrom same caste/religion are less likely to be cho-
sen compared to outsidersfrom same caste/religion. These are intriguing results that an insider
from same caste/religion will be discriminated against an outsiderto the firm from same caste/reli-
gion. However, this can be reconciled as follows. For these subsamples of firms, since a majority of
insiders(directors and key executives) are likely to be from family circles and friends and hence
likely to be from same caste/religion, firms may prefer to appoint those who are not from same
caste/religion when choosing professional CEO candidates. Such choices could be important to man-
age the morale of insiderswho are not from same caste/religion (and hence also not from within
family circles or friends) that they will have fair opportunity to be recruited for the highest position
without being discriminated against. Nevertheless, for the non-family firm subsamples, results sup-
port a bias in favor of same caste/religion insiders.However, this bias cannot still be construed as
evidence in favor of taste-basedinfluences (H2), although we cannot exclude taste-basedprefer-
ences, since this same caste/religion insidersinteraction variable helps only in providing discrimi-
nating evidence for the informationhypothesis.
There is also variability across these subsamples with regard to the net positive information
effect in case of outsiders(more than in case of insiders). The effects seem to be smaller for the
Indian Hindusubsample (column 5) compared to Indian non-Hindusubsample (column 7) and
also smaller for Vysya subsample (column 11) compared to Kshatriya/Jat subsample (column 10).
4.5 |Same caste/religion hires and firm performance
Hypotheses 3a and 3b are regarding the implications of positiveand negativediscrimination
effects of professional CEO hiring on firm value. If the caste/religion bias in hiring is a mechanism
for efficient hiring(H1), then same caste/religion CEO hiring should enhance firm value (H3a).
However, if the bias is due to taste-baseddiscrimination (H2), it leads to inefficient matching and
consequently should destroy firm value (H3b).
We test for the impact of same caste/religion CEOs on firm performance using the firm-year
observations when a CEO transition has occurred which had firm level information available.
In case of Kshatriya firm subsample as well, the positive bias for same caste/religion insiderhires could be driven by non-family
We checked and found that announcement dates of CEO appointments were not reported often enough to undertake an event study.
Table 6 displays the summary statistics and the correlation matrix for the variables used in these
regressions. The correlations are not indicative of collinearity problems among these variables. The
ROA of year following CEO change has a positive correlation, although very small, with same
caste/religion professional CEO hires, which is not suggestive of value destruction due to same
caste/religion CEOs.
We use ROA as the measure of firm value instead of Tobins Q since markets might adjust valu-
ations based on same caste/religion CEO hiring information and therefore it might be unlikely to
find a result with Tobins Q. ROA is less likely to adjust itself based on whether the CEO is chosen
from same caste/religion. In that sense, it mitigates concerns of endogeneity that are likely with
other variables in these settings. However, one could also argue that the choice of same caste/reli-
gion CEO may be influenced by expected ROA, and in that sense, the endogeneity problem may
still persist in terms of reverse causality.
Therefore, we first checked for reverse causality in our data, i.e., whether expected performance
influences the choice of same caste/religion CEOs. We control for the CEO characteristics and firm
characteristics (including the ROA of the year before CEO change) and use industry dummies and
robust standard errors for these regressions. Neither the ROA of the year of CEO change nor the
ROA of the year following CEO change exhibited any significant relationship with the CEO candi-
date being from same caste/religion as that of the firm owners/chairpersons. Therefore, reverse cau-
sality does not appear to be a concern in our data.
Panel A in Table 7 displays the results for the key variables, i.e., CEO candidates from same
caste/religion as firm owners/chairpersonsand CEO candidate an insider to the firmfrom the lin-
ear regressions for the effect of same caste/religion CEOs on firm performance. These regressions
control for several CEO and firm level characteristics. They also control for industry effects and use
robust standard errors.
Results suggest that same caste/religion professional CEOs do not affect firm performance. This
is the result for all subsamples. Results also suggest that the CEO being an insidercandidate also
does not impact firm performance. All these results hold after controlling for the previous years
ROA of the CEO candidates firm, which takes into account the effect of the quality of the individ-
ual belonging to any particular caste/religion that could have potentially impacted the performance
of the firm. These results are consistent in that if outsidercandidates are chosen based on same
caste/religion, there is likely to be information about them based on caste/religion. Therefore, perfor-
mance is not likely to be adversely affected. In the case of insiders,since there is already likely to
be information about them, again, performance is unlikely to be adversely affected.
To tease out the performance effects of same caste/religion CEO hiring more specifically, we
repeated the same linear regressions (as done for Panel A in Table 7), by including the interaction
term CEO from same caste as firm owners/chairpersons × CEO candidate insider to the firm.The
results for of these key variables of interest are displayed in Panel B of Table 7. The coefficients of
this interaction term are negative for most subsamples. In particular, they are negative with p= .138
and p= .114 for All non-PSUand Only Indianfirm subsamples (column 1 and column
2, respectively). These results suggest that information effects of caste/religion-based hiring are
lower in case of insidersthan in case of outsidersto the firm. The net information effects for
outsidersare positive, although small. Although these results lean towards supporting H3b than
H3a, we cannot conclusively establish taste-basedpreferences in case of hiring of insidersfrom
the same caste/religion as firm owners/chairpersons. The complete results of regressions for Panel A
and Panel B of Table 7 with all control variables displayed are in Appendix 4 in File S1.
These results are available from authors upon request.
TABLE 6 Summary statistics and correlation matrix among variables for performance regressions (non-PSU firms)
S. no. Variable n Mean s.d. Min Max 1 2 3 4 5 6 7 8 9 10 11 12 13
1CEO from same caste/rel. as firm
183 121.00 ——1.00
2CEO age 183 50.51 8.53 24.00 73.00 0.09 1.00
3CEO attended an elite Indian Institute 183 0.24 0.43 0.00 1.00 0.10 0.01 1.00
4CEO has masters or professional degree 183 0.65 0.48 0.00 1.00 0.08 0.04 0.19 1.00
5CEO has throughput background 183 0.10 0.30 0.00 1.00 0.15 0.02 0.10 0.01 1.00
6CEO insider to the firm 183 0.52 0.50 0.00 1.00 0.13 0.05 0.03 0.06 0.12 1.00
7CEO is also the chairperson 183 0.09 0.29 0.00 1.00 0.09 0.14 0.00 0.08 0.02 0.01 1.00
8Log of assets 183 6.20 1.44 2.20 11.31 0.01 0.18 0.11 0.11 0.01 0.01 0.01 1.00
9Debt to equity ratio 183 0.58 0.40 0.00 0.99 0.04 0.14 0.04 0.05 0.07 0.03 0.02 0.11 1.00
10 R&D expense to sales 183 0.003 0.01 0.00 0.12 0.07 0.08 0.05 0.08 0.10 0.07 0.00 0.05 0.22 1.00
11 ROA (lagged) 183 0.13 0.22 1.75 1.77 0.12 0.09 0.05 0.02 0.05 0.06 0.04 0.05 0.12 0.04 1.00
12 ROA (forward) 183 0.13 0.16 1.25 0.50 0.05 0.07 0.00 0.01 0.14 0.08 0.02 0.09 0.27 0.07 0.01 1.00
13 ROA current year 183 0.09 0.58 7.58 0.45 0.10 0.03 0.16 0.02 0.03 0.07 0.01 0.08 0.15 0.05 0.03 0.18 1.00
No high correlations.
A number of robustness checks were performed. The details regarding these robustness checks
are in Appendix 5 in File S1.
CEO selection is an important area of research and this study focuses on the social proximity
(or homophily) influences on CEO selection. In particular, we examine the role of social proximity
in the selection of professional CEOs using India as a context, where its numerous castes and reli-
gions serve as bases for social proximity. We hypothesize that social proximity based on caste/reli-
gion will lead to a greater incidence of in-groupCEO hires for two reasons. First, there may be
preference for in-groupCEOs in view of greater trust and information advantages. Second, in-
grouppreference can be for taste-based reasons (Becker, 1957). We also hypothesize regarding the
performance consequences of hiring same caste/religion CEO candidates based on these two
Results from the sample of professional CEOs hires over the 20012009 period by the top 1,000
publicly traded firms in India support the view that social proximity based on caste/religion matters.
This bias in favor of same caste/religion candidates holds in more controlled regression analyses
with matched samples of CEO candidates obtained using the Coarsened Exact Matching techniques
(with CEO candidates from same caste/religion as firm owners/chairpersons as the treatment
group and CEO candidates who are not from same caste/religion as firm owners/chairpersons as the
Furthermore, the overwhelming evidence from this study supports information/trustreasons
for same caste/religion CEO hiring. Evidence also suggests that tastepreferences are outweighed
by other factors in CEO selection, since in case of choice from among a pool of insidercandi-
dates, being from the same caste/religion as that of the firm owners/chairpersonshas an adverse
impact on being selected as a CEO. In addition, the results of this study find that there are no
adverse performance consequences of same caste/religion CEO hiring. Also, the results do not sup-
port adverse performance effects in case of insidersbeing from same caste/religion, i.e., there is
no evidence for taste-basedpreferences for same caste/religion professional CEO hiring.
This study also has its limitations. We lost observations when we moved from model free ana-
lyses as in Tables 2 and 3, to regressions involving controls for individual and firm-level characteris-
tics. Our results may be sensitive to such losses. This concern is alleviated, however, since our
sample sizes are still large. Our performance regressions were validated through 2SLS models in
which our instrument, although theoretically strong, did not pass the weak instrument test in models
where all relevant CEO characteristics were controlled for. Hence some of our robustness checks
could also have their limitations.
In terms of future research directions, the findings of this study can motivate investigation into
the effects of caste/religion on CEO compensation. There are, at least, two interesting questions that
can follow. The first is whether same caste/religion CEOs get paid more than those who are not
from same caste/religion as firm owners/chairpersons. While the results of our study lead to the
expectation that there will be no significant differences in CEO compensation based on same caste/
religion, it will be interesting to establish this empirically. The second question is whether there are
compensation differences across CEOs from different castes/religions. Again, ideally, there should
be no compensation differences, but it is an empirical question.
TABLE 7 Results from linear regressions for within caste/religion professional CEOs and firm performance (non-PSU firms with CEO changes)+
Model (1) (2) (3) (4) (5) (6) (7) (8) (9)
Sample category All non-PSU
Hindu family
non-Hindu family Vysya
DV: ROA prior to year of CEO change Panel A: Results without
CEO same caste/religion as firm owners/
0.011 0.008 0.016 0.032 0.012 0.012 0.159 0.146 0.006
[.599] [.812] [.720] [.732] [.832] [.886] [.127] [.544] [.947]
CEO insider to the firm 0.031 0.050 0.062 0.134 0.062 0.083 0.139 0.142 0.076
[.231] [.157] [.155] [.255] [.261] [.268] [.264] [.341] [.215]
Panel B: Results with same caste/religion
and insiderinteraction
CEO same caste as firm owners/chairpersons 0.038 0.086 0.084 0.075 0.111 0.083 0.083 0.045
[.309] [.151] [.259] [.383] [.445] [.690] [.690] [.563]
CEO insider to the firm 0.054 0.092 0.101 0.120 0.141 0.139 0.139 0.099
[.112] [.060] [.073] [.132] [.183] [.264] [.264] [.139]
CEO from same caste as firm owners/
chairpersons × CEO candidate insider to
the firm
0.084 0.142 0.126 0.178 0.194 0.076 0.076 0.101
[.138] [.114] [.245] [.188] [.349] [.768] [.768] [.284]
Observations 183 127 103 24 90 69 37 34 53
R-squared 0.594 0.617 0.679 0.996 0.672 0.729 0.927 0.926 0.834
+All models use industry controls and robust standard errors; these results also include controls for several other CEO level and firm level variables (Table 7A and B of Appendix 4 in File S1 are the
complete set of results for Panel A and Panel B, respectively); Robust p-values reported in brackets.
To summarize, this study supports informationreasons for same caste/religion bias in hiring
professional CEOs among the largest firms in India. There is no evidence from this study to support
taste-based preferencesin professional CEO appointments (Becker, 1957).
We are grateful to participants of Strategic Management Society conferences in Rome, Prague, and
Copenhagen, and to participants of the Academy of Management meetings in Philadelphia. We also
want to thank Uri Ben-Zion, Tomas Jandik, Kose John, and René Stulz for helpful comments. We
thank Scott Yonker for his valuable inputs for the prior version of this paper. We appreciate the
valuable feedback we received from faculty members at the Indian School of Business (ISB) on ear-
lier versions of this paper. Professor Bhimasankaram Pochiraju has been the backbone of support
for this work in all its phases. His associates, Ms. Manjusha Kancharla and Mr. Ashish Khandelwal,
have been of immense help with statistical analyses. Ashish also deserves a particular mention for
his insights into the phenomenon along with statistical support. Mr. Sridhar Sankranthi has been our
SAS special advisor and we could not have worked without him. Research associatesSaritha
Dasari, Shefali Mittal, Deep Maity, Ajit Kumar, Renu Chaturvedi, and Siva Shankar from the Learn-
ing Resource Center at ISB, have been strong contributors in the process of collecting and organiz-
ing data for this project and we are very thankful to them. Ms. Saritha Dasari was completely
devoted and dedicated to this project and this data gathering and organization would not have been
possible without her commitment and hardwork. Mr. Murugavel Janakiraman, CEO of, and his associates, Mr. Shankar and Mr. Jebamony, also deserve special
appreciation for their willingness and help in providing data and advice for this project. Dr. Rekha
Pandey and her associates from the University of Hyderabad gave us valuable insights regarding
caste classifications. Most importantly, we thank the anonymous reviewers for their highly valuable
inputs all through the course of development of this paper.
Ahuja, G., Polidoro, F., Jr., & Mitchell, W. (2009). Structural homophily or social asymmetry? The formation of alliances by poorly
embedded firms. Strategic Management Journal,30(9), 941958.
Anderson, S. (2011). Caste as an impediment to trade. American Economic Journal: Applied Economics,3, 239263.
Aral, S., Muchnik, L., & Sundararjan, A. (2009). Distinguishing influence-based contagion from homophily-driven diffusion in
dynamic networks. Proceedings of the National Academy of Sciences of the United States of America,106(51), 2154421549.
Aristotle (1934). Rhetoric. Nichomachean ethics.InAristotle in 23 volumes. Rachman transl. Cambridge, MA: Harvard University
Asch, S. (1951). Effects of group pressure upon the modification and distortion of judgment. In H. S. Guetzkow (Ed.), Groups, lead-
ership, and men (pp. 177190). Pittsburgh, PA: Carnegie Press.
Banerjee, B., & Knight, J. B. (1985). Caste discrimination in the Indian labor market. Journal of Development Economics,17(3),
Banerjee, A., Bertrand, M., Datta, S., & Mullainathan, S. (2009). Labor market discrimination in Delhi: Evidence from a field experi-
ment. Journal of Comparative Economics,37(1), 1427.
Banerjee, A., Besley, T., & Guinnane, T. W. (1994). Thy neighbors keeper: The design of a credit cooperative with theory and a test.
Quarterly Journal of Economics,109(2), 491515.
Becker, G. S. (1957). The economics of discrimination (2nd ed.). Chicago, IL: University of Chicago Press.
Bird, B. (1989). Entrepreneurial behavior. Scott, Foresman: Glenview, IL.
Blackwell, M., Iacus, S., King, G., & Porro, G. (2009). CEM: Coarsened exact matching in Stata. The Stata Journal,9(4), 524546.
Borooah, V. K., & Iyer, S. (2005). Vidya, Veda, Varna: The influence of religion and caste on education in rural India. Journal of
Development Studies,41(8), 13691404.
Bossard, J. H. S. (1932). Residential propinquity as a factor in marriage selection. American Journal of Sociology,38(2), 219224.
Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-motivational analysis. Psychological Bulletin,
86, 307324.
Chen, G., Chittoor, R., & Vissa, B. (2015). Modernizing without westernizing: Social structure and economic action in the Indian
financial sector. Academy of Management Journal,58(2), 511537.
Datta, D. K., & Rajagopalan, N. (1998). Industry structure and CEO characteristics: An empirical study of succession events. Strate-
gic Management Journal,15(7), 569577.
De Zwart, F. (2000). The logic of affirmative action: Class, castes and quotas in India. Acta Sociologica,43, 235249.
Fisman, R., Paravisini, D., & Vig, V. (2017). Cultural proximity and loan outcomes. American Economic Review,107(2), 457492.
Freitas, K. (2006). The Indian caste system as a means of contract enforcement (Unpublished manuscript). Northwestern University.
Gompers, P. A., Vladimir, M., & Yuhai, X. (2016). The cost of friendship. Journal of Financial Economics,119(3), 626644.
Guiso, L., Sapienza, P., & Zingales, L. (2009). Cultural biases in economic exchanges. The Quarterly Journal of Economics,124(3),
Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organization as a reflection of its top managers. Academy of Manage-
ment Review,9, 193206.
Hegde, D., & Tumlinson, J. (2014). Does social proximity enhance business partnerships: Theory and evidence from ethnicitys role
in US venture capital. Management Science,60, 23552380.
Hnatkovska, V., Lahiri, A., & Paul, S. B. (2012). Castes and labor mobility. American Economic Journal: Applied Economics,4(2),
Hoff, K., Kshetramade, M., & Fehr, E. (2011). Caste and punishment: The legacy of caste culture in norm enforcement. The Eco-
nomic Journal,121, F449F475.
Huston, T. L., & Levinger, G. (1978). Interpersonal attraction and relationships. Annual Review of Psychology,29, 115156.
Iacus, S. M., King, G., & Porro, G. (2011). Multivariate matching methods that are monotonic imbalance bounding. Journal of the
American Statistical Association,106(493), 345361.
Ibarra, H. (1992). Homophily and differential returns: Sex differences in network structure and access in advertising firm. Administra-
tive Science Quarterly,1, 422447.
Ishii, J., & Xuan, Y. (2014). Acquirer-target social ties and merger outcomes. Journal of Financial Economics,112, 344363.
Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and fiascos. Boston, MA: Houghton Mifflin.
Jackson, C. K., & Schneider, H. S. (2011). Do social connections reduce moral hazard? Evidence from the New York City taxi indus-
try. American Economic Journal: Applied Economics, 3(3), 244267.
Jodhka, S. S. (2008). Caste and the corporate sector. Indian Journal of Industrial Relations,44(2), 185193.
Kalnins, A., & Chung, W. (2006). Social capital, geography, and survival: Gujarati immigrant entrepreneurs in the U.S. lodging
industry. Management Science,52(2), 233247.
Khurana, R. (2002). Searching for a corporate savior: The irrational quest for charismatic CEOs. Princeton, NJ: University Press.
Lazarsfeld, P., & Merton, R. K. (1954). Friendship as a social process: A substantive and methodological analysis. In Freedom and
control in modern society (pp. 1866). New York, NY: Van Nostrand.
Lewis, K., Kaufman, J., Gonzalez, M., Wimmer, A., & Christakis, N. (2008). Tastes, ties, and time: A new social network data set
using Social Networks,30(4), 330342.
Lincoln, J. R., & Miller, J. (1979). Work and friendship ties in organizations: A comparative analysis of relational networks. Adminis-
trative Science Quarterly,24, 181199.
Marsden, P. V. (1988). Homogeneity in confiding relations. Social Networks,10(1), 5776.
McDonald, M. L., & Westphal, J. D. (2003). Getting by with the advice of their friends: CEOs advice networks and firmsstrategic
responses to poor performance. Administrative Science Quarterly,48(1), 132.
McPherson, M., & Smith-Lovin, L. (1987). Homophily in voluntary organizations: Status distance and the composition of
face-to-face groups. American Sociological Review,52, 370379.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in Social Networks. Annual Review of Sociol-
ogy,27, 415444.
Menning, G. (1997). Trust, entrepreneurship and development in Surat city, India. Ethnos,62(1/2), 5990.
Niehoff, A. (1959). Caste and industrial organization in North India. Administrative Science Quarterly,3(4), 494508.
Ocasio, W. (1994). Political dynamics and the circulation of power: CEO succession in US industrial corporations, 19601990.
Administrative Science Quarterly,39(2), 285312.
Parrino, R. (1997). CEO turnover and outside succession a cross-sectional analysis. Journal of Financial Economics,46(2), 165197.
Perez-Gonzalez, F. (2006). Inherited control and firm performance. American Economic Review,96(5), 15591588.
Plato. (1968). Laws. Plato in twelve volumes (Vol. 11. Bury translator). Cambridge, MA: Harvard University Press.
Quigley, T. J., & Hambrick, D. C. (2015). Has the CEO effectincreased in recent decades? A new explanation for the great rise in
Americas attention to corporate leaders. Strategic Management Journal,36(6), 821830.
Rajagopalan, N., & Datta, D. K. (1996). CEO characteristics: Does industry matter? Academy of Management Journal,39(1),
Rogers, E. M., & Bhowmik, D. K. (1970). HomophilyHeterophily: Relational concepts for communication research. Public Opin-
ion Quarterly,34(4), 523538.
Rudner, D. (1994). Caste and capitalism in colonial India: The Nattukottai Chettiars. Berkeley, CA: University of California Press.
Ruef, M., Aldrich, H. E., & Carter, N. M. (2003). The structure of founding teams: Homophily, strong ties, and isolation among
U.S. entrepreneurs. American Sociological Review,68(2), 195222.
Shrum, W., Cheek, N. H., Jr., & Hunter, S. M. (1988). Friendship in school: Gender and racial homophily. Sociology of Education,
61(4), 227239.
Siddique, Z. (2008). Caste based discrimination: Evidence and policy (Working Paper No. 3737). Institute for Labor Studies, Univer-
sity of Bonn.
Srinivas, M. N. (2003). An obituary on caste as a system. Economic and Political Weekly,38(5), 455459.
Stiglitz, J. E. (1990). Peer monitoring and credit markets. World Bank Economic Review,4(3), 351366.
Tian, J., Haleblian, J., & Rajagopalan, N. (2011). The effects of board human and social capital on investor reactions to new CEO
selection. Strategic Management Journal,32(7), 731747.
Vissa, B. (2011). A matching theory of entrepreneurstie formation intentions and initiation of economic exchange. Academy of Man-
agement Journal,54(1), 137158.
Yonker, S. (2017). Geography and the market for CEOs. Management Science,63(3), 609630.
Zajac, E. J. (1990). CEO selection, succession, compensation and firm performance: A theoretical integration and empirical analysis.
Strategic Management Journal,11, 217230.
Zhang, Y., & Rajagopalan, N. (2003). Explaining new CEO origin: Firm versus industry antecedents. Academy of Management Jour-
nal,46(3), 327338.
Zhang, Y., & Rajagopalan, N. (2010). Once an outsider, always an outsider? CEO origin, strategic change, and firm performance.
Strategic Management Journal,31(3), 334346.
Additional Supporting Information may be found online in the supporting information tab for this
How to cite this article: Damaraju NL, Makhija AK. The role of social proximity in profes-
sional CEO appointments: Evidence from caste/religion-based hiring of CEOs in India. Strat
Mgmt J. 2018;39:20512074.
ResearchGate has not been able to resolve any citations for this publication.
I examine the role of geography in the market for CEOs and find that firms hire locally five times more often than expected if geography were irrelevant to the matching process.This localmatchingbiasiswidespreadandexists evenamongthelargestU.S.firms. Tests reveal that both labor supply and demand influence local matching. Compensation and unforced turnover are lower for local than for nonlocal CEOs, and the compensation of local CEOs depends on local labor market factors, unlike that of nonlocal CEOs. These findings suggest the presence of market segmentation and contrast with much of the prior literature, which explicitly or implicitly assumes a single national market.
This analysis considers the impact of the top managers in an organization on the organization's outcomes, specifically strategic choices and performance levels. The focus is not on the chief executive alone, but rather on the entire top management team. Using a macro view, these organizational outcomes are perceived to be related to the values and cognitive bases of those high-power individuals in the organization. In developing the model, emphasis is on the background characteristics of the top managers as opposed to the psychological dimensions. A series of propositions that should be tested to support the upper echelons theory are presented. The topics of these propositions include age, functional track, other career experiences, education, socioeconomic roots, financial position, and group characteristics. The creation of this model is just the beginning of the work that is necessary to evaluate and understand the upper echelons theory. Further input is needed from areas such as the executive recruiting industry. Additionally, clinical and statistical studies are both necessary to fully develop this theory. (SRD)
Five thousand consecutive marriage licenses, in which one or both applicants were residents of Philadelphia, were tabulated according to distance between the residences of the couples. One-third of all the couples lived within five or less blocks of each other, and the percentage of marriages decreased steadily and markedly as the distance between residences of the contracting parties increased. The extent to which the role of residential propinquity is confined to social areas in which specific attributes or combinations of attributes are concentrated will be considered in the series of projects of which this was the initial step.
The attempt to superimpose western industrial organization upon less-developed societies has been only partially successful. The cultural values which eased the introduction and development of industrial capitalism in the West are usually quite different in nonwestern societies, and it is now recognized that industrialization in such societies will develop as some sort of a mutation rather than along western lines. The critical variables that affect economic development include family structure, religious systems, governmental forms, the structure of class and power in a given society, educational and skill level of the masses, and so on. In some societies unique barriers to industrialization and economic change exist. The caste system of India, with which the following analysis deals, provides a striking example.
This study examines the effects of five attributes - authority, education, sex, race, and branch assignment - on the proximities among the members of five professional organizations in networks of instrumental and primary relations. Sex and race are found to have greater influence on primary ties than instrumental ties. While there is some evidence in two organizations that authority and education affect instrumental ties more than primary ties, the data as a whole favor the hypothesis that these two attributes, which are associated with position in the formal division of labor, give rise to both instrumental and primary ties. These attributes generally serve to place high status persons in central network positions. In one organization where these attributes prove to have little predictive power, we examine the effects on network form of a conflict which led to the disintegration of the organization shortly after it was surveyed. The factions into which the organization was divided are readily apparent in a spatial representation of the network. In conclusion we review the implications for organizational theory and research of the attribute analysis in conjunction with the case study findings.
To explain patterns of political dynamics, this paper develops a model of the circulation of power and compares it with an alternative model, the institutionalization of power, in an event history analysis of CEO succession. The circulation of power emphasizes the internal contests for control and opposition to the CEO that emerge with increased executive tenure and under conditions of economic adversity. The study finds support for an increasing rate of CEO succession during the first decade of tenure, consistent with the model of circulation, followed by a slow decline afterward, consistent with institutionalization. The effects of economic adversity were found to trigger circulation when combined with long prior board tenure and large board size. Also, contrary to conventional views, under economic adversity, more inside board members increase CEO succession.
In seeking to understand whether the transition by Asian countries to market economies mirrors the path taken by the West, we ask how embedded network ties between equity analysts and the CEOs of the firms they follow in India influence the accuracy of analysts’ earnings forecasts. We contrast traditional institutions of caste and regional language with contemporary institutions such as universities as the locus for such ties. We posit that CEOs from the post-economic-reform generation in India are more likely to transfer material private information via their school ties while pre-reform generation CEOs favor caste or language ties. We then contrast domestic business groups (BGs) with western MNCs as organizational contexts and argue that BGs legitimate the transfer of private information along particularistic ties, whereas MNCs mitigate such transfers. Our conceptual framework is supported by analyses that draw on a sample of 1,552 earnings forecasts issued from 2001 to 2010 by 296 equity analysts. Our findings suggest that the embeddedness perspective should be broadened to incorporate the influence of larger historical social structures within which economic action is embedded, and to view BGs as carriers and repositories that blend modern management practices with particularistic behavioral patterns among top executives.