Primary status, complementary status, and organizational survival in the U.S. venture capital industry

Social Science Research 52 (2015) 588–601

Contents lists available at ScienceDirect

Social Science Research journal homepage: www.elsevier.com/locate/ssresearch

Primary status, complementary status, and organizational survival in the U.S. venture capital industry q Matthew S. Bothner a,⇑, Young-Kyu Kim b, Wonjae Lee c a

ESMT European School of Management and Technology, Schlossplatz 1, 10178 Berlin, Germany Korea University Business School, 145 Anam-ro, Seongbuk-gu, Seoul 136-701, Republic of Korea c Graduate School of Culture Technology, Korea Advanced Institute of Science and Technology, 291 Daehak-ro, Yuseong-gu, Daejeon 305-701, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 9 September 2009 Revised 12 March 2015 Accepted 14 March 2015 Available online 3 April 2015 Keywords: Status Networks Firm survival Venture capital

a b s t r a c t We introduce a distinction between two kinds of status and examine their effects on the exit rates of organizations investing in the U.S. venture capital industry. Extending past work on status-based competition, we start with a simple baseline: we describe primary status as a network-related signal of an organization’s quality in a leadership role, that is, as a function of the degree to which an organization leads others that are themselves well regarded as lead organizations in the context of investment syndicates. Combining Harary’s (1959) image of the elite consultant with Goffman’s (1956) concept of ‘‘capacity-esteem,’’ we then discuss complementary status as an affiliation-based signal of an organization’s quality in a supporting role. We measure complementary status as a function of the extent to which an organization is invited into syndicates by well-regarded lead organizations—that is, by those possessing high levels of primary status. Findings show that, conditioning on primary status, complementary status reduces the rate at which venture capital organizations exit the industry. Consistent with the premise that these kinds of status correspond to different roles and market identities, we also find that complementary status attenuates (and ultimately reverses) the otherwise favorable effect of primary status on an organization’s life chances. Theoretically and methodologically oriented scope conditions, as well as implications for future research, are discussed. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction ‘‘For each role, there is a certain complementary role . . . there can be no leader without followers.’’ [Turner and Shosid (1976)] Syndication networks present an opportunity to move beyond prevailing models of status by seeing status in more than just one dimension. Consider investment banks joined through syndicates (Rowley et al., 2005), scientists linked through coauthorship teams (Stuart and Ding, 2006), or Hollywood actors connected through casts (Zuckerman et al., 2003). In q This work was funded in part by the Ewing Marion Kauffman Foundation and the National Research Foundation of Korea (NRF-2013S1). The views expressed in this article are those of the authors and not necessarily those of the Ewing Marion Kauffman Foundation or the National Research Foundation of Korea. We have benefited from the advice of Peter Bearman, Ron Burt, Niko de Silva, Mark Mizruchi, John Padgett, Elizabeth Pontikes, Catalina StefanescuCuntze, Toby Stuart, and Ezra Zuckerman. ⇑ Corresponding author. E-mail addresses: [email protected] (M.S. Bothner), [email protected] (Y.-K. Kim).

http://dx.doi.org/10.1016/j.ssresearch.2015.03.010 0049-089X/Ó 2015 Elsevier Inc. All rights reserved.

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

589

the syndicates and syndicate-like structures that weave such networks together, organizations or individuals can play more than one valued role: leader or supporter. In banking syndicates, lead banks typically advise clients directly, while co-leads specialize in marketing. On coauthorship teams, first authors often craft new theories, while second authors run experiments and import new methods. In film, lead actors shoulder the narrative, while supporting actors play ancillary, though vital, parts. Using this simple difference in roles as our point of departure, in this article we introduce a distinction between two kinds of status and then examine their effects on organizational survival in a setting where syndication networks (Sorenson and Stuart, 2001) are a defining feature of market structure: the U.S. venture capital (VC) industry. We refer to the first kind of status, familiar from prior research (e.g., Podolny, 2005; Sauder et al., 2012), as primary status, because it is related to the enactment of a leadership role. Consistent with past work, an organization holds a desirable position in a latent primary status order to the degree that it assumes leadership positions—in syndicates or syndicate-like structures, such as alliances, groups, and teams—above other well-regarded lead organizations. Likewise, individuals endowed with this first kind of status, like a company president or CEO, reside near the top of a visible pecking order (Bonacich, 1987).1 In contrast, we refer to the second kind of status as complementary status, because it is related to occupying a supporting role. An organization occupies a valued position in a latent complementary status order to the extent that it is frequently sought out to fill auxiliary roles in syndicates by well-regarded lead organizations, that is, by those who possess high levels of primary status. Similarly, individuals with this second kind of status, like a CEO’s special advisor, certainly play a coveted role, but one that is ‘‘outside’’ the main pecking order. Thus, the two kinds of status on which we focus are analytically separate, but conceptually (and methodologically) nested: while each kind of status is related to a unique role in a system of exchange—leader or supporter—a focal actor’s complementary status is shaped by the primary status of those that the focal actor supports. We can summarize our main motivations for introducing this distinction by considering this question: What would status theorists miss if they were to overlook complementary status? Of course, for some social settings, nothing would be missed. In some domains, corporate or individual actors occupy just one important role—for instance, that of the producer—and the audience members who confer status agree strongly on the (narrow) criteria according to which recognition and esteem should be dispensed (Shils, 1994). Looking ahead to a later sketch of scope conditions on our hypotheses and results, a supporter at times has no chance to gain status-conferring recognition from a leader; the supporter is only lower-ranked or, worse, a sycophant.2 Conversely, in other contexts, leader and supporter roles are filled, and, accordingly, there are separate paths to the different kinds of status that concern us. Consider further a main feature of the labor market in the film mentioned earlier: among the actors and actresses comprising a cast, some get top billing and others jockey for auxiliary parts—often hoping for parts alongside bankable stars. Meryl Streep and Will Smith consistently enjoy top billing; Siobhan Fallon and Jason Alexander do not, though they nonetheless have garnered esteem from Hollywood’s denizens for their contributions as members of the supporting cast. Early in his career, Samuel L. Jackson played supporting roles with much success, but more recently he has improved his standing as a lead. Robert De Niro has long enjoyed considerable esteem in both kinds of roles. In our view, to overlook complementary status in contexts like film, and in related settings in which syndicate-like structures abound, is to bypass at least three important questions: First, although leaders cannot (by definition) esteem supporters in the same way that supporters esteem leaders, what might a plausible method look like for aggregating flows of recognition that do accrue from leaders to supporters in a measure of complementary status? Second, does complementary status affect life chances in a market, conditioning on primary status, and if so, to what extent? Third, does the known, favorable effect of primary status on market outcomes vary with complementary status, and if so, what are the implications for status theory? We explore these questions in the context of the U.S. venture capital industry by constructing yearly relational matrices that depict VC firms’ patterns of syndicated investment as leads or as co-investors. Using the primary and complementary status scores resulting from this analysis, we consider their effects on VC firms’ rates of exit, defined as seven or more years of failing to invest in any target company. Before turning to these measures and models, we develop two hypotheses about the main effect of complementary status and its interaction with primary status. Expectedly, our baseline expectation—and thus the implicit null for our first hypothesis on complementary status— reflects the wisdom of prior work on status-based competition: what we refer to as primary status is the only status dimension that matters for organizational survival. We nonetheless see this as an interesting null against which to argue for two reasons. One reason relates to the large amount of prior work tracing market outcomes to just one status dimension (see esp. Podolny, 2005). The notion that just one status ordering shapes outcomes in markets is almost taken for granted, and so exploring the possibility that two dimensions matter, at least in some markets, strikes us as important for advancing status 1 This claim prompts an interesting question: Does position in an asymmetric network virtually equate to status? Or, is working from an asymmetric sociomatrix merely a good strategy for approximating a social factor that is much harder to pin down? On the one hand, if one closely follows Podolny and Phillips’s (1996) conception of status as a ‘‘stock’’ built up from ‘‘flows’’ of deference, and if the act of one bank publicly ‘‘yielding’’ to another bank captures much of the deference exchanged between them, then Bonacich’s measure and status itself are closely aligned. On the other hand, one might argue (as one colleague did) that the opportunity to lead others in syndicates is made possible by status. One’s view here depends on (i) how much of the status–relevant deference or recognition is in fact captured in the cells of the relational matrix and, more generally, on (ii) how status is defined. 2 Separate from Bonacich status, a measure of sycophantic behavior can still be computed from the same sociomatrix. Using the weekly sociomatrices recording likability rankings in Newcomb’s fraternity, Bothner et al. (2010, pp. 962–963) present a weighted asymmetry measure, capturing the extent to which a given fraternity member returns deference to those from whom he receives disdain.

590

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

theory. Second, through exploring interactions between primary and complementary status, we also see an opportunity to better understand when status (of either kind) may end up being inconsequential for market outcomes. Since primary status and complementary status represent different dimensions on which status-holders are understood and coded, our intuition is that they ‘‘interfere’’ with each other in models of exit rates. We next sketch reasons for expecting a favorable main effect of complementary status and then turn to our reasons for anticipating this interaction. 2. Complementary status and life chances Our conception of complementary status draws mainly from the intuitions of Harary (1959) and Goffman (1956). (Harary’s, 1959, pp. 33–34) description of the role of a ‘‘consultant’’ to an organization’s leader highlights the fact that, at first glance, a member of a group can seem to lack status altogether if we see status one-dimensionally. In Harary’s discussion, the consultant lacks what we refer to as primary status, because he or she reports only to the leader and supervises no one. When viewing status solely as a function of location in a pecking order (Chase, 1980), his or her status is zero. But since the leader has exalted (and thus openly recognized) the consultant as a confidant, he or she has status of another kind—what we refer to as complementary status. Harary’s example is important because it points to the often dual nature of recognition, and thus to the dual nature of status, in settings where leaders meaningfully differentiate among occupants of supporting roles: while the consultant’s support of the leader advances the leader’s primary status, the leader’s exaltation of the consultant gives the consultant complementary status. In more abstract terms, A’s primary status grows as B supports A, but B’s complementary status rises as A recognizes and selects B. It is this process of acknowledging or differentiating among supporters by elite leaders that generates complementary status. Imagine further that A has more primary status than E and that B and F are supporters. In a status tournament for supporting roles, B ranks ahead of F under these conditions: B supports A, while F supports E; or, B and F jointly support A and E, but B is more strongly (less strongly) affiliated with A (with E) than is F. Either way, returning to Harary’s metaphor, B is the more prestigious ‘‘consultant’’ in the status contest with F. (Goffman’s, 1956, pp. 478–479) critique of classical models of status for overly privileging dominant social roles is also relevant to our conception of complementary status. Goffman highlighted the failure of early work to exploit theoretically the two-way nature of recognition or deference: ‘‘In thinking about deference it is common to use as a model the rituals of obeisance, submission, and propitiation that someone under authority gives to someone in authority. Deference comes to be conceived as something a subordinate owes to his superordinate. This is an extremely limiting view of deference. . . There are deference obligations that superordinates owe their subordinates.’’ These obligations owed to supporters by leaders differ in character from the most strictly defined (i.e., most subservient) flows of deference that flow from supporters to leaders. Yet it is also the case that the flows of commendation, recognition, and respect going from leaders to supporters matter for supporters’ standing in their role as supporters. (Goffman’s, 1956, p. 479) related discussion of ‘‘capacity-esteem’’ informs our view of complementary status as a market signal of otherwise undetectable quality in a supporting role and supports our hypothesis that complementary status favorably affects survival. Very simply, capacity-esteem refers to the recognition accruing from a leader to a supporter because of the specialized capabilities that the supporter supplies. Earlier work offers examples of two general types of capabilities likely to be associated with capacity-esteem in various empirical settings. One type is technical knowledge that is local and indispensable (Berger and Luckmann, 1966, pp. 77–78; Kanter, 1977), such as the deep expertise possessed by subcontracting firms serving general contractors in the housing market (Eccles, 1981) or the indigenous institutional knowledge of legislators working behind the scenes on congressional committees. The second type comprises softer, less tangible characteristics. For instance, past work has described proficient supporters as characterized by trustworthiness (Useem, 2001), and proactivity and independent thought (Gibbons, 1992); as having a facility for allaying organizational dysfunction (Riggio et al., 2008); and, more generally, as characterized by the discretion necessary to influence superiors without falling ‘‘out of role’’ (Kipnis and Schmidt, 1988). Thus, able occupants of ancillary roles, whether individuals or organizations, are thought to offer particular skills in a manner that preserves rather than challenges the legitimacy of their superiors and, in this sense, to comply with salient norms. Much as earlier research has envisioned (primary) status as a general signal of quality favoring elite firms in uncertain markets (Podolny, 1993; Podolny et al., 1996), we view complementary status as a more specific signal of difficult-to-discern quality in an ancillary role, and one that independently increases life chances—as long as audience members value capacities not only in traditionally prized, leading roles, but in auxiliary roles as well. Our main expectation, thus, is the following: Hypothesis 1. An organization’s life chances increase with its complementary status. Theoretical antecedents of our approach also lead us to anticipate that the signal strength, and thus the advantages for survival, of an organization’s primary status hinges on its complementary status. In particular, we expect the favorable main effect of primary status to wane as an organization’s complementary status rises. Primary status and complementary status orders are two different paths toward the development of separate kinds of ‘‘identities that can have value in exchange relations’’ (Podolny et al., 1998, p. 3, see esp. pp. 3–4). To the degree that an organization (or person) receives extensive recognition in its role as a supporter in syndicate-like structures, that organization will be known and classified for a specific set of corresponding capabilities, and thus come to benefit from a distinct identity-related asset (cf. Zuckerman et al., 2003). While

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

591

benefiting from the clarity of this identity, however, we also anticipate that the organization’s positioning over other lead organizations will fail to attract the attention of audience members or, worse, elicit negative attention. Getting coded as a ‘‘high-end second fiddle’’ is constraining: audience members settle on where the high-complementary-status organization can (and cannot) contribute, and thus discount, ignore, or even penalize its attainment in the other status hierarchy. Under these circumstances, the positive signals otherwise sent by primary status are clouded or even read negatively. Clouding can result from the shadow cast by the supporter’s known strength as a supporter. Negative reactions can result from audience members’ collective sense that the endorsements that a supporter receives in a leadership role are misplaced. Audience members may see these endorsements as illegitimate because of their specific beliefs about ‘‘trained incapacity’’ (Veblen, 1914) or their more general normative conviction that those known for strength in supporting roles ought not to remake themselves as leaders. Consequently, in models predicting organizational life chances, we expect that complementary status interferes with the advantageous main effect of primary status on survival. We therefore predict: Hypothesis 2. The favorable effect of primary status on an organization’s life chances declines with its complementary status. 3. The U.S. venture capital industry To test our predictions, we examined how complementary status affects VC organizations’ rates of exit from the industry. VCs have fueled many of the most consequential innovations of the modern economy, such as the PC, the operating system, retail medicine, genetic engineering, and online education (Gompers and Lerner, 2001; Zygmont, 2001). One of the VC industry’s main contributions to economic growth has been to back young concerns whose revenues are prohibitively low for entry into the public markets and whose levels of risk are too high for banks. In addition to supplying capital, VCs often perform a number of hands-on functions for the companies they support, such as recruiting and counseling executives, finding customers and suppliers, creating incentive packages, and selecting the underwriters and other professionals required for successful initial public offerings. Many trace the industry’s beginnings to the 1940s, but its expansion since the 1980s has been especially pronounced, with the estimated amount of money committed by VC firms growing from $700 million in 1980, to nearly $3.5 billion in 1990, to more than $81 billion in 2000 (Gompers and Lerner, 2001, pp. 72–73). Two considerations made the VC industry an attractive empirical setting for our analyses. First, VC organizations’ chances of survival serve as a clean measure of performance whose dependence on both kinds of status we could readily examine. Although other VC firms at times constitute an important audience for a focal VC firm (for example, peer firms can keep a chosen firm afloat by offering silver-platter deals and by serving as references), a focal VC firm fails principally because limited partners—its most consequential audience—lose confidence to such an extent that the focal VC cannot raise new capital and is thus forced to exit. Examples of limited partners are Adams Street Partners, CalPERS, and the New York State Teachers’ Retirement System, as well as universities, insurance companies, pension funds, endowments, and other intermediaries and investors who supply the capital that VCs channel into entrepreneurial start-ups. The VC industry is an appropriate setting for examining the effects of both kinds of status on rates of exit because in the fund-raising process, beyond relying on past financial performance, limited partners are known both to question VCs about prior instances of leading and co-investing and to scrutinize affiliations established in the context of syndicates. Second, the frequency with which VC firms invest jointly in syndicates enables us to devise yearly measures of primary and complementary status for the firms in our panel. Syndicates offer an opportunity for spreading risk across a range of target companies and for fostering diversity in perspectives and approaches, which can in turn enhance the quality of VCs’ judgments as they shepherd their targets toward maturity (Lerner, 1994; Kogut et al., 2007). Syndicates are often formed by a lead VC, which identifies the target company, invests in the first round, and then recruits other VC firms to participate in later rounds (Sorenson and Stuart, 2008).3 Although subsequently invited co-investors perform many important ancillary functions, the lead VC assumes the principal role in evaluating and managing targets—known in industry parlance as ‘‘grinding and minding.’’ Thus, the role differences within syndicates enable us to operationalize primary status as a function of the degree to which a focal VC firm leads other VC firms, which in turn lead others, and complementary status as a function of the extent to which a chosen VC firm is invited into syndicates by others that frequently occupy desirable lead roles.

4. Measuring primary status and complementary status We constructed our status measures using the VentureXpert database compiled by Thomson Financial (see Sorenson and Stuart, 2001; Podolny, 2001) and in doing so we relied on VC organizations’ patterns of syndicated investment in U.S.-based start-ups from 1980 onward. We began measuring status (and other time-varying covariates) in 1980 because of the comparatively complete nature of the data in VentureXpert starting in that year. 3 We identified a focal VC as a lead VC for a given syndicate using these two criteria: the focal VC firm invested by itself in the first round or, if more than one VC firm invested in the first round, the focal VC invested in that round and in the maximum number of rounds in which any VC firm participated (cf. Sorenson and Stuart, 2008, pp. 277–78).

592

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

Table 1 An example of a relational matrix used to compute primary status.

VCi VCj VCk VCq

VCi

VCj

VCk

VCq

Row sums

Primary status scores

0 3 1 0

3 0 0 1

2 1 0 0

8 0 3 0

13 4 4 1

1.58 1.07 0.55 0.22

Consider first the four-by-four relational matrix in Table 1 for a simple illustration of how we measured primary status using Bonacich’s (familiar) method. Starting with this brief review offers a useful backdrop for discussing the main features of our measure of complementary status. The matrix in Table 1 records instances of four imaginary VC organizations—i, j, k, and q—investing jointly in target companies in a given year. It distinguishes between lead VCs in the rows and co-investors in the columns, so that each entry counts the companies in which both organizations invested—as a lead or as a co-investor. The lead role is again the basis for primary status, while the co-investor (or supporter) role is the basis for complementary status. Reading across the first row in Table 1 for VC i, we see first that i and j jointly invested in three companies, with i as the lead and j as a co-investor. In a lead role, VC i also funded two companies with k, and eight companies with q—again with k and q as co-investors. Therefore, for any given row in the matrix shown in Table 1, the entries approximate the (latent) degree to which firms in the columns recognize or defer to—as co-investors—the lead VC in that row. That is, the row vectors proxy the interorganizational flows of recognition directed to leaders on which primary status rests. Keeping with this illustration, to capture time-varying levels of primary status for the organizations in our panel, we began by assembling asymmetric relational matrices Rt that mirror the hypothetical matrix in Table 1, for all organizations in each year. For a given annual matrix, cell Rijt records the number of target companies in year t in which organizations i and j jointly invested, where i was the lead VC and j was a co-investor. Using Bonacich’s (1987) infinite-sum method, we then collected yearly primary status scores as follows:

Pit ða; bÞ ¼

X ða þ bPjt ÞRijt

ð1Þ

j

where P it is an element of the vector Pt and denotes the primary status of organization i in year t.4 We selected the scaling parameter a so that, regardless of the size of the network, in each year the organization for which P it equals 1 does not possess an excessively high or low level of status (Bonacich, 1987, p. 1173). We thus chose a so that the squared length of Pt equals the number of firms in Rt , allowing for meaningful comparisons across yearly networks of different sizes. The parameter b determines the extent to which the status of a focal organization is shaped by the status levels of the coinvestors in its syndicate (Bothner et al., 2010, pp. 950–951). If b equals zero, status scores are perfectly correlated with organizations’ row sums in the underlying relational matrix and so do not take into account the status levels of those joining them as co-investors. Returning to Table 1, this is apparent in the equal row sums of j and k. In contrast, as b grows larger, organizations’ status levels are increasingly shaped by the status levels of their affiliates. We set b equal to 3=4 of the reciprocal of the largest normed eigenvalue of Rt (Podolny, 2005). Using this rule for b, the sixth column of Table 1 reports different primary status scores for j and k, with j now outranking k because j leads i—itself a high-primary-status organization—more frequently than k does. Turning now to complementary status, we induced this second status dimension from RTt , the transpose of the relational matrix used to measure primary status. The key substantive element in our measurement strategy is what we have referred to as the duality of recognition: as a focal co-investor joins a syndicate headed by a particular lead, that co-investor recognizes that lead (contributing to that lead’s primary status); yet, going the other way, by bringing in that co-investor, the lead also recognizes the co-investor (contributing to the co-investor’s complementary status). One way to see how this two-way process can be captured, at least partially, from network data is to consider cell (1, 4) in Table 1, which equals 8, from two different angles. When calculating primary status scores, we viewed this cell in the relational matrix as a proxy for the latent flow of (primary) status-conferring recognition emanating from VC q to VC i, with q investing in eight distinct target companies in syndicates headed by i. We view this cell in the matrix differently when measuring levels of complementary status. Consider now Table 2, which depicts the transpose of the matrix from Table 1 and, in particular, cell (4, 1) which of course equals 8. In Table 2, we can now view this entry as a proxy for the flow of (complementary) status-conferring recognition going from VC i to VC q. This is because we have evidence of i recruiting or inviting q into eight syndicates headed by i. Thus, for any chosen row in the transposed matrix in Table 2, the entries capture the degree to which firms in the columns, through inviting and recruiting, recognize—as lead VCs—the co-investor in that row. That is, the row vectors constituting the transposed matrix in Table 2 proxy the interorganizational flows of recognition directed to supporters on which complementary status rests. Reading across the fourth row in Table 2, for instance, VC q was recognized moderately by k and extensively by i. 4 Alternatively, Eq. (1) may be expressed in matrix form (and most easily computed using standard packages) as Pt ða; bÞ ¼ aðI  bRt Þ1 Rt 1, where I is an identity matrix and 1 is a column vector of ones.

593

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601 Table 2 An example of a transposed relational matrix used to compute complementary status.

VCi VCj VCk VCq

VCi

VCj

VCk

VCq

Row sums

Complementary status scores

0 3 2 8

3 0 1 0

1 0 0 3

0 1 0 0

4 4 3 11

0.47 0.61 0.52 1.77

Fig. 1. Venture capital organization’s levels of primary status and complementary status.

Consistent with this, to calculate time-changing levels of complementary status for the organizations in our panel, we worked from the transpose of each yearly relational matrix Rt . Using RTt allowed us to appraise the extent to which organization i is sought out for supporting roles by others who are well-regarded in the role of leader of syndicates. Thus, cell ði; jÞ of RTt tallies the count of distinct entrepreneurial ventures that both i and j financed in year t where i was a co-investor and j was the lead. We computed yearly complementary status scores as:

C it ða; v Þ ¼ a

X T v Rijt Pjt

ð2Þ

j

where C it is an element of Ct , denoting the complementary status of organization i in year t.5 We chose the scaling factor a as described above. With the weighing exponent v set to one rather than zero, organizations receive an added boost in complementary status insofar as those inviting them into syndicates are themselves high in primary status. Setting v > 0 is consistent with conceiving of complementary status as nested with primary status: although complementary status is conceptually and mathematically distinct from primary status (as long as Rt – RTt , as we later discuss when sketching scope conditions on our approach), complementary status is still intertwined with primary status. We conceive of organizations as contestants in tournaments for status in lead roles and in supporting roles. Two organizations with equal amounts of ‘‘activity’’ in supporting roles can nonetheless differ in standing based on the different (primary) status levels of those they support. In Table 2, and in our subsequent analyses, we set v equal to one. We recommend v ¼ 1 unless there are strong context-specific justifications for over- or under-weighting the influence of affiliation with elite leaders in the measurement of complementary status.6 We illustrate levels of primary status and complementary status for the organizations in our panel in Fig. 1. The upper-left zone of this space contains VC organizations well-known for originating deals. These organizations forge close ties with Eq. (2) may be expressed in matrix form as Ct ða; v Þ ¼ aRTt Pvt . Consider, for instance, a winner-take-most setting in which getting endorsements from the center of the elite core of the network (i.e., from those deep in the right tail of the primary status distribution) matters much more than receiving endorsements from those on the edge of the core. For such a context, an especially large v exponent is appropriate. 5 6

594

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

entrepreneurs from the earliest phases of companies’ growth, and then bring in syndicate partners. Draper Fisher Jurvetson is an example of an organization occupying this region of the space. Another is Austin Ventures. Conversely, in the lower-right region, we see a different approach to participating in the VC industry. Here, organizations rarely originate, but instead are known for adding their specialized capabilities and financial resources to existing syndicates. Intel Capital is one occupant of this area. Another is JP Morgan Partners, a New York–based prepublic investor. In addition, although the upper-right-most region, strictly defined, is unoccupied, a minority of organizations has nonetheless successfully played the roles of both leading and supporting. Some of the most well-known firms, such as New Enterprise Associates and Oak Investment Partners, occupy this area of Fig. 1. 5. Estimation and conditioning variables To test our hypotheses, we estimated the instantaneous hazard of exiting the VC industry. The hazard rate may be expressed as ki ðtÞ ¼ limDt!0 ðpi ðt; t þ DtÞ=DtÞ, with t denoting a focal VC organization’s industry tenure and pi signifying the probability of exit within the designated time interval. We estimated continuous-time models because of the availability of precise (to-the-day) information on the dates of VC organizations’ first investments in target companies (cf. Barron et al., 1994). To avoid imposing restrictive assumptions on the dependence of the hazard on industry tenure, we estimated Cox proportional hazards models of the following form:

ki ðtÞ ¼ k0 ðtÞ exp½h1 Pit2 þ h2 C it2 þ Xit1 b þ rkðiÞ þ crðiÞ þ st 

ð3Þ

which allow the baseline rate k0 ðtÞ to vary nonparametrically as a function of time-at-risk. We also split the spells for each organization into yearly spells (Tuma and Hannan, 1984), because many of our explanatory variables shifted over time, and we then updated these covariates each calendar year. Yearly spells were right-censored if in that year an organization was still participating in the industry (that is, if it had invested in at least one target company). To update our covariate recording industry tenure, we coded exits as occurring in the midpoint of a year, if that year was the first of seven or more years during which the focal VC organization failed to invest in any targets. Consequently, although we followed organizations’ life histories through 2006, the final year of our panel was 1999, allowing a full seven-year horizon in which to confirm exits. Of the 2021 organizations tracked in our fully specified models, we observe 565 exits (28%) from the industry. Turning to our explanatory variables, we expect that h1 and h2 will be negative, which is consistent with earlier literature and with our first hypothesis. Except for our measures of primary and complementary status, we lagged all time-varying covariates one year, matching a focal firm’s tenure with a predetermined level on each time-changing predictor. We lagged our status covariates an additional year for two reasons: first, to ensure that our effects of interest are not an artifact of a nonlinear effect of size, and second, since industry exit is, by construction, related to volume of investment activity. Using a two-year lag structure for status permits us to guard against the possibility that an unobserved process simultaneously weakens status and raises the hazard of exit. Our expectation is that, net of other factors likely to raise life chances, primary status and complementary status will discernibly reduce the rate of departure. Xit1 contains eight time-changing, organization-specific adjustments. Table 3 reports correlations and descriptive statistics for all variables included in our models. Starting with the fifth row of Table 3, we measured size as the count of the number of target companies in which the focal VC organization invested in that year. Earlier work has shown that size significantly affects survival (Baum and Mezias, 1992; Barron et al., 1994), and in the venture industry, size also serves as an important signal for limited partners and other constituents. When VCs back multiple targets, they are better positioned to justify their stewardship of the financial resources of limited partners as they raise new capital. Assuming a strong screening process, the number of targets funded is positively associated with the probability of strong financial returns. For members of the venture community, volume or ‘‘deal flow’’ is commensurate with VC firms’ ‘‘reach’’ or ‘‘ability to get into good investments with good entrepreneurs’’ and is generally seen as an indication of their financial ‘‘performance four years down the road’’ (Ellison, 2005). To proxy the extent to which the focal firm has been performing well in the eyes of limited partners, and thus has substantial capital under management, we include the variable funds, which counts the number of distinct funds from which a firm draws in a given year (Podolny, 2001). Our expectation is that funds will strongly and negatively affect the rate of exit. To adjust for the favorable consequences of initial public offerings, we included ipos, which tallies the number of target companies funded by the focal firm that went public that year. Using this measure, we can separate the consequences of status from the returns an investor enjoys as its companies enter the public markets. We expect ipos to lower exit chances. Our final set of time-changing covariates are firm-level diversification across the ten distinct industry categories listed in our database, proportions of investments in specific industry categories, and stagefocus. We expect diversification to raise survival chances given the environmental variation endemic to the VC industry (Freeman and Hannan, 1983; Péli, 1997) and in light of associated opportunities for scope economies and organizational search (Miller and Chen, 1994). We calculated a yearly Herfindahl-based measure for each firm as follows:

diversificationit ¼ 1 

10 X j¼1

p2ijt

ð4Þ

595

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601 Table 3 Descriptive statistics and correlations for variables in the analysis (N = 9653). Variable

Mean

1 2 3 4 5 6 7 8 9 10 11 12

Exit Tenure Primary Status Complementary Status Size Funds IPOs Diversification Computer Software and Services Internet Specific Medical/Health Stage Focus

6 7 8 9 10 11 12

Funds IPOs Diversification Computer Software and Services Internet Specific Medical/Health Stage Focus

SD

Min.

Max.

1

2

3

4

0.235 7.397 1.124 1.03 10.785 1.371 1.840 0.306 0.237 0.189 0.225 0.54

0 1.5 0 0 1 1 0 0 0 0 0 1

1 53.497 13.144 11.527 246 16 36 0.892 1 1 1 3

0.108 0.083 0.11 0.154 0.141 0.098 0.279 0.023 0.038 0.027 0.042

0.296 0.288 0.344 0.375 0.346 0.26 0.001 0.016 0.038 0.087

0.544 0.659 0.502 0.41 0.279 0.056 0.005 0.013 0.058

0.66 0.453 0.523 0.338 0.089 0.007 0.014 0.045

5

6

7

8

9

10

11

12

0.685 0.618 0.511 0.053 0.059 0.003 0.055

0.482 0.432 0.045 0.069 0.017 0.009

0.287 0.049 0.065 0.016 0.037

0.039 0.025 0.009 0.063

0.02 0.161 0.087

0.125 0.073

0.091

0.059 9.573 0.39 0.577 8.241 1.844 0.969 0.484 0.161 0.073 0.127 2.011

0

where pijt is the proportion of firm i s investments in industry category j at t.7 Separate from breadth of reach across market subsections, we also included adjustments for three market categories that, based on an initial investigation of exit rates, appeared especially consequential: Computer Software and Services, Internet Specific, and Medical/Health. We entered yearlyvarying proportions of total investments in each of these industry categories, expecting these controls to lower the hazard. Our measure of stagefocus is an investment-weighted average ranging from 1 (for investing exclusively in early-stage, rather than middle- or late-stage, ventures) up to 3 (for investing solely in late-stage targets). Adjusting for stagefocus is essential for ensuring that a favorable effect of status on survival is not a reflection of two factors that are likely associated with later-stage investing: a larger number of syndicate partners and lower risk. We finally included three kinds of fixed effects: indicators for firm type and region, denoted by rkðiÞ and crðiÞ respectively, as well as separate dummy variables for each calendar year, st . Using VentureXpert’s classifications, we first created 33 dummy variables to account for heterogeneity in the kinds of organizations participating in the VC industry. Together with classical VC firms (such as Kleiner Perkins Caufield and Byers, Sequoia Capital, Oxford Bioscience Partners, and others), our database includes angel investors, corporate venture programs, endowments, investment banks, merchant banks, and governmental organizations, among others (cf. Podolny, 2001, pp. 48–50). Organization-type fixed effects are important for separating the effects of status from the advantages (or disadvantages) of occupying particular subpopulations of the VC industry. Second, given the strong effects of spatial positioning on VC organizations’ strategic conduct established in prior research (e.g., Sorenson and Stuart, 2001), we include a separate indicator for each state in the United States in which VC organizations are located.8 Entering these state-level adjustments allows us to sweep out time-invariant regional differences in opportunities for investment and collaboration (Florida and Kenney, 1988; Elango et al., 1995) that possibly vary with status and affect life chances. Third, year-specific indicators st adjust for various aggregate fluctuations, such as the supply of targets, constraints on exit opportunities for portfolio companies, capital availability, limited partners’ inclinations, and related factors, allowing us to separate the impacts of status from the broader industry-wide factors. Using year indicators permits us to focus conservatively on cross-sectional effects, with all forms of aggregate temporal variation kept constant. 6. Results Table 4 presents estimates from Cox models of the hazard of exiting the VC industry. We start with the simplest possible specifications: in model 1, we enter only primary status, and in model 2, we include only complementary status. Model 1 concurs with the findings of many prior studies of the performance-related advantages of receiving recognition from esteemed peers. In model 2, the effect of complementary status is also strongly negative. Overall, we thus observe initial associations consistent with our expectations: status in both role domains is negatively associated with market exit. Importantly, no controls (beyond time-at-risk) are necessary for observing a significant coefficient on either type of status. 7 VentureXpert codes each target company as an occupant of one of the following ten broadly-defined industry categories: biotechnology, communications and media, computer hardware, computer software and services, consumer related, industrial/energy, internet specific, medical/health, other products, and semiconductors/other electronic. 8 We generated a total of 52 state-level indicators to accommodate two additional categories: location outside of the United States and unknown location. Similarly, one of the indicators created for firm type corresponded to unknown type.

596

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

Table 4 Cox estimates of exit rates. 1 Primary Status

2

3

5.48*** (1.126)

4

5

6

7

8

1.188* (0.541)

1.099* (0.527) 0.85*** (0.189) 0.447*** (0.069) 0.319* (0.16) 0.096 (0.076) 0.675* (0.287) 0.031 (0.139) 0.058 (0.256) 0.123 (0.157) 0.197** (0.061)

1.168* (0.533) 0.869*** (0.188)

1.939** (0.651) 0.985*** (0.215) 0.485*** (0.075) 0.23 (0.177) 0.114 (0.082) 0.567 (0.31) 0.029 (0.151) 0.031 (0.275) 0.1 (0.165) 0.21** (0.066)

1.187* (0.552) 0.864*** (0.189) 0.448*** (0.069) 0.32* (0.16) 0.096 (0.076) 0.671* (0.287) 0.031 (0.139) 0.057 (0.256) 0.123 (0.157) 0.197** (0.061) 0.389***

1.836*** (0.182)

Complementary Status

0.363*** (0.088) 0.422** (0.141) 0.085 (0.058) 0.688* (0.301) 0.111 (0.073) 0.133 (0.111) 0.054 (0.084) 0.134*** (0.033)

Size Funds IPOs Diversification Computer Software and Services Internet Specific Medical/Health Stage Focus

0.471*** (0.069) 0.304 (0.159) 0.16* (0.074) 0.682* (0.289) 0.01 (0.138) 0.094 (0.256) 0.13 (0.157) 0.194** (0.061)

0.307 (0.159) 0.097 (0.076) 0.097 (0.407) 0.03 (0.139) 0.047 (0.255) 0.115 (0.156) 0.198** (0.061)

Primary Status  Complementary Status

9

0.463*** (0.068) 0.323* (0.16) 0.126 (0.075) 0.664* (0.287) 0.014 (0.139) 0.07 (0.256) 0.123 (0.157) 0.198** (0.061)

(0.109) 3.947*** (0.688) 3.182*** (0.633) 2.683*** (0.645) 1.826** (0.687) 1.735* (0.698) 1.753* (0.722) 1.388 (0.825)

Size = 1 Size = 2 Size = 3 Size = 4 Size = 5 Size = 6 Size = 7

0.591** (0.185) 1.217* (0.59) 0.993 (1.023)

Low Primary/High Complementary High Primary/Low Complementary High Primary/High Complementary Firm type dummies Region dummies Year dummies N (firms) N (firm-years) Log pseudolikelihood

No No No 2029 9697 3601.7

No No No 2029 9697 3597.94

Yes Yes Yes 3794 15,150 12548.2

Yes Yes Yes 2021 9653 3057.93

Yes Yes Yes 2021 9653 3045.47

Yes Yes Yes 2021 9653 3042.02

Yes Yes Yes 1746 7068 2618.03

Yes Yes Yes 2021 9653 3045.33

Yes Yes Yes 2021 9653 3052.25

Robust standard errors reported in parentheses. *** p < 0.001. ** p < 0.01. * p < 0.05.

Model 3 includes the full set of conditioning variables described above, without primary or complementary status. In model 3, the estimates largely agree with our expectations. First, organizational size, measured yearly as the number of targets funded by the focal VC organization, discernibly reduces exit chances. With each additional target company funded, the rate drops by roughly 30% (exp[.363] = 0.696). The effect of the count of funds is also negative: organizations in good standing with limited partners are less likely to disband. With each additional active fund, the hazard drops by nearly 35% (exp[.422] = 0.656). We see further that, although success in the IPO market falls short of significance,9 a wider market

9 In other models (for instance, model 4), the effect of the IPOs covariate is significant. We speculate that the estimate on the IPOs covariate may reflect competing forces, apart from which its negative effect on exit would be stronger: on the one hand, success in the IPO market helps a VC firm survive; on the other hand, that success may change investors’ taste for staying in the industry: after seeing its target companies go public, satiated investors may grow less committed to staying.

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

597

position results in a lower rate of exit. Stagefocus is also statistically discernible, and contrary to our expectations, the hazard of exit is greater for organizations invested in late-stage target companies. Coefficients on firm type dummies and region indicators (not shown but available by request), as well as our year fixed effects, match generally held understandings of the industry.10 Model 4 adds to our baseline specification in model 3 the measure of primary status shown in Eq. (1). This model reports a statistically discernible negative effect of primary status (2.20 z-score). It is also substantively meaningful: a standard deviation increase in primary status depresses the rate of exit by roughly 75% (exp[1.19 ⁄ 1.12] = .26). Thus, this estimate reflects our baseline expectation, informed by much prior research, that status communicates firm-specific quality to relevant third parties and thus positively affects life chances. Next, model 5 establishes that complementary status from Eq. (2) independently affects the hazard of exit (4.50 z-score). A standard deviation increase in complementary status reduces the exit rate by approximately 60% (exp[.85 ⁄ 1.03] = .42). This effect is important for three main reasons. First, it provides evidence that, net of status in a leadership role, receiving recognition (in the form of invitations to join syndicates) in a supporting role uniquely raises survival chances. Second, on a more general, theoretical plane, it indicates that a dual conception of status is a fruitful approach for future research, even in empirical contexts in which researchers have so far assumed that just one status hierarchy matters. Third, viewed through a broader, methodological lens, the effect of complementary status suggests that novel insights can result from viewing relational matrices differently. Traditionally, most network-analytic work has examined social systems (and sociomatrices in particular) too restrictively—focusing only on the reception of social ties, or emphasizing only the sending of ties. In contrast, examining both ‘‘sides’’ of a social exchange brings into view a more complete image of a given network position and its consequences. We took two main steps to check the robustness of our core finding that complementary status, net of primary status and other factors, raises survival chances. First, in model 6, instead of entering size linearly, we fit the effect of size as a spline, using seven dummy variables for number of investments made. Eight or more target companies funded was the reference category. Taking this approach addresses the possibility that the effect of complementary status may in fact reflect a nonlinear effect of size. In model 6, we see the expected stepwise decline in the hazard of failure as volume of activity increases, and also that complementary status stays statistically significant (4.61 z-score). Second, in model 7, we ensured that our effect of interest did not hinge on our panel’s point of inception in 1980. While we know the dates of first investment for firms entering before 1980 and can therefore accurately determine time-at-risk, we examined the possibility that the inclusion of these particular firms in our panel affects our main result. In model 7, we estimated a version of model 5 that excludes them. In this model, with only those VC organizations whose dates of entry were after 1979, the effect of complementary status stays strongly significant, giving us added confidence in our effect of interest.11 Having assessed the robustness of our main result, we could assess our second hypothesis: that complementary status attenuates the favorable main effect of primary status on an organization’s life chances. Testing Hypothesis 2 is important because we have depicted each kind of status as related to a distinct market identity—not merely as a unique kind of social capital by which resources arise (cf. Stuart et al., 1999, pp. 321–322). Stated differently, in Podolny’s (2001) apposite terms, we view primary status and complementary status as ‘‘prisms’’ by which organizations are distinguished—not simply as ‘‘pipes’’ through which survival-related resources travel, such as information and opportunities. Consider as a backcloth the following alternative to Hypothesis 2: if the status positions we have proposed correspond only to resource-rich sites in social structure, an interaction effect in which both kinds of status amplify each other seems plausible. Under this alternative scenario, organizational failure would least likely occur in that space of the network where firms enjoy access to the information and opportunities of the strong leaders and of the strong followers with which they affiliate. Conversely, if primary status and complementary status—beyond the resources they make available—also send signals to third parties concerning capabilities in different role domains, and thus map to separate identity-related assets, then we anticipate an interaction in which complementary status interferes with the favorable effect of primary status.

10 More specifically, in model 3 we entered 32 firm-type indicators and 51 regional dummies (some of which were omitted from the analysis because of zero exits within the associated category). Corporate venture programs are, not surprisingly, more exit-prone than private equity firms that invest their own capital. Coefficients on regional dummy variables also concur with expected patterns: while VC firms headquartered in Arizona and Nebraska are comparatively apt to fail, those situated in California and Massachusetts enjoy better survival chances. Moreover, the year fixed effects reflect the salient peaks of the industry’s history, such as the favorable fundraising season of the mid-1990s (Devlin, 1995). 11 As further robustness checks, we estimated two additional versions of model 5. First, instead of entering primary status and complementary status as depicted in Eqs. (1) and (2), we entered primary status with b ¼ 0 and complementary status with v ¼ 0. Using this approach to the measurement of status (one that we find less attractive theoretically because it fails to incorporate the status levels of those recognizing a focal organization) our results were not meaningfully different. Coefficients on primary status and complementary status (thus measured) were .904 (2.41 z-score) and .730 (4.01 z-score) respectively. Second, we also estimated a version of model 5 in which, rather than including firm type dummies, we included only those firms classified as private equity firms that invest their own capital. Seeing whether or not our results hold in this subsample is important for ensuring that our findings do not require the other kinds of investing organizations in the VC industry, such as corporate venture programs, endowments, investment banks, and governmental organizations (cf. Podolny, 2001, pp. 48–50). Using only these private equity organizations, which we view as more committed to staying in the VC industry than their peers, our results were again strongly significant. Coefficients for primary status and complementary status were the following: 2.04 (2.71 z-score) and 1.05 (3.15 z-score).

598

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

We tested our second hypothesis from different angles in models 8 and 9. Model 8 includes a standard multiplicative term. Model 9 uses a dummy-variable approach to better understand this interaction. Model 8 supports Hypothesis 2.12 The positive coefficient on the complementary status-by-primary status covariate shows that the favorable effect of primary status is strongest when complementary status is minimal. This result is consistent with the premise that attaining complementary status elicits attention and congeals audience members’ expectations for a focal organization in a specific market role. As such expectations solidify, primary status no longer matters for survival. Using the descriptive statistics in Table 3, if complementary status equals zero, a standard deviation shift in primary status cuts the exit rate by over 70% (exp[1.187 ⁄ 1.12] = .26). Conversely, the same shift yields roughly a 50% drop in the rate when complementary status is a standard deviation above its mean (exp[(1.187 + .389 ⁄ (.577 + 1.03)) ⁄ 1.12] = .53). Moving out further, the impact of primary status drops to zero when complementary status is just over 3, and, from there, primary status works against survival chances.13 In model 9, we turn to a related specification that explores how exit chances vary across four quadrants of a space whose axes are our two status dimensions. Model 9 is useful for better understanding the interaction effect shown in model 8. We interpreted the estimates from model 8 as follows: when complementary status is low, an increase in primary status is beneficial, but when complementary status is high, an increase in primary status is inconsequential or even detrimental. The implication is that being high in just primary status may in fact be preferable to being high on both dimensions. Coefficients on three dummy variables entered in model 9 support this view. For simplicity, we split both status dimensions at their means reported in Table 3, allowing us to distinguish three status profiles from a baseline category in which the focal firm is low on both kinds of status: (i) low-primary/high-complementary, (ii) high-primary/low-complementary, and (iii) high-primary/high complementary. Of these prototypical identities, high-primary/low-complementary gives rise to the best survival chances: the estimate on its dummy variable is even better than the estimate for the category defined by ranking well on both status dimensions which, while negative, is still insignificantly different from the low/low baseline. Consequently, the pattern we observe suggests an interesting mismatch: organizations that play both roles well from the standpoint of their internal peers (i.e., the other VC firms collaborating with a focal VC firm) are nonetheless devalued from the standpoint of their external audience (i.e., the limited partners supplying, or choosing not to supply, a focal VC firm with capital). This is perhaps the most surprising result in our analysis. Our theoretical perspectives, and associated measurement approaches, assume that any organization (or person) that attains high levels of primary status and of complementary status has received peer recognition. This assumption is virtually definitional: one cannot have the ‘‘stock’’ of status without the ‘‘flows’’ of recognition that constitute that stock (Podolny and Phillips, 1996); and those upon whom these flows converge should enjoy greater life chances (cf. Bothner et al., 2012). Yet, maximizing the total ‘‘amount’’ of these flows from peers—in the form of peers yielding to the leadership of a focal firm, and in the form of peers inviting a focal firm into their syndicates— does not, in fact, maximize life chances. Peer recognition comes in different forms for different types of role performances, and to the extent that audience members see these roles as conflicting, faring well in both role domains nonetheless discounts the status-holder.

7. Discussion and conclusion Our aim was to introduce a new status dimension and show how it affects organizational survival. To bring into focus this important though often overlooked facet of social structure, we drew inspiration from an established vein of earlier work. Research in that vein has underscored the relational underpinnings and many consequences of status differences among diverse kinds of organizations, such as investment banks (Podolny, 1993), semiconductor firms (Podolny et al., 1996), biotech ventures (Stuart et al., 1999), wineries (Benjamin and Podolny, 1999; Roberts and Reagans, 2008), law firms (Phillips and Zuckerman, 2001), commercial banks (Jensen, 2003), and Formula One racing teams (Castellucci and Ertug, 2010). In these studies, organizations reside in a single status order. In contrast, using data on syndication networks, we have portrayed VC firms as participants in two role domains and thus incumbents of two status hierarchies. Our findings have shown that complementary status, adjusting for primary status, raises life chances, and that complementary status can cloud the beneficial signals otherwise sent by primary status. Our findings are important because they offer strong reasons for thinking about status two-dimensionally—with dimensions that are fully separate, yet methodologically nested. 12 Mirroring our sample-split check on model 5, we also estimated a version of model 8 using only private equity firms deploying their own capital. We found a similar set of effects for this subsample. Coefficients on primary status, complementary status, and their product, with z-scores in parentheses, were the following, respectively: 2.18 (2.85), 1.07 (3.20), and .58 (4.65). We also examined the possibility that, in addition to influencing exit chances, both kinds of status might also jointly affect the probability of failing to participate in any syndicate in a given year. Exploring this possibility is interesting: in addition to the limited partners who fund VCs, other VCs are also an important audience for a VC firm at risk of failure. Estimating a logit model based on model 8, in which failure to co-invest in a syndicate was the outcome variable, we found that only the effect of complementary status was significant (7.25 z-score). Neither the main effect of primary status nor its interaction with complementary status was discernibly different from zero. We see industry exit as a better outcome for testing our theory because it more closely reflects the perceptions and actions of limited partners who, more than other VC firms, almost certainly rely more closely on status signals in both role domains. 13 Going back to film, after this point in the complementary status distribution, one can no longer think of VC firms occupying a position that parallels Robert De Niro’s, in which greater stature in a lead role improves survival chances in the market. Later, in closing, we consider features of settings outside of venture capital in which we expect primary and complementary status instead to amplify rather than interfere with each other.

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

599

Nonetheless, a limitation of these findings is that they cannot inform future research in all networks in which (primary) status is measurable. We can meaningfully capture complementary status only in a subset of the networks that interest students of status. This brings us to a discussion of two main scope conditions, before indicating paths for new research: the first scope condition is methodological and follows from reexamining our measure of complementary status; the second arises from reconsidering a nonuniversal feature of the network of firms in the VC industry. 7.1. Scope conditions First, from a methodological angle, one condition that limits the generality of our approach is that the relational matrices—Rt and RTt —cannot equal each other. Observing asymmetric exchange is necessary for implementing our method. Otherwise, returning to Eqs. (1) and (2), as the underlying relational matrix grows increasingly symmetric, primary status and complementary status scores will converge. There are, however, at least two general constraints that make it unlikely that symmetric (or even closely symmetric) matrices summarize an exchange system as it evolves over time. One constraint is simply the cost of identifying the resources—in our setting, the portfolio companies—that are necessary to consummate a social exchange. Assume that one VC firm (A) ushers a peer (B) into three deals in a given year, and that A ‘‘considers them even’’ when in turn B brings A into three deals. It is unlikely that B can offer these deals as ‘‘countergifts’’ back to A in that year. Although the reciprocity norm is among the most binding and ubiquitous rules governing social interaction (Gouldner, 1960), instantaneous repayment is often difficult: countergifts are not always on hand. More generally, a second constraint—whose force is felt most strongly in purely social (rather than economic) exchanges—is strategic delay: the receiver of a gift deliberately waits before repaying the giver due to the receiver’s wish to appear polite (cf. Stack, 1974). If one individual (A) does another individual (B) a favor, and B repays A immediately, this rapid ‘‘settling up’’ signals B’s disrespect for A: B cannot stand to owe A and is therefore unwilling to let the debt simmer. Such displays of disrespect are socially costly, and so they rarely bring symmetry to the asymmetric networks that the measurement of complementary status requires. Second, for our approach to inform new research in other contexts, what we have measured as complementary status cannot approach what Harary (1959) called ‘‘contrastatus.’’ Defined informally, contrastatus is ‘‘the amount of status weighing down on an individual from his superordinates’’ (Harary, 1959, p. 23). Comparing an imaginary advice network with the syndication networks of the VC industry clarifies a main substantive difference between contrastatus and complementary status—despite their potential mathematical similarity. In particular, envision an asymmetric advice network, such as a set of social ties among physicians seeking one another out for guidance on best practices in their field of medicine (cf. Burt, 1987). Using a relational matrix that summarizes this network—where doctors in the rows receive counts of requests for advice from their peers in the columns—primary status is easy to compute: it is a function of the extent to which a focal doctor is sought out for advice by other physicians whose advice is also valued. Getting pursued for counsel by those deemed wise is an uncontroversial basis for status. Consider, in contrast, the upshot of using the transpose of this relational matrix of advice-seeking ties as an input when inducing an alternative status order. The result is clearly a measure akin to contrastatus rather than complementary status. Sending out, rather than receiving, ties to elite receivers would yield a high score on this metric—a metric that captures the opposite of status. Using the methodology shown in Eq. (2) would, in this case, assign high scores to those who seek out elite physicians for advice. While we suspect such a measure could prove valuable in its own right—possibly as a social–structural index of obsequiousness—it differs sharply from complementary status. Going back to our discussion of dual flows of recognition among VC firms in syndicates, the reason for this difference is that advice networks exhibit just a one-way flow of respect: from the advice-seeker to the advice-giver. A dual flow of recognition—from the co-investor to the lead VC (as the co-investor joins and supports the lead) and from the lead VC to the co-investor (as the lead identifies and recruits the co-investor)—is missing from an advice network. So, an essential condition for applying our approach is, again, the process of identifying and recruiting that underlies the construction of complementary status. Our response to concerns about generality is to note that the conditions necessary for our approach, while not universal, are nonetheless prevalent: actors (individuals or organizations) fill leading and supporting roles in syndicate-like structures; these structures emerge as leaders recruit supporters; and audience members, who value skills in leading and in supporting roles, rely on and reward relational signals of (partially unobserved) skills in these roles. Since these conditions are reasonably common, we expect complementary status to affect market outcomes in a wide set of cases in which syndicate-like structures regularly form: scientific coauthorship teams, political action committees, interfirm alliances, banking syndicates, and comparable structures. Consequently, one immediate next step is to investigate the effects of complementary status on organizational performance in a setting, such as banking, where researchers have so far focused on just one kind of status, but where Goffman’s capacity-esteem is both detectable for supporters and likely to affect organizational performance. 7.2. Paths for future research We see three additional avenues for new research: a general implication for network-analytic methods, plus two specific possibilities for new work on status-based competition.

600

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

First, a broad methodological implication is that perhaps the simplest of all matrix operations—taking a transpose—can cast new light on long-studied social processes. While earlier research has exploited information found in transposed sociomatrices (e.g., Gould, 1991, pp. 721–777; Wasserman and Faust, 1994, pp. 210–214; Bothner et al., 2010, pp. 964– 975), there is much important new work to be done in this area, especially as researchers offer substantive interpretations of newly emphasized aspects of social positions. We also suspect that working with different realizations of the same relational matrix will prove fruitful, even when the underlying social ties are only loosely related to recognition and status. Consider an asymmetric relational matrix recording who shares valuable information with whom. Perhaps the most obvious use of such data is to identify prominent individuals as those who receive information from central receivers. But it may also be valuable, particularly for diffusion models, to rearrange such a matrix and capture individuals’ propensity to transmit information to axial transmitters. A host of valuable new angles can clearly result just by looking at relational matrices dually. Second, although complementary status tempers the main effect of primary status for VC firms, an interesting follow-up study would both demonstrate that the opposite can occur in a different setting and offer a persuasive reason for the difference in interaction effects across the two settings. Imagine that, in an industry other than VC, organizations’ (or individuals’) primary and complementary status amplify, rather than counteract, each other’s positive effects on performance. One could think of this as a special case of the Matthew effect (Merton, 1968; Lynn et al., 2009; Bothner et al., 2011): an organization’s status on one dimension prompts its audience to monitor and value more significantly its status on the other (cf. Podolny, 2005, pp. 130–131). Consequently, as a new market forms, getting an early jump on one status dimension increases the returns to improving on the other, quickly bringing dominance to the competitor who is ‘‘quick off the blocks’’ in both status contests. Shifting to the individual level, one might imagine this happening among a group of status-homophilous scientists, for whom lead and supporting roles are equally valued: elites collectively pull ahead of others by trading roles on exclusive coauthorship teams. Moving back to the firm level, having high levels of primary and complementary status at the same time might also benefit performance if the audience recognizes that playing both roles well is very difficult and then rewards the few who do so. In other words, while recognizing that lead and supporter roles require different sets of skills, this audience might see the earning of esteem in both domains as the strongest of true quality signals in Spence’s (1974) original sense: only true superstars do both. Third, we argue in closing that future research can profit from more nuanced examinations of the temporal interplay between primary and complementary status. In future studies, the substantive differences between these two status dimensions can be successfully explored further by looking explicitly at distinct types of sequences of moves (Abbott, 1995, 2001) through a primary-status versus complementary-status space like the one in Fig. 1. We speculate that some sequences of moves—such as ascending early in status as a supporter, and later rising in status as a leader, or vice versa—are more congenial to organizational legitimacy than other sequences. Our observation that organizations occupy distinct positions in separate status hierarchies is a starting point for new portrayals of status and richer understandings of its consequences. References Abbott, A., 1995. Sequence analysis: new methods for old ideas. Annu. Rev. Sociol. 21, 93–113. Abbott, A., 2001. Time Matters: On Theory and Method. University of Chicago Press, Chicago. Barron, D.N., West, E., Hannan, M.T., 1994. A time to grow and a time to die: growth and mortality of credit unions in New York City, 1914–1990. Am. J. Sociol. 100, 381–421. Baum, J.A.C., Mezias, S.J., 1992. Localized competition and organizational failure in the Manhattan hotel industry, 1898–1990. Adm. Sci. Q. 37, 580–604. Benjamin, B.A., Podolny, J.M., 1999. Status, quality, and social order in the California wine industry. Adm. Sci. Q. 44, 563–589. Berger, P.L., Luckmann, T., 1966. The Social Construction of Reality: A Treatise in the Sociology of Knowledge. Doubleday, New York. Bonacich, P., 1987. Power and centrality: a family of measures. Am. J. Sociol. 92, 1170–1182. Bothner, M.S., Smith, E.B., White, H.C., 2010. A model of robust positions in social networks. Am. J. Sociol. 116, 943–992. Bothner, M.S., Podolny, J.M., Smith, E.B., 2011. Organizing contests for status: the Matthew effect vs. the Mark effect. Manage. Sci. 57, 439–457. Bothner, M.S., Kim, Y.K., Smith, E.B., 2012. How does status affect performance? Status as an asset vs. status as a liability in the PGA and NASCAR. Organ. Sci. 23, 416–433. Burt, R.S., 1987. Social contagion and innovation: cohesion versus structural equivalence. Am. J. Sociol. 92, 1287–1335. Castellucci, F., Ertug, G., 2010. What’s in it for them? Advantages of higher status partners in exchange relationships. Acad. Manage. J. 53, 149–166. Chase, I.D., 1980. Social process and hierarchy formation in small groups: a comparative perspective. Am. Sociol. Rev. 45, 905–924. Devlin, K., 1995. The statue of the venture capital industry: too darn hot? In: Weissberg, T. (Ed.), Pratt’s Guide to Venture Capital Sources. Securities Data Publishing, New York. Eccles, R.G., 1981. The quasifirm in the construction industry. J. Econ. Behav. Organ. 2, 335–357. Elango, B., Fried, V.H., Hisrich, R.D., Polonchek, A., 1995. How venture capital firms differ. J. Bus. Ventur. 10, 157–179. Ellison, S.A.W., 2005. Principal, Spectrum Equity Investors. Personal Communication, Menlo Park, CA. Florida, R.L., Kenney, M., 1988. Venture capital, high technology and regional development. Reg. Stud. 22, 33–48. Freeman, J., Hannan, M.T., 1983. Niche width and the dynamics of organizational populations. Am. J. Sociol. 88, 1116–1145. Gibbons, P.T., 1992. Impacts of organizational evolution on leadership roles and behaviors. Hum. Relat. 45, 1–18. Goffman, E., 1956. The nature of deference and demeanor. Am. Anthropol. 58, 473–502. Gompers, P.A., Lerner, J., 2001. The Money of Invention: How Venture Capital Creates New Wealth. Harvard Business School Press, Boston. Gould, R.V., 1991. Multiple networks and mobilization in the Paris Commune, 1871. Am. Sociol. Rev. 56, 716–729. Gouldner, A.W., 1960. The norm of reciprocity: a preliminary statement. Am. Sociol. Rev. 25, 161–178. Harary, F., 1959. Status and contrastatus. Sociometry 22, 23–43. Jensen, M., 2003. The role of network resources in market entry: commercial banks’ entry into investment banking, 1991–1997. Adm. Sci. Q. 48, 466–497. Kanter, R.M., 1977. Men and Women of the Corporation. Basic Books, New York. Kipnis, D., Schmidt, S.M., 1988. Upward-influence styles: relationship with performance evaluations, salary, and stress. Adm. Sci. Q. 33, 528–542.

M.S. Bothner et al. / Social Science Research 52 (2015) 588–601

601

Kogut, B., Urso, P., Walker, G., 2007. Emergent properties of a new financial market: American venture capital syndication, 1960–2005. Manage. Sci. 53, 1181–1198. Lerner, J., 1994. The syndication of venture capital investments. Financ. Manage. 23, 16–27. Lynn, F.B., Podolny, J.M., Tao, L., 2009. A sociological (de)construction of the relationship between status and quality. Am. J. Sociol. 115, 755–804. Merton, R.K., 1968. The Matthew effect in science. Science 159, 56–63. Miller, D., Chen, M.-J., 1994. Sources and consequences of competitive inertia: a study of the U.S. airline industry. Adm. Sci. Q. 39, 1–23. Péli, G., 1997. The niche hiker’s guide to population ecology: a logical reconstruction of organization ecology’s niche theory. Sociol. Methodol. 27, 1–46. Phillips, D.J., Zuckerman, E.W., 2001. Middle-status conformity: theoretical restatement and empirical demonstration in two markets. Am. J. Sociol. 107, 379–429. Podolny, J.M., 1993. A status-based model of market competition. Am. J. Sociol. 98, 829–872. Podolny, J.M., 2001. Networks as the pipes and prisms of the market. Am. J. Sociol. 107, 33–60. Podolny, J.M., 2005. Status Signals: A Sociological Study of Market Competition. Princeton University Press, Princeton, NJ. Podolny, J.M., Phillips, D.J., 1996. The dynamics of organizational status. Ind. Corpor. Change 5, 453–472. Podolny, J.M., Stuart, T.E., Hannan, M.T., 1996. Networks, knowledge, and niches: competition in the worldwide semiconductor industry, 1984–1991. Am. J. Sociol. 102, 659–689. Podolny, J.M., Roberts, J., Hannan, M.T., 1998. Organizational Inertia and Change. Adapted from Class Note in G205 Organizational Design. Stanford Graduate School of Business. Riggio, R.E., Chaleff, I., Lipman-Blumen, J., 2008. The Art of Followership: How Great Followers Create Great Leaders and Organizations. Jossey Bass, San Francisco. Roberts, P.W., Reagans, R., 2008. Locating Critical Influence Beyond the Margins of Status Hierarchies. Working Paper, Goizueta Business School. Rowley, T.J., Greve, H.R., Rao, H., Baum, J.A.C., Shipilov, A.V., 2005. Time to break up: social and instrumental antecedents of firm exits from exchange cliques. Acad. Manage. J. 48, 499–520. Sauder, M., Lynn, F., Podolny, J.M., 2012. Status: insights from organizational sociology. Annu. Rev. Sociol. 38, 267–283. Shils, E.A., 1994. Deference. In: Grusky, D.B. (Ed.), Social Stratification: Class, Race, and Gender in Sociological Perspective. Westview Press, Boulder, CO. Sorenson, O., Stuart, T., 2001. Syndication networks and the spatial distribution of venture capital investments. Am. J. Sociol. 106, 1546–1588. Sorenson, O., Stuart, T., 2008. Bringing the context back in: settings and the search for syndicate partners in venture capital investment networks. Adm. Sci. Q. 53, 266–294. Spence, A.M., 1974. Market Signaling: Informational Transfer in Hiring and Related Screening Processes. Harvard University Press, Cambridge, MA. Stack, C., 1974. All Our Kin: Strategies for Survival in a Black Community. Harper & Row, New York. Stuart, T.E., Ding, W.W., 2006. When do scientists become entrepreneurs? The social structural antecedents of commercial activity in the academic life sciences. Am. J. Sociol. 112, 97–144. Stuart, T.E., Hoang, H., Hybels, R.C., 1999. Interorganizational endorsements and the performance of entrepreneurial ventures. Adm. Sci. Q. 44, 315–349. Tuma, N.B., Hannan, M.T., 1984. Social Dynamics: Models and Methods. Academic Press, Orlando. Turner, R.H., Shosid, N., 1976. Ambiguity and interchangeability in role attribution: the effect of Alter’s response. Am. Sociol. Rev. 41, 993–1006. Useem, M., 2001. Leading up: How to Lead Your Boss so You Both Win. Three Rivers Press, New York. Veblen, T., 1914. The Instinct of Workmanship and the Industrial Arts. Macmillan, New York. Wasserman, S., Faust, K., 1994. Social Network Analysis: Methods and Applications. Cambridge University Press, New York. Zuckerman, E.W., Kim, T.Y., Ukanwa, K., von Rittmann, J., 2003. Robust identities or nonentities? Typecasting in the feature-film labor market. Am. J. Sociol. 108, 1018–1074. Zygmont, J., 2001. The VC Way: Investment Secrets from the Wizards of Venture Capital. Perseus Publishing, Cambridge, MA.