Christopher I. Rider – Georgetown University, McDonough School of Business
Article link: http://asq.sagepub.com/content/54/4/575
Question 1. The idea of brokerage has predominantly been studied through the mechanism of information benefits. That is, scholars have asked how ties and the network itself transmit information and what kind of benefits actors accrue or lose. You propose to shift the focus from the information benefits of brokerage. This is apparent already in your title: “Constraints on the Control Benefits of Brokerage”. What is the mechanism you propose and how is it different from the information benefits perspective? It would be interesting if you talked us through some empirical settings: how would prior work have understood your setting? What novel insights have your present explanation added to the literature?
Structural holes provide timely access to valuable information so, as Ron Burt says, brokers are at high risk of a “productive accident” – likely to detect and exploit good opportunities. A broker’s control benefit is being able to move early to exploit their information advantage. For example, Burt (1992: 31) describes tertius strategies like pitting two potential buyer against one another to increase sales price. Position affords vision but action produces favorable outcomes.
Focusing on the control benefits (and constraints on them) draws attention to what brokers do to generate returns and the costs they incur. Although one of Burt’s key insights is that there are both top-line and bottom-line benefits to brokers, we typically graph “returns” to brokerage with structural holes on the X-axis and a performance outcome that is not profit on the Y-axis. But, in other settings we don’t speak of “returns” without measuring investments or costs. So, the proposed shift is a plea to focus on the implied profit function of brokerage, not just “revenues.” My 2009 paper is intended to be an exception to the tendency to neglect brokers’ costs of generating returns to positions.
Take an executive search firm, for example. Search firms possess better information on the labor market’s supply and demand sides than employers or candidates do. Information benefits enable search firms to do well, on average, by collecting fees from clients. This is the focus of most brokerage research. Control benefits imply that search firms incur costs of screening candidates, conducting interviews, and generally determining which matches are likely to be value-creating for candidate, client, and firm. Ideally, search firms maximize the difference between those costs and the fees they collect and do so in a way that avoids disintermediation. The costs incurred represent risky investments in matches that might not materialize due to the unwillingness of a candidate or client to make a deal. In this way, search firms’ control benefits are constrained – they facilitate the matches they can but not always the ones they want.
Similarly, in venture capital fundraising, placement agents charge a standard fee (i.e., 1% of capital raised). Therefore, variance in their returns must be largely attributable to differences in their costs of “making a market” in a fund. Focusing on only the standard fee, we’d infer that the returns to brokerage vary only with capital raised. But, the cost of representing a fund is likely decreasing with its perceived quality and placement agents don’t necessarily realize their most preferable matches. Due to constraints on their control benefits, they represent neither the funds with the highest willingness-to-pay nor those with the lowest cost of service. In short, they compromise.
Question 2. A major part of your argument is that brokers and funds have conflicting motivations: brokers want to broker the best funds, but funds would prefer not to have to go through a broker. This ends up producing a matching between the best funds and the best brokers. What does this tell us about the relative power of brokers and the people they broker? How does this kind of role differentiation between the brokers and the brokered play out in network settings where actors do not have explicitly distinct roles (e.g. Aral & van Alstyne 2011)? Do you see your argument as a way to better predict a scope condition for the typical finding that brokers have better outcomes than their peers?
Please let me clarify two points. First, my argument applies specifically to settings in which the broker plays a liaison role – an intermediary between two disconnected actors (Gould & Fernandez, 1989). I’d be curious to know if others have applied the logic to other brokerage roles (e.g., coordinator, gatekeeper).
Second, my argument is that most brokerage occurs away from the tails of the quality distribution – where those who value brokers’ services most and those who are least costly to serve reside. So, best-best matches are not the norm. This is key to my emphasis on the brokerage profit function.
Earlier drafts included two stylized figures that are, admittedly, cartoonish but effectively convey the argument. Please see the figures below that depict the willingness of brokers to represent actors who differentially value representation based on the actor’s perceived quality. Excess supply or demand for brokerage services pushes the representation action towards the middle of the quality distribution but reputation moderates these constraints through positive assortative matching of brokers and the represented among the represented.
You raise a provocative point about outcomes for brokers and non-brokers. Most research demonstrates a positive correlation between brokerage and favorable outcomes. But, to my earlier point, we don’t typically know the costs that brokers incur to generate that correlation so I’m not ready to consider interpret the correlation as evidence of differential returns. Maybe brokers’ favorable outcomes reflect unearned rents on their positions? Or maybe the returns to brokers and non-brokers are roughly equivalent given the costs they incur? This logic suggests that “Is brokerage worth it and if so then why?” would make for a good research question. Middlemen get paid despite persistent efforts to disintermediate them. We know little about how they sustain their positions.
Question 3. Research often starts with some argument in mind and the search for data to test it. Unfortunately, the ideal dataset might not be available, or the data you do collect may sharpen and push your argument in a direction you did not expect so that you end up with data that cannot quite reach the argument you’d most prefer to make. What data or knowledge do you most regret not having in this study? What do you wish you could have known that would have most improved the article?
Conversations with venture capitalists (GPs) and placement agents (PAs) revealed two interesting insights. First, many GPs sought representation but could not secure the services of a reputable PA. Similarly, PAs acknowledged that many GPs didn’t need representation. Second, PAs were viewed unfavorably by many industry insiders.
I wish I had more data on the matching process to answer questions like “Who sought representation?” or “Which funds did each agent solicit?” or “Who was rejected by agents?” With such data, I could test more fine-grained predictions that account for the inverted U-shaped relationship between perceived quality and representation (e.g., whether the low representation rate is more attributable to the supply or demand side of the market). I theorize about likely matches and specific matches among the matched largely because I can’t observe the two-sided selection process that proceeds matching.
With process data, I might then understand why PAs – like so many intermediaries – are viewed unfavorably by so many market participants. I’m intrigued by the idea that brokers play a vital role in many markets but are often resented for doing so (e.g., middleman minorities). If brokers create value then why begrudge them a claim to some of that value?
Question 4. Young scholars might think that writing a solo-authored journal article is an asocial activity. What was your experience? Who did you get feedback from during the process? How many professional conferences did you present your results before you wrote it up? How did your article change under review?
Well, frankly, the asocial nature of writing and analyzing appeals to me. I like focusing so intensely on my work that I disengage from others. But, I certainly couldn’t publish any work by being anti-social.
This article was my job market paper, so I received extensive feedback on intermediate results from Berkeley faculty and students before others at conferences and job talks saw the full paper. The manuscript was presented extensively and benefited greatly from many thoughtful people – too many to mention in full but I’m especially grateful to Henrich Greve (the handling ASQ Associate Editor), the reviewers, and Managing Editor Linda Johanson. All were incredibly patient and constructive – I was clueless about how to write an academic paper and learned a great deal from the ASQ review and editing processes. The published article differs from the submitted version primarily in terms of framing and structure; it was exciting to see how much better the key results could be presented.
Even before submission, I also got great feedback from presentations in student seminars, job talks, and the INFORMS Dissertation Proposal Competition. I really enjoy presenting work-in-progress and, afterwards, incorporating feedback and addressing critiques to strengthen the study. After every presentation, I write up extensive notes and ideas for addressing all points raised and then I get back to work. Doing so makes it even more exciting to get back to the asocial part!
Question 5. 5. It has been seven years since your article got published; even more time has passed since you did the analysis and wrote the first draft. What has changed in this field since then? If you had a chance to rewrite the “recommendations for future research” section, would there be anything new to add? What particular current research streams are you excited about?
So much exciting work!
Organizational theorists have become more interested in studying network evolution and increasing interest among economists in studying networks has motivated many of us to contemplate causality more carefully. I think both developments are good for the field, although I’d like to see more network evolution studies that consider positional determinants outside of the network (e.g., pre-entry characteristics of firms) instead of lagged positions. I’m not a fan of putting lagged dependent variables on the specification for reasons other than accounting for omitted variable bias. And I believe that individual and organizational network positions are neither randomly nor mysteriously occupied.
Given these interests, I remain curious to know if typical “network effects” are more appropriately attributed to nodal position (e.g., centrality) or, alternatively to positional determinants like long-standing, inter-personal relationships of organizational members. Olav Sorenson and Michelle Rogan have a nice paper that provides a framework for considering, more broadly, how and when individuals provide organizations with social capital.
I also appreciate Ron Burt’s recent work on brokerage, especially his working paper on the contingent relationship between status and reputation. Olav Sorenson also addresses the similarities and differences between the two constructs. More work that distinguishes status and reputation effects and applies the distinction to brokers would enhance our understanding of how brokers claim value from their positions and, as I noted above, why brokers are often resented by those they seem to help.
Last, brokers do well. But, what about the returns to brokerage for the brokered? If we want to understand how brokers sustain their positions despite disintermediation efforts, then we need to study the benefits of brokered exchange for the other parties to such exchanges. That’s a line of inquiry motivated by my original presentations of this article. I was asked so many times whether or not it pays to pay the broker. Sampsa Samila and I are trying to answer that question in a current study. More soon…
Thanks so much for engaging my paper so thoughtfully and for the opportunity to revisit the work. I’ve enjoyed our exchange.