Pierre Azoulay – MIT Sloan School of Management
Nelson P. Repenning – MIT Sloan School of Management
Ezra W. Zuckerman – MIT Sloan School of Management
Valentina A. Assenova – Yale University
Kevin Lee – Leonard N. Stern School of Business, New York University
Question 1. You conducted a significant amount of fieldwork following the grounded theory approach espoused by Glaser and Strauss (1967). What challenges did you face in conducting fieldwork and analyzing the data that you collected, and how did you overcome them?
So to place this in context—all of this field data come from lengthy company visits I (Pierre) conducted when I was a grad student at MIT (in 1999 to be exact!). When the qualitative work was done, it was not supposed to stand alone. Rather, it was merely supposed to provide background institutional knowledge for a statistical exploration of the determinants of clinical trial outsourcing in drug development (as my dissertation advisers told me early and often, “know your phenomenon”).
Still, I did not know how to approach that process at all, having mostly been trained as an applied economist up to that point. With no shortage of ethnographers around among the set of MIT grad students at the time, I remember asking around where to start, and it is in this context that Glaser & Strauss came up. I remember liking the practical side of the book, and armed with that knowledge, I felt that I was not simply making it up as I went.
As for the challenges, the main one was simply to get access to these companies, and not to get bogged down in lengthy discussions with their legal departments. This took quite a while. Inside each of the companies, I also found that it was key to have buy-in from someone with the “just right” level of seniority. Too senior, and they have no idea what’s happening “down in the trenches” and whom I should actually talk to. If you are the guy that “important VP” wants you to meet, they are annoyed from the start (“don’t I have more important things to do?”) and you are less likely to get them to be candid in their responses. Too junior, and they don’t have the pull to organize a productive visit. I met almost all these folks at a conference for outsourcing managers. That was really the unlocking event for everything that followed.
The second challenge was that of interpreting the data. I got two-thirds of the way pretty quickly. The main thrust of the story I wanted to tell is: “organizations have a carrying capacity for embeddedness, which they can elect to pursue within their boundaries (with their employees), or in relationships with outside suppliers, but not both at the same time.” If this were right, my informants just needed to stop being frustrated with the terrible relationships they had developed with the Contract Research Organizations (CROs). These relationships were all that they could be, given the set of commitments they had made to their “insiders.”
That story was very comfortable for me as an applied economist with a fondness for “second best” ex post rationalizations. But just like my informants, I was also stuck with a story that felt incomplete, and did not make sense to organizational scholars. Why were these outsourcing managers so upset about the nature of their CRO relationships? Why couldn’t they just accept that this was part of the grand bargain (job security in exchange for greater commitment) they had struck with their own employees?
Choosing to collaborate with Ezra and Nelson on this project is exactly what got it unstuck. Before talking about it with them, it had never occurred to me that a vertical supply relationship could be “worse than arm’s length.” We do not have a vocabulary for that in economics. Once we realized these dysfunctional relationships were really in a class of their own, the rest of the story fell into place pretty naturally.
Question 2. The paper makes a fascinating argument that the relationships between CROs and pharmaceutical companies were underperforming in part because they were based on “negative commitment” (pp. 478). In what other settings do expect that negative commitment will produce or contribute to embeddedness failure?
There are examples galore: defense contracting, oil extraction services, automotive manufacturing. Accounts of dysfunctional supply relationships in these industries are plentiful, and can easily be recast as examples of negative commitment leading to embeddedness failure. What is harder (and also more interesting) to pin down is what all of these settings have in common. Because we have essentially an N of one in our study, what we can offer is at best an informed speculation. The one ingredient that is salient in our mind is tapered integration, i.e., for the same job requirement, you can find insiders and outsiders performing the same task, or range of tasks, for the buyer. It is probably not sufficient to set in motion the vicious cycle that leads to embeddedness failure, but it would appear to be necessary.
Question 3. You posit that moderately embedded ties will degenerate into “nasty, brutish, and short” relationships. Would you say that the relationship between the level of embeddedness of a tie (on a continuum) and the performance benefits of a tie are U-shaped, and if so, how might a firm determine the “optimal” level of embeddedness if it is strategically choosing market relationships?
That is an interesting question. A key part of our argument is that ‘moderately embedded ties’ between Pharmaceutical companies (“Pharma”) and clinical research organizations were fundamentally unstable. This doesn’t mean we didn’t find such relationships—we did. But they were difficult for the parties to maintain. The reason for this is that the organizational practices supporting arm’s length exchange in this setting are fundamentally at odds with those required to support even moderate degrees of embeddedness. In short, because Pharma could not commit to strongly embedded relationships with the CROs, the CROs responded by developing strategies that were predicated on their being able to flexibly move across Pharma clients. And organizationally, the latter strategy involves a very different set of roles and routines than one that is geared to adding value to a particular client. Our system dynamics model illustrates this as two competing feedback loops. They can cancel each other out to create a moderately embedded equilibrium, but it is an unstable one. And to preserve it, performance targets must be lowered accordingly—something that was impossible in our case; thus consigning the Pharma-CRO relationship to the “nasty brutish and short” equilibrium. This does not mean that moderate embeddedness is impossible- just that they are likely to be unstable in a setting where the organizational routines for embeddedness are so different from those that support arm’s-length exchange. Moreover, in other settings, other factors may weigh more heavily. For instance, Cheris Chan has an interesting 2009 AJS paper (which we read after ours was in the publication process) that suggests how moderate embeddedness may be most effective in life insurance sales to Chinese consumers. Chen argues that transactions with intimates—highly embedded ties—worked less well once information about commissions became public due to norms against exploiting close ties for profit. This is an interesting finding, but one can immediately see why various contextual factors—(a) that embeddedness there refers to friends and family; (b) that norms against profit-seeking are more salient when the seller is an individual (see Haran, Management Science, 2013); (c) that such norms may be in more rapid flux in a traditional economy; and (d) that shifting individual routines may be more feasible than organization routines—make this a very different case. This is a long way of saying that while we are confident about the mechanisms in our theory and how they interact than we are about the specific prediction that moderately embedded ties are problematic. The key thing is to look very carefully at the commitments that any “mixed mode” of relationship entails and whether it is feasible to mix the associated organizational routines.
Question 4. The paper is framed in a slightly unconventional way, around an empirical puzzle. Why did you choose this framing and what challenges did you face in writing the paper in this style?
This was a very challenging paper to set up rhetorically. And indeed it was not set up this way in prior drafts and iterations, including the first two rounds of this paper. Ultimately, we came to the view that the main reason the paper was interesting was that the fieldwork documented a phenomenon that did not fit into the existing literature. This led to three challenges:(a) to lay out very clearly what the existing theory (at the level of organizational practices generally, and more specifically in terms of interorganizational relationships) predicts such that our case clearly emerges as an impetus for theoretical revision; (b) to make a convincing argument that our case was not idiosyncratic; and (c) to foreshadow our main findings in the introduction without saying too much. We were very happy with how the paper ended up but it is worth noting that the managing editor was not happy with the introduction and pushed us to make it shorter and more conventional; we resisted, but it’s possible she was right (the paper’s impact has been quite modest). It’s also worth noting that it is highly unlikely such an unconventional setup would have worked as an initial submission. By the time we arrived at this setup, it was in the final round and we had an associate editor who liked the paper and seemed disposed to publishing it. This is our take, however; you’ll have to ask him.
Question 5. Do you know of any other papers that are jointly written by an applied econometrician, a systems dynamics scholar, and a sociologist? How did such a rare combo cone to work together and what lessons can we learn from this to promote more work across the disciplines?
Excellent question! :-). In short, this is a very MIT Sloan paper. At Sloan, we are highly committed to disciplinary rigor and to contributing to disciplinary conversations that further general models and methods for better capturing social and economic processes. But we are equally committed to using any and all tools to make progress in understanding and addressing real-world phenomena/problems as would be appreciated by seasoned practitioners. And once we see that colleagues who have been trained differently have something valuable to bring to bear, we have the basis for very productive conversations and collaborations. Of course it is not always kumbaya nor should it be. But unusual synergies occur often enough to keep us excited about the place we are privileged to call our academic home, and the prospect of such synergies were ultimately responsible for Pierre turning to Ezra (for his expertise in theory relating to interorganizational relationships) and for them both to turn to Nelson (for his expertise in modeling complex feedback processes). We thoroughly enjoyed working on this paper together, and we are heartened by the fact that you appreciated the result.