Michelle Rogan – INSEAD
Olav Sorenson – Yale School of Management
Jessica Santana – Stanford University
Pete Aceves – University of Chicago
Article link: http://asq.sagepub.com/content/59/2/301
Question 1. Your paper is about the effect of indirect relationships on the success of acquisitions in the global advertising industry, moving beyond an analysis of average effects by explaining the variation in acquisition outcomes. You operationalized indirect relationships by analyzing clients in common between the acquiring and the acquired firms, but in your paper you also mention other types of indirect ties such as shared suppliers and service providers as well as the hiring of each other’s employees. Each of these types of indirect ties leads to different information flows between the firms, which brings us to a general question. Since not knowing the exact content of a tie is a common challenge that network studies encounter, what would be your advice to budding network scholars in selecting the most theoretically relevant ties?
Observing tie content is a challenge in most network studies, but observational challenges typically inform the choice of empirical context but not theory. In our study, we developed arguments for when indirect ties would affect partner choice and subsequently harm performance in mergers. The arguments could apply to a range of types of ties. Our archival data allowed us to clearly and accurately observe the client ties of merging firms, and so we used these for our empirical tests. An area for future research would be additional empirical work testing the theory on different types of ties like employee relationships or supplier relationships.
Question 2. In the paper you analyze both the emergence and retention of network ties. In the first case, you analyze whether a firm is acquired (tie here is defined as acquisition) conditional on clients in common. In the second case, you evaluate whether a client will sever their relationship with the acquired or acquiring firm as a result of a merger. A major challenge for organizational network scholars is analyzing network mechanisms without ignoring the complexity and dynamism of social networks. Both of you have written on such topics as multiplexity and network dynamics. Rogan (2014), for example, describes how multiplex ties between two firms, in which one firm provides multiple services to the other, reduce the loss of clientele following executive departure. Sorenson et al. (2006) explain how moderately complex knowledge diffuses more slowly to socially distant actors. What needs to change in organizational theory to account for these perspectives? How would you describe a new science of organizational networks?
I don’t think that organizational theory needs to change to account for these perspectives, but clearly opportunities exist for extending theory by investigating networks taking into account the complexity and dynamism of networks. Fortunately a number of scholars are conducting research in these areas like Maxim Sytch, Andrew Shipilov, and Jason Davis to name a few. Networks research is also taking a multi-level focus. For example, as described in a paper Olav and I co-authored in ARS in 2014, research has begun to recognize that the ties we observe between organizations in many cases are actually maintained and controlled by individuals. This has implications for the retention of exchange partners when individuals leave.
Finally, more studies that acknowledge the endogeneity of networks are needed. On one hand, endogeneity is an issue that requires econometric attention. Adam Kleinbaum’s 2012 paper in ASQ is a great example of this and he goes to great lengths to deal with the endogeneity of networks of employees in his setting. However, endogeneity also merits theoretical attention. Understanding when an organization’s current ties affect the formation or dissolution of its future ties (as in this paper) is an important part of network theory. For example, a study that considers both exogenous and endogenous mechanisms of network evolution is Rosenkopf and Padula’s 2008 Organization Science article.
Question 3. In order to evaluate the performance of merged entities following acquisition, you measure client retention and billings per client. You explain that cumulative abnormal returns (CAR) and return on assets (ROA) are not appropriate metrics for an industry dominated by privately owned firms with few assets. Performance appears to sometimes be a vague concept that can be measured in a variety of ways depending on industry, organizational age, and other distinctive factors. Thus, the choice of performance metric can sometimes be a contentious issue. How do you recommend scholars select the best metrics for firm performance, given an array of alternatives?
Your choice of performance metrics should be guided primarily by your empirical context and theory but also by the precedent set in prior literature. We used the metrics that the advertising industry looked to – annual billings and client retention. And these were validated by the interviews of executives in the industry we conducted. Furthermore, as noted in the paper, performance metrics for publicly traded asset heavy firms did not make sense in our empirical context. As you note, performance is a broad concept. So it is worth thinking carefully about the performance outcomes suggested by your theory, i.e. whether short-term financial performance is the right measure or if you should consider other measures such as new tie formation, new product introductions, patents, efficiency gains, and so on.
A rich area for research is investigation into the performance trade-offs individuals and organizations experience. Often an action that improves performance in the near term can have negative long term benefits. It is also important to consider the processes that give rise to performance differences and how biases in these processes can lead to false inferences about the causes of performance. Olav’s 2006 ASQ with Dave Waguespack is a wonderful example of research in this line.
Question 4. You acknowledge compelling reasons to acquire firms with clients-in-common, including information access and trust. Despite these potential explanations, you find that acquisitions of firms with clients-in-common negatively impacts performance of the merged firm. Given your findings, when might a merger actually benefit from common clients?
The pattern of results in our study suggests that the negative performance effects occurred because common client ties constrained the acquirer’s search and so they may have selected lower quality targets than if they had searched more broadly. So a merger could benefit from common clients if the common client ties provided the benefits of information access and trust without the constrained search. Some firms in the advertising industry use search firms when seeking targets, however, I expect that the likelihood that an acquirer would use these search firms is lower when it already has connections to potential targets via indirect ties, and so the adverse selection problem would persist.
Clearly in instances in which search costs are very high for acquirers, then searching through common clients could make sense in the near term. The savings in search costs might offset (in part) the lower performance of these mergers. Yet, over the long term, the costs of lower post-merger performance would likely overshadow the search costs savings.
Question 5. This paper does a great job of linking the macro context with micro social-psychological mechanisms; what advice do you have for effectively integrating macro empirics and micro behavioral foundations/mechanisms?
The best approach is to seek empirical tests at the macro level that provide a pattern of results consistent with the micro-behavioral mechanisms you propose. Identify the conditions under which the relationship you propose should strengthen and weaken if your arguments are valid and then conduct the empirical tests to demonstrate this. For example, we argued that the reason acquirers would prefer potential targets to which they were indirectly tied was uncertainty reduction. We then were able to show that the preference for indirectly tied targets increased when uncertainty was higher, i.e. when the target was more geographically distant, and decreased when uncertainty was lower, i.e. when the target served clients in the same industries.
Your question brings to mind one of my biggest frustrations as a reader and reviewer – reading papers that include several hypotheses that are disconnected in which the “whole is not worth more than the sum of its parts”. Studies with one or two interesting hypotheses and several related tests to support them tend to be much more elegant and impactful. Hedstrom and Swedberg’s writings on Social Mechanisms (1996 article and 1998 book) are worth (re)reading to understand a mechanisms-based approach to social theory.