Shavin Malhotra – University of Waterloo
Taco H. Reus – Rotterdam School of Management, Erasmus University
PengCheng Zhu – University of San Diego
Erik M. Roelofsen – Rotterdam School of Management, Erasmus University
Xena Welch Guerra – University of St. Gallen
Stevo Pavićević – University of Zurich
Question 1. In your article, you propose a novel explanation of corporate acquisition behavior. You find that extraverted CEOs are more likely to engage in corporate acquisitions, do so more frequently, and conduct larger deals than other CEOs. You further find that the mechanism through which these effects materialize is the inclination of extraverted CEOs to possess larger board networks. Moreover, you find that the effects of extraversion on acquisition behavior are particularly pronounced in “weak situations” (weakly competitive industries and high managerial entrenchment). Finally, you reveal a positive effect of CEO extraversion on acquisition returns. Your paper has many elements that make the overall story convincing and interesting. Could you share what motivated this study and how did the paper evolve over time?
We are a diverse author team, coming from the fields of strategy (Shavin and Taco), finance (Phil) and accounting (Erik). What unites us is an interest in the psychology behind consequential strategic decisions. In the past, we considered how for example biases, emotions, behaviors, and fairness perceptions could influence forming and implementing such decisions. Acquisitions are interesting in this regard because they generally are highly consequential for the companies involved, and the outcomes are highly uncertain. One important assumption we all shared was that quite a lot of subjectivity can go into consequential strategic decision-making, making it interesting and important to study the influence of the psychology behind these decisions.
So, when Erik mentioned that he had been working on gauging CEO personality through novel techniques in machine learning using unscripted text spoken by the CEOs during quarterly conference calls, the others quickly were interested in considering the role of CEO personality in acquisition decisions. Most scholars and practitioners already agreed that personality of CEOs matters. However, we felt extant research had not yet fully considered core personality dimensions, such as the Big Five dimensions. We felt that this was an important gap, particularly when it comes to the influence of CEO extraversion because of the lengthy stream of psychology research that emphasizes extraversion in the study of leadership. A detailed focus on a central personality dimension in an effort to look at its effect and underlying mechanisms with a large sample of CEOs seemed like a great opportunity.
We were fortunate to be able to build on a rich extraversion literature that spans over 100 years. In our first draft, we relied perhaps too much on old seminal pieces of extraversion. The reviews we received were excellent and spot on – so for the revision we relied on much more current extraversion research. Moreover, the reviewers helped us strengthen the logic further and pushed us to explore mediating mechanisms more explicitly. We also built much more on the situational strengtharguments to make clearer when personality matters and when it does not.
Question 2. Prior research has often highlighted that the entire top management team – not only the CEO – may drive acquisition decisions. How do you see this in light of your paper? In the discussion section of your paper you acknowledge that, although CEOs play the central role they also rely on top management teams in acquisition decision making. What future research do you see emerging from the interplay between the personalities of CEOs and other top managers?
This is a very good point! We assumed that CEOs are central players in consequential strategic decisions of their firms. While a lot of advisors and other top management team members are involved in these decisions, ultimately the decision is for the CEO to make. However, the extent to which the effects of CEO personality is mitigated (or intensified) by the personalities of fellow team members is a very interesting question. Also, a person’s personality can affect behaviors of others – for example, people that are more extravert tend to trigger others to “behave” more like an extraverts. And, from a different perspective it would be interesting to see whether opposites attract or whether similarity in personality across team members work better. Research on this team-level could help us get a better idea how firms may want to look at top team composition or could help specify different needs top management teams might have depending on their composition of personalities.
Question 3. You provide joint reasoning for hypotheses related to the effect of CEO extraversion on M&A likelihood and M&A deal size. We were wondering to what extent these outcomes are driven by the same mechanisms. While the M&A likelihood related findings – in particular the mediating effect of CEO network size – may be associated with an inclination to identify more (and potentially better) acquisition opportunities, the deal size-related findings seem to point to an inclination for higher risk taking. What is your comment on that?
You are right that extraversion has sometimes been linked with risk-taking. However, the link is not as clear-cut (see for excellent reviews also on this topic, Wilt and Revelle 2009; 2016). In particular, scholars are now more often treating risk-taking behaviors and impulsivity as a dimension that is distinct from extraversion. Also, risk-taking behaviors can stem from a wide variety of other personality traits, such as hubris and narcissism.
So for us, it was actually important to see whether CEO extraversion influences M&A behavior above and beyond risk-taking. We therefore controlled for risk-taking both at CEO- and firm-level. We did this by controlling for whether the CEO had a pilot license and whether the CEO grew up in the great recession period (who scholars found generally tend to take less risk), and the firm’s long-term debt to asset ratio.
We very much agree that the size of board networks are only part of the story. It would be great if we can explore different mechanism that could further help understand the underlying reason for the acquisitive nature of extraverted CEOs. We do think that the extended board network is an important mechanism for spotting deals both small and large. Some of the features of extraversion could further help explain the effect. It is likely that with extraverted CEOs’ increased levels of energy, assertiveness and desires for social engagement, extraverted CEOs would be more likely to make more but also larger deals when they spot those through their extended networks.
Question 4. You use textual analysis of CEO speeches from quarterly earnings conference calls to predict CEOs’ Big Five personality dimensions. What are the main advantages and disadvantages of using textual analysis for such a purpose? Do you expect to see more wide-spread use of linguistic techniques to assess personality in the future and why (or why not)?
The main advantage of textual analysis is that it does not require the cooperation of CEOs. Moreover, with recent developments in machine learning it has become feasible to train algorithms to automate the analysis of texts. This methodology therefore allows us to obtain a larger dataset of CEOs’ personality traits than most alternative methods. The fact that we do not need the cooperation of CEOs also reduces potential selection bias that may arise when data collection depends on whether CEOs volunteer. A further benefit of the method is that language is one of the most proximate reflections of personality traits one can get data on. Many alternative methods use more distant proxies that bring along more methodological issues.
In order to train our algorithm we had to be sure that the texts that we use are actually the CEO’s own words. Many texts, such as the letter to the shareholders, are less appropriate because they have likely been prepared by others than the CEO, or there is a large involvement of others in editing the texts. We use the question-and-answer part in quarterly earnings conference calls, because this is a setting for which we may reasonably assume that the words spoken by the CEO are her/his own words.
However, a disadvantage of the method is that we gauge the CEOs personality traits as s/he presents herself/himself in a public setting in front of a specific audience. To the extent that certain CEOs are systematically acting out to achieve a desirable response from analysts, our construct of extraversion would be biased. We have addressed this in our research by validating our construct with data from other sources. Based on this we are confident that we do pick up actual extraversion of the CEO. However, whether this equally holds for other personality traits remains to be revealed by future research.
We expect that over the next years, the developments in linguistic techniques to assess personality traits will be tremendous. Our research may be considered one of the first attempts to use this methodology in a management research context, but we expect that many others will follow suit. Apart from the benefits of the method described above, we believe this will happen because the methodology does not require access to firms or people and can therefore be implemented by anyone. Moreover, we expect to see large progress in the accuracy of the algorithms over the years to come. This progress is not only driven by academic research but also by the vast resources that are invested in similar algorithms by firms like Amazon, Facebook and the like, who use the algorithms for business purposes.
Question 5. In the development of arguments for your hypotheses you touch upon agency research, network research, as well as research on investor sensemaking. How did this affect the writing of the paper? Could you also share some insights on the review process that has led you to the final version of the paper? What was the main challenge? What, if any, aspects of your study that did not end up in the final publication?
This is a broad question – we had quite a journey! The reviewers were instrumental in the development of the paper. They pushed us to empirically examine mediating mechanisms, and frame our contributions more clearly in personality research. For example, while we already tried to explain situational conditions when extraversion mattered more or less, the reviewers helped us bring in the situational strength hypothesis much more directly.
One challenge we faced was that the reviewers rightly pointed out that current research is clear that extraversion has two sub-dimensions: (1) an affiliative dimension (tendency to socialize and to achieve close connections for their own sake, regardless of whether they help with achieving organizational or work-related goals) and (2) an agentic dimension (tendency to achieve social connections not only for their own sake but also to influence and persuade others to pursue collective goals). Ideally, we would capture these distinct sub-dimensions both, and see whether they would have varying influences in the M&A decision-making process. However, the algorithm used for our measure attempts to capture the higher-order construct extraversion, and could not capture the two sub-dimensions.
In our revision, we therefore discuss the two dimensions as clearly as possible, and while doing that we could rely on much prior work that indicates that for top leaders such as CEOs, the agentic dimension is much more important than the affiliative dimension. This is because CEOs are unlikely to socialize just for the sake of socializing. We also found some limited support that our measure is more likely to capture the agentic dimension of CEO extraversion – while gender differences tend to be small for extraversion scores in general, women usually score higher on affiliative facets and men score higher on agentic facets; since men scored higher than women on CEO extraversion we find some evidence that the measure captures agentic rather than affiliative features of extraversion.
One aspect that did not make it to the final version is that we also consider how CEO extraversion is related to more diversifying M&As. In an earlier version, we presented the results for high scores on the moderators separately from the results for low scores on the moderators. This nicely showed the stark difference in the effects of CEO extraversion between weak and strong situation, including effects on making more diversifying M&As. However, since the sample was a bit smaller for the study of the effects on making diversifying M&As, the effect was a bit less stable when we used moderated regression analyses.
One aspect we did not have in the first draft but the reviewers asked us to bring in was the focus on M&A performance. We felt the theory was not as clear-cut on this, so instead stating the effect in a formal hypothesis, we felt it was better stated as question, which we were happy to be able to answer. Extraverted CEOs did not only make more and larger acquisitions but also seem to be better at it, at least in the eyes of investors.
Thanks again for the questions! Hope the answers give a bit of peak behind the scenes!