Aleksandra Kacperczyk – MIT Sloan School of Management
Christine M. Beckman – Robert H. Smith School of Business, University of Maryland, College Park
Thomas P. Moliterno – Isenberg School of Management, University of Massachusetts Amherst
Article link: http://asq.sagepub.com/content/60/2/228
Question 1. A great contribution of your study is to clarify theoretically and disentangle empirically two important constructs: risk and change. We found this fascinating, and would like to get some further thoughts along these lines. First, the entrepreneurship literature, starting from the classic contribution by Frank Knight, distinguishes between risk and uncertainty. Both assume variation in the distribution of possible outcomes. However, risk is a situation in which such distribution is known, while uncertainty is a situation in which such distribution is unknown. Does this distinction play any role in your theorizing? If so, would that change some of the findings? Second, we found your operationalization of risk – based on the CAPM, following a long tradition in finance – to be an excellent measure of the risk associated with the firm, but less so of the decisions the firm makes to take on risk. In other words, that seems more of an outcome than a decision. If you were to measure risky decisions directly, without looking at the risk outcomes, how would you go about that?
As you note, we focus on the distinction between risk and change in our paper. We did this because we are bringing together (and disentangling) two literatures that have used these constructs (behavioral theory and prospect theory). For that reason, we avoided a discussion of uncertainty in this paper. Our theory applies to intentional risk taking and situations in which the distribution of outcomes is known, or at least anticipated, ex ante (as Knight defines risk). Employees can assess risk ex-ante and choose to take actions associated either with lower or higher risk. Our focus was on accounting for risk taking more directly than it has been done by prior studies and distinguishing this from change.
A key challenge, however, was to operationalize risk cleanly. It occurred to us that the financial sector offers a unique context in which it is possible to quantify intentional risk taking because ex ante estimations of risk are widely known. In particular, any security has specific financial risk, manifest in both the firm’s sector and prior performance. Hence actors may choose to adopt a more or less risky investment strategy. Note that the decision makers in our context are the mutual fund managers. They decide on the composition of the portfolio, and therefore make explicit decisions about risk in the portfolio. We also consider a set of “risky change” decisions, and these are actual decisions to increase the risk in the portfolio (by changing the portfolio to include more low-cap stocks or more growth stocks, which are higher risk). You are right that CAPM is a not necessarily a measure of the risk taken by the firm, but we are arguing that the decision to hold that stock is a decision about the risk taking of the mutual fund manager.
As you note, our paper is silent on the question about whether poor performance would result in more uncertain behavior, but it is an interesting question to contemplate. The inability to estimate a risk distribution ex ante is, in fact, uncertainty. In those contexts, actual risk is more likely to be an incidental rather than an intended consequence of managers’ decisions. We believe that our findings might be stronger when measuring uncertainty. Uncertain actions (actions with unknown probabilities or potential outcomes) may be frequently chosen by decision makers and firms performing poorly or under threat. Indeed, we read much of the early performance feedback literature as concerned with the uncertainty embedded in the change decisions associated with negative performance feedback. These decision makers may be so desperate to improve their performance that they leap into the unknown. This is in fact the world that most organizational decision makers live in, and Knight talks about the uncertainty of business decisions. Risk, variation in outcomes with known probabilities, is much less common in the organizational world. This posed a challenge for our measurement, as we discuss above, but we think made for a more conservative (and accurate) test of the theories. We all agree that this would be a profitable direction for future studies!
Question 2. Your study proposes two theoretical mechanisms: individual concern (triggered by internal performance comparison) and organizational concern (triggered by external performance comparison). Could you tell us a bit more about your reasoning in building the theoretical framing for this study? You also mentioned there may be a more complex theoretical argument (p245, first paragraph after Table 4). Can you tell us a bit more about that? Do you think other mechanisms in the organizational learning literature could play a role in explaining some of the findings?
The origin of our contribution lies in mapping individual and organizational concerns to two different sources of social performance feedback. Prior research and theory in the behavioral tradition had never considered whether social performance feedback could have its origins inside the firm: work in this domain has always seen social performance feedback occurring through comparison to other firms. Thus, these different sources of social performance feedback were in fact the motivation for this paper. We were intrigued by the question of when decision makers will attend to the performance of one set of social referents versus another. This led us to theorize that social comparisons can trigger individual concerns when made to referents inside the firm because employees experience career concerns when they underperform relative to internal peers. At the same time, such comparisons can trigger organizational concerns when made to referents outside the firm because referents external to the firm are the central source of feedback about organizational problems.
As we discussed it, and considered the multi-level issues, we realized that there would be both career reasons for attending to a reference group and organizational reasons – but that attention might be focused on different groups and with different outcomes. This insight was motivated by our observation that researchers have, over time, incorporated prospect theory into behavioral theory. They assumed that the associations and mechanisms predicted by prospect theory will occur at both the individual and organizational levels of analysis – even though this implicit assumption has never been tested. In fact, multilevel theory would suggest that we cannot simply assume that a theoretical association seen at one level of organizational analysis (i.e., prospect theory’s focus on individual-level risk taking) is necessarily isomorphic with associations at different levels of analysis (i.e., behavioral theory’s focus on firm-level decision making). We suggest that it is crucial therefore to go back to the seminal conceptualizations, in which these two theories predict outcomes at different level of analyses, individual and organizational. This led us to the hypotheses that we have in the paper. So we started with social comparison, and in the process of thinking it through we disentangled risk and change.
For your second question, the more complex theoretical model that we allude to refers to how we might prioritize our performance relative to internal and external referents. We did not hypothesize that one type of feedback would be more important than another. But the results suggest that our decision makers are more responsive to internal feedback than external feedback (and thus perhaps internal career considerations dominate organizational needs). This is also consistent with Robert Frank’s argument about being a big fish in a small pond (Robert Frank, Choosing the right pond: Human behavior and the quest for status, 1985). Decision makers may be more concerned with their internal status than their global status. This direction proved to be too complicated for an already complex paper, but – again – hopefully another area for future research.
Question 3. Your study proposes that risk decision is the result of individual loss aversion and change decision is the result of organizational problemistic search. However, in this empirical context, it would appear all decisions are made by the fund manager (or the fund management team). While the manager’s career concern in the individual-level loss aversion mechanism is perfectly natural, the organizational mechanism was less clear to us. Would you tell us more about the individual-organization link in the process of organization-level problemistic search? Would you say your theory suggests that the distance of performance reference group leads the decision-maker to switch from individual concern to organizational concern?
Extant behavioral theory on social performance feedback posits that organizational problems are identified when managers assess organizational performance by making comparisons with similar organizations or competitors in the market. Through these external comparisons, organizational decision makers recognize the firm’s poor relative position and engage in problemistic search to address the problem. While change occurs at the organizational level, it is organizational decision makers that drive it. Your question hinges on what makes individual decision makers consider organizational level referents and engage in organizational-level search—this takes us back to the complexity of the multilevel theorizing inherent in these conceptual arguments.
In our context, we can identify changes made by managers at the fund-level, which is a division within an asset-management company. The fund performance is an important factor in the fund manager’s compensation and promotion in the company. It is also a component of the asset-management company’s overall performance. To the extent that fund performance influences both individual and firm performance, mutual fund managers should be cognizant of fund performance for both of these reasons. Our logic is that when comparing a fund with an external fund, the contribution of the fund to the overall firm performance is clear (if a similar external fund is performing better, the internal fund is dragging down firm performance). This should have organizational-level consequences and lead to change because organizational-level concerns are triggered. Although this is partly a function of distance (external funds are more distant than internal funds), we think of it more as a function of the level of analysis at which a decision maker is making decisions. Distance matters more for the salience of the reference group than the actual distance. To that end, one of supplementary analyses looks at co-location of funds (to argue that internal comparisons will be stronger when funds are co-located in a firm because of proximity).
Question 4. You theorize that, when both organizational and individual concerns exist, the two mechanisms work together resulting in risky change. That suggests there are cases in which we have change only and cases in which we have risk only. While change-only decisions are easier to imagine, it is less clear what risk-only decisions would look like in more general contexts. Do you think risk-only decisions only involve situations in which the firm chooses not to make any move? If not, what other types of decisions qualify as such? We found that imagining risk-only decisions becomes harder we try to generalize to other industry contexts. What could be good examples of risk-only decisions in other contexts, like manufacturing or technology? And to what degree could we generalize the findings in this study to other industries?
Our study is the first to propose that risk and change need not be equivalent: one can occur in absence of the other. This indeed implies that risk-only decisions involve no change. Such situations are most common when decision makers decide to continue with the current status quo. Although we might think about such decisions as risk-free, a lack of adaptation and change may often be risky for the firm. In high-velocity markets, for example, firms may become obsolete precisely because they fail to adapt to changing environments. Consider the examples of Kodak and Polaroid: both became obsolete because they failed to adapt to the changing environment. More generally, when firms fail to change in response to competitive pressures, they risk losing market share or ultimately going out of business.
Hence, we believe that our findings extend beyond the finance sector: they have important implications for other industries and are highly relevant in fast-paced environments. However, this raises your earlier question about uncertainty and risk. We focused on risk (and financial risk) precisely because of our ability to look at risk-taking without change. But it is harder to imagine other contexts where decision makers know ex ante that they are taking risks. These situations may be more usefully characterized as uncertain.
Question 5. How does this paper fit into your larger research agenda? After completing this project, did your following work evolve in certain directions rather than others in response to your findings and your experience in this study? What would be the primary follow-up research questions? If you were to advise a PhD student who loved this paper and wanted to do a follow-up study, what would be your suggestions?
The study raised many important questions. Most importantly, it emphasized the need to revisit the existing theory in the context of organizations with multiple units. This contrasts with the common approach adopted in past research, which treats an organization as a monolithic entity. Our study takes a step in documenting that the conventional theoretical mechanisms need revisiting when applied to organizations in which each unit has its own goals, performance benchmarks, and aspirations.
Hence, an attractive opportunity may lie in leveraging a multilevel framing to examine the core propositions in the extant research. For example, we believe that future conceptual and empirical research should explore more explicitly where and how theoretical associations are isomorphic across levels in behavioral theory and should engage in a full examination of cross-level effects. One example of a promising research avenue might be a deeper understanding of when threat-rigidity effects are theorized to be isomorphic across the individual, group, and organizational levels. Thus, although our study offers a step in this direction, our hope is to encourage scholars who wish to examine behavioral theory in general, and performance feedback theory in particular, through a multilevel lens.