Authors:
Anne Bowers – Rotman School of Management, University of Toronto
Matteo Prato – Universita della Svizzera italiana

Interviewers:
Mara Guerra – Imperial College Business School
Margeum Kim – Yale School of Management

Article link: https://journals.sagepub.com/doi/full/10.1177/0001839217727706

Question 1. Your paper takes a longitudinal view on status and examines how exogenous changes in market categories affect unearned status. We are very fascinated by your paper and the methodology you used.

Question 1(a). Can you tell us more about how the idea for this paper emerged and how it fits with your broader research interests?

Anne:  Matteo and I decided to work together shortly after he arrived at USI in Lugano and I was there for a conference on organization theory.  We both had done work on analysts so we were both familiar with the data from IBES, although we had different interests.  I’ve always been interested in the structure of rating systems, and Matteo’s work has focused more on analyst behaviour.   This seemed like an opportunity to put the two together.

Matteo: Our collaboration started after we met at the Lugano Conference on organization theory. I had recently joined the faculty at the “Universita’ della Svizzera italiana”, the organizing institution. As both Anne and I are organization theorists and work on analyst data, we thought it could be an opportunity to work together.

After the conference, we had several ideas on potential projects but we ultimately decided to explore something Anne mentioned: that the categories that the magazine Institutional Investor (II) uses to produce the coveted All-star award change over time, and that there might be some strategic action beneath these changes. Thus, our project initially started as a study of the antecedents of category changes. Our initial question was: “why do intermediaries (such as II) change categories over time?” It took a few years before it became a paper about the consequences of category changes.

Question 1(b). A key purpose of this blog is to share insights about how a paper developed and give a view of what happens behind the scenes. Can you tell us more about the paper evolved as you collected and analysed the data?

Matteo: I would say that the project evolved in a serendipitous process.  After we decided to focus on the antecedents of category changes, we started collecting and analysing the data to identify the role of ranked actors (i.e., analysts/brokerage firms), intermediaries (i.e., II) and audiences (i.e., investors) in the formation and dissolution of categories. We presented several versions of that paper at different conferences, but one consistent piece of feedback we received was about the challenge identifying the causality of our claims. This critical feedback on our original study was quite fundamental to discovering the untapped potential of our data. Following this feedback, I started discussing with Anne whether we should refocus the paper on the consequences of seemingly arbitrary changes in categories rather than on their antecedents. All of a sudden, I remember, it seemed like low-hanging fruit.

After we agreed that studying the consequences of category changes had potential, we faced another critical question: do we write two separate papers or just one studying both antecedents and consequences of category changes? We thought it might be a risky strategy but, we ultimately decided to opt for the second option. So, our question became: “why do intermediaries change categories and what happens when they do it?” Our first submission at ASQ addressed this question.

Question 1(c). We would also like to know more about the process of crafting and publishing the paper. What was the most challenging part of this process?

Anne:  As Matteo hinted, originally, the paper had a very different structure!  We wondered whether third parties might configure the categories of the rating systems—adding and subtracting categories—if there wasn’t enough churn in the rating system and the same people kept winning time after time in each category.  So that was part 1 of the original paper.  It was Matteo who said that we should also consider the implications of this.  So that was part 2.  We thought the two worked well together—but, it was clear from the beginning that the editors and the reviewers thought that part 2 was the more interesting and meaningful part.  While they saw how the two parts fit together, they didn’t think that part 1 was necessarily the only question that could lead to part 2.  So after the happiness of the R&R letter wore off and we started doing the work, a big question we faced was whether we should make a case for keeping the two parts together, or simply focus on the part that everyone thought was the more interesting contribution.  As you see in the published paper, we chose to remove part 1 and focus on part 2.  Perhaps not surprisingly, when you cut out half the paper you get more space to develop the remaining pieces, and I think the paper is much better for focusing deeply on this one specific question about the status consequences of seemingly arbitrary behavior.  With that said, I presented the “part 2 only” version of the paper after we had finalized the decision and no one at the place where I presented it seemed to think the contribution was as interesting and exciting as we did.  Ultimately, you have to make the paper what *you* want it to be, realizing that everyone will have an opinion about it.

Matteo: Besides the endogeneity challenge with “part 1” and the challenge to integrate “part1” and “part 2” (and write a coherent paper on both antecedents and consequences of category changes), a third major challenge we faced was responding to the editor’s and the reviewers’ feedback. They were extremely helpful but also demanding. Given what they asked for, it was clear they had a lot of expertise, which proved fundamental in improving our research design and empirical analysis, which changed between the first and second versions. Like Anne said, they also hinted that we should focus on the consequences of category changes only, so after our first submission our question finally became: “How do category changes affect status?” 

Question 2. We are fascinated by your findings and would like to know more about their implications for different players in the market and the market itself.

Question 2(a). You find that individuals can lose status due to structural changes in market categories. Do you have any suggestions regarding how your findings can help individuals to structure their strategic work? We are curious to know, did you observe any cases when analysts sensed that a change was coming and any strategies that they used to face it?

Anne:  From interviews I did for my dissertation, I got the sense that brokerage firms certainly knew which categories had more action.  One research director said something like, “If we just wanted to win a lot we would cover the airline industry, but airline stocks are not good investments, so we don’t have any analysts who cover them.”  Meaning, they know which categories have a lot of competition and which don’t.  I’m not sure how much ability they have to lobby for changes in their favour, though, and have it be meaningful.  There’s some evidence that intermediaries pay attention to powerful firms in classification—for instance, Lounsbury and Rao (2004) show that mutual fund categories linger long after they are useful if the category is dominated by powerful incumbents.  But that’s in horizontal category groupings with no ordering.  In our setting, the nature of rankings provides a true ordering of high and low, so you have to assume that any category change that would, for example, advantage Merrill Lynch would at the same time disadvantage another powerful firm, like Goldman Sachs. 

Matteo: This is an interesting observation. We did explore the possibility that brokerage firms (and analysts) acted strategically to predict (or even influence) category changes. Unfortunately, our data didn’t allow us to provide a definitive answer, but, in our preliminary analysis, there seemed to be no robust evidence that that was the case. In our follow-up paper (forthcoming at Organization Science), we were able to find some robust correlational evidence that II itself might act self-interestedly in determining when and what particular category should be changed.

Even if I cannot answer whether analysts are able to sense or influence category change by II, I would be inclined to think that brokerage firms and analysts do react strategically to category changes. For example, brokerage houses might open up a new research unit (and hire new analysts) if a new category is introduced by II. Similarly, analysts might change their stock coverage in the wake of a category change to increase the probability to be ranked. As Anne mentioned, there might be some categories that in terms of pure profitability might not be that relevant for investors (e.g., Airlines), but because they are legitimized by II through its ranking, they get covered by brokerage firms with specialized analysts.

Question 2(b). How long is the effect of category shock on the market power of analysts? In some sense, the status gains and losses due to the change in classification systems create noise in the market. How long do you think such a noise would last?

Matteo: This is a question that is worth exploring. Our study focuses on short-term “noise” because it is what our data allows us to identify. But we cannot discard that that noise is dissipated in the long term. However, many intervening mechanisms might cement the status gain/loss that structural changes impart. For example, if we elaborate on a Mertonian view of status, it might be reasonable to expect that by gaining (or losing) status, analysts gain (or lose) access to resources, thus increasing the chance to reinforce their (initially arbitrary) status change. In other words, what might be initiated as an exogenous shock can trigger a self-reinforcing cycle of status consolidation.  But it might be also reasonable to expect that those analysts who raise on fragile foundations (analysts who become high-status because of a category change) are those who will fall in disgrace in the long term. For example, think of the (counter-)bandwagon effect studied by Rao, Greve and Davis (2001) in their Administrative Science Quarterly publication on analysts. They showed that analysts who initiated coverage on stocks following a bandwagon effect, penalize those stocks after they realize that their choice was not fully rational. By analogy, one might expect that investors that feel “misled” by an exogenously determined change in status counter-react in the long term and penalize the actors who were arbitrarily enhanced in status, if they realize that such status changes did not reflect objective changes in quality. Thus, instead of dissipating in the long term, noise might reverberate in markets causing further noise.

Question 2(c). Would you expect the effect of category shock to change if the whole ranking was released (for example, in the context of university rankings)? If so, what do you think would happen?

Anne: I’m not sure what would happen to institutional investors.  They have their own in-house analysts, and are mostly looking for unique information they don’t have in-house.  It’s possible that this might have some impact on firms going public and so on, in that they might pay attention to different banks, since they want certain analysts to write about their firms.  But, interestingly, the brokerage firms themselves do buy information from II about their results, so they know how their analysts are doing relative to others.  In Dan Reingold’s book “Confessions of a Wall Street Analyst”, he talks about getting feedback from his boss that he’s not doing well with institutional investors from a certain geographic area in the II voting.  So the brokerage firms know a lot about where their analysts are each year.

Matteo: I think that the clear-cut between high- and low-status actors produced by truncated rankings play an important role in our context. Had the full ranking be released, we might expect a more nuanced effect. Whereas a “short” ranking such as the one produced by II makes immediately visible who changes status and creates a gulf between those who are ranked and those who are not, a “long” ranking might blur status changes in the eyes of the ranking audience. In other words, in fully-expressed rankings, only big changes in status might grab audience attention. Moreover, it might be that attention is not uniformly distributed across the status hierarchy in these long rankings. I think there’s some interesting work focusing on the latter issue right now.

Question 3. In this study you look at the Institutional Investor All-Star award and the status it confers to analysts. Your empirical setting provides a very nice set-up for the study of status in that you observe the outcomes of status shocks holding the quality of analysts constant pre- and post- category changes. Important to this identification strategy is the fact that certain categories disappeared, and others were born. We are curious to know what drives the birth and demise of market categories.

Question 3(a). When interpreting your findings, how should we factor in the conditions that led to the structural change (i.e. some industries were demising in size, innovation, and public attention while others were up-and-coming)?

Anne:  After our paper got accepted at ASQ, we took part 1 and turned it into a separate paper, now forthcoming.  In that paper we show associations between the likelihood of category change for what you might call “functional” reasons—the categories aren’t capturing the current structure of the stock market well because industries change over time or the internet shows up.  We also show that stability in categories—that is, when the same analysts keep winning over time, is associated with changes in the categories.  So there seem to be both practical and strategic reasons.  One thing that’s important to remember, though, is that you can’t say whether any changes in a system are “wrong,” even if they are strategic.  That is, consider the oil and gas industry.  You could divide firms by their position in the value chain or by geography or by market segment—all of these systems of categories are legitimate, and depending on how you divide them, different analysts will be considered “best.”  But if you switch from value-chain to market segment, is it necessarily wrong?  I don’t know that we can say that.  Most category systems are ultimately pretty arbitrary.

Question 3(b). Regarding the analysts who gained status due to the category addition, do you think they would have attained status eventually, even without the change in classification system, as the market favors the segment that they cover over other segments?

Anne:  We were able to get some very fine-grained data about the ordering of analysts directly from II, and we’re looking at that now.  We’ll keep you posted!

Matteo: We cannot discard that some analysts might have climbed the status hierarchy eventually, but, as our results suggest, being ranked in II might make a significant difference in the analyst’s career. I think that at the very least, our result suggests that this process is sped-up by category changes. In any case, as Anne mentioned we were able to collect some new data that we hope can provide some further insights on the process of status construction in markets.

Question 4. What future work on this topic do you foresee? What advice might you share with PhD students interested in studying status?

Anne:  It’s been neat to see what has resonated with people about the paper and also what areas seem to be attracting attention.  There seems to be a growing interest in specifically studying the structure of rating and ranking systems, which is very exciting because I don’t think we pay enough attention to the process of rating.  So seeing papers that take the structure of the rating and the process of rating seriously is validating, and I’m hopeful more people will consider it.  For status, it’s one of those fields that seems like it should be mature but in fact is still very vibrant.  I think some of the most interesting work is trying to take seriously the fact that we are typically members of multiple status systems (for example, individual prestige or organizational prestige)—more so if you include inequality more generally—and so how does that work?   

My advice to students interested in studying status is to think carefully about where you’ll study status, to make sure that you are properly capturing the relevant audience for status and actually measuring status.  For example, do investors think that firms that win the “best places to work” awards are necessarily high status?  Or would they focus on something else?   Also, present your work a lot to a wide variety of people, which is good practice for anyone.

Matteo: I think that your questions open up some promising avenues for future research: what is the long-term effect of category changes? Do they trigger a self-fulfilling prophecy whereby actors who see their status enhanced by category changes consolidate their status in the long term? Or do they seed the roots for future status falls as they create the condition for temporary increases in status? What is the role of the status structure (e.g., long vs short rankings) in driving audience attention and shaping status beliefs? I think these questions are worth exploring. But even if we (like many others in the field) do focus on categories (partly because we can leverage a consolidated literature on the topic), classification systems are just one instance of the many socially constructed architectures that shape social life. As these architectures continue to channel our attention and shape our judgments of what is worthy, status dynamics may unfold through novel, perhaps more volatile, processes that are worth exploring.

Question 5. Are there any questions you wished we asked you about the article? Or, are there any lessons that you would like to share with us?

Anne:  A lot of people ask me if there’s anything left to be done using analyst data.  My answer to that is yes, in part because it’s such a good data source.  The IBES data can be quite tricky to work with, but as a field don’t have a lot of data sources where we can actually see someone’s work product, and its accuracy, and in the industry, we can watch their career progression.  Finance does have its peculiarities, and that has to be acknowledged in terms of generalizability, but even within those limitations there are interesting questions still to be answered.

The other thing I’d say is, as someone who does a lot of research alone, if you’re going to collaborate, pick someone who a) you actually like and b) you think you can be honest with.  It’s much easier if you can just say, “I think that’s wrong” or “I’m not interested in that idea” or “I don’t understand what that sentence means” rather than worrying about whether you can or should say something, or committing to an idea that you’re not entirely excited about.  Because we can do this I think we’ve come up with more interesting ideas and better research than we would have otherwise.  It’s also allowed us to keep our own identities given we also both write papers using the analyst data with other coauthors.  There are some issues we will likely never resolve, however, like Matteo’s preference for writing in LaTeX and my preference for Word.

Matteo: As Anne mentioned, collaboration is another important part of the process. I think Anne and I have developed an idiosyncratic way of collaborating that is based on trust for each other’s (at time diverging) ideas and forbearance for extensive revisions of each other’s text. Trust and Forbearance that I am about to put to a test as I am changing operating system and moving to Linux, but please don’t tell her yet.