Countercountersignalling

Doing What Others Don't Do Because Others Do

Reading Length: 10 Minutes

Key Words: Information Economics, Game Theory, Sociology, Psychology

---

Adam Neumann, founder of WeWork, barefoot in NYC

I have a number of super-successful Silicon Valley clients who dress in ripped denim, Vans shoes, and T-shirts. They are worth hundreds of millions, even more, but it’s a status symbol to dress like you’re homeless to attend board meetings.

- Tom Searcy, CBS Moneywatch

An academic paper on nonconformity opens with this amusing anecdote on countersignalling.¹ But what actually is countersignalling? What is plain old signalling to begin with?

Signalling

Signalling has a special definition in the subdiscipline of information economics and game theory. Despite its name, information economics is about how parties interact when there is incomplete, or asymmetric, access to information. If someone knows something that someone else doesn’t, then there is an information asymmetry.

If everyone had access to the complete, perfect set of information in an environment than there would be no need for signalling. For example, a job applicant would not need to signal that that they are hard-working, intelligent, etcetera to an employer in this world because the employer would somehow already know.

Of course, this is entirely unrealistic. In the real-world there is virtually always the problem of information asymmetry, and so signalling is a ubiquitous phenomenon across contexts.

The job market has a privileged place in the signalling literature because a foundational paper in the field was titled ‘Job Market Signaling’. The author, Michael Spence, was one of the first researchers to rigorously apply the theory of signalling to a real-world context in this 1973 paper.²

As already noted, the context of signalling in a job market is fairly obvious. Spence also hoped to show, however, that signalling had important ramifications in many areas:

By the end I hope it will be clear (although space limitations will not permit an extended argument) that a considerable variety of market and quasi-market phenomena like admissions procedures, promotions in organizations, loans and consumer credit, can be usefully viewed through the conceptual lens applied to the job market.

Spence showed in a mathematical model how it can be possible for education to be used as a signal for job candidates to signal their productivity to employers. An interesting consequence is that sometimes this signalling game can lead to an over or -underinvestment in education, solely for the sake of signalling:

Looked at from the outside, education might appear to be productive. It is productive for the individual, but, in this example, it does not increase his real marginal product at all.

Systematic overinvestment in education is a distinct possibility because of the element of arbitrariness in the equilibrium configuration of the market.

For example, an employer may consider a graduate degree to be a signal of employee productivity. Even if pursuing that graduate degree does not increase the ability of the candidate, they will pursue that degree to properly signal to the employer.

Countersignalling

Picking up where Spence left off after a few decades, a trio of researchers extended the analysis of signalling in the job market to include countersignalling.³ This paper set another landmark for formalizing and exploring the development of countersignals.

Signalling, generally speaking, is a way for high-status/high-quality parties to distinguish and differentiate themselves. Countersignalling, then, is the intentional rejection of opting into such signals. The authors give a handful of illustrative examples:

The nouveau riche flaunt their wealth, but the old rich scorn such gauche displays. Minor officials prove their status with petty displays of authority, while the truly powerful show their strength through gestures of magnanimity. People of average education show off the studied regularity of their script, but the well educated often scribble illegibly. Mediocre students answer a teacher’s easy questions, but the best students are embarrassed to prove their knowledge of trivial points. Acquaintances show their good intentions by politely ignoring one’s flaws, while close friends show intimacy by teasingly highlighting them. People of moderate ability seek formal credentials to impress employers and society, but the talented often downplay their credentials even if they have bothered to attain them. A person of average reputation defensively refutes accusations against his character, while a highly respected person finds it demeaning to dignify accusations with a response.

But why do people countersignal? And under what conditions should we expect to see countersignalling arise? These are the questions pursued in the paper.

The authors create a simple mathematical model in which there are 3 types of students applying for a job: low, medium, and high. The student makes a binary choice of disclosing grades to a potential employer or not.

Intuitively, we should expect that if a student has good grades (of the ‘high’ type), they will disclose this information so that the employer knows they of the high type. Importantly, however, the employer has some a priori information about the distribution of types in the student population, albeit with some ‘extra noisy information’. An example of extra noisy information could be the student’s alma mater, since that is merely a proxy for student quality and could be unreliable.

The introduction of this extra, noisy information can lead to the counterintuitive result of countersignalling. If the student has high confidence in the extra information the employer has, although they can’t know for sure how that information is assessed by the employer, it could be rational for them to ‘pool’ with the low quality students by not disclosing grades. Why is this?

It is assumed that low quality students do not disclose their grades because the risk of getting caught for lying is too high. High quality students do not have to worry about being dishonest, so why would they choose not to disclose?

The crux of the issue is that medium students are not certain about the extra information the employer has about them, so they are motivated to disclose their grades to prove they are not a low student. Because of this, employers will assume that when a student discloses their grade, they are a medium student. Because a high student does not have to worry about the extra information, they can confidently mimic the behavior of the low students without having to worry about being assessed as a low student for mimicking that behavior:

Such countersignalling by high types can be seen as a sign of confidence. Signalling proves the sender is not a low type but also reveals the sender’s insecurity that she would be perceived as such if she did not signal. In contract, countersignalling indicates the sender’s faith that whatever pother information the receiver has on her will probably be consistent with her being of high quality.

Adam Neumann, pictured at the top, can be confident that he will not be mistaken for a ‘low’ person that would typically walk around New York City barefoot. Furthermore, he can signal that he is not a ‘medium’ person that would walk around New York City in a suit by countersignalling with bare feet.

Do people consciously countersignal, or are they just being quirky? The three researchers conducted a laboratory experiment with subjects playing a countersignalling game like the one described above for 90 minutes. They found that over time subjects that were assigned a ‘high’ identity did in fact learn to countersignal to their benefit, proving that the strategy can be learned by humans.

Example: Strategic Disclosure of School Rankings

A pair of researchers put the theory of countersignalling to the test in 2015 by looking at business school admissions websites.⁴ They noticed that some schools disclose their rankings, while others do not.

Intuitively, we would expect that low-ranked business schools would choose not to brag about their ranking, while high-ranked schools would. According to the countersignalling theory, however, it is possible that high-ranked schools could countersignal by not disclosing their rankings, even if they are stellar. After looking at the data, they found this is exactly what happens:

Turning to the data, we find large amounts of voluntary disclosure of rankings; 65% of the schools in our sample publish or mention at least one ranking on their main websites. We find that disclosure decisions are non-monotonic as a function of a school’s rank: top schools are least likely to display their rankings, mid-ranked schools are most likely to display their rankings, and bottom-ranked schools fall in between. These findings are largely consistent with the idea of countersignaling.

Recently, prestigious law schools have been in the news for opting out of rankings.

While the schools claim that they are opting out of rankings for reasons such as unfair and iniquitous criteria, it is also possible that they see this as an opportunity to countersignal and differentiate themselves even further from lower ranked schools. They can be confident that the ‘extra information’ (e.x. their brand, alumni) will not confuse them with low ranked law schools, so they do not have the need to participate in such rankings like medium law schools do.

Middle-Status Conformity

While models of countersignalling are a relatively new innovation, sociologists have been writing about nonconformity and its relationship to status for many decades. It was loosely theorized that there was an inverted U-Shape (IUS) relationship between low status and conformity. Low status members did not feel that it was worth it to conform, while high status members could afford to be nonconforming.

Turning the IUS curve sideways, the sociologists Zuckerman and Phillips graph this dynamic as:

Although over-complicated, the graph shows that as rank increases, the tendency to conform at first increases but then decreases towards the top ranks.

The pair of sociologists do not construct a quantitative model like the prior game theorists, but they do look at two empirical contexts for evidence of the theory.⁵ They found that Silicon Valley legal firms are more likely to offer ‘law status’ legal services if they are more prestigious, and Wall Street analysts are more likely to offer negative stock rankings if they work for a prestigious firm.

Does countersignalling actually work?

In 2013 three psychologists tested how nonconformity impacts the assessments of an audience.⁶ While the rationale for why one may be tempted to countersignal has been explained, it has not been determined if countersignalling actually works.

In this interesting paper, the psychologists conducted 4 studies to see how people view the act of countersignalling. In one study, for example, shopkeepers at an Italian fashion store were surveyed about their opinions of a shopper that came in poorly dressed in a sweatsuit. The shopkeepers had a more favorable impression of such a customer compared to one dressed nicely:

Specifically, shop assistants at luxury boutiques perceived a client to be more likely to make a purchase and to be a celebrity when she was wearing gym clothes or a Swatch than when she was earing an elegant dress or a Rolex.

In another study, it was found that students expected their professor to be more prestigious and competent if they wore a t-shirt and were unshaven, compared to dressing professionally and being well-shaven.

It seems as if countersignalling is indeed an effective strategy to pursue.

Countercountersignalling

Towards the end of the countersignalling paper written by Feltovich et al., they close with a tantalizing suggestion:

Both signalling and countersignalling equilibria may coexist, along with mixed-strategy equilibria where some high types signal and others countersignal. When there are more than three types, counter-countersignalling equilibria are also possible in which the very highest types signal to separate themselves from countersignalling high types.

It is hard to think of real-world examples of counter-countersignalling, and there are none recorded in the literature. Perhaps there is a multibillionaire that dresses professionally to distinguish themselves from billionaires that dress casually as a countersignal, but this could also be attributed to simply a traditional attitude towards clothing.

Signalling, countersignalling, and counter-countersignalling express the rich dynamics that are possible when information is asymmetric, which is a predominating condition of real life.

1. Bellezza, Gino, & Keinan, ‘The Red Sneakers Effect: Inferring Status and Competence from Signals of Nonconformity’, 2013.

2. Technically Spence wrote his doctoral disseration on the same topic the year prior in 1972.

3. Feltovich, Harbaugh, & To, ‘Too cool for school? Signalling and countersignalling’, 2002.

4. Luca & Smith, ‘Strategic disclosure: The case of business school rankings’, 2015.

5. Phillips & Zuckerman, ‘Middle-Status Conformity: Theoretical Restatement and Empirical Demonstration in Two Markets’, 2002.

6. Bellezza et al., supra.