Economics, Literature and Scepticism

Powered by Blogger.

About Me

My photo
I am a PhD student in Economics. I am originally from South Africa and plan to return there after my PhD. I completed my M. Comm in Economics and my MA In Creative Writing (Poetry) at the University of Cape Town, where I worked as a lecturer before starting my PhD.

Tuesday, August 03, 2010

Illusory Superiority and Signalling

Posted by Simon Halliday | Tuesday, August 03, 2010 | Category: , , , |

My sister is an actress, doing more stage acting and dancing than screen work.  My brother is a musician.  I dabbled in acting and singing for much of my youth, even heading up the University of Cape Town choir for a while when I was a student.  Now why is this relevant? As a consequence of these affiliations and activities I occasionally end up watching programs like So You Think You Can Dance (which my brother-in-law did quite well in previously) and IdolsI explained to my students that these two programs create problems for economists and some of our theory.  Why? I shall try to explain below.

Consider signalling theory and let's take the classic example of education.  The idea here is that you try to create a set of incentives such that high ability individuals and low ability individuals self-select into groups such that high ability individuals get an education to signal that they are high ability because to get such an education would be too costly for low ability individuals (this is summarizing greatly).  What this assumes is that people know their own abilities.  But, as Idols and So You Think You Can Dance so readily exemplify, many people do not know that they have low ability.  They believe, falsely, that they are high ability individuals. They are victims of illusory superiority.. 

So, I explained to my students that in the context of Idols and So You Think You Can Dance, that humans are funny creatures.  We're subject to all kinds of cognitive biases.  One particular bias, appropriate for these programs, is the Dunning-Kruger effect.  The Dunning-Kruger effect exemplifies a cognitive bias by which individuals often think that their abilities are better than they are, that is people suffer from illusory superiority.  Consider two people: one of low ability, another of high ability.  If the Dunning-Kruger effect (and, specifically, illusory superiority for low ability individuals) holds, then the individual with low ability does not realise that they have low ability.  This creates a problem for signalling theory.  Why? The low ability person will lack the metacognitive abilities to evaluate their own cognition.  Consequently, they may incorrectly evaluate the costs that they would incur to attend university (get an education).  They might then attend university and fail at a cost to themselves and to society.  The might also, by some fluke, get through university and send the signal of 'high ability'.  But, once they enter the workplace, their employer might then discover that the signal was untrue and they are indeed a low ability individual, which undermines the credibility of the signal for others.

But, the Dunning-Kruger effect has another side.  Intelligent, or high ability, individuals generally have the metacognitive abilities to evaluate their own abilities relative to others.  Consequently, they are often better at measuring when they have high ability in one subject and low in another, but they are also often victims of incorrectly evaluating the incompetence of others: they take their own ability as representative of an average ability to complete a task, therefore underestimating their own abilities.  This adds a second problem: imagine that a high ability individual, falsely believing that she is a low ability individual chooses not to attend university.  This further complicates the signals for companies.  A high ability individual who is hired and found to be a high ability individual undermines the 'low ability' signal of no education. 

Signalling theory in economics consequently faces a host of problems, at least as far as it tries to grapple agents who suffer from cognitive biases, i.e. humans.  I have simplified the problems here, but it does give us a first blush of some of the problems of reconciling actual human thought processes with economic theory.

[Note: I haven't ever read anything about Dunning-Kruger and Signalling theory together, it just struck me that the one contradicted the other, what I know about Dunning-Kruger and other cognitive biases is predominantly self-taught so I may have phrased things slightly loosely.  My apologies.] 

Currently have 4 comments:

  1. Nice...

    One point about the DK effect is that the incompetent think they're better than they are, and the more competent think they're worse than they are. So it's even WORSE for signalling theory.

    Empirical evidence > a theory some economist with physics envy came up with without much testing...

  2. I'm not sure I see a contradiction, unless signalling theory requires the ability to recognize one's own signals.

    Credentials are at best an imperfect signal of ability compared to actual performance. We all know the stereotypes of absent-minded Ph.D.s and brilliant autodidacts.

    Even in your own example, the DK victim who successfully completes an education is accurately signalling an ability to put in enough consistent effort to complete a degree program, vs. the DK victim who fails out, who signals an inability to do so. But employers don't generally think that ability to complete a degree program necessarily translates into high performance at a job in the same field--rather, it's some minimal baseline indicator.

  3. meta cognitive abilities are very important. It is interesting to note that empathy and concepts of other only evolves at ages 3-4 years.

    my guess is that metacognitive traits such as empathy probably have many forms similar to Gardners multiple intelligences.

    another interesting point is the social expectations vary from culture to culture in terms of what or how far knowledge should be stretched. asking for directions is a simple case in point.

    Warren Buffett as an investor does a great job with his concept of a "circle of competance" which he rigidly won't invest outside of. It is an internally and extrenally proscribed mental barrier limiting his actions and has served him well.

    Thanks for the post.

  4. As far as signalling theory is concerned a requirement of making the decision to signal your type (high or low ability) is that the agent knows his or her type (level of ability) beforehand because they should know the costs of effort, time and money that the signal will cost, and therefore also understand the signal. Consequently, it's a problem if you don't know your type. My point is that the theory fails because of what you say - the stereotypes of brilliant autodidacts and absent-minded PhDs (unless these people are signalling something other than education, which they might be).

    I'm not saying that people don't know their type in all circumstances, but in at least some circumstances where economists assume that people know their types, people probably don't. I'd say that education could be one of these.

    I also wasn't saying that employers in real life actually take education as a signal of ability, but rather that the economic theory of signalling has a problem when it comes to DK. In fact, I'd say that real world employers have come up with all kinds of ways independent of education in an attempt to screen candidates for jobs themselves because of education failing as a signal and, possibly, because of DK.