Saturday, October 14, 2006

Signaling at the AEA Job Market

If you are an econ grad student about to go on the job market, this post is for you.

A PhD candidate from the University of Michigan sends me an email:

Hi Greg,

As a possible blog idea that would be well received by many econ grad students around this time each year, perhaps you could offer some thoughts as to the econ job market. How to best go about it, how to reduce noise in the process, how to sell oneself appropriately, ... Some insight as to the new signalling system offered by the AEA would also be most welcome! The latter is raising some real questions in my neck of the woods among this year's outgoing cohort (of which I'm one). Some like the idea as helping to reduce noise in the process by identifying some individuals' strong preferences. Others fear the inevitable interview question: "Why didn't we get a signal from you?" which can best be avoided by not sending any signal!

We'd all your insight on this whole process!

[name withheld]

On the job market in general, please see the links in a previous post.

Before receiving this email, I didn't know anything about the new AEA signaling system. I guessed (correctly, it turns out) that my Harvard colleague Al Roth would have something interesting to say. I emailed Al and got this in reply:

Hi Greg:

Indeed, I'm the chair of the AEA ad hoc committee on the Economics job market that put the new signaling mechanism into effect. (Last year we introduced the "scramble" web page in March, which will also operate this year.)

Just as the idea of the scramble was to add some thickness to the late (March-April) part of the market, the idea of signaling is to reduce some of the congestion in the thick, January interviews part of the market.

(Incidentally, Muriel Niederle is giving a seminar on it this Tuesday, in the behavioral/experimental seminar. She is on the committee too, and is doing some related theoretical work with Peter Coles.)

In terms of advice, we wrote the (attached) document, which also appears on the web now. [See the first link given above.] I'm also enclosing the scramble document which appeared last year and which isn't now on the web, but a revised version will reappear later in the market.

The basic idea of a signaling mechanism is that there is a big part of the market in which departments, in allocating scarce interview slots, have to form an assessment not only of how promising a student looks, but also of how likely that student is to be interested in them. (Harvard doesn't spend a lot of time worrying about that, but at Pitt, where I spent many happy years before coming to Harvard, we definitely factored that into our decisions.)

Of course students can send any signals they want in their cover letters, but because every cover letter expresses interest, that may be of limited help to departments in separating the signals from the noise. To some extent that may also apply to information in emails and letters from advisors. Those channels can all convey valuable signals of interest, of course. The new signaling mechanism is just a supplement to the traditional ways of signaling interest, and may be of most help to students who are interested in places to which they don't have other reliable means of conveying their interest. Because they can send a maximum of two signals through the AEA mechanism, the signals may convey some information.

So, what information should students try to convey? In our "advice to applicants" paragraph, we said

"The two signals should not be thought of as indicating your top two choices. Instead, you should think about which two departments that you are interested in would be likely to interview you if they receive your signal, but not otherwise (see advice to departments, above). You might therefore want to send a signal to a department that you like but that might otherwise doubt whether they are likely to be able to hire you. Or, you might want to send a signal to a department that you think might be getting many applications from candidates somewhat similar to you, and a signal of your particular interest would help them to break ties. You might send your signals to departments to whom you don't have other good ways of signaling your interest."

Basically we don't think students will often want to signal MIT or Princeton or Stanford or Chicago or other very competitive departments very often, because those departments can somewhat safely presume that they'll have a reasonable chance of being attractive to any students they interview. And we expect that departments will understand that they may not get signals from applicants who can demonstrate clear interest in other ways. So, in our advice to departments, we wrote

"Applicants can only send two signals, so if a department doesn't get a signal from some applicant, that fact contains almost no information. (See advice to applicants, below, which suggests how applicants might use their signals). But because applicants can send only two signals, the signals a department does receive convey valuable information about the candidate's interest.

A department that has more applicants than it can interview can use the signals to help break ties for interview slots, for instance. Similarly, a department that receives applications from some candidates who it thinks are unlikely to really be interested (but might be submitting many applications out of excessive risk aversion) can be reassured of the candidate's interest if the department receives one of the candidate's two signals."

If you wanted to give a very toy model of why signaling might be useful, you might want to start with the two-firm, two-applicant example in which on one even cares who works for whom, but each firm has only one interview slot, and can only hire someone they have interviewed. Then the symmetric equilibrium involves randomization (each firm randomly chooses one applicant to interview), and there is coordination failure half the time (when both firms interview the same applicant, so only one hire is accomplished). But if applicants can first send one signal, then even if they randomize to whom they send the signal (since they don't care in this simple example), then coordination failure is cut in half, if each firm adopts the strategy of interviewing the applicant whose signal they get in case they get exactly one signal. (Now, coordination failure is only a possibility when both applicants signal the same firm, in which case both firms randomly choose who to interview, which is the situation that existed in the absence of any signal...)

Al Roth

Thanks, Al.