Job Market Signaling

What it is

Spence published the model in The Quarterly Journal of Economics (volume 87, issue 3). It starts from the hiring problem: an employer cannot observe productivity before the hire, but can observe traits that the worker has chosen or can alter. Education is the example the paper develops.

Workers invest in the trait to signal their productivity, and employers set wages on the basis of what they observe. In one worked example education does not raise productivity at all, yet it still determines wages because it sorts workers. A year of schooling costs a low-productivity worker twice what it costs a high-productivity worker.

In effect

The paper states the condition on which everything rests. A signal separates applicants only if the cost of signalling is negatively correlated with productive capability. Spence called this a critical assumption.

He added that the correlation is necessary but not sufficient, because there must be enough signal levels within the relevant cost range. The same condition appears in biology in Alan Grafen's 1990 model, where the marginal cost of advertising must be lower for the better signaller, and Grafen's paper discusses economic signalling models alongside it.

What it does not say

The model does not say that education is worthless or that it only signals. Its worked example is one in which education adds no productivity, chosen to show that a signal can sort workers without any such effect, and the paper does not claim that real schooling works this way. It also says nothing about luxury goods or animal displays; those applications come from later authors, including Nelissen and Meijers (2011) and Simler and Hanson (2018).

---