Signalling is basically a means to indicating turning points for certain economic concepts in the context of economic thinking. This article is a review of the original article by Michael Spence.

The problem of signalling is the inability of investors to carry out investment decisions as a result of hoarded information about the investment. Interpreting signals thus become difficult when an investor has expectations about the market, but is not equipped with the requisite information about the market, thus making the investment decision one under uncertainty. Hiring can also be viewed as an investment decision, in that it could be compared to a lottery in which an investor [the employer] acquires labour, but without prior knowledge about his acquired investment. Employees possess certain characteristics from which the employers can make probable deductions about their capabilities. Some of these characteristics, the employees can manipulate, some others, they cannot. The characteristics at the disposal of the employees include education attainment, years of experience and so on for which the employee can invest more time and money in to acquire more. Other characteristics like sex, race cannot be altered by the employees. Those which can be altered are referred to as Signals, while those which cannot be altered are referred to as Indices. This would then mean that employees possess both signals and indices.

Investment under uncertainty could take the form of Hiring.​​​​​​​

Applicants knowing fully well they have abilities of manipulating the signals go ahead to do this, though most times at some cost. Improving one’s educational status involves costs. These costs are seen as signalling costs. It is thus assumed that a signal would not effectively differentiate an applicant from another unless the cost associated with signalling are correlated negatively with productivity – the wage-offer per applicant. Its is on this premise that an observable alterable characteristic can be accorded a persistently informative signal in the labour market. This would also mean that some characteristics may serve as signals for some job types and not for others. Employers modify their conditional probabilistic opinions to wage-offer after being exposed to signals from applicants in one round of employment, such that new entrants face a different wage schedule. With changes as such in the labour market, there would be the need to identify points of consensus between what signals are agreeable between employers and employees, thus requisite for specific wage-offers. This is regarded as a Signalling Equilibrium. A Signalling Equilibrium would thus be defined as a set of employer beliefs which generate offered wage schedules, applicant signalling decisions, hiring and new market data over a given period of time, consistent with the initial beliefs.

For some level of equilibrium, since the years spent in education is negatively correlated with the workers’ productivity, employers may be able to perfectly predict productivity for employees with respect to the job type. In some other cases, the employer may not be able to predict this. This would occur in the case of measure some degree of welfare for the employees, such that if given two groups of employees – I and II, increase in the optimal years of education maybe beneficial to one group and hurtful to the other. To this end, if there is no signalling, an employee maybe paid his unconditional expected marginal productivity (i.e. not predicated by education). Indices may also play some role in information provision. An index such as say sex, is uncorrelated with productivity. This would mean that an employer cannot tell of an employee’s productivity by mere observation of the sex. But then, if sex were to have any informational impact, it would be through its interaction with the educational signalling mechanism. This would also mean that for two groups, men and women, they would both have same signalling costs since as an axiom, people faced with the same preferences ad opportunity sets will make similar decisions and end up in similar situations. They would thus maximise their income net of signalling costs so that their preferences are same.

With same signalling costs they would also be exposed to same opportunity set, thus leading to an assumption that sex possesses no informational quality. But once again, this would be spurious since there are also possibilities of externalities, in that, one person’s signalling strategy affects the market data which also affects the employer’s conditional probabilities. The offered wages and rates of return on education are determined by these, for various education levels. If employers’ distributions are predicated on sex and then education, then the externa impact of a group’s [say man’s] signalling decision are felt by other men. Same goes for the women. And if members of either groups are not investing in education in similar ways, the returns to education for members of either groups would be different in the next round as will their opportunity sets. Given these, there is the possibility of the existence of equilibrium for models of signals and then interaction of signals and indices.

The above article is a review piece of the article below. Ideas contained therein are also credited to the authors cited within in base article.

Spence, Michael (1973). Job Market Signalling. The Quarterly Journal of Economics, 87(3); 355-374