Reconciling the signaling and human capital models of education?

Economists have long debated the “signaling theory of education” versus the “human capital” or “value-add” models. In the former case, students benefit from degrees because degrees are a credible signal of underlying traits such as intelligence or conscientiousness; in the latter case, students benefit because because they learn useful skills, go through personal development, etc., experiences which improve their long-run earning potential. Here I lay out some preliminary thoughts on how one might begin to reconcile these two models by looking at the role of compulsory education in the transition from a primitive to a developed economy.

A warning to the reader: Here I am operating well beyond my expertise; the post should be interpreted not as an authoritative statement, but rather as a work-in-progress toward a more unified understanding of a complex phenomenon.


The reason why I choose to analyze this question through the lens of national development is because there is a certain tension between the signaling model of education, in which improvement of human capital plays a minimal role, and the long-appreciated importance of compulsory education in development economics, which appears to be a substantial determinant of whether or not a given country successfully escapes the “middle-income trap.” It is not as though this question has been ignored by the literature; for example, Lang (1994) in the American Economic Review (Does the Human-Capital/Educational-Sorting Debate Matter for Development Policy?) points out that even if the signaling model is completely accurate, efficient sorting of labor can still produce economic gains, an argument which goes back to Stiglitz (1975). Nevertheless, there is something not quite intuitively ‘right’ about attributing the development gains of compulsory education entirely to signaling.

My understanding of the argument for the importance of education (regardless of which model one prefers) in a developing economy largely comes from the book Invisible China, which is written by and based off the work of the economist Scott Rozelle at Stanford. I trust that its overview is more-or-less standard. It goes as follows:

  • In a poor country with low levels of industrialization, returns on education are poor and most people work as unskilled laborers
  • However, as an economy industrializes, general education is necessary to transmit basic skills such as literacy, computer usage, basic writing ability, and so on
  • Whether or not a country successfully educates the bulk of its population, even before the clear and obvious need to do so, is a very strong predictor of whether that country successfully “makes it through” the middle-income trap

In this setting, the human capital improvement model certainly seems plausible. Even if a factory worker has very high IQ, it is just intuitively difficult to imagine such a worker going home and reskilling into an accountant by learning how to use Microsoft Excel in order to keep up with the times (for instance). How would they obtain computer access to begin with? Perhaps they have low levels of literacy and struggle to read a standard accounting textbook. Or perhaps their job is physically exhausting, and they have children to tend to, so even finding enough time to re-skill is something of a foregone cause. Developing countries also likely have poor social safety nets, which certainly does not help.

There is a fair amount of corroborating evidence for this model. Take, for example, the recent paper by Card et al. (2022), The Intergenerational Transmission of Human Capital: Evidence from the Golden Age of Upward Mobility, which uses discontinuities in teacher pay in 1940s America to argue that teacher quality is a significant determinant of upward mobility, therefore lending support to the human capital model. I will not review the entire literature here, but the existence of multiple levels of evidence (both international and intranational) generally makes this fairly compelling to me.

Yet how does one square these results with the equally thorough literature in support of the signaling model? Consider Stephens and Yang (2014) in the American Economic Review, Compulsory Education and the Benefits of Schooling. Here they find that a more careful analysis of the causal effects of increasing compulsory education makes the positive results in prior literature disappear. Similarly, Clark and Cummins (2020), Does Education Matter? Tests from Extensions of Compulsory Schooling in England and Wales 1919-22, 1947 and 1972 show that the effects of increasing years of compulsory schooling seem to have almost no long-term benefit whatsoever. Notably, these results seem to hold across quite a wide range of time periods.

There is also, of course, our own lived experience. I believe most adult readers of this blog will be able to admit to themselves that most of what they learned in school (here I am including every level of schooling, from elementary to university) is largely irrelevant to their present careers. The notable exceptions might be engineering and quantitative analysis; however, these only constitute a small minority of all jobs in the developed world. For myself, I cannot say I remember the last time I ever applied density functional theory or algebraic geometry in a professional context! That is not to say that I did not enjoy learning those topics; however, I retain only a scattering of the details, and I will likely never have to call upon those skills again for the rest of my life, hobbyist endeavors aside. Overall, it is hard to deny the facts presented in Bryan Caplan’s The Case Against Education about the dismal retention of formal education across the entire population.


Now, to speculate a little bitー

Let’s think step-by-step about what it would be like to be a factory owner in 1990s China. Imagine yourself as a foreign capitalist trying to hire a floor manager to supervise the assembly lines. You are not particularly familiar with the culture and cannot easily distinguish, on the basis of verbal ability alone, which candidates are more or less intelligent. Or say you are a native factory owner trying to hire an accountant: the industry is not very developed, so you are yourself somewhat unfamiliar with standard accounting practices and their difficulty (or lack thereof). Your goal is to simply get things running. In both cases, one can imagine that hiring would proceed largely on the basis of easily verifiable skills, such as observable fluency with computer systems and familiarity with the basic technical terminology of accounting.

Alternatively, suppose you are a factory owner in 1940s America. You are American yourself, fairly well off and acculturated into a high social class. You likely have a strong sense of what level of intrinsic talent it takes to do a given job; furthermore, given one or two interviews with any given person, you can generally pick up on their levels of intrinsic ability pretty well. Because you are hiring for a back-office position, you want your hire to be someone you will get along with well, so you subconsciously weigh factors like verbal fluency, status markers (e.g. a degree from a prestigious institution, as opposed to a technical degree specializing in the exact job you are hiring for), and so on relatively higher; consequently you are maybe somewhat less concerned with whether or not a candidate knows the ins and outs of accounting.

What I am trying to propose is that there is perhaps some notion of preexisting social stratification and development that is underappreciated here. People have a natural desire to hire people who are smart and hard-working (whether they admit such a preference for intrinsic traits is a different matter, but I think most would not disagree with this characterization). To assess these traits, one needs a certain level of ambient ‘cultural maturity,’ so to speak: prestige rankings among different universities must be established; the social mores and cultural hallmarks of the intelligentsia need to be relatively stable; hiring managers must have themselves developed the ability to assess (subconsciously at least) these traits and come to some judgment about the overall talent of any given candidate. When this is not possible, however, we fall back on the evaluation of explicit skills, which is in the long-run a noisier test of candidate fitness than an evaluation of deeper, more intrinsic psychological and cognitive traits.

America and Britain (and other Western countries) industrialized somewhat slowly, and so arguably the value of education has always had a high signaling component, even in the early 20th centuryーbecause even at that time sociocultural stratification (perhaps one might say differentiation) had occurred to a great extent. In contrast, nations playing “catch-up” with the West, such as South Korea, Taiwan, and so on undergo extreme social upheaval and instability from a starting point where the lower classes were largely merged together as one undifferentiated mass. Education helps ‘bootstrap’ the process of candidate-to-job matching via teaching explicit skills that can be readily evaluated, but it also accelerates the process of sociocultural class differentiation in and of itself as peripheral, non-job related parts of the school curriculum take on stronger associations with underlying intellect and class.

When mass education is not present, therefore, job matching remains inefficient: most people do not have skills that can be evaluated (in a manner that distinguishes better candidates from worse ones), but the underlying culture also remains sufficient undeveloped that it is very hard for employers to easily look for signals of high intellect and work ethic.

February 3rd, 2023 | Posted in Society

Leave a Reply