College-Going Behavior as Signaling

  • Kazuhiro Arai


There is a theory of college-going decision making which is quite different from human capital theory. This alternative theory is based on the idea that the role of a college degree is not to certify the knowledge and skills acquired in college, but simply to convey information to society about the degree holder’s productivity. This productivity may be innate, or may have been acquired (at home) by the time of the start of college education. According to this theory, those who have high productive capabilities go to college so that firms can identify degree holders (or highly educated people) as more productive and pay them more. Here, college education does not enhance the productivity of the students, but only plays a role of judging how productive each individual originally is. This role of education has been called signaling, screening, filtering, or sorting, by different authors (in slightly different problem settings). This “signaling” theory is very powerful and competes with human capital theory to explain college-going behavior.


High Education Female Worker Rich Family Productive Capability Statistical Discrimination 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    In fact most signaling models assume that education does not enhance the productivity of students at all. As we will see in Sect. 3.6, however, it is possible to build a model in which education plays a signaling role and at the same time enhances student productivity. On the other hand, even if education does not enhance individual productive capabilities, it may help enhance productivity because it reveals aptitude, and this information enables correct matching of individuals to firms or jobs. (The productivity of an individual tends to be low if he/she works in an environment which does not suit him/her.) This is the problem of matching, in which education helps allocate heterogeneous individuals to their most productive use. See Jovanobic (1979) and MacDonald (1980) for job matching.Google Scholar
  2. 2.
    Even in human capital theory, all additional costs of higher education, whether mental or time costs, are investment costs of higher education. As we have seen in Chap. 2, however, human capital theory tends to make light of unmeasurable costs.Google Scholar
  3. 3.
    When using the latter assumption, the firm is assumed to be risk-neutral. A risk-neutral firm is supposed to behave so as to maximize expected profits.Google Scholar
  4. 4.
    Each of these values should be interpreted as the total present value of the products that each type of individual can produce from the time of hiring to the time of mandatory retirement. Strictly speaking, those values should be called marginal value products, but they are called productivity in the following for simplicity, because it is implicitly assumed here that marginal productivity is constant regardless of the number of workers employed and because the price of the product can be assumed to be unity. This set of assumptions can be rephrased in accordance with Fig. A.1 in Appendix A: a special case is assumed here in which the demand curve (like D1D,or DD) for each group of workers becomes horizontal if the two groupsGoogle Scholar
  5. 5.
    Bar edistinguishable. (Because only one firm exists in this economy, D,D,is identical with DD.) Of course, the demand curve for GI is different from that for GII, the height of the former being 1 and that of the latter 2. If the two groups are not distinguishable, the demand curve for labor becomes a horizontal line with a height equal to the average productivity of the two groups. Because workers are paid wages equal to their productivity, the firm’s (expected) profits become zero, as far as its beliefs are correct. In other words, the firm behaves so as to maximize (expected) profits, but maximized profits are zero. It is possible to consider a case in which maximized profits are positive, but this causes the model to become complex. According to Shigeno and Matsuura (1995), parents’ income, wealth, and social status significantly affect their children’s purchase of private school education and cram school education at the stage of compulsory education.Google Scholar
  6. 6.
    Spence (1974a, Chap. 6) mentions the case where some productive individuals cannot go to college because their parents are poor. He allows, however, that the basic pattern of the signaling equilibrium in this case is much like the one obtained in Spence (1973). This will be discussed in Sect. 3. 3. 2.Google Scholar
  7. 7.
    The signaling equilibrium in this model is obviously disadvantageous to G21 members. This disadvantage may be removed by introducing a financial aid system in which those from poor families can use loans at a low interest rate to receive higher education. Under such a system, the members of G21 may be able to go to college in the same way as those of G12 and G22. Then not only the welfare of G21 members but also that of G12 and G22 members will increase, because the average productivity of the whole group of college graduates will increase. (To simplify the story, we assume that receiving a financial aid does not become a signal.) In contrast, non-college graduates will be only GI1 members and their wage will decrease to 1. However, such a large-scale financial aid system can exist only in very mature welfare states. Other countries cannot provide such generous financial aids. Moreover, if there is competition for admission as in Japan, those from poor families may not succeed in this competition and thus cannot obtain financial aids which are offered on condition that they are admitted to college.Google Scholar
  8. 8.
    The benefits of potlatching and conspicuous consumption are mainly psychological, whereas those of investment in higher education in the above models are pecuniary. However, the above models also provide insight into the case where individuals purchase higher education to obtain psychological benefits. Suppose that displaying wealth by consuming higher education brings about psychological benefits. Then, wealthy individuals, for whom education costs are low in comparison with psychological benefits, purchase higher education. These are benefits of present and future consumption according to the classification of the benefits of education in Chap. 2.Google Scholar
  9. 9.
    Higuchi (1992) shows in his empirical study that students enrolled in universities with keen competition for admission tend to be from high-income families and the graduates of those institutions tend to obtain jobs which offer high lifetime salaries.Google Scholar
  10. 10.
    See Phelps (1972), Akerlof (1976), Borjas and Goldberg (1978), Aiger and Cain (1977) for statistical discrimination.Google Scholar
  11. 11.
    Those whose pre-college record is 0.6 actually lose from the existence of filtering, because though their expected productivity is 0.6, the expected productivity of those who have a college degree is regarded as 0.556.Google Scholar
  12. 12.
    This fact is contrary to Arrow’s idea of filtering, but firms in his model are assumed not to know job applicants’ pre-college records.Google Scholar
  13. 13.
    This section presupposes that the reader has elementary knowledge of partial differentiation.Google Scholar
  14. 14.
    When time changes continuously in an economic model, the rate of growth of a variable x which is a function of time t can be expressed as (dxldt)lx This is because when time changes by one unit, x grows by dxl dt and the rate of growth equals the ratio of dx/dt to x. If you lend 1 yen at an annual interest rate of r, it grows at the annual rate of r and will become e“ yen in t years. This is because (de”ldt)/é ` = r and e“ certainly grows at the rate of r. Since 1 yen at present and e” yen t years later have the same economic value, the present value of y yen t years later equals e - “y yen.Google Scholar

Copyright information

© Kazuhiro Arai 1998

Authors and Affiliations

  • Kazuhiro Arai
    • 1
  1. 1.Hitotsubashi UniversityKunitachi, TokyoJapan

Personalised recommendations