The Sheepskin Effect

Critics often point out, for example, that employers in the U.S. of any size have been effectively barred by law from testing job applicants for their cognitive skills if these tests have a disparate impact on protected groups. (The Ricci case that has figured so prominently in Judge Sotomayor’s nomination to the Supreme Court illustrates the legal situation well.) But, the critics have argued, the law does not bar colleges and universities from such testing. These institutions have available to them a whole array of aptitude and skills tests like the ACT, SAT, academic achievement tests, and so on that employers themselves cannot use to evaluate prospective employees. Employers take advantage of this fact, according to the screening or credentialing theory of higher education, letting a college degree serve as a proxy for those skills. Employers reason that if a job applicant has been admitted by a college, has completed the degree requirements, and has the credential, she is more likely than a job applicant who does not have the credential to have the kinds of skills that the employer needs and wants. The credential will therefore have considerable value to the degree holder and to the employer even if college has added nothing to the labor force productivity of the college graduate. It has value to the individual and to the employer purely as a screen or as a filter. (The value to the general society that has thereby conferred value on this otherwise meaningless credential is less obvious.)

Skepticism of this sort about the value of a college degree has grown louder in recent years. However, it is not a new criticism. It dates back to at least the late 60s and early 70s. In fact, there is by now a very large body of theoretical and empirical research on this question. Virtually all of it has been done by economists—which is probably why this research seems to have gone largely unnoticed in most discussions about credentialing in higher education, most of which has been conducted by those who are not trained in economics. It is unfortunate that debate about the value of the college degree has been conducted almost entirely without reference to this large and very interesting body of research by economists. In this essay I will try my best (being a non-economist myself) to bring this body of research into the discussion.

In economics, research on the credentialing issue developed against the backdrop of human capital theory. This fact is of some interest in itself, because many (not all) of the critics of the academy in recent years who have argued that the college degree has little or no intrinsic value have been on the conservative or libertarian side of the political spectrum. From what I have been able to glean from the literature in economics on the issue, debate over the sheepskin effect in economics is not seen as a political issue of liberal versus conservative, or left versus right. Nevertheless, it is of some interest that the human capital theory in labor market economics was developed originally by economists who are often described—and to some degree have been self-described—as conservative or libertarian. Many of the principal figures in the human capital movement have been associated with the Chicago school of economics. Gary Becker and Milton Friedman (1912-2006) are two of the names prominently associated with the University of Chicago and the Chicago school of economics, which continues to be prominently associated with human capital theory. Sometimes, the human capital theorists are referred to more generally as the Columbia-Chicago school, principally in order to include Jacob Mincer, widely regarded as the founder of labor economics, and one of the most important proponents of human capital theory, who taught at Columbia.

Gary S. Becker has defined human capital in the following way:

To most people, capital means a bank account, a hundred shares of IBM stock, assembly lines, or steel plants in the Chicago area. These are all forms of capital in the sense that they are assets that yield income and other useful outputs over long periods of time.

But such tangible forms of capital are not the only type of capital. Schooling, a computer training course, expenditures on medical care, and lectures on the virtues of punctuality and honesty are also capital. That is because they raise earnings, improve health, or add to a person’s good habits over much of his lifetime. Therefore, economists regard expenditures on education, training, medical care, and so on as investments in human capital. They are called human capital because people cannot be separated from their knowledge, skills, health, or values in the way they can be separated from their financial and physical assets.

A prediction of human capital theory is that earnings will increase as a function of years of training or education. This prediction has been consistently borne out by a large body of research past and present, and has always constituted one of the principal arguments for the intrinsic value of education, including higher education.

Beginning in the late 1960s and early 1970s, another group of economists began to develop theories of markets that are characterized by “asymmetric information.” Prominent economists who have worked in this area of economic theory include George Akerlof, Kenneth Arrow, Michael Spence, and Joseph Stiglitz. (Like Gary Becker and Milton Friedman, these have been awarded the Nobel Prize in economics; Akerlof, Spence, and Stiglitz were awarded the prize principally for their work in this area.) Work in the economics of asymmetric information led fairly quickly to questioning the assumption that education must have intrinsic value.

A prediction of human capital theory is that education increases human capital, with each year of schooling adding to that capital until the diminishing marginal utility of education is cancelled out by the opportunity costs involved in getting the education. This view was reflected in the labor market equations developed by Jacob Mincer and adopted by many other labor economists. The asymmetric information school of economics challenged this view of the relationship between years of schooling and labor market earnings by proposing that education served a signaling or filtering function instead. These economists have focused on the imperfections in markets that arise when actors (in this case, employers) are forced to work with imperfect information. For a variety of reasons, these economists reasoned, employers do not have all the information on prospective employees they need to make fully rational employment decisions. College education and college degrees serve as useful signals in this kind of market.

According to one widely discussed model, what is really happening is that employees signal their respective skills to employers by acquiring education. This is costly to all employees, but more costly to some than to others. Employers will pay higher wages to more educated employees, because they know that the proportion of employees with the abilities they want is higher among the more educated, as it is less costly for them to acquire education than it is for employees with low abilities. (Individuals with lower abilities will not persist as long in school, for a perfectly rational reason: they realize that the returns on education given the costs will be much less for them than for more able students.) According to this model of the labor market, it is not necessary for education to have any intrinsic value. All that is necessary in order for a credential to have value is that it convey valuable information about the employee to the employer that the employer would not otherwise have.

The screening or credentialing theory is an empirical hypothesis that needs to be, and can be, tested against the data. It predicts, for example, that employees with a credential—for example, a BA or a BS—will have higher earnings in the labor market than individuals with the same number of years of education but no degree. Alternatively, researchers who do not have data available to them about degree completion, but do have data about years of education, can look for discontinuities (called “knotting”) in the earnings curves. The screening or filtering theory predicts that the employees’ wages will rise faster with extra years of education when the extra year also conveys a certificate. Economists call this prediction the sheepskin effect—or sometimes the “program effect.”

It is very important to be clear about the strength and the weaknesses of this research approach. One weakness is that demonstrating the existence of a sheepskin or program effect cannot, in itself, prove the screening or filtering theory. The pure screening or credentialing theory of education does predict the sheepskin effect, as defined above, and it would be disproved by its absence. However, demonstration of the sheepskin effect cannot, in itself, establish the pure credentialing theory. As Barry Chiswick pointed out as early as 1973 (see the bibliography below), the sheepskin effect (assuming it exists) is not inconsistent with human capital theory. The sheepskin effect might indicate that the credential holders are more productive, as human capital theory holds, even though we do not know why. It might be that all the relevant variables are not in the database, and are therefore invisible to the researcher, although they might be very well known and visible to the employer. The data might also show a complicated, mixed picture that combines intrinsic value and signaling (credentialing).

Empirical work on the sheepskin effect can be divided very roughly into three periods. In the late 1960s through the 80s and the first half of the 90s, economists worked with databases that included (besides the background variables needed to match individuals) only information on earnings and years of education. In the early 1990s, some important databases began to include information on degree completion as well as the background variables and years of education. Economists began using this data in the late 1990s. More recently, research has been aided significantly by the advent of databases that include information on three relevant types of information: years of schooling, degree completion, and measures of cognitive skills involving both numeracy and literacy.

Even when they were forced to rely on databases that included only years of schooling, some researchers found significant sheepskin effects. A good example is the work in the late 1980s of Hungerford and Solon (1987; see bibliography below.) They hypothesized that the years when most students are expected to complete degree programs—8, 12 and 16, in both the U.S and Canada—would show special effects. They reasoned that there would be discontinuities in those years. Each year would show a certain gain in earnings, but students finishing 8, 12 and 16 years of schooling would show significantly larger gains than those who completed only, say, 7, 11, or 15 years of schooling. (For those who are mathematically literate, as I am not: this involved treating the relationship between wages and the education variable S—i.e, the highest grade completed—as either a discontinuous spline function or step function, with the discontinuities at S = 8, 12, and 16.) Using certain mathematical techniques, Hungerford and Solon found significant sheepskin effects:

All of our results point to the existence of sheepskin effects in the returns to education. This finding suggests, first, that treating the log wage as a smooth function of years of education, as is conventionally done in the earnings function literature, gives an inferior fit to the data.

This finding suggested to Hungerford and Solon that the previous dismissals of the screening hypothesis by many economists in the 1970s and early 1980s had been premature.

Research using databases with information on degree completion in addition to years of schooling (Jaeger & Page and Jin Heum Park)

In 1996, Jaeger and Page used data from the 1991 and 1992 March Current Population Survey (CPS), which in those years included, for the first time, information on both years of education and diplomas received. Earlier research, including that by Kane and Rouse (1995)—who had worked with two other databases that included information from postsecondary transcripts, work experience measures, standardized test scores, high school performance, family background, and program completion—had found the credentialing effects of both two and four year college degrees to be small. However, using new data from the CPS, Jaeger and Page (J&P) found (1996: 738-739) significant sheepskin effects:

Bachelor’s and post graduate degrees are valued by the labor market at least as much as years of education. Returns to individual years of schooling are small relative to the estimated sheepskin effects of these degrees, which do not suffer from the bias inherent in previous research.

J&P found that completing a bachelor's degree was worth more than the human capital acquired during three years of college. Their findings for white men are displayed graphically in Figure 1 (p. 736) of their paper, which plots the returns with and without diploma effects against completed years of education. On the same page, J&P express their findings this way:

A substantial part of the total return to education appears to be due to sheepskin effects. ...Sheepskin effects explain approximately a quarter of the total return to completing 16 years of education and more than half of the return to completing 16 years relative to 12 years. … The marginal effect of completing a Bachelor’s degree over attending ‘some college’ is 33%, conditional on attending school for 16 years.

These findings are for regression-controlled data. That is, after controlling for background variables of the individuals, the earnings of individuals holding a BA credential were 33% higher than matched individuals who had the same years of schooling but no BA credential.

Several years later, Jin Heum Park (1999; see bibliography below) found similar results using data that included degree completion, though his estimates of the sheepskin effect were slightly lower than those found by J&P. Park found earnings gains of approximately 9%, 11%, and 21% for obtaining a high-school diploma, an associate’s degree, and a bachelor’s degree, respectively.

Newer databases have made it possible to approach the matter from a different direction. The research discussed so far in this essay, and virtually all the research listed in the bibliography below, have been directed towards testing the human capital theory of education against the screening/filtering/sheepskin hypothesis. These tests have involved using the available databases to predict the effect of years or schooling and (much more recently) degree completion on earnings.

This approach is appropriate if one wants to test the sheepskin effect directly and specifically. However, implicit in the human capital theory is the claim that education increases productivity by developing marketable skills, including cognitive skills. A more direct and natural way of testing the human capital theory itself would be, then, to assess the impact of education on the generation of those skills, and then to see what impact those skills have on labor market earnings. Newer databases, fortunately, support these research inquiries.

There is another important advantage to bypassing the issue of the sheepskin effect issue and testing for gains in cognitive skills directly. There might be a signaling effect simply for attending college. Employers might value such job applicants over those who never attended college, even if the former did not complete a degree. (It is for this reason that J&P included "some college, no degree" as a separate "degree" category in their analysis.) Databases that support estimates of the effect of education on cognitive skills are in a position to estimate the intrinsic value of education more directly, thereby avoiding the questions and issues raised by the possibility that any college education at all has a sheepskin effect. Such databases also support estimates of how education affects earnings by developing these skills. This is done by estimating the indirect effect that education has on earnings via the development of cognitive skills.

So far, the only research using this approach comes from the analysis of data from a Canadian database, the 2003 International Adult Literacy and Skills Survey (IALSS). For those of us who are interested specifically in higher education in the U.S., this is less limiting than it might seem. Canadian researchers have done extensive research on U.S. and Canadian data, and have found a high level of comparability for the data from the two countries. Later research on the U.S. data, therefore, can be expected to largely duplicate the Canadian findings.

The IALSS is the Canadian component of an international survey, involving six countries (including the U.S. and Canada), called the Adult Literacy and Life Skills Survey (ALL). The IALSS and ALL evolved from an earlier international test, first developed in the 1990s, called the International Adult Literacy Survey (IALS). The main purpose of these surveys has been to find out how well adults used printed information to function in society. Survey data include background information (demographic, education, language, labor force, training, literacy uses, information and communication technology, income, etc.) and psychometric results of respondents' proficiency in four skill domains: prose and document literacy, numeracy, and problem-solving.

Two Canadian researchers, David A. Green and W. Craig Riddell, who have used the IALSS data, have found that skills do increase with schooling but at a decreasing rate. (See Figure 1, p. 28.) One extra year of schooling, evaluated when the individual already has 12 years of education, increases cognitive skills by 3.2%. For those with 16 years of schooling, an additional year of education raises cognitive skills by 2.1%. Cognitive gains at lower levels of education are much higher—about 6% for each additional year of education.

These findings might mean that there are diminishing returns to formal schooling in producing cognitive skills. Another possibility, however, is that the skills gains are underestimated by the IALSS data. The IALSS is not aimed specifically at college-level skills, but rather at the general population in Canada (ages 16-65), less than 20% of whom have a college degree. Thus, the skills that are assessed by the IALSS are not high level skills. As Craig Riddell points out (2008: 9), a good high school graduate should do well on the IALSS skill tests. A good university graduate should do better, but perhaps not in proportion to the additional amount of formal education.

Corroboration of the claim that the finding of diminishing returns for later years of education in the IALSS database is an artifact of the limitations and weaknesses of that particular data set, and that the IALSS underestimates the skills gains from higher education, comes from another, quite similar, performance assessment test used in the U.S., called the Collegiate Learning Assessment (CLA), which I have discussed in an earlier posting. Unlike the performance assessments in ALL and the IALSS, the CLA is aimed specifically at college-level skills. It is therefore relevant and useful to compare the available preliminary findings on the CLA with the Canadian IALSS findings.

A preliminary study of the CLA by the Social Science Research Council found that students’ average spring 2007 CLA scores were 0.18 standard deviations higher than their original Fall 2005 performance. A gain of 0.18 standard deviations is about a 16% gain in two years, or an average 8% gain in each of the first two years covered by the SSRC preliminary study. This 8% skills gain estimate is much higher than the 3.2% skills gain estimated in the IALSS data for an individual who already has 12 years of education. It is also much closer to the related 6% estimated skills gains from the IALSS data for lower levels of education in Canada, for which the IALSS might be a more appropriate test.

Databases like the IALSS that include information on earnings, measures of cognitive skills, and years of education also support estimates of how earnings are affected by education through the development of cognitive skills. David A. Green and W. Craig Riddell found that the impact of literacy and numeracy skills on earnings is substantial. Riddell has summarized their findings as follows (taken from an email to me dated July 7, 2009):

Specifically we find that about 20% of the return to additional education is due to the impact of additional education on cognitive skills (at least those cognitive skills measured in the IALSS survey) and the impact of cognitive skills on earnings.

This 20% estimate is based on ordinary least squares analysis. Another estimation technique, which might be less affected by unobserved factors, would be to use instrumental variables (IV). On this analysis, the estimate is higher: 28%.

The above estimates are for education in Canada generally. The findings for higher education in Canada are as follows (from the same email):

In Canada male university grads earn about 40% more than high school grads after controlling for labor market experience; the comparable figure for females is 50%. Our estimates would imply that about 8 percentage points of the 40% male earnings gap is due to the greater literacy and numeracy skills of male university grads (10 percentage points of the 50% gap for females).

This is important new evidence that higher education contributes to higher earnings through the indirect effect of enhancing cognitive skills. It is also important to keep in mind that the 8% and 10% figures may be underestimates, because there is reason to think that the skills gains for higher education are themselves underestimated in the IALSS database. If they are, similar databases to the IALSS that provide better measures of the skills gains of college-level students (like the CLA) can be expected to show higher earnings gains related to the skills developed in college when work is done on them in the future.

A finding of a sheepskin effect is not, in itself, evidence for the pure credentialing or screening theory of higher education. The pure credentialing or screening cannot be true unless there is a sheepskin or program effect, but the credentialing theory cannot be proven by evidence for that effect alone. It is possible, for example, that the B.A. has signaling value because credential holders are more productive in ways that are not visible to the researcher but are visible to the employer. This feature of the credentialing theory has been widely noted by researchers, including some of those notable for having demonstrated the existence of the sheepskin effect. Two examples may be cited.

After arguing that their own findings indicated that the dismissal by previous authors of the screening hypothesis on the ground that sheepskin effects do not exist was premature, Hungerford and Solon (1987: 177) said:

On the other hand, it should be noted that evidence of sheepskin effects need not be interpreted as corroboration of the screening hypothesis. For example, an alternative interpretation due to Chiswick (1973) is that dropouts are disproportionately comprised of inefficient learners who leave school when they realize how little their productivity is augmented by education. Graduates are disproportionately comprised of efficient learners who complete their diploma programs because their productivity is much enhanced by education. Statistical comparisons of wages of graduates and dropouts then appear to show large diploma effects because the graduates are much more productive. Under this interpretation, education's effect on wages arises solely from its effect on productivity and not from any screening function.

Similarly, after concluding that his estimates indicate that degrees at all levels are associated with extra returns to schooling, Jin Heum Park said (1999: 240, fn. 8):

This statement is not meant to indicate that there is a definite causal relationship between degrees and earnings. The results only indicate that those who have degrees earn more than those who do not have them conditional on having the same number of years of schooling. It is possible that those who have degrees are inherently more able than those who do not have degrees and, hence, the earnings differential may only reflect their differential abilities.

What was previously conjectural has now been partially confirmed by research using data from the IALSS data base. In a still unpublished paper that is available online (“Understanding ‘Sheepskin Effects’ in the Returns to Education: The Role of Cognitive Skills—Preliminary”), W. Craig Riddell reports, as many economists had suspected, that those who complete education programs are more skilled than those who drop out of programs, after controlling for years of completed education. (Riddell says, “To my knowledge this is the first evidence that graduates are more skilled than non-graduates with the same years of schooling.”) Riddell also found that the higher literacy and numeracy skills of program completers is a factor in the higher earnings of those completers—i.e., that the cognitive skills are a factor in the sheepskin effect itself.

For education overall, the decline in the estimated sheepskin effect when controls are introduced for cognitive skills is fairly large—20%. This means that estimated sheepskin effects are large even when the IALLS measures of cognitive skills are entered into the equations.

Riddell also found that the decline in the size of the estimated sheepskin effects is largest at the high school level (about 22%), smaller for the completion of a non-university degree (about 11%), and smallest for the estimated effects associated with a college diploma or university degree (about 6-8%). This decline seems to mirror declines in the gains in cognitive skills for those with more years of education. As I pointed out above, these declines in the gains in cognitive skills for higher years of education that are found in the IALLS data base may be due to a limitation of the IALSS as a measure of cognitive skills at the college level, since the IALSS is targeted at the general population in Canada rather than the college-educated population. Since Riddell used the same data set in his estimates of the impact of cognitive skills on earnings and the sheepskin effect, the same limitation may be operative here.

The lifetime earnings advantage of college graduates is one of the most persistent and well-established findings of social science research. The earnings advantage is also very large. According to the College Board (p. 9; see also Figure 1.2), in 2005 the typical full-time year-round worker in the United States with a four-year college degree earned $50,900—62 percent more than the $31,500 earned by the typical full-time year-round worker with only a high school diploma. Two economists, Kevin M. Murphy and Finis Welch, estimated that the earnings premium on getting a college education in 1997 was even higher—1.75 times that of high school graduates.

The view that higher education has intrinsic value, both in terms of earnings generation and the development of marketable skills that improve labor market productivity, has been challenged, but not undermined, by the credentialing/screening theory of higher education. The view that higher education has intrinsic value has survived that challenge in a number of important respects.

These results indicate that both years of completed schooling and credentials are important determinants of earnings. When both measures of educational attainment are available, both the restricted human capital earnings function—in which earnings depend only on years of schooling and experience—and the pure credentialist model—in which earnings are a function only of experience and degrees, certificates and diplomas received —are clearly rejected by the data. This provides a rationale for including both dimensions of human capital in economic and labour market surveys. To the extent possible, both measures should also be incorporated in comparisons of human capital investments over time and across countries.

2.   Another reason for rejecting the pure credentialing hypothesis in favor of a human capital view of higher education is the finding that college graduates are more skilled than non-graduates with the same amount of schooling. This is exactly what one would predict on the human capital theory. An additional reason for rejecting the credentialing hypothesis is that these skills have now been shown to be a factor in the higher earnings of college graduates, as well as a factor in the sheepskin effect itself.

      In recent years, there has been a spate in the development of performance assessment measures. To my knowledge, the only one to date that assesses important skills with a focus at the college level is the Collegiate Learning Assessment (CLA). Unfortunately, the CLA is not yet in a position to estimate the impact of college-generated skills on earnings in the labor market. However, even the IALSS, which is aimed at the general population and is therefore a questionable measure of skills generation at the college level, has shown an earnings effect on college-generated cognitive skills in Canada. Given the high comparability of the Canadian and U.S. data on higher education, this finding is very likely to show up eventually in U.S. research on the ALL database. For the reasons mentioned, longitudinal follow-up studies using the CLA could also be expected to show higher earnings gains connected with college-generated cognitive skills than the IALSS data.

3.   The development of cognitive skills is not the only way that college might increase earnings. Another possibility—which is mentioned in many of the articles discussed in this paper and that are listed in the bibliography below—is that the higher earnings of college graduates might be due to their possession of non-cognitive personal and social skills. None of the research considered in this paper has considered this possibility; this research might therefore be omitting variables that have important indirect effects on lifetime earnings: ambition, drive, leadership, personal and social qualities, emotional maturity; independence, and so on. College graduates might possess more of these kinds of skills, which are also important in the labor market. (There is a large research literature on this question that I hope to address in a future essay.)

For some critics, empirical findings supporting the intrinsic value of higher education like the ones above are beyond counterintuitive: they are unbelievable. A good example of this reaction is a comment on one of my earlier postings (mentioning the CLA) by Kevin M. James:

The CLA study looks interesting and I shall give it a read (despite its gaudy color and layout scheme that hurt my eyes and dredge up unpleasant memories of the late 1970s).

But I must confess that, regardless of the strength of the evidence contained therein, I shall have a hard time believing that virtually any student, of virtually any level of native intelligence and academic aptitude, can benefit from a college education. Unless said benefit is defined downward to the point where, say, a semester in remedial English that brings a student's writing from a 6th-grade to a 9th-grade level is counted as success in collegiate learning.

The experiences I had in my own college career, and that I now have every day in dealing with today's undergraduates, simply cannot be reconciled with this thesis.

In such an environment and in the face of such criticisms, it is important not to lose sight of our own experiences of college, however long ago that was. That is why part of James’ comment surprises me. It has been decades since I have taught in a college classroom, and I am therefore not in a position to judge today’s college students. However, I must question James’ statement that the experiences he had in his own college career cannot be reconciled with the thesis that virtually any student, of virtually any level of native intelligence and academic aptitude, can benefit from a college education. That was certainly not my experience of college, and I dare say it was not the experience of any of my friends in college. Was it yours? Think back and consider what your intellectual skills and general fund of knowledge were when you graduated from high school, and compare them with what they were when you graduated. Was there no value added in your college experience? I would be surprised if very many of those who read this essay would agree with James and say Yes. I certainly wouldn’t. In fact, the proposition that there is no intrinsic value to higher education, at least when I apply it to my own experience, seems patently false.

My own experience has been that higher education can have enormous value to the individual and to the culture and society. It is important for those of us who have had this experience to keep it well in mind. At the same time, personal experience and recollections are not enough. The defense of higher education requires much more than this, including good objective data in the present. That is why studies like the ones considered in this essay are important.

As difficult as it might be for some critics of higher education to admit, the evidence is clear that for a very large segment of our society, a high school education is not enough. If we are to have a literate, informed, intellectually competent electorate and work force in an increasingly complicated and confusing world, in which cognitive skills will be at a very high premium, pre-collegiate education will not be sufficient.

As low as the cognitive skills and literacy of college students are, the skills and literacy levels of high school students are even worse. Furthermore, getting large numbers of high school students trained for middle-skills jobs, as Skills2Compete has urged, while highly desirable, is no substitute for the training in cultural literacy and cognitive skills that can and should be obtained at the college and university level. The kinds of skills that the CLA measures are precisely the skills that are needed for even entry-level management jobs in business. These are also skills that even the liberal arts, which have been much maligned recently, develop. Students enrolled in liberal arts programs have shown gains on the CLA as large or larger than those in science and math, the social sciences, business, and other fields.

At the same time, empirical research confirming the intrinsic value of higher education in generating higher earnings and labor market skills is no justification for self-congratulation or insouciance in higher education. Reform—serious reform—is clearly needed. Fortunately, there does appear to be a growing sense that higher education is reaching a defining moment in the present troubled economy, as tuition increases are meeting price resistance from consumers. (For an interesting list of ways that universities might respond to the changed circumstances, see the Boston Globe op-ed by Joseph Aoun, president of Northeastern University, that was posted recently to NASNET.)

The development of performance assessment measures like ALL and the CLA are surely a sign of the times. Higher education itself sees the need for greater accountability on this score, and will need to concentrate more resources on developing the kinds of cognitive skills measured by these tests. In the process, and as part of the need to lower costs, colleges will probably have to shed many of the frills, particularly those associated with “student services” at elite residential colleges.

In the present climate, the proper attitude would seem to be a recognition of both the value of higher education and the need for reform. The empirical findings reviewed in this essay and in earlier postings show what can be accomplished by serious reform. If the academy, when it is in such a sorry state, can produce significant results on the tests like the CLA, one can only imagine what the results would be if the academy were in better shape. This, I think, is the real importance of the kind of research data considered here. In the absence of such findings, native skepticism about the value of higher education might lead us down the false path of thinking that reform of higher education is not even worth attempting.

Image: "Sheep" by Carl // CC BY-SA

2008 Riddell, W. Craig. “Understanding ‘Sheepskin Effects’ in the Returns to Education: The Role of Cognitive Skills--Preliminary.” Paper for presentation at the CLSRN Workshop, University of Toronto, November 18-19, 2008.

2008 Riddell, W. Craig. “Cognitive Skills: Determinants and Labour Market Outcomes,” Power Point presentation, WISE conference, Xiamen, China, December 12-13, 2008 (obtained from the author).

2007 Riddell, W. Craig and Green, David A. “Literacy and the labour market: the generation of literacy and its impact on earnings for native born Canadians.” Minister, Statistics Canada.

2002 Ferrer, Ana M. and Craig W. Riddell. “The Role of Credentials in the Canadian Labour Market” (Canadian Journal of Economics, forthcoming) Department of Economics, University of British Columbia, May 2002.

2001 Ferrer, A. and Riddell, W. Craig. “Sheepskin Effects and the Returns to Education.” Unpublished paper, University of British Columbia.

1999 Arkes, Jeremy. “What Do Educational Credentials Signal and Why Do Employers Value Credentials?” Economics of Education Review 18 (1999) 133–141.

1999 Park, Jin Heum. “Estimation of sheepskin effects using the old and the new measures of educational attainment in the Current Population Survey.” Economics Letters, Volume 62, Issue 2, 1 February 1999, pp. 237-240.

1997 Belman, Dale and Heywood, John S. “Sheepskin Effects by Cohort: Implications of Job Matching in a Signaling Model.” Oxford Economic Papers, New Series, Vol. 49, No. 4 (Oct., 1997), pp. 623-637.

1996 Jaeger, D. & Page, M. “Degrees matter: new evidence on sheepskin effects in the returns to education.” The Review of Economics and Statistics, Vol. 78, No. 4 (Nov., 1996), pp. 733-740.

1995a Kane, T. & Rouse, C. “Comment on W. Norton Grubb: ‘The varied economic returns to postsecondary education: new evidence from the class of 1972.” The Journal of Human Resources, Vol. 30, No. 1 (Winter, 1995), pp. 205-221.

1995b Kane, Thomas J. and Cecilia Elena Rouse. "Labor-Market Returns to Two- and Four-Year College" American Economic Review 85 (June 1995) pp. 600-14.

1995 Weiss, Andrew. "Human Capital vs. Signalling Explanations of Wages" Journal of Economic Perspectives 9 (Fall 1995) pp.133-54.

1994 Heywood, J. “How widespread are sheepskin returns to education in the U.S.?” Economics of Education Review, Vol. 13, No. 3, pp. 227-234, 1994.

1993 Becker, Gary S. Human capital: a theoretical and empirical analysis, with special reference to education. Chicago: The University of Chicago Press. Third edition. Chapter 2: Human capital revisited.

1993 Frazis, Harvey. “Selection Bias and the Degree Effect.” Journal of Human Resources 28 (Summer 1993): 538-554.

1991 Belman, D. and Heywood, S. “Sheepskin effects in the returns to education: an examination of women and minorities.” The Review of Economics and Statistics, Vol. 73, No. 4 (Nov., 1991), pp. 720-724.

1987 Hungerford, T. & Solon, G. “Sheepskin effects in the returns to education.” The Review of Economics and Statistics, Vol. 69, No. 1 (Feb., 1987), pp. 175-177.

1983 Weiss, Andrew. “A Sorting-cum-Learning Model of Education.” The Journal of Political Economy, Vol. 91, No. 3 (Jun., 1983), pp. 420-442.

1979 Willis, R. & Rosen, S. “Education and self-selection.” The Journal of Political Economy, Vol. 87, No. 5, Part 2: Education and income.

1979 Riley, J. “Testing the educational screening hypothesis.” The Journal of Political Economy, Vol. 87, No. 5, Part 2: Education and income.”

1977 Olneck, Michael “The effects of education,” in Christopher Jencks (ed.), Who gets ahead: The determinants of economic success in America (New York: Basic Books, 1977), 159-160.

1977 Wolpin, K. “Education and screening.” The American Economic Review, Vol. 67, No. 5 (Dec., 1977), pp. 949-958.

1975 Stiglitz, J. “The theory of ‘screening,’ education, and the distribution of income.” The American Economic Review, Vol. 65, No. 3 (Jun. 1975), pp. 283-300.

1974 Layard, R. & Psacharopoulos, G. “The screening hypothesis and the returns to education.” The Journal of Political Economy, Vol. 82, No. 5 (Sep. – Oct., 1974), pp. 985-998.

1973 Arrow, K. “Higher education as a filter.” The Journal of Public Economics 2 (1973) 193-216.

1973 Chiswick, Barry R. "Schooling, Screening and Income," in Lewis Solmon and Paul Taubman, eds., Does College Matter? Some Evidence on the Impact of Higher Education, New York: Academic Press, 1973, pp. 151-159.

1973 Spence M. “Job Market Signalling”, Quarterly Journal of Economics 87, 355-374.

1973 Taubman, P. & Wiles, T. “Higher education, mental ability, and screening.” The Journal of Political Economy, Vol. 81, No. 1 (Jan. – Feb., 1973), pp. 28-55.

1970 Akerlof, G. “The Market for Lemons: Quality Uncertainty and the Market Mechanism," Quarterly Journal of Economics 84, 485-500.

Markets with Asymmetric Information

Nobel Prize: Economics: Laureates: October 10, 2001

George A. Akerlof: Prize Lecture: The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001

A. Michael Spence: Prize Lecture: The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001

Joseph E. Stiglitz: Prize Lecture: The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2001