[Conclusion]
![[ Table of Contents ]](/icons/up.gif){kind=icon}
[Bibliography]
Almost two-thirds of 1980 high school sophomores in the High School and Beyond Survey (64.5 percent) had enrolled in a postsecondary institution by 1992. However, a considerably smaller proportion (42.7 percent) had attained any postsecondary degree by that time.
The internal quote is from Frank Levy, who was interviewed by Harwood. Levy is coauthor, with Richard J.Murnane, of Teaching the New Basic Skills (Murnane and Levy 1996).
Page 654. Robert Samuelson (1998) also sees a "glut of bad students" in college, but posits a different cause. Postsecondary education, he believes, is oversubsidized, causing "too many colleges to chase too few good students. To survive, colleges scramble to get bad students. . ."
Since data are not available for October 1972 and October 1980, we used data from October 1973 and October 1981. For the sake of consistency, we also used data from October 1993 rather than October 1992. The 1993 data are very similar to 1992 data, differing only 0.6 percent for 2-year colleges and 0.2 percent for 4-year colleges.
There were some declines between 1972 and 1980 in the percentage of high school seniors who said they expected to complete postsecondary education. This change no doubt corresponded to the decline in college premiums in the 1970s, which we discuss later.
The data are only suggestive, because, as the previous footnote indicates, enrollments in part "c" occurred a year later than those planned (in part "b"). The individuals involved in parts "b" and "c" of the table are not the same.
These were members of the 1980 sophomore cohort, interviewed 2 years later.
It is sometimes argued that because college graduates make up a larger proportion of their age cohorts than before, they must include more graduates of lower ability than before. That may be true if we assume that the ability referred to is native ability, that its level and distribution have not changed over time, and that colleges start selecting at the high end of the distribution and work downward. However, the key to performance in college and elsewhere is functional ability, rather than native ability as such. Standardized achievement tests measure functional cognitive ability and predict performance in college, especially for freshmen. The math, verbal, and other skills they assess are the result of a combination of native ability, family, schooling, and a variety of other factors. Even if the mean native ability of college-age age cohorts has remained constant over the years, family circumstances and schooling have changed in ways that affect the mean achievement levels of youth. As we saw earlier, young people today have much more schooling than they did at the turn of the century, and levels of parental education, almost certainly a factor in student achievement scores, have increased correspondingly. These are only two of the many factors that can cause average achievement scores to change over time. There can be little question that if a standardized achievement test such as the National Assessment of Educational Progress (NAEP) had been administered to all 17-year-olds at the turn of the century, the mean scores would have been much lower than they would be if it were administered to a similar cohort today. So it's not necessarily true that enrolling a larger proportion of a youth cohort in college will cause the mean functional ability levels of college students to decrease over time, whatever their native ability levels may be.
James Maxey at the ACT Program provided the reports from which the data for this analysis were drawn.
John Bishop (1989) discusses the broad decline in test scores from 1967 to 1980.
The pattern over time is different for college freshmen and high school seniors. The ACT scores of seniors declined between 1970 and 1980 and then began a recovery. However, their recovery did not reach the levels of 1970 (1997 Digest of Education Statistics, table 134). The same is true of the SAT tests (Digest, table 129).
It is unclear whether Bishop's freshmen include those in 2-year colleges. They seem to.
Astin et al. (1996) point out that elements of ACT's methodology tend to inflate the estimates. For example, ACT weighted private colleges, which tend to have higher graduation rates, the same as the larger public colleges, which have lower rates.
An increase in the proportion of such students would cause the bachelor's degree completion rate to fall, but would not necessarily cause the dropout rate to rise. These students would not be counted as dropouts because they were still in school or because they had attained certification other than a bachelor's degree.
See indicator 11 and table 11-1 ofThe Condition of Education 1996 (Smith et al. 1996).
This finding does not contradict the NCES finding that the proportion of young people taking longer than 6 years after high school to get a BA increased from 25 percent to 32 percent between 1977 and 1990. The starting point for the NCES figures is high school graduation; the starting point for Astin's figures is entry into college.
One feature of community colleges that is associated with lower completion rates is that they are not residential institutions. Campus life in residential institutions strengthens a student's attachment and commitment to the institution and thus tends to increase persistence.
In this particular analysis, Hecker is not specific about the period of increase. It seems to be approximately 1970?1990. However, the pattern of change over this period is complex and does not easily translate into an argument that both premiums and proportions of college graduates in non-college jobs increased. The greatest increase in the non-college job proportion occurred in the 1970s, at a time when the college premium was declining. In the 1980s, when the premium was again ascendant, the proportion of college graduates in non-college jobs was leveling off.
The "bumping" of high school graduates out of the higher-paying non-college jobs is consistent with Lester Thurow's (1975) "job competition" model of employment and wages. Thurow argues that employers use education level as a proxy for ability and select the most able job-seekers available, whatever the skill level of the job.
The authors call them "high school" jobs, though they use Hecker's classification.
In occupations where the average education level was more than high school but less than 4 years of college, Pryor and Schaffer found a similar but less pronounced pattern.
Other explanations are possible, but we think they are not supported by the weight of empirical evidence. For example, if Spence (1973) is right, colleges just screen out less able students and graduate the more able ones. Then, following Thurow's (1975) theory, firms would prefer to hire the more able applicants, regardless of the skill level required by the job. The college premium could increase if the real earnings of high school graduates fell for reasons unrelated to their own skills and the skills needed for the job. Such reasons might include reduced union bargaining power and a diminished real minimum wage. The section entitled The Value of a College Degree in Perspective examines the evidence for de-unionization and reduced minimum wages as factors in the growing difference between college and high school earnings.
Data are from the Bureau of Labor Statistics Internet web site. Estimated weekly earnings for male college graduates were $800 in 1994, $794 in 1995, and $796 in 1996. Weekly earnings for females were $621 in 1994, $616 in 1995, and $608 in 1996.
The purpose of this analysis is to compare premiums in the 1970s with premiums before and after the 1970s. If the study in question did not include the 1970s, the year of the low rate was not shown in table A-3.
In technical terms, the internal rate of return is measured by the discount rate that equates the present value of the difference between the lifetime earnings of college graduates and the lifetime earnings of similar high school graduates with the present value of the costs of college.
Freeman and McCarthy (1982) also show that humanities and social science majors suffered the greatest decline in earnings premiums in the early 1970s. One explanation for the higher earnings of occupational majors is that occupational education may give graduates an earnings advantage in entry-level jobs because employers, who do not have to train them extensively, are willing to pay more for their services. Over longer periods, graduates of other majors may gain the training and work experience they need to command comparable earnings. However, Fox (1988) emphasizes supply factors. He shows that increases in the size of graduating classes from a major, relative to population, tend to depress starting salaries. Since there are more arts and science majors than business and engineering majors, their starting salaries are lower. If Fox is right, why don't more students shift from the less rewarding arts and sciences to the more rewarding science and engineering majors? One possible explanation is that liberal arts majors may be less motivated by monetary incentives than business and engineering majors, on average. Another is that many students may regard business and engineering majors as more difficult and intrinsically less rewarding (e.g., less interesting) than arts and science majors.
One of the striking findings in the inequality literature is that the growth of wage inequality has been greater within groups of the same gender, education, race, and experience than between such groups (Katz and Autor 1998, Gottschalk 1997).
Katz and Autor (1998) actually use a supply/demand/institution framework. In addition to supply and demand, institutional interventions in the market such as union bargaining and minimum wage setting are factors in variation in wages.
Indeed, as Goldin and Katz (1998) pointed out, the demand for skilled labor has been growing for most of the century, a fact passed over in much of the literature.
Snyder et al. (1998), table 244.
U.S. Bureau of the Census (1974, 1982, 1992, 1993a, 1993b, 1993c, 1996).
See figures 3 and 4 in Topel's article.
Another view is that the oversupply of college graduates in the 1970s actually contributed to the increased premiums of the 1980s. See discussion below and Acemoglu (1998).
We need to keep in mind that "technology" includes more than computers and information management. The 1980s were also a decade of major change in the organization of work in business and industry. Organizational innovations such as the "high-performance workplace? based on principles pioneered by W. Edwards Deming in Japan tended to require higher-level skills and to improve productivity. How large a role such changes played in the growth of inequality is unknown, but they should not be overlooked.
Not all of these between-industry shifts are due to globalism; but on the other hand, some within-industry shifts are due to more open global markets. Outsourcing—a firm's contracting to have some of its work done by others—has grown in recent years. A good deal of outsourcing goes to contractors in other countries. To the extent that U.S. firms contract abroad for low-skilled labor, they change the domestic skill mix within the firm, driving up the relative demand for skills. How large an effect outsourcing abroad has had on the growth of demand for skills is a subject of debate. Berman et al. (1994) conclude that the effect is small, while Feenstra and Hanson (1996) conclude that it is an important factor in the growth of demand in the 1980s.
| [Conclusion] |
|
[Bibliography] |