Arts United States. Congress. Senate. Committee on Labo.

Education's impact on economic competitiveness : hearing before the Subcommittee on Education, Arts, and Humanities of the Committee on Labor and Human Resources, United States Senate, One Hundred Fourth Congress, first session ... February 2, 1995 online

. (page 9 of 15)
Online LibraryArts United States. Congress. Senate. Committee on LaboEducation's impact on economic competitiveness : hearing before the Subcommittee on Education, Arts, and Humanities of the Committee on Labor and Human Resources, United States Senate, One Hundred Fourth Congress, first session ... February 2, 1995 → online text (page 9 of 15)
Font size
QR-code for this ebook

Mr. Bishop. Thank you for giving me this opportunity to testify.
I have longer remarks that I would like to be included.

Senator Jeffords. They will be included in the record.

Mr. Bishop. Thank you.

I will concentrate on the quality dimension as opposed to the
number of years of schooling dimension which the two previous
speakers spoke about.

Compared to seniors in Northern Europe and East Asia, we lag
behind substantially in math, science, foreign languages, and geog-
raphy — ^but surprisingly, maybe, not in reading, and also maybe
surprisingly, given the way the discussion goes, our primary school
students are not lagging behind in reading and science, and they
do lag somewhat behind in math.

So to the extent that we have a quality problem, at least com-
pared to the rest of the world, the problem is a secondary school

Second, we no longer lead the world in terms of proportion of age
cohort graduating from high school. Much larger proportions of age
cohort graduate from high schools in Northern Europe than in our

We still lead in terms of B.A.S as a proportion of the age cohort
getting B.A.S, but a larger share of our B.A.s are in nonscientific
and nonengineering fields. So if you look just at engineering and
science, computer science, the bachelor's and master's degrees, in
those areas we do not lead; we are sort of in the middle of the pack,
in fact, slightly below the average.

Now, does this matter? It matters a lot. We have heard about the
effect of years of schooling. I would like to mention just a few
things about the future of these returns.


In the past when we have had very strong markets for college
graduates, that causes a huge increase in the number of students,
and that has then dumped a lot of people on the market. That hap-
pened in the 1970's, and a lot of them could not find jobs. Will that
happen again? I have looked at that, and it is not happening, and
it will not happen.

The reason is in fact the supply has not gone up. In 1974, the
ratio of the number of people getting B.A.S to employment was 1.09
percent. The most recent year, 1992, it is .96 percent. It has fallen
by about 15 percentage points. And it is projected to fall in the fu-
ture, by the year 2000 and the year 2005.

Why is that happening? The reason is the age cohort has shrunk.
The size of the 20 to 30-vear-old age group is a lot smaller now
than it was in the past relative to population, and so even thought
that group of people is going to college in greater proportion, there
are fewer of them.

The second factor is that more and more retirees are college
graduates now, because we are now retiring that cohort of people
who got a college degree under GI bill.

So consequently, in fact, the growth of supply of college grad-
uates is slowing, and that, put together with a continuing rapid in-
crease in demand for college graduates on the part of business has
meant that that is what has caused this rise in the return to col-
lege in the last 15 years, and it is likely to continue to cause fur-
ther increases.

Not only has the return to college gone up; the return to knowing
more, holding years of school constant, has gone up. So that the re-
turn to knowledge of math is now higher than it was 10, 15 years

Let me give you a little information about how big those returns
are. If we imagine an experiment where the competence in terms
of test scores of people graduating from high school went up by one
grade level equivalent, essentially the difference between the aver-
age score of 11th -graders and 12th-graders on a test, that raises
their earnings on average over the course of their lives by about
$1,000 a year, and that nas a present discounted value, including
the compensation effects, in that they also get higher fringe bene-
fits and so forth, of $20,000 to $30,000, is the present discounted
value at 5 percent real or 8 percent nominal interest rates, which
is roughly the current bond rate of such a benefit.

The effects are small when you are very young, but they grow
quite rapidly and they are quite substantial by Uie time you are

Another kind of evidence is productivity growth effects. I have
looked at the effects of the test score decline on our productivity
growth during the 1980's and 1990's and calculated that essentially
GDP was lowered by 2 percent because the strong positive trend
in test scores that had been in existence prior to around 1965

So what is there that you can do to improve our education? Let
me just point to a couple of things. One is to move further down
the path we have already started of establishing standards and en-
couraging States to set up curriculum-based examinations. New
York State currently is the only State with a system of curriculum-


based examinations, the Regents exams. And guess what — when
you look at the SAT scores of New York State students, holding
constant their family background, they are higher than any other

Canada has some provinces with curriculum-based exams, other
provinces without them. The provinces with curriculum-based
exams at age 13, 4 years before the students would be taking these
exams, the provinces with exams holding family background con-
stant, the students perform about six- to eight-tenths of a grade
level equivalent higher.

The effects of this kind of stimulus are very substantial. The
strategy that needs to be followed particularly is to increase the re-
wards for achievement for youngsters and for schools. This strategy
is already working. Essentially, that is what has happened in the
last 15 years. The payoff to colleges doubled, the payoff to math
has gone up. Employers are paying more attention to academic
achievement when they are making hiring decisions. Colleges are
taking a look at how tough the courses are that students are tak-
ing. That has caused a major change in the culture of our second-
ary schools.

In 1980, 33 percent of 13-year-olds said they either did not get
assigned homework or did not do it. Now that is only 9 percent. In
1980, 32 percent of the kids said they did one or more hours of
homework each night; now it is 66 percent.

On top of that, we have had big increases in the proportion of
students taking algebra-II, geometry, chemistry, physics, all the
tougher college prep courses. And this has had an important effect
that I do not think people realize. The NATE math scores are now
1.06 g^ade level equivalents higher than they were 10 years ago in
1982. The science scores are 1.22 grade level equivalents higher. In
other words, we have already accomplished essentially a $1,000 in-
crease in the earnings of our graduates in just 10 years. Now, that
is coming back from a terrible decline of about that magnitude in
the 1970's, so we are just undoing the damage that was done in
the 1970's, but it has been accomplished in the face of a substantial
increase in the minority share of student bodies and a lot of other
difficulties that schools have had to do with.

So that by following along this path of increasing rewards for
achievement, that has induced the kids to work harder, the schools
to take the academic side of their mission more seriously; that has
resulted in greater achievement. And the numbers that I present
in the paper suggest that if this had not happened, and we had
stayed essentially one grade level equivalent below what we would
have otherwise been — that thought experiment, what is the benefit
to the society — discounted at 5 percent real, 8 percent nominal in-
terest rate, it is $2 to $3.4 trillion.

Thank you.

Senator Jeffords. Thank you very much.

[The prepared statement of Mr. Bishop follows:]


John H. Bishop

Cornell University




"The fate of empires depends on ihe educadon of youth'


If your plan is for one year, plant rice. For ten year plant trees.
Far a hundred years, educate women and men.

The Problem

The NadoD^ Assessment of Educadcnal Progress (NAEP) reports that 92 percent of high school
seniors cannot 'integrate specialized «-i^arifir information' and do not have 'tiie capacity to apply marJiemacical
operations in a variety of problem settings.' (NAEP 1988a p. 51, 1988b p. 42) According to the 1992 National
Adult Literacy Survey, only 23 percent of adults are able to reliably determine correa change using information
from a menu (National Center for Education Statistics, 1994 Table 13).

Secondary school completers in Northern Europe and East Asia are considerably better prepared in
mathematics, science and foreign languages than their American counterparts. Figures 1 to 4 plot the scores
in Algebra, Biology, Chemistry and Physics during the early 1980s against the proportion of the 18-year old
population in the types of courses to which the interaaiional test was administered (Postlethwaite and WUey,
1994). The Americans who participated in the Second International Mathematics Study were high school seniors
in college preparatory math courses. This group, which represented only 13 percent of American 17 year olds,
was thought roughly comparable to the 15 percent of youth in Finland and Che SO percent of Hungarians who
were taking college preparatory mathematics. In Algebra, the score of 40 percent correa for this very select
group of American students was about equal to the score of the much larger group of Hungarians and
substantially below the Finnish score of 79 percent correa (McKnight et al 1987).

The finHing* of the Second International Science Study are similar. Take Finland and Canada, for
example. The 41 percent of the Ftooish students who were taking some biology in their senior year of secondary
school got 50 percent correct The 28 percent of English speaking Canadians taking biology got 43.7 percent
correct The 12 percent of Americans taking a second biology course in senior year got 38 percent correct The
16 percent of Finns and the 25 percent of ^^"giitti speaking r^^narfianc taking chemistry knew almost as much

(Figure 1-4 about here)
as the 2 % of American high school seniors who were taking their second year of chemistry (many of whom were
in 'Advanced Placemenf ) (Fostlethwaite and Wiley, 1994).

It is sometimes said that low achievement is the price one must pay for greater a cc ess. While the share
of all adults with high school diplomas is higher in the U.S. than in other nations, this is no longer the case for
young adults. Table 1 presents data on the ratio of secondary school diplomas awarded to population for a
variety of industrialized countries. The ratio is over 100 percent in Denmark, Hniand and Germany, 90 percent
in Japan, 85 percent in France md 65 percent in England. Despite the minimal standards for getting a diploma


S £ S

s s


- C

5» 2«






r _










20 2»




?• r





in Che United States, the ratio of secondary school diplomas awarded to population 18 years of age was only 73.7
percent in 1988, slightly below its level in 1968.' Standards were lowered in the 1970s, but completion rates did
not improve.

^^ Table 1

Graduation Rates for Secondary and Postsecondaiy Education

Sec Dipl Bachelors Sd,Hng,Math
/PoDlS /Pqd22 Deg/Pop2534

Sec Dipl Bachelors Sd,Hng,Math
/Popl8 /Pop22 Deg/Pop2534

Australia — 24.4%

Canada 73% 33J%

Denmark 100% 16J%

Fmland 125% 17.2%

France 76% 16J%

Germany 117% 133%

Ireland 51% 16.0%



























United Kingdom





United States




Source: OECD, Education at a Glance . 1993, p. 176, 179 & 185. Column 1 is the ratio of secondary school
diplomas and credentials awarded in 1991 to population 18 years of age. It exceeds 100 percent in Denmark,
Fmland and Germany because older individuals &om larger birth cohorts are completing cheir secondary
schooling and because some individuals obtain two secondary level credentials (eg. In Germany some recipients
of the Abitui pursue 3 year apprenticeships which yield vocational quali5cations). The liiird column is 10
multiplied times the ratio of science, mathematics, computer science and engineering degrees awarded in 1991
at all levels (BS, MS and PhD) to the labor force 25 to 34.

Participation in postsecondary education is higher in the United States (see Table 1), but most college
freshmen are studying material that European university students studied in secondary school Many Europeans
doubt that BAs from second rank American universities are equivalent to the French Licence or the Dutch
Doctomal examen.

In the economically critical fields of science, mathematics, computer science and engineering, degree
production relative to population exceeds U.S. levels in Japan, Norway, France and Ireland, inland, Canada,
Denmark, and Germany produce proportionately just about as many people trained in these fields as the United
States. Only Italy, the Netherlands and Sweden are distinctly below the U.S. Many observers beheve that the
abundance and quahty of scientists and engineers has historically been an important source of competitive
advantage for American companies. This advantage is diminishing

^ If G£D certificates were counted as diplomas, American secondary school graduation rates would be
about 10 percentage points higher. The labor market, however, does not view the GEO as equivalent to a high
school diploma. GED certified high school equi\'alents are paid 6 percent more than high school dropouts but
8 to 11 percent less than high school graduates. Most GED test takers spend little time preparing for the exam.
The median examinee spent 20 hours preparing for the exam and 21 percent did not prepare in any way. Their
ASVAB test scores are above those of other high school drop outs but significantly below those of high school
graduates. (Cameron and Heckman 1993). Hence, the OECD did not think GED certificates should be counted
as high school diplomas.



1.1 The EfTect of the Quantity of Schooling on Wages

Educational attainment is the single most important determinant of a person's success in the labor
market. According to the most recent Census report, persons over the age of 18 without a high school degree
earned only S1Z809 on average in 1992, a poverty level standard of living for a family of three. High school
gndiutes emmed 46 percent mora than drop-onts or S18,737 on average. Auodate degrees holders earned 30
percent more than high school graduates or S24,398. on average. Bachelors degree holders earned 31 percent
more than those with AA degrees, or S32,629. on average. PhDs earned 68 percent more than BAs aad those
with professional degrees earned 128 percent more than BAs (Census Nov. 1994). Only a third or so of these
wage differentials are caused by pre cnsting differences in ability, motivation and family background.' The lions
share of the gains represent the real value added of extra schooling. In the 50 years we have tracking it, Lhe
payoff to schooling has never been higher.

Is there a danger of over doing the expansion of higher education? Newspaper stories about laid off
managers and professionals led some to mistakenly announce the end of the strong labor market for college
graduates. While the 1991-92 recession saw a cut back in the hiring of recent college graduates, yotmg high
school graduates suffered even more. Even at the height of the recession unemployment rates of college
graduates never exc e eded 3J percent Their unemployment rates are now less than 2J percent? Those who
completed their BA in 1994 were quite successful in getting good jobs.

What about the future, however? Let us begin by looking at projections of the supply of college
graduates. The high economic payoffs to college during the late 1980s and 1990s resulted in a big increase in
the ratio of BAs awarded to the number of 22 year olds- from 21.6 percent in 1980 to 293 percent in 1992. This
ratio is projected to mcrease further to 33.8 percent in the year 2000, a 56 percent increase over 1980 (NCES
Jan. 199S). The proportionate increase in the total number of BAs awarded, however, is much smaller because
the low birth rates of the 1960s and 70s means that there are significantly fewer individuals in the 20 to 30 year

' Correaed estimates of private returns to schooling can be obtained by including measures of ability
in the model (Griliches and Mason 1972; Taubman and Wales 1975. Hause 1975) or by using sibling data to
match people on ability and socioeconomic faaors (Behrman et al 1977; Olneck 1977). Corrected estimates of
rates of return must also take into account downward biases introduced by errors m measuring schooling (Bishop
1974; Griliches 1979) and the probability that those who choose to continue schooling face higher rates of return
than those who do not (Willis and Rosen 1979). When models correcting for omined variables and selection
effects were estimated in the 1970s, impacts of years of schooling were typically smaller than in simpler models
but the effects were still quite strong. Ashenfelter and Krueger's (1992) recent studies employing comparisons
of identical twins which correct for the biasing effects of measurement error in schcwling found the effect of
schooling to be about as large as the standard cross section relationship.

^ Unemployment rate of managers and professionals, which was 2.0 percent in the first quarter of
1989, rose to 3J percent La September 1992 and have since fallen to 2.2 percent by November 1994. The
unemployment rate of operatives and laborers, which was 7.7 percent in the first quarter of 1989, rose to 11.4
percent in July 1992 and has returned to 7.7 percent in November 1994.


old age cohort that typically receives most of the BAs. As a result, tbc ratia of the number of BAs awanicd
to total employineat fell firom IM percent in 1974 to 0^ percent in 1980 and 0S6 percent in 1992. It is
projected to fall even ftirther to 0.88 percent in 2000 and 0J6 percent in 2005. Thus, despite tbc technology
driven shift in enpioycr demand in favor of college educated worlcers, the flow of new graduates into the labor
maricet has declined. To make matters worse the number of college graduates retiring trom the labor force
is increasing every year (as the veterans who went to college under the GI bill retire from the work force).
As a result, the ratio of workers with a college degree to those with a high school degree or less is projected to
grow at only 19 percent per year between 1988 and 2000, significantly below the 3 J percent per year growth of
this ratio between 1980 and 1988 and the 4.9 percent per year growth between 1972 and 1980 (Bishop 1992).

Now let us examine projections of the demand for college educated workers. In 1991 Shani Carter and
I published two papers forecasting a continuation of upskilling trends (Bishop and Carter 1991; Bishop 1992).
These papers employed a regression analysis of changes in occupational employment shares during the 1972 to
1991 period to project future occupational employment shares. The variables found to tiave significant effects
on occupational shares were: a simple trend, the unemployment rate, the merchandise trade surplus as a
proportion of GDP, and the ratio of personal computers used in business to total empioymenL The personal
computer variable captures the accelerated introduction of computer technology during the 1980s as well as the
direct effects of microcomputers. The preferred model containing all four variables predicted that managerial,
professional and technical jobs will account for 68 percent of growth of occupational employment between 1990
and 200S. Dropping the variable representing the share of the work force with a PC on their desk lowers the
projeaed high skill share to 57 percent and dropping both the trade deficit and PC share lowers it to 52.5
percent So far these projections are pretty much on track. Managerial, professional and technical Jobs
accounted for 59 percent of the 6,728,000 increase in jobs between November 1989 and November 1994.*

If, as predicted by our models, the relative demand for college educated workers continues to grow at
rates similar to those that prevailed in the 1960s, 70s and 80s, current very high wage premiums for college
education will continue and may even escalate further. The latest data (presented in Figures 5a ami 5b) support
the predictions made 4 years ago of cnntiniring escalation of the wage (Ufferential between college graduates and
high school graduates. The earnings of male college graduates fell slightly from 1989 to 1992. But the wages
of high school graduates fell even more, so the payoff to getting a college degree grew dramatically. For females,
there were increases in both the wimings of college graduates and the differential between high school and
college graduate. The present discounted value (PDV) at age 21 of the '■gming* plus fringe benefit gains

* The BLS projects that managerial, professional and technical jobs will account for 40.9 percem of
job growth to the year 2005.. However, the BLS method of projecting occupational changes has consistently
under predicted the growth of managerial and professional jobs. They start with an assimiption— the occupational
composition of employment in individual industries will aot be radically different in the year 2005— that is
manifestly wrong. A few ad hoe adjustments are made to the occupational compositions projections for 2005,
but most of these parameters are taken as fixed. This results in a substantial understatement of upskilling trends.
In 1981 the BLS projeaed that professional, technical and managerial jobs would account for 28 perceu of
employment growth between 1978 and 1990. Dau from the Current Population Survey indicate that these
ocoipations. in fact, accoumed for 53.6 percent of 1978-90 job growth.

88-005 0-95-3


Figure 5a

Median Real Earnings By Education
Ml Male FTYR Workers 25+





Figure 5b

Median Real Earnings By Education
All Female FTYR Workers 25*





HS Srad

BA or Mors I '■ College/HS Ratio

I9a9 tloursi or* 1980 Caniui w«iahti.


associated with a college degree are 5360,000 of which at least S240,(XX) represents real value added.' A four
year college education entails approximately S37^00 m student time costs and S54,500 in instructional costs
(OECD 1993 Table P6). Consequently, the social benefits of a college education are at least 15 times the social

Clearly increases in the quantity of schooling have high sodal payoffs Most of the policy debate,
however, is about the quality dimension. How would improvements in student achievement affea the economy?
It is to this I now turn.

12 The Effect of Education Quality on Wages and Productivity

The mathematical, scientific and technical competencies of workers have big e£Fects on:

* their wages and earnings,

* their productivity on-the-job and

* the nation's standard of living and competitiveness.
Each of these will be taken up in turn.

L2.1 - Consequences for Wages and Earnings

Academic achievement has major effects on the wages of adults even when years of schooling are held
constant In the Department of Education's literacy survey, high school graduates who are in the top 5 percent
in quantitative literacy earn more than twice as much as high school graduates who are in the bottom IS percent
of the literacy distribution. Holding years of schooling constant, a grade level equivalem increase in quantitative
literacy raised 1992 annual r.amings by 5.7 percent or about S900. annually (NOES 1995 Table 4.7). The present
discounted value (PDV) of the increase in compensation resulting from a one GLE increase m quantitative
literacy is about S20,100 (see Table 2).

Analysis of a higher wage group, household heads m the Panel Study of Income Dynamics, found that
increases in general academic achievement raised ^""'"gn by about 4 percent or S1400. annually per GLE
(Bishop 1989). The PDV of the increase in compensation resulting from the one GLE achievement gain was

1 2 3 4 5 6 7 9 11 12 13 14 15

Online LibraryArts United States. Congress. Senate. Committee on LaboEducation's impact on economic competitiveness : hearing before the Subcommittee on Education, Arts, and Humanities of the Committee on Labor and Human Resources, United States Senate, One Hundred Fourth Congress, first session ... February 2, 1995 → online text (page 9 of 15)