View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

June 2013, EB13-06

Economic Brief
Implications of Risks and Rewards
in College Decisions
By Kartik Athreya and David A. Price

Despite a large and growing earnings premium for college graduates, growth
in college enrollment and especially college attainment in the United States
has been quite slow. The labor market’s apparent lack of responsiveness to
the earnings premium may be driven in part by risks that marginally prepared
students face when they go to college. Failing or dropping out could leave
them with low wealth, high debt, and low earnings. Recent research indicates
that neither further increases in the earnings premium nor reductions in costs
are likely to produce large increases in the college completion rate. And if
technological change continues to increase the demand for skilled labor,
both the earnings premium and income inequality will continue to grow.
Completion of a college degree generates, on
average, a large and growing wage premium. In
one longitudinal analysis, people who completed
college earned, over their lifetimes, more than
twice as much as those who did not.1 The college
premium increased substantially from the late
1970s onward; workers with bachelor’s degrees
received 1.4 times the wages of high-school-only
graduates in 1980, a multiple that climbed to
1.75 by 2005.2 Even taking into account the rising
costs of higher education, investing in a bachelor’s degree remains highly lucrative for most
college-ready students.
The growth of the college premium has given rise
to an economic puzzle: In view of the substantial
payoff to higher education, why have college enrollment and completion rates been growing so
slowly? Although college enrollment is high, approximately one-third of students who complete
high school either delay college or never go, and
the share of people who complete four-year deEB13-06 - Federal Reserve Bank of Richmond

grees by the age of 25 has increased over the past
two decades by a mere four percentage points. In
short, the response of the supply side of the labor
market to changes in the college premium has
been surprisingly weak. Recent research by Richmond Fed economist Kartik Athreya and Northwestern University economist Janice Eberly seeks
to pinpoint the extent to which various factors
are deterring students from investing in college.3
At the center of Athreya’s and Eberly’s analysis is
the role of risk in a student’s decision-making.4
Studies looking at the rate of return on a college
investment commonly address the value of a successfully completed college education. Such an
approach disregards the significant uncertainties
that a student must consider at the threshold of
his college decision: What is the probability that
he will fail academically? And even if the student
earns a degree, what is the probability that the
high average payoff from a degree will not be
realized in his particular case? Moreover, because
Page 1

college programs are normally two or four years in
length, the student may consider these questions
anew when deciding whether to return to school for
each new semester.

is also more opaque given their greater tendency
toward price discrimination in the form of high sticker
prices combined with individualized financial aid
packages.

Athreya and Eberly argue that both “failure risk” and
“rate-of-return risk” are important determinants of
an individual’s actual return from a college investment. Around half of students who begin college
do not complete it, and the return to attending
college without earning a degree appears to be low.5
For many students, college is a costly investment.
It entails years of foregone earnings. Finally, a wide
range of empirical work in economics suggests that
even after completion, there is risk of a disappointing payoff from college. Wages and earnings seem to
have significant heterogeneity.6 These risks arise from
the possibility that college graduates may suffer longlasting shocks to employment opportunities from,
say, changes in the structure of the economy that render certain occupations less valuable. In the current
context, an additional risk is that college premia may
shrink, as they have done at times in the past.7

With regard to available resources, the literature
indicates that family savings are typically minimal;
indeed, one study finds that among families saving
for college, only 25 percent of those with high school
seniors had set aside more than $10,000.10 Athreya
and Eberly assume a distribution of family resources
with a median of $3,000.11 In addition, in view of federal loan-guarantee programs, they assume that students are able to borrow the full cost of college if they
wish, since this falls within the parameters of the loan
programs. For the college premium, the researchers
assume a path for earnings that would approximate
the premium prevailing from 1993 through 2005—
that is, a multiple of 1.75 for those who received their
bachelor’s degrees compared to those who only
earned high school diplomas, and 1.2 for those who
completed some college.

Lastly, for many students, college is a highly leveraged
investment.8 Moreover, unlike a financial or physical
asset, it cannot be resold, and the loans often used to
finance a college degree—U.S. government-guaranteed student loans—are not dischargeable in bankruptcy.9 Hence, students who fail or drop out face a
triple threat to their financial futures—low wealth,
high debt, and low earnings.
To estimate the extent to which risk and other factors
influence enrollment decisions, Athreya and Eberly
construct a model to predict the flow of new college
enrollees on the basis of education costs, resources
available to pay those costs, the expected college
premium, the risk of failure, and the risk to subsequent earnings. They base education costs on College Board estimates of tuition at all public two-year
and four-year institutions of higher education. They
focus on public colleges because some 75 percent of
students attend those institutions and because public
colleges normally are a feasible option for students
who choose private institutions (while the reverse
often would not be true). Pricing at private colleges

In assessing failure risk, the researchers note that
students and their families estimate the risk of not
completing college on the basis of numerous factors, “including prominently a combination of family
background, high school performance, and standardized test scores.” In the researchers’ model, the proxy
for failure risk—or, put differently, the level of preparation for college—is academic achievement tests
given to high school students as part of longitudinal
studies that began in 1972 and 1988.12
In assessing rate-of-return risks, specifically to the
wages households will earn, the authors employ
estimates of risk that take account of differences in
educational attainment as well as differences in the
way both taxes and public social insurance systems
affect people of different educational levels.13
Athreya and Eberly find that any further increase in
the college premium—within reasonable bounds—is
unlikely to motivate a large number of additional students to attend college. Figure 1 shows the model’s
implications across a wide range of multipliers for the
college premium.14

Page 2

Among the one-third of high school graduates who
do not immediately go on to college, and among
those who now attend, few are on the fence. The
reason that increases in the college premium lead
to a minimal number of additional enrollments and
completions appears to arise from the fact that
enrollment does not automatically lead to reaping
the benefit of the premium. The risk of failure seems
to influence both the decision whether to enroll and
the decision whether to drop out. In effect, a higher
college premium improves the expected return for
the well-prepared students, who were already likely
to attend, but not for poorly prepared students, who
were not likely to attend. As a result, increases in the
premium do not change the decisions of many students. Moreover, even though the researchers’ model
does not place constraints on borrowing, a student’s
wealth and preparedness appear to substitute for one
another in the decision whether to enroll. For those
who are well-prepared, family financial resources
do not greatly affect that decision, but as the level
of preparation declines, the importance of family

resources becomes greater in light of the risk of failure. Reductions in the cost of college—such as those
resulting from subsidies—correspondingly matter
primarily for the poorly prepared, and so again may
not substantially change the overall rate of college
completion. This finding suggests, consistent with a
growing body of evidence, that resources aimed at
earlier ages (pre-K and K-12) are more likely to yield
benefits than any modest increase in the level of aid
to college enrollees.
These results also have implications for the future of
earnings inequality in the face of any technological
changes that favor skilled workers and increase the
college premium. Because college enrollment and
especially completion may not respond strongly to
such changes, the premium may continue to rise and
remain persistently high. As a result, all else equal,
income inequality arising from the gap in earnings
between those able to take advantage of the college
premium and those unable to do so will continue
to grow.

Figure 1: As the College Premium Increases, the Model Predicts that More Students Would Enroll
but a Lower Percentage Would Graduate
90
80

Percent of Students

70
60
50
40
30
20

Enrollment Rate
Graduation Rate
Attainment Rate

10
1.4

1.5

1.6

1.7
1.8
College Premium

1.9

2.1

2.0

Note: The college premium is the average lifetime earnings of four-year-college graduates expressed as a multiple of the average lifetime
earnings of high-school-only graduates. The enrollment rate is the percent of high school graduates who enroll in two-year or four-year
colleges. The graduation rate is the percent of those enrollees who earn bachelor’s degrees. The attainment rate is the percent of all high
school graduates who earn bachelor’s degrees (the enrollment rate multiplied by the graduation rate).
Source: Athreya and Eberly (2013)

Page 3

Kartik Athreya is group vice president for microeconomics and research communications in the
Research Department of the Federal Reserve Bank
of Richmond, and David A. Price is senior editor of
the Bank’s economics magazine, Econ Focus.

11

The true resources available to an enrollee are not directly
observable because they depend on the willingness of parents
and others to make transfers to the student. Researchers have
used data that attempts to track the actual “within-lifetime”
(as opposed to bequest) transfers. See, for example, Abbott,
Brant, Giovanni Gallipoli, Costas Meghir, and Giovanni L.
Violante, “Education Policy and Intergenerational Transfers in
Equilibrium,” NBER Working Paper No. 18782, February 2013.
Athreya and Eberly seek to remain conservative in estimating
resources, as higher numbers would ensure, once the remainder of their model is “calibrated” to match current enrollment
rates, that even more households would be insensitive to the
college premium.

12

The tests were given during the National Longitudinal Study
of the High School Class of 1972 and the National Educational
Longitudinal Study of 1988. See Bound, John, Michael F.
Lovenheim, and Sarah Turner, “Why Have College Completion
Rates Declined? An Analysis of Changing Student Preparation
and Collegiate Resources,” American Economic Journal: Applied
Economics, July 2010, vol. 2, no. 3, pp. 129–157. Family wealth,
which correlates with test scores, still will vary across households and generally matter above and beyond any effect on
test scores. Athreya and Eberly account for this in their model
by distributing failure risk such that the model’s enrollment
rates by test score, when averaged across all wealth levels of
households within a given test-score quartile, match those in
the data.

13

Hubbard, R. Glenn, Jonathan Skinner, Stephen P. Zeldes,
“Precautionary Saving and Social Insurance,” Journal of Political
Economy, April 1995, vol. 103, no. 2 pp. 360–399.

14

It is important to note that the model aims to study enrollment on the basis of a premium that is known by the student
to be fixed over his or her lifetime. In reality, college premia
have changed over shorter periods. As a result, the model
does not allow for a direct comparison with a given historical
moment when a particular college premium prevailed.

Endnotes
1

This calculation is based on the present values of lifetime
earnings using a sample of 33- to 48-year-old workers from
the Panel Study of Income Dynamics. See Restuccia, Diego,
and Carlos Urrutia, “Intergenerational Persistence of Earnings:
The Role of Early and College Education,” American Economic
Review, December 2004, vol. 94, no. 5, pp. 1354–1378.

2

See Archibald, Robert B., and David H. Feldman, Why Does
College Cost So Much? New York: Oxford University Press,
2011, p. 55.

3

See Athreya, Kartik, and Janice Eberly, “The Supply of CollegeEducated Workers: The Roles of College Premia, College Costs,
and Risk,” Federal Reserve Bank of Richmond Working Paper
No. 13-02, March 5, 2013.

4

Earlier work that has examined the role of risk in college
decision-making includes Altonji, Joseph G., “The Demand
for and Return to Education When Education Outcomes Are
Uncertain,” Journal of Labor Economics, January 1993, vol. 11,
no. 1, pp. 48–74; Elizabeth M. Caucutt and Krishna B. Kumar,
“Higher Education Subsidies and Heterogeneity: A Dynamic
Analysis,” Journal of Economic Dynamics and Control, 2003,
vol. 27, pp. 1459–1502; and Chatterjee, Satyajit, and Felicia
Ionescu, “Insuring Student Loans Against the Financial Risk
of Failing to Complete College,” Quantitative Economics,
November 2012, vol. 3, no. 3, pp. 393–420.

5

See Athreya and Eberly, p. 5.

6

For example, see Guvenen, Fatih, “Macroeconomics with
Heterogeneity: A Practical Guide,” Federal Reserve Bank of
Richmond Economic Quarterly, Third Quarter 2011, vol. 97,
no. 3, pp. 255–326.

7

See Goldin, Claudia, and Lawrence F. Katz, The Race Between
Education and Technology, Cambridge, Mass.: Harvard University Press, 2008.

8

For more on student debt burdens, see Lee, Donghoon,
“Household Debt and Credit: Student Debt,” Federal Reserve
Bank of New York, Presentation, February 28, 2013.

9

The feature of non-dischargeability should be associated with
lower interest rates on these loans since they are not collateralized. Also, income-based repayment programs reduce the
effect of non-dischargeability and provide more risk-sharing.

10

See Table 13 of Kane, Thomas J., “College-Going and Inequality: A Literature Review,” Russell Sage Foundation Working
Paper, June 30, 2001.

This article may be photocopied or reprinted in its
entirety. Please credit the authors, source, and the
Federal Reserve Bank of Richmond, and include the
italicized statement below.
Views expressed in this article are those of the authors
and not necessarily those of the Federal Reserve Bank
of Richmond or the Federal Reserve System.

FEDERAL RESERVE BANK
OF RICHMOND
Richmond Baltimore Charlotte

Page 4