The
dataset is U.S Department of Education’s The Integrated Postsecondary Education
Data System (IPEDS) analytics from 1987 to 2010.[1] It
is a panel data with a total of 202800 observations on 932 variables derived
from institutional characteristics, finance expenditure, enrollment,
completions, graduation rates, student financial aid, human resources IPEDS
survey components and a limited number of outside sources.
Expenditure variables are derived to
present the function of expenditure in the broader context of different
institutional purposes. Instruction, student services, and the associated share
of overhead costs are grouped into education and related expenses; research and
the associated share of overhead costs are grouped into research and related
costs; and public services and the associated share of overhead costs are
grouped into public service and related costs. These three categories along
with net scholarships and fellowships combine to be education and general
spending. The expenditures that are largely self-supporting, including
independent operations, auxiliary enterprises, and hospitals are aggregated
into a separate category. Variables have also been derived to put expenditures
into the context of completions to show an estimate of what an institution
spends for each degree or completion in a given year. In addition, variables to
adjust the financial information to 2010 constant dollars have been included
for Consumer Price Index (CPI) and Higher Education Price Index (HEPI) scalars.
A word of caution about IPEDS data set is
necessary. From 1990 to 1996 the financial accounting standard for all
institutions was Government Accounting Standards Board (GASB). Starting from
1997, however, private institutions also had to report using Financial
Accounting Standards Board (FASB). To be consistent, this paper uses data reported
under GASB, but some technical differences between the two accounting standards
might lead to different results. In addition, expenditure on depreciation is a
new concept for higher education institutions since the FASB change. Starting
from 1998, private institutions began to report their spending on depreciation,
while for public institutions, it was not until 2002 that a considerable amount
of them did the report.
For the 1626 institutions in my sample,
real total expenditure per full time equivalent (FTE) student increased about
2.6 percent annually from 1990 to 2010. Over the same period, the CPI increased
at an annual rate of 2.7 percent, while the HEPI increased by 3.5 percent. The
implication is that only 0.8 percentage point of the 2.6 percent per year
increase in real expenditure per FTE student can be accounted for by the
inflation in inputs used intensively by colleges and universities. Thus,
expenditure increases are more due to institutions’ own decisions.[2]
The previous paragraph shows that CPI and
HEPI scalars have different deflation effects. In the rest of the paper, our
expenditures are in 2009-2010 dollars, deflated by the CPI rather than the HEPI
scalar, which was designed to deflate higher education related expenditure. The
reason I didn’t use this scalar is because it might overstate or understate the
inflationary effect. The prices of certain inputs with large weights
in the HEPI, such as faculty salaries and fellowships of graduate teaching
assistants, are partially determined by decisions of institutions themselves.
If these institutions compete fiercely for capable professors, it will bid up
price and thus HEPI. From this perspective, the HEPI doesn’t represent well as
the uncontrollable increase in cost of higher education. [3]
A summary of data shows that all four types
of institutions experienced an increase in charging and spending over years. For
public research universities, net tuition per full time equivalent (FTE)
student in 2010 dollars increased 3.98 percent annually. In the meantime, average
yearly increase of community colleges is 3.49 percent, while 1.44 percent and
1.49 percent for private research universities and private bachelor
institutions, respectively.[4]
From this rough calculation, some people might
argue that private institutions are not the ones that experienced the most
drastic change of tuition charging. However, percentage change of net tuition
for private institutions can be small for two reasons. Firstly, private
institutions on average have a higher level of tuition than public peers, and
thus a same amount of tuition increase yields a smaller percentage change. In
addition, net tuition by definition is the amount of money institution takes in
from student after institutional grant is provided (note that this doesn’t
include external discounts and allowances), and it implies that private
institutions might use scholarships and fellowships to attract students and
lower their costs for attending.
Data suggest that both reasons hold. Table
2 tabulates FTE real gross tuition of four types of institutions, and it is
clear that private institutions charge tuitions that are many multiples of tuitions
in the public sector, and in the aggregate they continue to thrive. Table 3
illustrates institutions’ trends of grant expenditure, and it is obvious that
for private institutions, there is an increasing importance of scholarships
spending. Private institutions have a steady increase in scholarship
expenditure and in 2010 they spent more than 10000 dollars recruiting each
student. On the other hand, public institutions’ trends of grant expenditure
are rather flat and even decreased since 2000. Given that private institutions
provide more scholarships on average, they can attract students with better
academic performance and motivation. Since a good student body provides
positive peer effect on learning (Winston & Zimmermann, 2004), being able
to spend more on grants tends to motivate private institutions’ student
selectivity in admission and further enhance their quality. Table 5 summarizes
average score and standard deviation of SAT math 75 percentile for each type of
institutions from 2002 to 2010, and it clearly shows that private research
institutions recruit students with the best performance. One thing interesting
is that average scores of private bachelor institutions are lower than public
research institutions’, but we should notice that there is a big standard
deviation in terms of score distribution. In my preliminary summary of grant
expenditure, private bachelor institutions sector also has the largest standard
deviation, which implies that among these colleges, difference of scholarship
spending lead to different student bodies.
As for total expenditure, the same story
holds: all four types of institutions experienced continuous increase over
years (see Table 6): private research universities have a 10 percent increase,
with public research 30 percent, public associate 16 percent, and private
bachelor 11 percent increase. A decomposition and comparison of expenditure by
functions can tell us what functions have had priority among college executive
boards over the last two decades. We can also draw inferences about quality
change by tracking their spending records on instruction and research.
The functional categories used by IPEDS
encompass broad areas of the American higher education enterprise.
One category of great interest is
education related expenditure (Table 7), which consist of spending on instruction,
student services, and the education share of spending on central academic and
administrative support, and operations and maintenance.[5] Although the increasing role of administrative
spending has been criticized as exacerbating cost problems and pulling
universities and colleges away from their core mission of educating students
and expanding knowledge,[6] it
is too radical to conclude that administrative investment has absolutely no
impact on quality improvement.[7] For
private research universities, though education related expenditure soared over
37000 dollars in 2000 and continues to thrive since then, in the meantime its
share of total expenditure shrank from 57 percent to 46 percent in 2010 (Table
8). This suggests that private research institutions are spending a lot more in
other fields in addition to this field. For public research institutions, their investment
in this area seems to keep the pace with their total expenditure over years,
reflected by a rather flat slope. As for private bachelor institutions, share
of education-related expenses keeps increasing over years and consisted of
roughly 75 percent of total expenditure in 2010. This is not a surprising
percentage since private bachelor institutions don’t have to spend on
facilities oriented for research and graduate school training, and they can
concentrate on undergraduate instruction. Community colleges have a relatively
flat education related spending trend, but the share peaked at 82 percent in
2002 and dropped since then, implying that they may also be investing money in
other fields.
Now we turn to have a look at schools’
pattern of instruction expenditure. For community colleges, their real FTE
spending on instruction decreased some years between 2000 and 2010, but at the
same time their net tuition and total expenditure keep increasing. So over
years community colleges somewhat cut their investment in instruction while
keep students paying more for their maintenance, indicating that they may have
some revenue pressure. Data on government grants support this claim. From 2000
to 2010, federal, state and local support plummeted. Since private endowment
has never been a major source of revenue for community colleges, they face a
lot of revenue pressure if government constrained their investment.
As for the other three sectors of
institutions, they don’t have such worry. Public research institutions’ total
expenditure and net tuition keep increasing, while instruction and research
expenditures are nearly constant over years. Government support is also flat
during the time period, and thus it shows that public research institutions
have been investing money in other areas expect for instruction and research. For
private research institutions, their spending is increasing in nearly every
aspect, ranging from instruction and researching to operation and maintenance. For
private bachelor institutions, government support is decreasing over years, but
instruction expenditure increases somehow.
A brief examination of expenditure trend
shows that postsecondary institutions all experienced an increase of total
expenditure and net tuition. For public associate institutions, expenditure on
instruction decreases somehow and it is because the shrinkage of government
funding puts financial constrains upon their budget. Research institutions don’t
have such worries on revenue because their endowment and investment income are
robust overall expect year 2009. As a result, they increased expenditure in literally
every aspect. For private bachelor institutions, the main increase lies in
grant expenditure. From this trend analysis, it seems public associate
institutions don’t have much input in terms of producing better students while
other types of schools might experience a higher increase of quality over
years. To further examine the contributing effect of quality related
expenditure upon total expenditure, I will use regression method, namely the
fix effect model.
Table5. SAT Math 75
Percentile Score
Academic year
|
Public research
|
Private research
|
Private bachelor
|
Community colleges
|
2002
|
614 (53)
|
664 (74)
|
601 (83)
|
494 (24)
|
2003
|
616 (52)
|
667 (74)
|
599 (80)
|
499 (26)
|
2004
|
619 (52)
|
670 (71)
|
600 (82)
|
509 (32)
|
2005
|
620 (53)
|
670 (75)
|
598 (82)
|
515 (40)
|
2006
|
621 (52)
|
676 (72)
|
598 (83)
|
518 (36)
|
2007
|
624 (53)
|
673 (77)
|
597 (82)
|
530 (43)
|
2008
|
622 (52)
|
669 (80)
|
592 (83)
|
536 (47)
|
2009
|
625 (53)
|
671 (82)
|
593 (85)
|
534 (48)
|
2010
|
625 (55)
|
677 (81)
|
590 (85)
|
526 (52)
|
[1] IPEDS dataset is maintained and updated by The National Center for Education Statistics (NCES) and is downloadable from http://nces.ed.gov/ipeds/deltacostproject/.
[2]
The FTE count is included in IPEDS dataset. The formula is used by the U.S.
Department of Education to produce the full-time equivalent enrollment data
published annually in the Digest of Education Statistics. I did the percentage
calculation based on raw data.
[3]
Getz and
Siegfried (1991) used gross national product (GNP) implicit price deflator
instead. They claimed that the GNP deflator provided a standard economy-wide
production index to assess change in the costs of higher education, while the
CPI is more often used to evaluate tuition inflation since it reflects price of
goods and services that people would have instead of a college education. Since
CPI scalar is already included in the dataset, I used it as deflator for
convenience. However, results might differ if GNP deflator is used instead.
[4]
Data summary tables, figures and regression results are in Appendix.
[5]
Percentage share was done by previous researchers and stored in IPEDS data set.
[6]
American Association of University Professors (AAUP) publishes annual faculty
compensation surveys and one of its main arguments is that massively
disproportionate growth in the number of administrative employees, coupled with
the continuing shift to an increasingly precarious corps of mostly temporary,
underpaid, and insufficiently supported instructors, represents a real threat
to the quality of academic programs. For the latest survey, refer to http://www.aaup.org/our-work/research/annual-report-economic-status-profession. However, we have to think carefully
before accepting AAUP’s argument because it has a vested interest in finding
that too little money is going to faculty and too much is to other areas like
administration.
[7]
American
higher education has been attracting more applications from different
countries, and numerous sources of revenues entails more reporting, so hiring
more administrative staff is necessary and help faculty members dedicate to
teaching and researching to some extent. This is one example that
administrative spending can help enhance quality of higher education.
No comments:
Post a Comment