On assessing the impact of federal student aid

On assessing the impact of federal student aid

of Education Economics 0272-7757188 $3.00+ 0.00 PergamonPressplc Review, Vol. I, No. 1, pp. II-84.1988. Printedin Great Britain. On Assessing the...

750KB Sizes 28 Downloads 69 Views

of Education

Economics

0272-7757188 $3.00+ 0.00 PergamonPressplc

Review, Vol. I, No. 1, pp. II-84.1988.

Printedin Great Britain.

On Assessing the Impact of Federal Student Aid* MICHAEL Economics

Department,

Fernald

House,

S. MCPHERSON Williams

College,

Williamstown,

MA 01267, U.S.A.

Abstract-Most empirical research on student aid has examined its effects on enrollment demand. This paper argues that a proper assessment of the net impact of federal student aid must include examination of its impact on institutional behavior as well. No systematic empirical work is presented, but selected data and examples are used to argue that the potential effects of aid on supply and on institutions are

significant. INTRODUCTION

made. The data may be time series of enrollments for an institution, a state, or a nation; they may be aggregate cross section data for a set of states; or they may reflect choices made by individual students as described in microdata. Many of these studies simply estimate the effect of price on enrollment, holding other things constant, and assume that increases in aid are equivalent to price reductions. Some of the studies measure aid effects separately. The most recent prominent study, and in many ways the most thorough and sophisticated, is that by Manski and Wise (1983). The other class of studies might best be described as “before and after” studies. The essence of these studies is to contrast enrollment levels or patterns before and after some major event - such as the introduction of Basic Economic Opportunity Grants in 1973 - and attribute the changes observed to the event. Much discussion of enrollment data implicitly utilizes this approach: we look at the trend or lack of trend in some enrollment measure and try to relate what we observe to major changes in the environment over the relevant time period. W.L. Hansen (1984) is one of the few analysts actually to use this “before and after” methodology explicitly to assess the effects of federal student aid programs. In a study published in 1984, Hansen compared the relative enrollment rates of lower and upper income students in 1972 and 1980, to judge the effects of the large expansion in federal student

THIS

PAPER is concerned with the problem of assessing the effects of student aid on enrollment behavior, and more especially with the problem of measuring the impact of the federal student aid effort on enrollments in the 1970s. Everyone knows that it is difficult to disentangle the influences of a single variable like federal student aid in a complex reality where many things change at once. For example, we usually think of student aid as simply reducing the net price facing students, and set about isolating that price reduction effect from all the other influences on student behavior. That is difficult enough. But federal aid may also have indirect effects on other aspects of the higher education system, for example, by influencing the prices schools charge or the amounts of aid they offer from their own resources. Any full-scale attempt to measure the net impact of aid on student behavior must also take these indirect effects into account.

ALTERNATIVE

APPROACHES

There are two main strands in the literature on measuring aid effects. The first consists of econometric studies of enrollment demand. Such studies draw on data that describe students’ enrollment choices and the environment - including both price and non-price factors - in which those choices are

*Research underlying this paper was partly supported through a grant to the Brookings Institution from the Mellon Foundation.

77

78

Economics of Education Review

aid in the 1970s on lower income enrollments. As is well known, he found little or no systematic change in the ratio of lower and upper income enrollments over that period, and used that finding to conclude that the aid programs had no apparent effect on equalizing opportunity. LIMITATIONS Neither of these approaches is fully adequate to assess the effects of federal aid programs on enrollment. Turning first to the “before and after” genre, it displays three very important limitations. First is the failure to control for other variables that may have changed over the period in question. In the case of student aid in the 1970s we know that a great deal besides federal aid changed over those years. To name only a few such developments: the war in Vietnam ended, the oil crisis came along, inflation went up and economic growth fell compared to the 1960s and markets for educated labor were flooded with unprecedented numbers of college graduates. Without controlling for these variables, it is difficult to identify with confidence the effects of student aid. Second (and related) is the issue of what benchmark should be used to compare the effects of programs. Hansen used the relative enrollment rates of families at different income levels, relying on the assumption that student aid was intended for, and delivered to, lower income families. But in fact federal aid has been delivered to students with widely varying incomes for various purposes. Moreover, the distribution of aid changed quite importantly during the 1970s especially following the Middle Income Student Assistance Act of 1978. It is not clear that we should have expected such diffused aid to have affected enrollments of lower income students more than that of other groups of students. Finally, ambiguity remains about what constituted the real changes in federal aid effort during the 1970s. During the 197Os, as the main “Title IV” higher education programs expanded (BEOG/Pell, GSL, and campus-based), grant aid to veterans and to surviving children of Social Security recipients fell in real terms. As a result, the peak year for realdollar grant aid in all federal student aid programs combined was 19751976. A picture of “little aid in 1972, lots in 1980” is oversimplified. The econometric approach to assessing effects of aid seems - and is - considerably cleaner. Indeed,

a main purpose of econometric work is to avoid pitfalls of “before and after” studies like those just described, by “controlling” statistically for the effects of other variables. It is a relatively simple matter to convert econometric estimates of demand responsiveness into estimates of the impact of federal aid programs on enrollment. Leslie and Brinkman (1985) conclude from an extensive literature survey that a consensus exists that a price cut of $100 (in 1982 dollars) nationally would raise enrollment by about 1.8%. If we assume that a price cut has the same magnitude of effect as an equivalent grant increase, we can estimate that by 1979 the Pell program should have boosted enrollments by roughly lo-15%, compared to what enrollments would have been in that year without the program.’ Manski and Wise (1983) simulate the effects of the 1979 Pell program in a more elaborate way, and get a somewhat larger estimated effect an enrollment increase of about 21%) almost all of it at 2 year colleges. There is, however, some reason to think that they have overestimated the effect at 2 year colleges and underestimated the effect at 4 year colleges. Allowing for these possible biases in their results, they are not so different from the consensus.2 Similar exercises could be used to estimate the effects of other federal aid programs on enrollment. For example, expanded eligibility for guaranteed student loans in the 1970s allowed an extra 18% of all students to obtain GSLs. If we suppose that a loan’s impact on enrollment is equivalent to the impact of a grant equal to the subsidy value of the loan, we can estimate that this expansion should have led to roughly a 4 or 5% increase in enrollment over the decade. Calculations of this kind can be subjected to various empirically based criticisms. Those above, for example, neglect the possibility that lower income students may be more responsive to price than higher income students. There is another, often little noticed, conceptual limitation to such exercises. It centers on the question of what should be held constant in assessing the effects of federal aid on enrollment. The above analyses assume that as federal aid changes, the only effect is a subtraction in the net price facing students. The implicit assumption is that other aspects of state and institutional behavior, such as pricing, levels of state budgetary support, offers of institution-based aid, and so on, are unaffected by federal policy change. It is this assumption that

79

Federal Student Aid

allows studies to only governmental or to be responsive

demand -

focus on econometric analyses of family and student behavior, not institutional behavior, is assumed to federal policy.3

INDIRECT EFFECTS OF FEDERAL AID For some purposes the artificial assumption that institutions and states are passive in response to federal policies may be useful. But for purposes of assessing the net effects of federal initiatives on enrollment outcomes, it is important to try to take these indirect effects into account, to see if they matter. Figure 1, which might be called my “Rube Goldberg” diagram, is a means of beginning to sort out these indirect effects - or at any rate of recognizing their complexity. The direct link from federal aid to net price reductions, emphasized in the econometric studies, is present, running through the center of the diagram. But a great number of other potential connections are present as well. The arrows at the top and bottom of the diagram call attention to possible responses by both public and private institutions to federal aid policy,

responses that may affect students through changes in admissions policies, tuition levels, or the college’s own aid offers. (A further complexity, ignored here, is the proprietary sector, many of whose institutions might not exist without federal aid.) Other connections are largely self-explanatory. Some will be elaborated below. It is useful to note that both “before and after” and “econometric demand” studies can be characterized by the features of this diagram that they omit. The econometric studies essentially mask all the linkages between federal aid and student decisions except the one running directly through net prices, indicated by the cross-hatching. They also take account of socioeconomic variables that may affect student outcomes. Before-and-after studies, on the other hand, compress all the channels connecting aid and student decisions together, treating the whole complex of state, institution, and student decisions as a black box. Indirect effects are, in a sense, captured, but with no means of assessing their relative importance. Importantly, before-and-after studies do not control for the effects of socioeconomic variables. Any effects from changes in these vari-

Student decisions Admission policy

Figure 1

I

80

Economics of Education Review

Federa student aid

I

Socio-economic

Figure 2

Federal.

Figure 3 ables are simply mixed in with the direct and indirect effects of changes in aid. Figure 2 shows that in such a study all the connections are masked. A systematic assessment of the significance of these indirect effects will require considerable basic research. What is required, and not easily obtained, is a good understanding of the behavior of the institutions that are influential in the student aid system. These institutions include governments that supply aid and schools, both public and private, whose policies are influenced by the kinds of federal student aid policy that exist. What we need, in a nutshell, is a theory of the influence of student aid on the supply of higher education which will complement our relatively well developed theory of the influence of student aid on higher education demand. Such a theory must build on theories of governmental behavior and of the behavior of non-profit institutions. Obviously some theoretical work of this kind exists, in the theory of public choice and in writings about the economics of non-profit institutions, but it is in need of further development for this application. Empirical study of linkages between student aid and institutional behavior is likewise called for. One useful kind of study would be a comparison of states. States differ significantly in the level and

form of student aid they offer to those who attend private institutions. What impact do such differences have, if any, on the pricing, admissions, and other decisions of private colleges in different jurisdictions? There may similarly be something to be learned from analysis of time series of data on the behavior of private institutions, over periods during which state or federal aid policies changed significantly. Empirical studies that shed light on how institutions allocate their own resources for student aid might also provide a basis for predicting the effects on their behavior of changes in federal policy. EXAMPLES

OF EFFECTS

Little systematic work of this kind has been undertaken to date in the field of higher education, and I have no systematic findings to report here. There may, however, be value in reviewing some of the more interesting possible connections and reporting on some selected data that suggest the role federal student aid may play in institutional decisions. 1. Other Federal Spending on Higher Education Does more federal spending on student aid induce less federal spending on other federal programs that

81

Federal Student Aid

benefit colleges and universities? We do know that growth in other such spending was curtailed in the 1970s as the major student aid programs expanded. This curtailment holds both for so-called “specially directed” aid to students - GI Bill and Social Security benefits - and for federal spending on basic research at colleges and universities. Rapid growth in federal spending on academic research and development actually stopped earlier, at the end of the 196Os, some years before the major funding of federal student aid, so the causal connection is unclear. The question is whether expansion in R&D funding might have resumed had resources not been going into student aid. 2. State Spending Do states cut back on their support for higher education as federal student aid expands? There is some fairly weak evidence of this. In constant dollar terms, state appropriations to public colleges and universities rose moderately from 1965 to 1973 and were then roughly constant from 1973 through 1979. This, however, roughly followed the pattern of change in real costs of public higher education. Moreover, the share of state support for public higher education did not change much. State programs of scholarship aid grew rapidly in real terms in the 1960s and continued to grow relative to inflation through the late 1970s. (The federal program of State Student Incentive Grants was specifically designed in the 1970s to encourage such growth). 3. Prices Perhaps the most obvious link to expect is between federal student aid and the tuitions and other charges at colleges and universities. By analogy to medical cost inflation, many observers have suggested that “third party payments” of higher education costs tend to induce rapid price increases. This worry in fact received considerable discussion during Congressional debate over the student aid programs in 1972.4 The Secretary of Education has also recently raised this concern on several occasions. It is thus quite interesting to discover that there is no evidence in the aggregate data of any such relationship - in fact, if anything, the numbers point the other way. As Table 1 shows, the real rate of increase in student charges (tuition, fees, and room and board) was around 2% per year from 1963 to 1970 and was actually negative from 1970 to 1980,

Table 1. attendance

Year

Annual real rates of increase in costs of at selected categories of colleges and universities, 1963-1983 Private university W)

1963-1970 1970-1980 1980-1983 Source: Gillespie Quincy, 1984.

Public 4-year college

2.2 -0.5 6.3 and Carlson,

Community college

WI

WI

1.5 -0.8 4.6

3.2 -0.9 2.6

1983 and Gillespie

and

failed to keep pace with inflation. Since 1980, prices have gone up quite rapidly. This pattern is actually inverse to movements in the constantdollar amounts of aid available in the major programs. It would obviously be imprudent to claim an inverse causal link between aid and prices, but this evidence does cast doubt on the hypothesis of a direct link. And in fact a closer look at the structure of the Federal aid programs does not suggest that they contain strong incentives to raise prices. In traditional third party medical care payment schemes, the provider can receive more revenue from the third party by raising prices (either directly, or sometimes with a lag through formulas that determine reasonable charges). But this is rare in the federal aid programs. Although the Pell grant formula is in principle sensitive to institutional costs, in fact the operative award ceiling for most Pell recipients is determined by family size and income rather than by institutional costs. There is a link between an institution’s costs and the amount of money it receives through the campus-based programs, but that link is highly indirect and would only operate strongly if appropriations for those programs were rising rapidly, which has not been the case for many years. Legislation in place since 1982 linking the maximum GSL that high income recipients can receive to their cost of education does provide some incentive for private colleges to raise prices, and this could conceivably have played some role in recent price increases. as prices

4. Institution-based Student Aid The largest and most interesting changes in institutional behavior have been in the use of their own resources to offer aid to students - either through price discounting or direct expenditure.

Economics of Education Review

82

There is evidence that the period prior to the 1972 education amendments was one of rapidly increasing institutional expenditure on student aid, while the period from 1973 to 1980 was one of noticeable decline. Aggregate data are not yet available for the most recent years, but anecdotal evidence suggests that (at least in private higher education) institutional expenditure on student aid has been rising rapidly again in the 1980s as the Federal commitment has declined. Table 2 provides data on institution-based aid per student in public and private higher education over the period in question.5 Inflation-adjusted aid per student grew at 5.4% in public institutions and at 5.7% at private institutions from 1966 to 1973. From 1973 to 1980, it fell at 7% per year in public

institutions and at 3% per year in private institutions. Total real spending on institution-based aid was lower in both sectors in 1980 than in 1973. These are large changes. Had real per student spending on institution-based aid at private colleges continued to grow from 1973 to 1980 as it had earlier, spending per student would have been higher by $344 (1980 dollars) in 1980 than it in fact was. This compares to Pell grants per student in private higher education of $544 in 1980.6 Again, many factors other than student aid policy were at work over this period, but the pattern present in these numbers is highly suggestive. The most straightforward interpretation is that institutions cut back on their own aid efforts as the

Table 2. Institution-based

Year

All

1966-1967 1969-1970 1972-1973 1973-1974 1974-1975 1977-1978 1978- 1979 1980-1981 1981-1982 1982-1983

429 701 1067 1133 1196 1526 1600 1831 2138 2333

Current dollars Private Public 156

273

537 581 718 717 779

596 623 825 909 1079

federal government expands its aid. That interpretation, however, raises as many questions as it answers, for example: Could institutions have found the resources to fund expanded levels of aid in the 1970s had the federal government not intervened as it did? The response by institutions might instead have been to cut back on their commitment to need-based aid, with the resources instead used for other purposes - perhaps merit aid or perhaps activities unrelated to student finance. If institutions had continued to expand their aid budgets during the 1970s at the rate they did in the 1960s who would have borne the costs of greater aid? The most obvious possibility is that other students and their families would have done so in the form of more rapid increases in tuition. There may in this sense be a link between the historical pattern of aid expenditures and of tuition, at least for private institutions. When federal aid resources are rising, tuition increases may slow, because the cost of expanded aid is picked up by the national government. But when they are low or falling - as in the late 1960s (low) and in the 1980s (falling) tuitions may have to rise to finance increased aid from institutional revenues. Would institutions have directed aid to the same set of students the federal government did? Even if institutions continued to aim their dollars at needy students, they might have distributed them differently than federal programs did. One possibility is

scholarship aid All 442 638 852 851 810 841 819 742 785 806

selected years, 1966-1982*

1967 dollars Public Private 160

281

403 393 396 367 316

448 422 455 465 437

Per student7 1967 dollars Public Private All 69 80 92 89 79 75 73 61 63 65

37

138

54 49 45 42 33

205 189 187 188 166

*Total institutional scholarship expenditures less supplemental educational opportunity grant revenues, as reported by institutions in Higher Education General Information Survey and by the Federal Government’s Office of Student Financial aid. tTota1 scholarship aid divided by total enrollment (including graduate and part-time). Source: National Center for Education Statistics, Financial Statistics of Colleges and Universities, various years; Digest of Educational Statistics, 1983-1984, and Office of Student Financial Aid, OSFA Data Bank, various years.

Federal Student Aid that institutions competing for students might have diverted their aid budgets toward “low need” middle class students, for whom relatively small awards might make a real difference in choice of school. This would have restricted opportunities for needier students, relative to the outcome with federal intervention. Questions like these are important for determining the net import and hence effectiveness of the federal aid effort and obviously require further thought and research. To suggest directions such analysis might go, I will conclude with a highly speculative interpretation of developments in the 1970s that suggests the policy relevance of the issues introduced here.

A SPECULATIVE

INTERPRETATION

During the latter part of the 1960s both public and private institutions undertook major efforts to provide expanded educational opportunity for needy and disadvantaged students. In the private sector, these efforts resulted in the expansion of need-based aid, already noted, and in programs for recruitment of minority and disadvantaged students. A similar expansion of need-based aid occurred in public higher education, but the development of community colleges and other open access institutions, often in urban areas, was probably more important in stimulating enrollment of lower income students in public institutions in the late 1960s. Although data on student income distributions for these years are weak, there is some evidence that these varied efforts had a noticeable effect. Davis and Johns’s (1982) analysis of data from ACE survey of freshmen suggests that the percentages of freshmen from families below the U.S. median income jumped between 1966 and 1971 by 35% at public 2 year colleges (from 40 to 54%), by roughly 30% at various categories of private 4 year colleges, and by somewhat lesser amounts at

public and private universities. Since then, these percentages have been roughly constant.’ Perhaps colleges and universities overreached themselves in making these efforts in the late 1960s. In response to a climate of opinion in the 1960s that greatly valued efforts toward more equality, they may have undertaken efforts they could not - or anyway would not - sustain at growing cost in the longer run. Without federal support, private colleges might have backed off from their commitments to recruit and help finance the education of and state governments disadvantaged students, might have backed off from their commitments to support open-access institutions. If so, federal intervention may have had the effect of sustaining the gains made in the late 1960s in the face of financial pressures that would otherwise have eroded them. The fact that low income and minority enrollments have declined in the 1980s as federal support has been cut back, provides circumstantial evidence in favor of this interpretation. This conjectural story is consistent with Lee Hansen’s “before and after” data for the 197Os, but it would lend those data a very different normative significance. For it would suggest that the federal programs, rather than failing because they did not raise enrollments instead succeeded by preventing enrollments of lower income students from falling. But this is just a conjecture, and obviously needs much more investigation. The real point of the story, and of this paper, is that we cannot properly understand the effects of the federal student aid programs unless we take account of the effects of those programs on the institutional structures through which they operate. Acknowledgements

- This paper was prepared while the author was Senior Fellow in Economic Studies at the Brookings Institution. An earlier version appeared in the Proceedings of the Second Annual NASSGP/NCHELP Research Conference, May 31-June 1, 1985 (Illinois State Scholarship Commission, 1985). Useful comments were provided by Henry Aaron, Rob Meyer, Stephen Hoenack, and an anonymous referee.

NOTES 1. This assumes an average

83

Pell award of about $1000 in 1979 dollars and that about half of all freshmen would have been eligible for Pell awards under 1979 rules. Compare Manski and Wise (1983). 2. The large effect on two year enrollment results from a mysterious finding that 2 year enrollments are much more sensitive to aid changes than to equivalent price changes - a result the authors are puzzled by and which is not found in other studies or at other categories of schools in their study. If the aid effect is set equal to the price effect, the effect of the Pell program on total enrollment drops to 6%. However, in estimating the effect of the Pell program on 4 year college enrollment, the authors

Economics of Education Review

3.

4.

5.

6. 7.

assume that the availability of Pells will not induce any increase in applications to 4 year colleges; the only effect they allow for is increased enrollment among students who (without Pells) apply to and are admitted at a least one four year school but do not enroll. Since, in their sample, about 65% of the eligible population never are accepted for admission at a four year college, this assumption sharply limits the possible effect they could have found. Manski and Wise (1983) take account of one possible indirect effect of federal aid by performing a set of calculations that assumes institutions reduce their aid awards by an amount equal to the Pell award made (but never setting the institution’s aid award below zero). However, because most institutionbased aid is offered by four year institutions, and because they find essentially no responsiveness of enrollment to price at four year institutions (partly for reasons discussed in note 2 above), this adjustment has no effect on their results. Spokespersons for public colleges and universities argued for direct institutional aid in preference to student aid partly on grounds that the latter would encourage legislatures to raise public tuitions and turn more costs back-& students (see Gladieux and Wola&, 1976). These numbers are derived from the HEGIS survev of institutions conducted bv the federal government. They are not ideal, since there is apparently some variation in how institutions interpret the survey instructions, and a small amount of federal fellowship money is probably included in the estimate of institution-based aid. It would also be desirable to have numbers on aid per recipient, but that is not available. I have checked these numbers against independent numbers developed by the Carnegie Council up to the mid-1970s, and they are in rough agreement. See Carnegie Council for Policy Studies in Higher Education, (1979). All per student figures are per total enrollment, including graduate and part time. Undergraduate full-time-equivalent enrollment data are not available for some years covered in the data. The Carnegie Council (1979) reports a similar trend based on Census data. REFERENCES

Next Steps for the 1980s in Student Financial Aid. Washington, DC: Jossey Bass Publishers. DAVIS, J.S. and JOHNS, K. (1982) Low family income: a continuing barrier to college enrollment? J. Student Financial Aid 12 (February). GILLESPIE. D.A. and CARLSON, N. (1983) ~ , Trends in Student Aid: 1963 to 1983. Washington, DC: The College’Board. GILLESPIE, D.A. and QUINCY, L. (1984) Trends in Student Aid: 1980 to 1984. Washington, DC: The College Board. GLADIEUX, L.E. and WOLANIN, T.R. (1976) Congress and the Colleges: The National Politics of Higher Education. Lexington: D.C. Heath and Company. HANSEN, W.L. (1984) Economic growth and equal opportunity: conflicting or complementary goals in higher education? In Education and Economic Productivity (Edited by DEAN, E.). Cambridge: Ballinger Publishing Company, Inc. LESLIE, L. and BRINKMAN, P. (1987) Student price response in higher education: The Student Demand Studies. .I. Higher Educ. 58, 181-204. MANSKI, C. and WISE, D. (1983) College Choice in America. Cambridge: Harvard University Press. CARNEGIE COUNCIL FOR POLICY STUDIES IN HIGHER EDUCATION (1979)