As we start the new century, the nation's first and oldest higher education system faces a set of conditions that resemble those of the first half of the twentieth century more than those associated with the periods ushered in by the GI Bill following World War II or the Higher Education Act of 1965. Government grant assistance for students represents a minimal percentage of revenue for most private institutions, which are again dependent on the ability of families to pay or finance tuition. Unlike the first half of the century, however, a much smaller percentage of families are able to pay fully today's tuition rates.

Prior to these landmarks of public policy for higher education, private colleges and universities charged the tuition necessary to generate the revenue and enrollment required to operate their academic programs in general, these tuition rates were affordable for the growing middle-income population of the country.

Public institutions had not yet begun their massive postwar enrollment expansion; their similarity to private liberal arts colleges was reflected in frequent athletic contests in which state schools and private liberal arts colleges were athletic peers. Indeed, most small, private liberal arts colleges can reflect on victories as recently as the 1940s over state schools that are National Collegiate Athletic Association Division I athletic powers today.

The limited scholarship aid that was available to private college students at that time was financed largely by private donors and was administered without the benefit of a formal needs analysis. Scholarships largely went to academically worthy students who could not otherwise afford to attend. Private colleges and universities of the day were largely cash businesses; the notion of a generally available financial aid program, much less the idea of aid as a discount to tuition, is a modern concept.

Since World War II, increased participation in postsecondary education generally has been fueled by demographic factors, such as the earnings premium associated with educational attainment, and by higher education finance policy. Census figures for 1995 indicate that average income for a household headed by a baccalaureate degree holder is $73,334, compared to $43,182 for a household headed by a someone with a high school diploma.
Federal and state need-based grant aid and loans, the extension of the GI Bill, and increased tuition subsidies in the public sector also have contributed to the high level of higher education participation we see today.

This combination of public policies created a new set of economic realities for private colleges and universities. With increased federal grants and loans available to families, as well as state-funded financial aid that could be used at private colleges in some cases, private sector institutions expanded and diversified their student populations. Although still largely dependent on tuition, these institutions and their students could count on a significant share of that tuition to be financed by government sources.

Increased participation by middle- and lower-income families using the help of federal and state financial aid led to increased enrollments and revenues. Freed from total reliance on family tuition payments, and with government as a financing partner, independent colleges could raise tuition and strengthen academic programs without reducing demand. In raising tuition away from what the majority of the population could afford to pay out of current income and toward a level that only very affluent families could afford without assistance, private institutions engaged in a progressive or means-tested pricing strategy in which the contribution families made toward the cost of attendance was determined by their financial resources.

Increasingly, private colleges and universities added institutional grant dollars to government aid to broaden and strengthen their applicant pools, increase enrollment, and become more socially and economically diverse. Indeed, through the 1970s, it was possible for the professional associations serving admissions and financial aid officers- the National Association of College Admissions Counselors and the National Association of Student Financial Aid Administrators - to establish a culture and prescribe professional practices that support the concept of need-blind admission coupled with a commitment to meet the full financial need of admitted students. Most colleges and universities embraced these norms.

The revenue growth that resulted from increased enrollment and higher tuition permitted growth in faculty and their salaries. It also increased academic innovation and investment in academic and co-curricular programs, including new initiatives, such as women's or black studies; new student support efforts, such as career development and various counseling services; and new women's intercollegiate athletic programs as mandated by Title 9.

Until the early 1980s, government aid increased and family incomes grew along with tuition. As a result, private colleges and universities in general did not suffer badly in their competition against the highly subsidized public sector, though their market share was deteriorating, and middle- and upper-income families were moving gradually toward the public sector, particularly public flagship universities. Indeed, even as greater numbers of students with demonstrated need enrolled in private colleges, the conventional wisdom on these campuses - and throughout higher education - was that public colleges and universities provided access for the masses through low tuition, whereas private colleges continued to serve a largely affluent population.

In this environment, forecasting revenue and enrollment was a relatively straightforward exercise in professional judgment. And although it would be misleading to suggest that colleges experienced no variation in application volume or yield, higher education was essentially a seller's market, rather than the buyer's market we know today, and there existed substantial market stability. Without a substantial public investment, variations in enrollment and revenue were linked more to economic conditions than to public funding and policy. Most colleges operated at a known capacity, had a historical sense of their retention rates and new student enrollments, and could project revenues and enrollments for budgeting and planning purposes in an unscientific but reliable manner. Revenue was simply gross tuition plus auxiliary income. The financial aid budget was established and predictable, part of the overall Education and General Expense budget. Attrition would occasionally depart from historical ranges, but in the main, the greatest difficulty lay in predicting residence hall utilization, a problem mainly created by liberalized college policies permitting greater numbers of students to live off campus.

Today, everything has changed. Private colleges and universities are forced to become acquainted with the new realities that challenge their pricing and finance structures. Among these are stagnating family income, minimal family savings, declining government support for grant programs, tuition levels that represent roughly twice the proportion of median household income as a decade ago, and a widening price gap with public sector counterparts. A brief exposition of these trends is useful to understand the ways in which private colleges and universities have chosen to predict and manage enrollments and revenues.

Since the late 1980s, family income for most population segments has stagnated and in some cases fallen in real terms. For private colleges and universities serving traditional age students, the economic anxiety affecting parents ages forty-five to fifty-four is apparent in the increased need demonstrated by most students, increased parental efforts to negotiate financial aid packages, the migration of students from affluent families to the public sector, and application trends favoring very selective institutions whose imprimatur is perceived as the greatest guarantor of future prosperity. Statewide studies performed since 1992 in Minnesota, Oregon, and Florida have documented not only this migration of affluent families to the public sector but also the minimal level of effort families of all income brackets make to save for college (Minnesota Private College Research Foundation, 1992; Florida Postsecondary Education Planning Commission, 1994; Oregon State System of Higher Education, 1995).

One of the common assumptions behind most higher education finance policy is that families bear the first responsibility to pay for college. Although poor families cannot save for college, these studies indicate that from one-third to one-half of the most affluent families do not save for college.

As federal and state grant aid have receded relative to inflation, replaced by increased federal loan eligibility, the dominant share of grant aid provided in private college financial aid packages comes from institutional sources, a trend documented by National Association of College and University Business Officers reports and many statewide and consortium studies. Indeed, at most private colleges, institutional grant aid exceeds state and federal sources combined.

At the federal level, this trend is accelerated by the federalization of the needs analysis. This process removed the needs analysis function from its historical, independent, nonprofit providers - the College Scholarship Service (CSS), a unit of the College Board, and American College Testing - and eliminated the student fee.

More important, the government liberalized the needs analysis formula itself, eliminating consideration of home equity, for example, and reducing the traditional needs analysis to a rationing device for federal aid. By defining more families as having need and as having greater need, the net result of federal policy has been to create a burgeoning unfunded mandate to meet student need. This situation is particularly vexing for private colleges and universities for several reasons, as most of their students demonstrated need under the old formula.

Some of these colleges continue to use the older, more rigorous methodology, which demands an additional effort from students. But their public competitors use federal methodology, which results in smaller family contributions, placing private institutions at a further disadvantage to their public sector competitors.
Moreover, many independent colleges historically have met the full need of students without "gapping" students, that is, failing to provide enough grant aid in the financial aid package to avoid surpassing- the maximum Stafford loan. Increasingly, these colleges have been forced to slow, stop, or cannibalize academic investment to increase the financial aid budget to maintain enrollment.

Beyond shifting the burden for funding grant aid to students from government sources to the tuition and voluntary gift resources of the institution, these recent changes in the federal needs analysis, and resulting adjustments in college aid policies, have further complicated family- financing of college, making the prediction of enrollment and revenue even harder. The three statewide studies have shown that for all but a small number of families, the federal needs analysis neither predicts nor defines family contribution.

As a result, the entire financial aid application process, an onerous endeavor to begin with, lacks public credibility. This credibility is further eroded as some colleges negotiate financial aid packages outside of the needs analysis or implement financial aid strategies that lack an explicit and intelligible philosophical foundation.

The main result of this combination of economic and policy circumstances has been to drive the price gap between public and private institutions to historical levels, both in published tuition rates and in the cost of college net of financial aid. In response to this price gap, increasing numbers and percentages of the most affluent families are- enrolling their children in public sector institutions, especially flagship -universities, taking advantages of state government subsidies that offset 50 to 80 percent of the cost of instruction. This trend is documented in national data and most demonstrably in each of the existing statewide family finance studies.

Beyond the public policy concerns about the equity and efficiency of higher-income families taking advantage of taxpayer tuition subsidies at public colleges, this oversupplied and increasingly price-competitive market means that private colleges and universities no longer can operate as relatively simple cash businesses. Though most college communities are loath to link their academic missions and public purposes to business comparisons, it simply is not useful to deny that higher education, like other goods and services in the economy, must "meet the market" and prove its value to those who pay the tab. In this way, college enrollment and revenue managers have more in common with marketing and inventory managers in the automobile, airline, and health care industries than they might care to admit. It is no surprise, in this context, to see that financial aid has become a price-discounting mechanism. Although colleges are now managing price strategically through financial aid, what they are unable to do with any flexibility is to manage capacity in the ways other industries do.

What are the specific results of these historical trends? Price again has become a barrier to a wide range of students attending private colleges. Because government grant aid is generally available only to the poorest private college students, the typical private college now enrolls a bifurcated student population-the very poor and the wealthy-with middle-income students taking up residence in the public sector.

Many private colleges are now locked in a vicious cycle of tuition discounting. They must provide more and more institutional grant aid in an attempt to maintain market share and the enrollment of middle-income families. To generate this aid, many colleges have pursued high-tuition strategies designed to generate the revenue necessary to support their programs, as they provide increased discounts in the form of grant aid. As higher tuition drives up family need, they must increase grant aid or lose enrollment and revenue. As their discount rates rise with the provision of more unfunded institutional aid, these colleges will inevitably reach a point at which any additional revenue generated by increased tuition will be committed to financial aid. At this point, when diminishing returns become negative returns, colleges will have to pursue alternative strategies to raise funds for needed academic investments. Today, the average tuition discount rate for small, private liberal arts colleges is nearing 40 percent, but many are well past the 50 percent mark.

The operating objective for many private colleges now is not just to bring in an able freshman class or a class of a particular size, but rather to achieve enrollment that will produce a required level of net revenue. To do this, colleges now routinely differentiate grant aid in financial aid packages, rather than to simply address need equitably across the population of all admitted students, in an exercise that has become known as financial aid leveraging.

Financial aid leveraging implicitly recognizes that families will exhibit different levels of willingness-to-pay depending on their student's academic ability, the family's resources, and the reputation of the college. In a leveraging environment, institutional grant aid is awarded partly on the basis of need, partly on the basis of academic ability, and increasingly on subjective characteristics, such as leadership or citizenship, or on demographic grounds, such as state residency.

For example, a college or university may choose to award more generous grants to in-state students if they are located in states where state aid is available to students attending private colleges. Or, a college might provide scholarships for county residents if the school would like to enroll more local students but is priced too high for local incomes.

Given the range of colleges, the competitive market conditions described above, and the very loose determination of need offered by the federal needs analysis, the assessment of need and ability-to-pay varies widely from school to school. In the days of need-based aid, when merit scholarships were rare, ability-to-pay and willingness-to-pay were assumed to be the same thing. Now they are two different but equally relevant variables in solving the matriculation equation.

The financial aid community, in particular, laments the prevalence of differential financial aid awarding policies, but nearly all colleges, even those able to maintain purely need-based awarding, have had to recognize the penalty that rising financial aid commitments impose on their ability to fund academic investments.

Thus, even those colleges whose aid is need driven may "gap" a segment of their freshman admits. In order to preserve their ability to meet the full need of admitted students, an increasing number of colleges make ability-to-pay a factor in late admission decisions.

From an economist's point of view, however, a market in which willingness-to-pay is the chief determinant in enrollment decisions benefits society as the ultimate source of equity by producing the most value, for the most people, at the lowest possible price. Because more and more families are not simply accepting a stated price, they exert greater power in the market, and colleges must be sensitive to the assessment these families make of the academic value offered by competing schools.

Given this environment, methodologies for predicting enrollment and revenue have become increasingly complex. Current predictive tools are statistically based, reliant on the ability to access and manipulate individual student-record data, and driven by the distribution of financial need among continuing and entering students as well as the net revenue contribution each will make.

For large, public institutions-particularly open-enrollment schools being able to reliably predict revenue and enrollment has been a serious matter for some time. Dependent on taxpayer support, without the ability to manage class size through the admission process, these institutions must expand or contract programs based on the number of students who enroll year to year. With the majority of their revenues limited in the short-term by a set state appropriation and with tuition itself set far below the actual cost of instruction, these schools have been forced to develop an early warning system in order to manage their resources appropriately. In addition, for many of these schools, enrollment fluctuates dramatically with the health of the economy. Given their size, small percentage changes in student participation can cause major swings in costs and revenues.

To address these issues, public institutions have long used statistical models that consider such variables as tuition cost, high school graduation numbers, the health of the economy, and labor-market demand for various skills and occupations as they attempt to predict demand.

For private institutions-as for public ones-the complexity of predictive tools and models differs with the student population served (undergraduate versus graduate and professional), degree of selectivity, and the extent to which admission and financial aid notifications adhere to a single date or are presented on a rolling basis. For example, undergraduate liberal arts colleges with significant waiting lists experience a different and more manageable predictive challenge than do institutions that make admission decisions on a rolling basis and do not have the type of applicant pools to sustain a waiting list.

For highly selective liberal arts colleges, predicting enrollment and revenue is hard to separate from managing enrollment and revenue. Until recently, the challenge for these institutions was to predict the yield rate of admitted students. These colleges continue to manage this variable by establishing waiting lists of admitted students to provide replacements for those who decline offers of admission. Given the interlocking nature of the admitted student pool at many such colleges, managing yield to result in a class of a specific size, academic, or revenue profile can be a dicey exercise: one school's waiting list may be another school's regular admits.

More recently, as they recognized that the growth of their aid commitments was unsustainable, some highly selective liberal arts colleges have sought to manage enrollment yield through admission decisions and waiting lists and to manage revenue by linking admission policy with financial aid policy. For example, Carleton College and Smith College have recommitted themselves to meeting the full need of admitted students. But they also have established a policy of need-sensitive admission for a fraction of their entering classes in order to assure that they will have the revenue needed to finance their financial aid commitments and maintain academic quality at the margin.

These colleges generally have not assigned a particular percentage of the class for need-sensitive consideration. To do so might needlessly erode the role and benefit of need-blind admission in building the strongest and most diverse class possible. Instead, they make this calculation on the run by assessing the amount of aid committed and likely to be accepted by matriculants and determining the amount of net revenue likely to result, At some point, need sensitivity is triggered to assure that budget targets are met -a process that places a greater premium on predictive models in the management of yield and budgets.

In general, the emphasis for these institutions is on managing to an enrollment objective that produces the desired revenue results, usually based on historical understandings of the composition and behavior of the entire admitted student population, not on developing an econometric statistical capacity to predict enrollment and revenue based on characteristics of particular student segments.

For other venerable but less selective colleges, budget predictions in the past were simply made in this conservative way, based on extrapolations of history Although net revenues have become the defining objective of price and financial aid policies, many private college leaders who have a fixation on competition with other private colleges have failed to understand the effects of declining market share and the migration of students to the public sector.

For this reason, they have set tuition relative to these private competitors and have concentrated on aid-focused competition with these colleges rather than attempting to reacquaint the larger market with the academic value they offer, Thus, as they consider next year's tuition increase, they often confront the reality that financial aid will effectively consume all of the revenue generated by the tuition hike.

The new and inescapable reality in projecting enrollment and revenue for private colleges and universities is that one cannot be considered independent of the other. Participation in higher education has always been a function of price and value. But now, with from 50 percent to 90 percent of enrolled Students receiving need- or merit-based financial aid, for most private colleges enrollment is now a function of net price, which differs from student to student based mainly on need and academic ability.

Correspondingly, net revenue is a function of the ability and willingness of families to pay for a particular college based largely on its reputation. In order to predict enrollment and revenue, colleges have had to develop better understandings of the enrollment behavior of their continuing and entering students, assessing the role of many variables, especially academic characteristics, need, and grant aid.

As a by-product of efforts to maximize net revenue by leveraging financial aid or by reining in the financial aid budget, in recent years many private colleges have chosen to assess their admitted student pools by segmenting them based on levels of financial need and academic desirability. In doing so, colleges examine the number of students they are admitting and matriculating in each cell of the need-ability matrix and, based on the financial aid provided, calculate net revenue and attempt to manage and project enrollment and revenue. Here, financial aid leveraging means any strategic differentiation of institutional grant aid to affect the matriculation behavior of specific student groups to produce the desired results for net revenue or student profile (see Figures A, B, & C).

FIGURE A

FIGURE B

FIGURE C

These matrices provide valuable information about the distribution of students and allow colleges to observe differing price response characteristics by student segment. To some extent, they allow colleges to see clearly for the first time the actual distribution of students that, taken together, produce such measures as the average Scholastic Assessment Test verbal score that is reported to college guide publications annually.

To the extent that a college can admit and then enroll the number of students who fall within particular cells that it needs, an enrollment management matrix can provide assurance that the college will meet its revenue and enrollment goals. Viewing a matriculation matrix over several years (which is essential to avoid charting a course based on an unusual year) allows a college to understand the stability of its admitted student population, economically and academically. It also can observe trends in matriculation rates relative to changes in grant aid.

Though matrices are essentially descriptive statistics -a picture of an entire entering cohort- in the hands of enrollment managers, they are potentially a more critical tool than even the most sophisticated econometric analysis. This is true for several reasons.

An enrollment management matrix in which admitted students are segmented by need, academic ability, and residency (in- and out-of-state residency) provides several important advantages that econometric models have difficulty providing. First, they describe the behavior of the entire population or large portions of it. This is important because for most private liberal Arts colleges, the admitted student pool is too small to provide a robust basis for modeling behavior from one year to the next for smaller segments of the pool, such as likely orchestra members or African Americans.

Second, although the parametric matrix view does not provide a statistically tested basis for setting specific grant and matriculation targets for particular student segments, it does provide a framework within which a financial aid policy can be rationalized and its performance evaluated. Some practitioners of matrix-based financial aid analysis have suggested that a matrix segmentation that describes quality of financial aid package offered and resulting matriculation rates amounts to a calculation of willingness-to-pay. This is not so. Only econometric modeling and simulation can provide this calculation. The difference between these two complementary tools is the difference between bracketing an aperture setting and using a light meter to take a picture in uncertain light.

But even when such a calculation is not available, the information it provides for setting targets for grant aid and matriculation cell by cell still must be evaluated with the help of the matrix analysis. And the matrix becomes essential for fitting this information into a coherent award policy and framework. The matrix is essential, too, for the management and evaluation of experiments a college may wish to take into the market to sound out particular student segments.

Because the relationship between grant aid and matriculation is not linear and differs for particular students, matrices cannot provide a statistically project table estimate of price response -that is, they cannot tell you how much matriculation rates will change, if at all, given changes in awards levels. This is the role of econometric modeling.

Beyond its important role in setting targets and suggesting aid experiments within a matrix management framework, econometric modeling provides a predictive tool for matriculation rates in response to chosen grant aid targets.

The descriptive model provided by the matrix analysis and the projective model provided by econometric analysis complement each other. Each gives the other additional power. Whereas the limitations of the matrix are straightforward, the shortcomings of econometric analysis are more arcane and potentially dangerous. These limitations relate mainly to the small number of observations available and the need to cross-validate the matriculation significance of particular student characteristics over multiple years and data sets. In this process, researchers start with exploratory analysis of issues such as the multi-colinearity of variables in order to understand which characteristics offer the best basis for model specification.

Observation shows that matriculation rates differ across student segments. In general, matriculation rates decline as academic ability increases: the stronger a student's academic qualifications, the greater his or her options. Matriculation rates will usually increase with the availability of grant aid. Correspondingly, as a rule, matriculation rates decline as need increases. But these differences are not uniform or consistent, either within a particular cohort or year to year. In fact, for many colleges, there is simply no linear correlation between the quality of financial aid package and matriculation (see Figure D. For example, as indicated in Figure D, there are low matriculation rates (less than 40 percent) at all levels of percentage of need met, whether it is 50 percent or almost 100 percent.

Because students with different characteristics demonstrate different elasticities of demand, some colleges are turning to more sophisticated statistical models to predict student behavior and thereby project enrollment and net revenue.

Built on logistical (logit) regression techniques, these models explain matriculation as a function of multiple variables, such as grade point average, college entrance test scores, class rank, the rigor of high school curricula, gender, status as an alumnus or alumna son or daughter, parental education attainment, state of residence, extracurricular participation, outside interests, and various indicators of family financial capability, including whether an aid application was made, the projected level of need, and the estimated family contribution.

As the regression analysis illuminates the role that each of these variables plays in matriculation, the college is in a better position to predict the behavior of students sharing a given characteristic. Indeed, with a sufficiently reliable model, colleges can identify the marginal effect on matriculation of, say, each additional $100 in grant aid. Understanding the effect of financial aid at the margin, a college can then calculate the net revenue available from a change in matriculation.

For this reason, multivariate modeling provides not only a predictive tool but also an important though limited addition to the matrix management tool. By running simulations of the effects of changes in grant aid in various student segments based on the outputs of the logit regressions, colleges may fine-tune their financial aid awarding strategies to achieve a combination of objectives relative to net revenue, enrollment, and student profile objectives, including diversity.

Although the analytic tools for econometric modeling are probably available at most colleges, it is helpful to establish a framework to guide the analytic exercise. For example:

• Define enrollment capacity, by major or discipline, if possible.

• Ascertain the sufficiency of the databases required to support the analysis.

• Understand the state policy and economic environment.

• Develop a historical understanding of the cumulative changes in state median household income compared with college tuition.

• Achieve a historical, matrix-based understanding of admission, matriculation, financial aid packaging, and net revenue, by student segment.

• Develop a multivariate analysis from tested variables that explains matriculation at a level at least 20 percentage points better than a random guess.

• Simulate market response to incremental changes in grant aid, by segment.

• Subject this analysis to rigorous interpretation and professional judgment based on historical understandings of student behavior across the entire population; set targets for aid and matriculation.

• Rationalize a financial aid policy and award strategy to achieve college objectives and provide a coherent, legitimate treatment of all admitted students.

• Establish targets for admission, matriculation, and net revenue.

• Monitor, by drawing data into a matrix-based spreadsheet, the resulting composition of admits, matriculants, and net revenue on a weekly basis.

To a great extent, this kind of econometric analysis provides a backward-looking needs analysis, providing information that the current federal needs analysis no longer offers. It opens a window on willingness-to-pay that assesses the price-value calculation of students of various academic and economic backgrounds based on the past response of similar students to financial aid offers.

The kind of multivariate analysis and econometric modeling described above is the product of standard statistical tools but is controversial in its application to college pricing. To statisticians, logit regression is a tried-and-true arrow in the analytic quiver. To economists, the combination of private college tuition and financial aid policies is a classic example of price discrimination that occurs when the full price of desirable products is affordable only for a relative few, forcing providers to strive for an average net price and sales volume that will support provision of the product or service by utilizing discounts, financing options, and other pricing options.

This strategy is dangerous when the market deems a product less valuable than its price. And though the market may recognize that a college education pays dividends, increasingly families perceive that these dividends differ from college to college and are certainly not guaranteed. At the same time, higher education nationally suffers from overcapacity. There are many substitute products for most private colleges, including, for affluent students, the opportunity to capture sizable benefits through the tuition subsidies at public colleges and universities.

For all of these reasons, statistically based econometric models and simulations can provide private colleges with invaluable predictive and management tools. They also can be a source of poor and damaging judgments if colleges do not understand their limitations or do not utilize them within strong policy principles.

Although models and simulations are helpful, they are inherently mindless and heartless. They are fueled simply by the variables they consider, rendering simple situations needlessly complex or vice versa. The knowledge they provide will raise policy issues as well as questions of right and wrong.

For example, if the application of these tools includes such variables as whether a student visited a campus, and analysis suggests that campus visits are statistically significant in explaining matriculation, then the college might perceive the opportunity to limit financial aid offers to students who have toured the campus. The model itself does not and cannot address the ethical, policy, or pragmatic issues this knowledge raises. So, knowledge about the factors that influence behavior, useful for predictive purposes, may be applied in ways that are damaging in the long run to the trust people have in the integrity of the college.

Because it is conventional wisdom in private college admissions operations that a campus visit is one of the better indicators of eventual matriculation, the question of whether aid offers should be differentiated on this basis has been discussed and resolved on most campuses. Most colleges see differentiation on this basis as unethical at worst and counterproductive at best, but the point remains that this kind of modeling demonstrates for colleges "the price of their principles," as one president has put it. Once they pursue the kind of knowledge models pen-nit, colleges should be prepared to push interpretation of results to the policy level and to establish a strong set of principles from which to operate.

As private colleges and universities approach the turn of the century, they must once again operate independently of government support. To make the necessary adjustments in such areas as price, cost and financial structure, financial aid policy, and academic program delivery, colleges will have to understand the relatively new concept of enrollment management in increasingly sophisticated ways, bringing increasingly sophisticated tools to bear on the task.

In this process, colleges and universities increasingly will call on the institutional research office to provide predictive and management tools in the areas of marketing survey research, retention research, analysis of peer financial and market positions, and, of course, enrollment and net revenue modeling.

Clearly, such predictive tools are necessary in the current environment. Just as clearly, this environment creates distortions in market behavior that require greater sophistication than in the past if colleges are to achieve the required predictive capability. Like all data-based statistical tools, models of this kind are only as good as the information on which they are built and the institutional values that they support.

In short, these tools are never more powerful than when interpreted and employed by ethical, intuitive professionals. More than ever, institutional researchers and enrollment managers will have to work collaboratively in teams that also include a college's data management personnel. Successful modeling cannot be a purely intellectual exercise, nor do most colleges today have the consistency in their databases required to support this kind of analysis. Building the necessary capacity will be a team endeavor.

Florida Postsecondary Education Planning Commission. How Floridians Pay for College. Technical Report of the Florida Family Funding Study. Tallahassee: Florida Postsecondary Education Planning Commission, July 1994.

Minnesota Private College Research Foundation. Ways and Means: How Minnesota Families Pay for College. St.Paul: Minnesota Private College Research Foundation, Nov. 1992.

Oregon State System of Higher Education. Oregon Family Resource Study. Eugene: Oregon State System of Higher Education, Aug. 1995.