Revealing Objectives in University Financial Aid: An Inverse Optimization Approach

Revealing Objectives in University Financial Aid: An Inverse Optimization Approach | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Revealing Objectives in University Financial Aid: An Inverse Optimization Approach Marc Bara This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6778626/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This study uses inverse optimization to uncover the implicit objectives driving financial aid allocation decisions across U.S. higher education institutions. Analyzing data from 3,130 institutions in the 2021–22 IPEDS dataset, we estimate the strategic weight each institution places on four competing goals: access (coverage), equity (need-based targeting), prestige, and cost control. Private nonprofit universities prioritize prestige (54.8%), awarding an average of $ 23,155 per student; public universities emphasize broad coverage (59.7%) with more moderate awards; and for-profit institutions adopt mixed strategies, balancing access and need. By recovering these revealed institutional priorities from observed behavior, we provide the first large-scale quantitative evidence that financial aid is frequently used to enhance institutional reputation rather than expand access. These findings highlight the equity implications of strategic aid allocation and offer a methodological framework to evaluate whether institutional priorities align with access goals. Decision Sciences financial aid higher education policy institutional behavior equity in access inverse optimization Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction In 2021-22, U.S. higher education institutions distributed over $ 234 billion in financial aid, grants, loans, work-study, and tax benefits (Ma, Pender, & Welch, 2022 ). These resources play a vital role in shaping college access and affordability, but institutions vary widely in how they distribute their share of aid. Private nonprofit universities awarded an average of $ 23,155 per student in institutional grants, compared to just $ 6,344 at public institutions (IPEDS, 2022). Such disparities raise a fundamental question: what goals are institutions actually optimizing when deciding how to allocate aid? While institutions universally claim to prioritize student need and access, their revealed behaviors suggest more complex motivations. Private institutions' high-aid, high-tuition model, combined with relatively low percentages of Pell grant recipients (34% versus 59% at for-profit institutions), hints at strategic considerations beyond pure need-based allocation. However, no prior research has mathematically proven what objectives institutions are truly optimizing. This paper fills this gap by applying inverse optimization — a technique that infers objective functions from observed optimal decisions—to financial aid allocation. Inverse optimization has been successfully applied in transportation (Bertsimas et al., 2015 ), healthcare resource allocation (Chan et al., 2019 ), and energy markets (Saez-Gallego et al., 2016 ), but never before to educational finance. By analyzing the aid decisions of 3,130 US institutions, we reveal the hidden weights institutions place on competing objectives: maximizing coverage, focusing on high-need students, building prestige through large awards, and maintaining cost efficiency. Our findings provide the first mathematical evidence that private institutions optimize primarily for prestige (54.8% of their objective function), while public institutions focus on coverage (59.7%). This quantitative proof of strategic aid allocation has significant implications for policy interventions aimed at improving equity in higher education. 2. Literature Review and Theoretical Framework 2.1 Financial Aid in Higher Education The literature on financial aid spans multiple disciplines, examining its effects on access (Dynarski & Scott-Clayton, 2013 ), persistence (Goldrick-Rab, 2016 ), and institutional choice (Avery & Hoxby, 2004 ). A consistent finding is that aid significantly impacts enrollment decisions, with each $ 1,000 in grants increasing enrollment probability by 3–4 percentage points (Deming & Dynarski, 2010 ). However, the distribution of institutional aid has grown increasingly strategic. Winston ( 1999 ) documented the rise of "high-tuition, high-aid" policies at selective institutions, while Ehrenberg ( 2000 ) linked these policies to competition for rankings. McPherson and Schapiro ( 1998 ) showed that institutional aid increasingly flows to middle- and upper-income students rather than those with the highest need. Recent work has examined algorithmic approaches to aid allocation. Phan et al. (2019, 2021) developed optimization models for financial aid distribution, but these assume institutions maximize enrollment yield or revenue—objectives specified a priori rather than revealed from behavior. Similarly, Sugrue et al. ( 2006 ) applied linear programming to aid management decisions. However, all these studies employ forward optimization approaches that require researchers to specify institutional objectives in advance, rather than inferring what institutions actually optimize from their observed behavior. 2.2 Institutional Competition and Strategy The higher education market has become increasingly competitive, with institutions viewing students as both inputs to and consumers of education (Winston, 1999 ). This dual role creates complex strategic considerations. Theories of positional competition (Hirsch, 1976 ; Frank, 1985 ) suggest institutions use resources primarily to improve relative standing rather than absolute educational quality. Empirical evidence supports strategic behavior. Monks and Ehrenberg ( 1999 ) demonstrated that rankings significantly influence institutional policies, while Bowman and Bastedo ( 2009 ) showed how rankings create self-reinforcing cycles. However, these studies rely on correlation rather than revealing underlying objective functions. 2.3 Inverse Optimization Theory Inverse optimization, pioneered by Ahuja and Orlin ( 2001 ), provides a framework for inferring objective functions from observed decisions. The key insight is that if decisions are optimal, they must maximize some objective function subject to constraints. By analyzing patterns in decisions, we can recover the implicit objectives. The approach has evolved from early linear programming applications (Heuberger, 2004 ) to handle non-linear objectives (Keshavarz et al., 2011 ) and multiple objectives (Roland et al., 2013 ). Recent advances allow for noise in observations (Aswani et al., 2018 ) and partially observed constraints (Esfahani et al., 2018 ). 2.4 The Forward vs. Inverse Optimization Distinction The optimization literature on financial aid allocation has exclusively focused on forward optimization approaches that assume known institutional objectives. Studies like Phan et al. ( 2022 ) and Sugrue et al. ( 2006 ) develop sophisticated models to help institutions optimize their aid distribution given stated goals such as maximizing enrollment, minimizing cost, or improving access metrics. While valuable for institutional planning, these approaches cannot reveal what objectives institutions actually pursue in practice. Inverse optimization offers a fundamentally different perspective by working backward from observed decisions to infer the underlying objective functions. This distinction is crucial: forward optimization assumes we know what institutions want to achieve, while inverse optimization discovers what they actually optimize for. No prior study has applied inverse optimization techniques to analyze institutional behavior in higher education finance, creating the methodological gap this research addresses. 2.5 Theoretical Predictions Three theoretical perspectives offer competing predictions: Human Capital Theory (Becker, 1964 ) predicts institutions should maximize social returns by investing in students with the highest marginal benefit—typically those from disadvantaged backgrounds who lack alternative funding sources. Positional Goods Theory (Hirsch, 1976 ) predicts institutions will allocate aid to enhance relative prestige, focusing on students who improve institutional metrics rather than those with the greatest need. Resource Dependence Theory (Pfeffer & Salancik, 1978 ) suggests behavior reflects the need to secure resources from key stakeholders. Private institutions dependent on donations and rankings would thus optimize differently than public institutions dependent on state funding. Our inverse optimization approach allows empirical testing of which theory best explains observed behavior. 3. Data and Institutional Context 3.1 Data Sources This study employs data from the Integrated Postsecondary Education Data System (IPEDS), the comprehensive federal data collection system administered by the National Center for Education Statistics. We construct our analytical dataset by merging three IPEDS survey components for academic year 2021-22: Student Financial Aid survey (SFA2122): Institutional grant recipients and award amounts Institutional Characteristics survey (HD2022): Institutional control, enrollment size, and Carnegie classification Fall Enrollment survey (EF2022): Student demographic characteristics and enrollment patterns We focus our analysis on 2021-22 as it represents the most recent academic year with complete financial aid reporting across all institution types. Our sample selection criteria exclude institutions with incomplete financial aid reporting and those enrolling fewer than 100 full-time equivalent undergraduates to ensure data quality and institutional viability. The resulting analytical sample comprises 3,130 degree-granting institutions, representing approximately 95% of total undergraduate enrollment in US higher education. 3.2 Variable Construction We construct our analytical variables following established conventions in the higher education finance literature. Our dependent variables capture the key dimensions of institutional aid allocation decisions: Primary Outcome Variables : Average institutional grant per recipient (UAGRNTA): Mean need- and merit-based institutional aid awarded to recipients Institutional grant coverage rate (UAGRNTP/100): Proportion of enrolled students receiving institutional aid Total institutional aid expenditure: Aggregate spending calculated as recipients × average award Institutional Control Variables : Institutional control: Public (n = 1,552), Private non-profit (n = 943), For-profit (n = 635) Federal grant recipient proportion (UPGRNTP): Percentage of students receiving federal need-based aid (Pell Grants), serving as our primary proxy for student financial need Full-time equivalent enrollment (SCUGFFN): Total undergraduate enrollment adjusted for part-time status Institutional-to-federal aid ratio: Relative emphasis on institutional versus federal aid programs 3.3 Descriptive Evidence Table 1 presents financial aid patterns by institution type: Metric Public (n = 1,552) Private (n = 943) For-Profit (n = 635) Average Institutional Grant Mean $ 6,344 $ 23,155 $ 5,114 Std Dev $ 2,922 $ 11,325 $ 1,866 Median $ 5,832 $ 22,140 $ 4,956 Coverage Rate (%) Mean 68.0% 84.5% 67.9% Std Dev 21.2% 15.3% 19.4% Federal Grant Recipients (%) Mean 32.1% 33.9% 58.5% Std Dev 13.4% 17.1% 19.2% Institutional/Federal Aid Ratio Mean 1.40 4.73 1.15 Std Dev 0.55 2.23 0.35 Total Aid per FTE Student Mean $ 4,312 $ 19,546 $ 3,470 Std Dev $ 2,456 $ 12,331 $ 1,654 The patterns reveal striking differences. Private institutions provide awards 3.65 times larger than public institutions while serving similar proportions of federal aid recipients (33.9% vs 32.1%). The institutional-to-federal aid ratio at private institutions (4.73) far exceeds public (1.40) and for-profit (1.15) institutions, suggesting different optimization strategies. Figures 1 – 3 provide comprehensive visual evidence of the fundamental differences in institutional aid strategies. Figure 1 reveals the core trade-off institutions face between coverage and generosity. Public institutions (green circles) cluster along the high-coverage, low-generosity frontier, optimizing to serve as many students as possible within budget constraints. Private institutions (orange squares) spread across the high-generosity space, with many achieving both high coverage and high awards—a pattern only possible with substantially larger per-student aid budgets. For-profit institutions (blue triangles) concentrate in the moderate coverage, low-generosity region. Figure 2 exposes a crucial contradiction in the relationship between institutional resources and student need. For-profit institutions, despite having the smallest aid budgets, serve the neediest student populations with median federal aid coverage around 60%. In contrast, both public and private institutions serve similar proportions of high-need students (~ 30%), yet deploy vastly different aid strategies. This pattern suggests that prestige considerations, rather than need-based targeting, drive private institution behavior. Figure 3 quantifies these strategic differences through distributional analysis. Private institutions provide aid amounts nearly four times larger than public institutions (median ~ $ 22,000 vs ~ $ 6,000) while also achieving superior coverage rates, with most private institutions exceeding 85% coverage compared to more variable coverage at public and for-profit institutions. Together, these patterns demonstrate that private institutions operate with fundamentally different resource constraints and optimization objectives than their counterparts. 4. Methodology: Inverse Optimization Framework 4.1 Theoretical Model We model institutional financial aid allocation as a constrained optimization problem where each institution maximizes a weighted combination of competing objectives subject to budget and operational constraints. The institutional decision-maker's problem can be formalized as: Forward Optimization Problem : max Σₖ wₖ · fₖ(a, c, p) a,c subject to: a · c · N ≤ B (budget constraint) 0 ≤ c ≤ 1 (coverage constraint) a ≥ 0 (non-negativity constraint) Ω(a, c) (institution-specific constraints) where the decision variables are average aid amount per recipient (a) and coverage rate (c), defined as the proportion of enrolled students receiving institutional aid. The parameter vector includes the proportion of high-need students (p), measured as federal grant recipients; total undergraduate enrollment (N); and the institutional aid budget (B). The functions fₖ represent distinct institutional objectives, with weights wₖ capturing the relative importance assigned to each objective. 4.2 Objective Function Specification Drawing from institutional mission statements and higher education finance theory, we specify four potential objectives that capture the primary dimensions of institutional aid allocation: Coverage Maximization: f₁(c) = c Economic interpretation: Maximize the proportion of students receiving aid Theoretical foundation: Equity-oriented institutional missions emphasizing broad access Need-Based Targeting: f₂(c,p) = c · p Economic interpretation: Maximize aid allocation to financially disadvantaged students Theoretical foundation: Social mobility and human capital development objectives Prestige Signaling: f₃(a) = a/ā, where ā = $30,000 Economic interpretation: Signal institutional quality through generous award amounts Theoretical foundation: Positional competition theory and reputation-building strategies Normalization rationale: The normalization constant ensures scale comparability across objectives while preserving relative differences in award generosity Cost Efficiency: f₄(a,c,N,B) = 1 - (a·c·N)/B Economic interpretation: Minimize aid expenditure relative to available budget Theoretical foundation: Resource constraint optimization and financial sustainability 4.3 Inverse Optimization Approach Given observed aid allocation decisions (a , c ) for each institution, we employ inverse optimization techniques to recover the implicit objective function weights w = [w₁, w₂, w₃, w₄] that rationalize these choices. Our approach integrates revealed preference theory with variance-based weight identification. Step 1: Calculate Objective Values For each institution i, we calculate realized objective values: Coverage: f₁ᵢ = cᵢ Need-based targeting: f₂ᵢ = cᵢ · pᵢ Prestige signaling: f₃ᵢ = aᵢ / 30,000 Cost efficiency: f₄ᵢ = 1 - (aᵢ·cᵢ·Nᵢ)/Bᵢ Step 2: Revealed Preference Analysis Within each institutional control type g, we apply revealed preference logic: objectives receiving higher optimization weight should exhibit greater cross-institutional variation as institutions make different trade-offs along these dimensions. We compute the sample variance σₖ² and mean μₖ for each objective k. Step 3: Weight Recovery Following the variance-proportionality principle, we recover institutional weights as: wₖ = (σₖ² · μₖ) / Σⱼ(σⱼ² · μⱼ) Where σₖ² is variance and μₖ is mean of objective k. This specification ensures that objectives showing both high variation (indicating active optimization) and high mean achievement receive proportionally higher weights. Step 4: Validation We validate recovered weights through multiple empirical tests: (i) cross-sectional regression analysis examining whether institution type predicts aid generosity after controlling for enrollment size, (ii) within-type heterogeneity analysis testing whether private institutions with different aid strategies serve systematically different student populations, (iii) bootstrap resampling with 1,000 replications to assess statistical precision and generate confidence intervals for key institutional metrics, and (iv) for each institution we compute the coefficient of complementarity (ρ; Chan et al., 2019) to quantify how closely observed allocations satisfy the KKT conditions: ρ ranges from 0 (no fit) to 1 (perfect optimality). 5. Results 5.1 Revealed Institutional Objectives Table 2 presents the recovered objective weights by institution type: Objective Public Private For-Profit Coverage 0.597 (59.7%) 0.298 (29.8%) 0.434 (43.4%) Need Focus 0.201 (20.1%) 0.150 (15.0%) 0.277 (27.7%) Prestige 0.179 (17.9%) 0.548 (54.8%) 0.105 (10.5%) Cost Efficiency 0.023 (2.3%) 0.003 (0.3%) 0.185 (18.5%) Interpretation Coverage-focused Prestige-focused Balanced Primary Trade-off Coverage vs. Budget Prestige vs. Access Need vs. Cost Figure 2 presents the recovered objective weights from the inverse optimization analysis. The results demonstrate significant variation in optimization strategies across institution types, with each sector exhibiting distinct priority structures. Private institutions exhibit the highest weight on prestige signaling (54.8%). This objective receives substantially greater emphasis than coverage (29.8%) or need-based targeting (15.0%). Cost efficiency receives minimal weight (0.3%), indicating limited consideration of spending constraints in aid allocation decisions. Public institutions prioritize coverage above other objectives (59.7%). This emphasis on broad student access is accompanied by moderate attention to need-based targeting (20.1%) and prestige signaling (17.9%). The relatively balanced approach across coverage and need-based objectives suggests optimization under resource constraints. For-profit institutions demonstrate a more balanced objective structure. Coverage (43.4%) and need-based targeting (27.7%) receive comparable emphasis, while this sector shows the highest weight on cost efficiency (18.5%) among all institution types. This pattern indicates greater attention to operational efficiency in aid allocation. The recovered weights reveal systematic differences in institutional priorities. Private institutions optimize primarily for prestige outcomes, public institutions emphasize access maximization, and for-profit institutions incorporate efficiency considerations more prominently than other sectors. These differences suggest that institutional type serves as a strong predictor of aid allocation strategy. 5.2 Statistical Validation To validate our inverse optimization findings, we conduct two empirical tests examining whether the recovered objectives predict institutional behavior. 5.2.1 Cross-Sectional Analysis We first test whether institution type predicts aid generosity, controlling for institutional size. Estimating the model: log(aid_amount) = β₀ + β₁Private + β₂ForProfit + β₃log(enrollment) + ε yields the following results: Private institution coefficient: β₁ = 1.323 (p 0.05) Model R² = 0.665 The private institution coefficient of 1.323 indicates that private institutions provide 275% more aid per recipient than public institutions (e^1.323 ≈ 3.75) after controlling for enrollment size. This substantial difference, explaining 66.5% of variation in aid amounts, confirms that institution type fundamentally determines aid allocation strategy. The non-significant for-profit coefficient suggests these institutions behave similarly to publics in terms of award size, consistent with their balanced objective weights. 5.2.2. Within-Type Heterogeneity To examine whether prestige optimization varies within private institutions, we partition them into quartiles based on average aid amounts and analyze their characteristics: Aid Quartile Average Award Coverage Rate Federal Aid Recipients Q1 (Lowest) $ 9,887 82% 49% Q2 $ 19,056 87% 34% Q3 $ 25,283 90% 31% Q4 (Highest) $ 38,401 82% 23% A striking pattern emerges: institutions providing the largest awards serve substantially fewer high-need students. The percentage of federal aid recipients decreases monotonically from 49% in the lowest quartile to 23% in the highest quartile—a 26 percentage point decline. This inverse relationship between award generosity and student need provides additional evidence that high awards serve prestige rather than access objectives. Institutions in the top quartile effectively purchase prestige by concentrating resources on students who require the least financial assistance. Figure 4 visualizes these patterns across the three key dimensions of aid allocation. The top panel demonstrates the dramatic escalation in average awards, with the most generous institutions providing nearly four times larger awards than the least generous. The middle panel reveals coverage rates that peak in the third quartile before declining as the highest-award institutions become more selective about recipients. Most critically, the bottom panel confirms the systematic decline in federal aid recipients as award generosity increases. This within-type variation demonstrates that even among private institutions sharing similar organizational structures and missions, those that weight prestige objectives more heavily systematically serve different student populations than those prioritizing access or coverage. The heterogeneity in revealed preferences within the private sector thus provides additional validation of our inverse optimization approach by showing that objective weights vary meaningfully even within institutional types. 5.2.3. Goodness-of-Fit via Complementarity The coefficient of complementarity confirms that the recovered objectives explain observed behavior. Median ρ across all 3,130 institutions equals 0.65, indicating allocations lie within 35% of perfect optimality under the recovered weights. ρ varies systematically by sector —private = 0.726, public = 0.647, for-profit = 0.608 — with private institutions exhibiting the tightest fit to their prestige-focused objective. This level of fit is well within the range typically reported in inverse-optimisation studies involving real-world human or institutional behavior, where values between 0.6 and 0.7 are considered strong (Chan, Lee & Terekhov, 2019 ). Institution Type Median ρ Interpretation Private 0.726 73% optimality Public 0.647 65% optimality For-Profit 0.608 61% optimality Overall 0.650 Key model-fit statistic 5.3 Heterogeneity Analysis Our aggregate findings may mask important variation within institution types. We examine heterogeneity along two dimensions: geographic region for private institutions and institutional size for public institutions. 5.3.1. Regional Patterns in Private Institution Behavior Analysis of private institutions by region reveals systematic geographic variation in aid strategies: Region Average Award Inst/Fed Ratio N New England $ 30,642 6.19 89 Far West $ 25,944 5.29 74 Mid East $ 25,408 5.08 207 Great Lakes $ 23,512 4.92 156 Plains $ 22,037 4.74 104 Southeast $ 20,381 4.12 222 Rocky Mountains $ 19,568 4.04 13 Southwest $ 18,244 3.73 58 The geographic gradient is pronounced: New England institutions provide average awards 68% higher than those in the Southwest ( $ 30,642 vs. $ 18,244). The institutional-to-federal aid ratio similarly varies from 6.19 in New England to 3.73 in the Southwest, indicating that competitive dynamics shape optimization strategies. Figure 5 visualizes this geographic gradient through a dual-axis presentation showing that average awards (bars) and institutional-to-federal aid ratios (line) follow remarkably similar patterns across regions. Both metrics peak in New England and decline systematically toward the Southwest. This correlation between award generosity and relative emphasis on institutional versus federal aid suggests systematic regional differences in aid allocation strategies. The geographic clustering of similar behaviors is consistent with positional goods theory's prediction that competitive dynamics influence institutional strategy, though the underlying mechanisms driving these regional patterns warrant further investigation. 5.3.2. Size and Public Institution Objectives Among public institutions, we examine how institutional scale affects revealed objectives: Size Category Institutions Coverage Rate Average Award Small ( 20,000) 0 - - The data reveal that medium-sized public institutions achieve modestly higher coverage rates (70.4% vs. 68.2%) while providing substantially larger awards ( $ 11,340 vs. $ 6,184). This pattern suggests that economies of scale in aid administration allow larger institutions to pursue coverage objectives more effectively. However, the limited number of medium-sized institutions (n = 48) and complete absence of large public institutions (> 20,000 students) in our sample significantly constrains analysis of size effects. The directional relationship observed between the two size categories supports our interpretation that operational capacity influences objective weights, though definitive conclusions require broader institutional representation. 5.4 Robustness Analysis We conduct three robustness tests to assess the stability and statistical reliability of our findings. 5.4.1 Statistical Inference Through Bootstrap Resampling To quantify uncertainty in our revealed objective estimates, we implement bootstrap resampling with 1,000 replications. For each iteration, we resample institutions with replacement and recalculate key metrics: Private Institutions : Institutional/Federal aid ratio: 4.734 [95% CI: 4.591, 4.865] Standard error: 0.070 Public Institutions : Coverage rate: 0.682 [95% CI: 0.672, 0.693] Standard error: 0.005 The narrow confidence intervals indicate precise estimation of institutional behavior. The private institution confidence interval for the inst/fed ratio (4.591, 4.865) lies entirely above the public institution mean of 1.40, confirming that the differential optimization strategies are not artifacts of sampling variation. Bootstrap distributions show no evidence of multimodality, suggesting our revealed preference approach identifies unique, stable objective functions rather than averaging across multiple equilibria. 5.4.2 Hypothesis Testing for Differential Objectives To formally test whether private and public institutions pursue different objectives, we compare their institutional-to-federal aid ratios using a two-sample t-test: Test statistic: t = 56.262 Degrees of freedom: 2,493 p-value < 0.000001 The extreme test statistic provides overwhelming evidence against the null hypothesis of equal objectives. The probability of observing such different aid strategies if institutions truly shared common objectives is effectively zero. This statistical significance, combined with the economic magnitude of the differences (private ratio is 3.38 times the public ratio), confirms that institution types fundamentally differ in their optimization strategies. Figure 6 provides intuitive visual evidence for the statistical differences documented in our hypothesis testing between private and public institutions. The dramatic separation between these two institution types is immediately apparent: public institutions (blue) cluster tightly around a median ratio of 1.4, with most institutions falling between 1.0 and 2.0. In stark contrast, private institutions (orange) show both higher central tendency and greater dispersion, with the majority of institutions operating at ratios between 3.0 and 6.0—entirely above the public distribution. This complete separation of distributions explains the extreme t-statistic (56.262) in our formal hypothesis test and demonstrates that the differential optimization strategies between private and public institutions are not marginal differences but fundamental distinctions in institutional behavior. 6. Policy Implications 6.1 Transparency Through Mathematical Clarity Perhaps the most significant contribution of this analysis is providing mathematical clarity to previously opaque institutional decisions. By recovering objective weights through inverse optimization, we transform abstract discussions of "institutional priorities" into concrete, quantifiable metrics. Institutional leaders can now answer specific questions: What percentage of our aid optimization focuses on prestige versus need? How does our revealed objective function compare to peer institutions? What would be the concrete impact of shifting our allocation weights? This transparency extends to other stakeholders as well. Students and families can better understand the forces shaping aid offers. Donors can assess whether their contributions support stated institutional missions. Policymakers can evaluate whether current incentive structures promote desired social outcomes. 6.2 A Tool for Institutional Analysis The methodology developed here offers institutions a tool for self-examination. By applying inverse optimization to their own historical aid data, institutions can uncover their revealed priorities and assess alignment with stated missions. This approach moves beyond anecdotal evidence or good intentions to provide rigorous, quantitative insight into actual institutional behavior. For institutions genuinely committed to expanding access, our findings highlight the magnitude of change required. Moving from prestige-dominant optimization (54.8% weight) to need-focused allocation would represent fundamental transformation rather than incremental adjustment. 7. Discussion 7.1 Theoretical Implications Our empirical findings provide strong evidence for the applicability of Positional Goods Theory to understand financial aid allocation in higher education. The recovered objective weights reveal that private institutions allocate 54.8% of their optimization effort toward prestige enhancement, mathematically confirming what Hirsch ( 1976 ) theorized about competition for relative position. This finding challenges the prevailing assumption that educational institutions, particularly those with substantial endowments and expressed commitments to access, operate according to Human Capital Theory's prescription of maximizing social returns. The contrast between institution types proves particularly illuminating. Public institutions, with 59.7% weight on coverage and only 17.9% on prestige, demonstrate behavior more aligned with their statutory missions and public accountability structures. This differential supports Resource Dependence Theory's prediction that institutional behavior reflects stakeholder constraints. For-profit institutions present an unexpected finding: despite their profit motive, they show greater balance between coverage (43.4%) and need-focus (27.7%) than private non-profits, suggesting that market pressures may sometimes align better with social objectives than philanthropic governance structures. 7.2 Methodological Contributions This study advances the application of inverse optimization methods to social science questions. While inverse optimization has proven valuable in engineering and operations research contexts, its application to institutional behavior in education opens new methodological possibilities. Our approach differs fundamentally from traditional econometric methods that require researchers to specify behavioral models a priori. Instead, we allow the data to reveal the objective function, providing a more inductive path to understanding institutional priorities. The revealed preference framework adapted here addresses a persistent challenge in studying organizational behavior: the gap between stated and actual objectives. By mathematically recovering the weights institutions place on competing goals, we move beyond reliance on mission statements, strategic plans, or administrator surveys. This methodology could extend to numerous educational contexts where institutions face multi-objective optimization problems, from admissions decisions to resource allocation across academic programs. 7.3 Limitations Several limitations warrant consideration when interpreting our findings. First, the aggregated nature of IPEDS data prevents observation of individual student aid packages, potentially masking within-institution heterogeneity in allocation strategies. Institutions may apply different objectives to different student segments, a nuance our institution-level analysis cannot capture. Second, our model assumes institutions freely optimize their stated objectives, but unobserved constraints may influence allocation decisions. Restricted donations, state-mandated aid programs, and historical commitments create path dependencies not reflected in our optimization framework. While our robustness checks suggest these constraints do not fundamentally alter the revealed objectives, they may explain some variation in institutional behavior. Third, our analysis focuses on absolute dollar amounts rather than aid as a proportion of institutional costs. While institutional grants represent different proportions of total costs across institution types, our focus on revealed optimization priorities through aid allocation patterns captures fundamental institutional behavior regardless of underlying cost structures. Fourth, our cross-sectional analysis captures a snapshot of institutional behavior rather than dynamic evolution. Market pressures, leadership changes, and policy interventions may alter institutional priorities over time scales our data cannot observe. Finally, the functional form of objective specifications influences results, though our robustness tests using alternative formulations yield qualitatively similar findings. The choice to normalize prestige by $ 30,000 and cost efficiency by budget size reflects reasonable but ultimately subjective scaling decisions. 7.4 Future Research This work establishes a foundation for multiple research trajectories. Longitudinal analysis could trace how institutional objectives evolve in response to market conditions, policy changes, or competitive pressures. The COVID-19 pandemic's disruption to higher education financing provides a natural experiment for studying objective stability under stress. Access to student-level data would enable more nuanced inverse optimization, potentially revealing how institutions apply different objectives to different populations. Such analysis could test whether the aggregate patterns we observe result from uniform policies or averaged heterogeneous strategies. Comparative international research could examine whether the prestige optimization we document reflects uniquely American competitive dynamics or broader patterns in marketized higher education systems. Countries with different financing models and governance structures provide natural contrasts for testing the generalizability of our findings. Experimental or quasi-experimental studies of institutions that have reformed their aid policies could validate whether stated changes in priorities translate to shifted objective weights in practice. Such research would move beyond revealing current behavior to understanding the conditions under which institutions can successfully transform their allocation strategies. The methodological framework developed here invites application to other domains where institutions balance multiple objectives. From hospital resource allocation to public school funding formulas, inverse optimization offers a tool for revealing the true priorities governing institutional decisions that affect social welfare. 8. Conclusion This study provides the first mathematical evidence that private US universities optimize financial aid for prestige rather than access or need. Our analysis of 3,130 institutions reveals that private universities allocate 54.8% weight on prestige signaling while dedicating only 15.0% to need-based targeting—a nearly 4:1 ratio that stands in stark contrast to public institutions, which prioritize coverage (59.7%) over prestige (17.9%). The inverse relationship we document between award generosity and student need among private institutions demonstrates that prestige optimization fundamentally conflicts with access objectives, quantifying patterns that prior research suggested but never measured precisely. The significance extends beyond higher education to organizational behavior more broadly. By demonstrating that institutional actions can reveal true priorities regardless of stated missions, this research offers a methodological foundation for understanding decision-making in any context where organizations balance multiple competing objectives. As higher education faces increasing scrutiny over affordability and equity, this research contributes quantitative evidence about institutional aid priorities that can inform policy discussions and institutional self-reflection about the $ 185 billion distributed annually in financial aid. Declarations Data Availability Statement All data used in this study are derived from publicly available IPEDS datasets for the 2021–22 academic year. The processed dataset—including recovered objective weights and validation metrics for 3,130 institutions—along with full replication code and figures, is archived at: https://doi.org/10.5281/zenodo.15496894 References Ahuja, R. K., & Orlin, J. B. (2001). Inverse optimization. Operations Research, 49(5), 771-783. Aswani, A., Shen, Z. J. M., & Siddiq, A. (2018). Inverse optimization with noisy data. Operations Research, 66(3), 870-892. Avery, C., & Hoxby, C. M. (2004). Do and should financial aid packages affect students' college choices? In College choices: The economics of where to go, when to go, and how to pay for it (pp. 239-302). University of Chicago Press. Becker, G. S. (1964). Human capital: A theoretical and empirical analysis, with special reference to education. University of Chicago Press. Bertsimas, D., Gupta, V., & Paschalidis, I. C. (2015). Data-driven estimation in equilibrium using inverse optimization. Mathematical Programming, 153(2), 595-633. Bettinger, E. (2015). Need-based aid and college persistence: The effects of the Ohio College Opportunity Grant. Educational Evaluation and Policy Analysis, 37(1_suppl), 102S-119S. Bourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education (pp. 241-258). Greenwood. Bowman, N. A., & Bastedo, M. N. (2009). Getting on the front page: Organizational reputation, status signals, and the impact of US News and World Report on student decisions. Research in Higher Education, 50(5), 415-436. Chan, T. C., Lee, T., & Terekhov, D. (2019). Inverse optimization: Closed-form solutions, geometry, and goodness of fit. Management Science, 65(3), 1115-1135. Deming, D., & Dynarski, S. (2010). College aid. In Targeting investments in children: Fighting poverty when resources are limited (pp. 283-302). University of Chicago Press. Dynarski, S., & Scott-Clayton, J. (2013). Financial aid policy: Lessons from research. The Future of Children, 23(1), 67-91. Ehrenberg, R. G. (2000). Tuition rising: Why college costs so much. Harvard University Press. Esfahani, P. M., Shafieezadeh-Abadeh, S., Hanasusanto, G. A., & Kuhn, D. (2018). Data-driven inverse optimization with imperfect information. Mathematical Programming, 167(1), 191-234. Frank, R. H. (1985). Choosing the right pond: Human behavior and the quest for status. Oxford University Press. Goldrick-Rab, S. (2016). Paying the price: College costs, financial aid, and the betrayal of the American dream. University of Chicago Press. Heuberger, C. (2004). Inverse combinatorial optimization: A survey on problems, methods, and results. Journal of Combinatorial Optimization, 8(3), 329-361. Hirsch, F. (1976). Social limits to growth. Harvard University Press. Keshavarz, A., Wang, Y., & Boyd, S. (2011). Imputing a convex objective function. In 2011 IEEE International Symposium on Intelligent Control (pp. 613-619). Ma, J., Pender, M., & Welch, M. (2022). Trends in college pricing and student aid 2022 . College Board. https://research.collegeboard.org/media/pdf/trends-in-college-pricing-student-aid-2022.pdf McPherson, M. S., & Schapiro, M. O. (1998). The student aid game: Meeting need and rewarding talent in American higher education. Princeton University Press. Monks, J., & Ehrenberg, R. G. (1999). The impact of US News and World Report college rankings on admission outcomes and pricing decisions at selective private institutions (No. w7227). National Bureau of Economic Research. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. Harper & Row. Phan, Vinhthuy; Wright, Laura; & Decent, Bridgette. (2022). Optimizing Financial Aid Allocation to Improve Access and Affordability to Higher Education. Journal of Educational Data Mining , 14(3), 26–51. Roland, J., De Smet, Y., & Figueira, J. R. (2013). Inverse multi-objective combinatorial optimization. Discrete Applied Mathematics, 161(16-17), 2764-2771. Saez-Gallego, J., Morales, J. M., Zugno, M., & Madsen, H. (2016). A data-driven bidding model for a cluster of price-responsive consumers of electricity. IEEE Transactions on Power Systems, 31(6), 5001-5011. Sugrue, P. K., Mehrotra, A., & Orehrovec, P. M. (2006). Financial aid management: an optimisation approach. International Journal of Operational Research, 1(3), 347-362. Winston, G. C. (1999). Subsidies, hierarchy and peers: The awkward economics of higher education. Journal of Economic Perspectives, 13(1), 13-36. Zhang, L. (2013). Effects of college educational debt on graduate school attendance and early career and lifestyle choices. Education Economics, 21(2), 154-175. Additional Declarations The authors declare no competing interests. Supplementary Files AppendixA.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6778626","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":463781180,"identity":"5eae0509-1f39-4974-b11c-29e1bc3effef","order_by":0,"name":"Marc Bara","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA80lEQVRIiWNgGAWjYLACxgYgwd5AshaeAyRrkUggUrVu+9mHHxh32OXzz3z88NPNHYftGdibHz5gqKjDqcXsTLqxBOOZZMsZt9OMpXPPHE5s4DlmbMBw5jBuLQfS2BgY25gNGG7nMEjnth1OALrQTIKx7QBuLeefgbTUG8jfPMP8G6jFnkH++fcfjP/wOOwG2JbDBgY3eNhAtjA2SPCYAQOEGY+WZ8wSiWeOGxieSTOzzj2TntjGk1MskXAMj1/OpzF++Lij2kDu+OHHt3N3WNvzsx/f+OFDDW6HgUECjAGKIDYUEYIAHKejYBSMglEwCtAAAIBEUqzgDsEsAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0009-0005-1480-5760","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Marc","middleName":"","lastName":"Bara","suffix":""}],"badges":[],"createdAt":"2025-05-29 17:34:54","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-6778626/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6778626/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":83661065,"identity":"989394ce-282e-4a46-82ba-12d72a3571a5","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":283455,"visible":true,"origin":"","legend":"\u003cp\u003eThe Coverage-Generosity Tradeoff in Financial Aid Strategy.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/9899cab32e7abcd65d87daef.png"},{"id":83661071,"identity":"123deb4b-1f14-43cf-9b08-2ac34edb710c","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":91989,"visible":true,"origin":"","legend":"\u003cp\u003eStudent Need Levels by Institution Type.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/d95027dbbf8e4ea4f5a297d8.png"},{"id":83661770,"identity":"5b78773e-0b98-45e3-8660-8c6ca1c8125b","added_by":"auto","created_at":"2025-05-30 10:09:05","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":70272,"visible":true,"origin":"","legend":"\u003cp\u003eFinancial Aid Distributions by Institution Type.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/2183217849690883a24947e2.png"},{"id":83661066,"identity":"a324653c-6948-40d6-afd1-4dfc566383fa","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":98431,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 2. Revealed Institutional Priorities from Financial Aid Allocation\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/0686b71bb40557402dbc3cc9.png"},{"id":83661070,"identity":"d37981ae-58f4-4be1-b44d-ecac76060981","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":117847,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4. Private Institution Selectivity Analysis by Aid Quartile\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/772cb25b332906c2336e9beb.png"},{"id":83661068,"identity":"4f13ccad-43c9-4d28-930c-d9d63a32abac","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":83852,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 5. Regional Variation in Private Institution Aid Strategies.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/f656ab13be6f36a0e6ade407.png"},{"id":83661067,"identity":"4471272a-c65f-42e1-9379-674bcf6b5af0","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":62235,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 6. Institutional/Federal Aid Ratio Distributions: Private vs. Public Institutions.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/7e0931d73195dd473817233e.png"},{"id":83661965,"identity":"099063a9-03f9-4ed3-9267-1ba1c8c3a49c","added_by":"auto","created_at":"2025-05-30 10:17:06","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2259573,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/9044d5da-86d9-4879-b15f-c31d6203ce6e.pdf"},{"id":83661063,"identity":"73a3b599-bc9f-42c0-b8bf-e226e27ff586","added_by":"auto","created_at":"2025-05-30 10:01:05","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":15866,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixA.docx","url":"https://assets-eu.researchsquare.com/files/rs-6778626/v1/3bfdacc3f1bcb6966d758b62.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eRevealing Objectives in University Financial Aid: An Inverse Optimization Approach\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eIn 2021-22, U.S. higher education institutions distributed over \u003cspan\u003e$\u003c/span\u003e234\u0026nbsp;billion in financial aid, grants, loans, work-study, and tax benefits (Ma, Pender, \u0026amp; Welch, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). These resources play a vital role in shaping college access and affordability, but institutions vary widely in how they distribute their share of aid. Private nonprofit universities awarded an average of \u003cspan\u003e$\u003c/span\u003e23,155 per student in institutional grants, compared to just \u003cspan\u003e$\u003c/span\u003e6,344 at public institutions (IPEDS, 2022). Such disparities raise a fundamental question: \u003cb\u003ewhat goals are institutions actually optimizing when deciding how to allocate aid?\u003c/b\u003e \u003c/p\u003e \u003cp\u003eWhile institutions universally claim to prioritize student need and access, their revealed behaviors suggest more complex motivations. Private institutions' high-aid, high-tuition model, combined with relatively low percentages of Pell grant recipients (34% versus 59% at for-profit institutions), hints at strategic considerations beyond pure need-based allocation. However, no prior research has mathematically proven what objectives institutions are truly optimizing.\u003c/p\u003e \u003cp\u003eThis paper fills this gap by applying inverse optimization \u0026mdash; a technique that infers objective functions from observed optimal decisions\u0026mdash;to financial aid allocation. Inverse optimization has been successfully applied in transportation (Bertsimas et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), healthcare resource allocation (Chan et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), and energy markets (Saez-Gallego et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), but never before to educational finance. By analyzing the aid decisions of 3,130 US institutions, we reveal the hidden weights institutions place on competing objectives: maximizing coverage, focusing on high-need students, building prestige through large awards, and maintaining cost efficiency.\u003c/p\u003e \u003cp\u003eOur findings provide the first mathematical evidence that private institutions optimize primarily for prestige (54.8% of their objective function), while public institutions focus on coverage (59.7%). This quantitative proof of strategic aid allocation has significant implications for policy interventions aimed at improving equity in higher education.\u003c/p\u003e"},{"header":"2. Literature Review and Theoretical Framework","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Financial Aid in Higher Education\u003c/h2\u003e \u003cp\u003eThe literature on financial aid spans multiple disciplines, examining its effects on access (Dynarski \u0026amp; Scott-Clayton, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), persistence (Goldrick-Rab, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), and institutional choice (Avery \u0026amp; Hoxby, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2004\u003c/span\u003e). A consistent finding is that aid significantly impacts enrollment decisions, with each \u003cspan\u003e$\u003c/span\u003e1,000 in grants increasing enrollment probability by 3\u0026ndash;4 percentage points (Deming \u0026amp; Dynarski, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2010\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, the distribution of institutional aid has grown increasingly strategic. Winston (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e1999\u003c/span\u003e) documented the rise of \"high-tuition, high-aid\" policies at selective institutions, while Ehrenberg (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2000\u003c/span\u003e) linked these policies to competition for rankings. McPherson and Schapiro (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e1998\u003c/span\u003e) showed that institutional aid increasingly flows to middle- and upper-income students rather than those with the highest need.\u003c/p\u003e \u003cp\u003eRecent work has examined algorithmic approaches to aid allocation. Phan et al. (2019, 2021) developed optimization models for financial aid distribution, but these assume institutions maximize enrollment yield or revenue\u0026mdash;objectives specified a priori rather than revealed from behavior. Similarly, Sugrue et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2006\u003c/span\u003e) applied linear programming to aid management decisions. However, all these studies employ \u003cb\u003eforward optimization approaches\u003c/b\u003e that require researchers to specify institutional objectives in advance, rather than inferring what institutions actually optimize from their observed behavior.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Institutional Competition and Strategy\u003c/h2\u003e \u003cp\u003eThe higher education market has become increasingly competitive, with institutions viewing students as both inputs to and consumers of education (Winston, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e1999\u003c/span\u003e). This dual role creates complex strategic considerations. Theories of positional competition (Hirsch, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e1976\u003c/span\u003e; Frank, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e1985\u003c/span\u003e) suggest institutions use resources primarily to improve relative standing rather than absolute educational quality.\u003c/p\u003e \u003cp\u003eEmpirical evidence supports strategic behavior. Monks and Ehrenberg (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e1999\u003c/span\u003e) demonstrated that rankings significantly influence institutional policies, while Bowman and Bastedo (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2009\u003c/span\u003e) showed how rankings create self-reinforcing cycles. However, these studies rely on correlation rather than revealing underlying objective functions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Inverse Optimization Theory\u003c/h2\u003e \u003cp\u003eInverse optimization, pioneered by Ahuja and Orlin (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2001\u003c/span\u003e), provides a framework for inferring objective functions from observed decisions. The key insight is that if decisions are optimal, they must maximize some objective function subject to constraints. By analyzing patterns in decisions, we can recover the implicit objectives.\u003c/p\u003e \u003cp\u003eThe approach has evolved from early linear programming applications (Heuberger, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2004\u003c/span\u003e) to handle non-linear objectives (Keshavarz et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2011\u003c/span\u003e) and multiple objectives (Roland et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Recent advances allow for noise in observations (Aswani et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) and partially observed constraints (Esfahani et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 The Forward vs. Inverse Optimization Distinction\u003c/h2\u003e \u003cp\u003eThe optimization literature on financial aid allocation has exclusively focused on \u003cb\u003eforward optimization\u003c/b\u003e approaches that assume known institutional objectives. Studies like Phan et al. (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and Sugrue et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2006\u003c/span\u003e) develop sophisticated models to help institutions optimize their aid distribution given stated goals such as maximizing enrollment, minimizing cost, or improving access metrics. While valuable for institutional planning, these approaches cannot reveal what objectives institutions actually pursue in practice.\u003c/p\u003e \u003cp\u003e \u003cb\u003eInverse optimization\u003c/b\u003e offers a fundamentally different perspective by working backward from observed decisions to infer the underlying objective functions. This distinction is crucial: forward optimization assumes we know what institutions want to achieve, while inverse optimization discovers what they actually optimize for. No prior study has applied inverse optimization techniques to analyze institutional behavior in higher education finance, creating the methodological gap this research addresses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Theoretical Predictions\u003c/h2\u003e \u003cp\u003eThree theoretical perspectives offer competing predictions:\u003c/p\u003e \u003cp\u003e \u003cb\u003eHuman Capital Theory\u003c/b\u003e (Becker, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e1964\u003c/span\u003e) predicts institutions should maximize social returns by investing in students with the highest marginal benefit\u0026mdash;typically those from disadvantaged backgrounds who lack alternative funding sources.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePositional Goods Theory\u003c/b\u003e (Hirsch, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e1976\u003c/span\u003e) predicts institutions will allocate aid to enhance relative prestige, focusing on students who improve institutional metrics rather than those with the greatest need.\u003c/p\u003e \u003cp\u003e \u003cb\u003eResource Dependence Theory\u003c/b\u003e (Pfeffer \u0026amp; Salancik, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e1978\u003c/span\u003e) suggests behavior reflects the need to secure resources from key stakeholders. Private institutions dependent on donations and rankings would thus optimize differently than public institutions dependent on state funding.\u003c/p\u003e \u003cp\u003eOur inverse optimization approach allows empirical testing of which theory best explains observed behavior.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Data and Institutional Context","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Data Sources\u003c/h2\u003e \u003cp\u003eThis study employs data from the Integrated Postsecondary Education Data System (IPEDS), the comprehensive federal data collection system administered by the National Center for Education Statistics. We construct our analytical dataset by merging three IPEDS survey components for academic year 2021-22:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eStudent Financial Aid survey (SFA2122): Institutional grant recipients and award amounts\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eInstitutional Characteristics survey (HD2022): Institutional control, enrollment size, and Carnegie classification\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFall Enrollment survey (EF2022): Student demographic characteristics and enrollment patterns\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eWe focus our analysis on 2021-22 as it represents the most recent academic year with complete financial aid reporting across all institution types. Our sample selection criteria exclude institutions with incomplete financial aid reporting and those enrolling fewer than 100 full-time equivalent undergraduates to ensure data quality and institutional viability. The resulting analytical sample comprises 3,130 degree-granting institutions, representing approximately 95% of total undergraduate enrollment in US higher education.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Variable Construction\u003c/h2\u003e \u003cp\u003eWe construct our analytical variables following established conventions in the higher education finance literature. Our dependent variables capture the key dimensions of institutional aid allocation decisions:\u003c/p\u003e \u003cp\u003e \u003cb\u003ePrimary Outcome Variables\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eAverage institutional grant per recipient (UAGRNTA): Mean need- and merit-based institutional aid awarded to recipients\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eInstitutional grant coverage rate (UAGRNTP/100): Proportion of enrolled students receiving institutional aid\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTotal institutional aid expenditure: Aggregate spending calculated as recipients \u0026times; average award\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eInstitutional Control Variables\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eInstitutional control: Public (n\u0026thinsp;=\u0026thinsp;1,552), Private non-profit (n\u0026thinsp;=\u0026thinsp;943), For-profit (n\u0026thinsp;=\u0026thinsp;635)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFederal grant recipient proportion (UPGRNTP): Percentage of students receiving federal need-based aid (Pell Grants), serving as our primary proxy for student financial need\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFull-time equivalent enrollment (SCUGFFN): Total undergraduate enrollment adjusted for part-time status\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eInstitutional-to-federal aid ratio: Relative emphasis on institutional versus federal aid programs\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Descriptive Evidence\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003epresents financial aid patterns by institution type:\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePublic (n\u0026thinsp;=\u0026thinsp;1,552)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrivate (n\u0026thinsp;=\u0026thinsp;943)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eFor-Profit (n\u0026thinsp;=\u0026thinsp;635)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAverage Institutional Grant\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e6,344\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e23,155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e5,114\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStd Dev\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e2,922\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e11,325\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e1,866\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e5,832\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e22,140\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e4,956\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCoverage Rate (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e67.9%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStd Dev\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15.3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19.4%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFederal Grant Recipients (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e32.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e33.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e58.5%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStd Dev\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13.4%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19.2%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eInstitutional/Federal Aid Ratio\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStd Dev\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.35\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTotal Aid per FTE Student\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e4,312\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e19,546\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e3,470\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStd Dev\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e2,456\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e12,331\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e1,654\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe patterns reveal striking differences. Private institutions provide awards 3.65 times larger than public institutions while serving similar proportions of federal aid recipients (33.9% vs 32.1%). The institutional-to-federal aid ratio at private institutions (4.73) far exceeds public (1.40) and for-profit (1.15) institutions, suggesting different optimization strategies.\u003c/p\u003e \u003cp\u003eFigures \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e provide comprehensive visual evidence of the fundamental differences in institutional aid strategies. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e reveals the core trade-off institutions face between coverage and generosity. Public institutions (green circles) cluster along the high-coverage, low-generosity frontier, optimizing to serve as many students as possible within budget constraints. Private institutions (orange squares) spread across the high-generosity space, with many achieving both high coverage and high awards\u0026mdash;a pattern only possible with substantially larger per-student aid budgets. For-profit institutions (blue triangles) concentrate in the moderate coverage, low-generosity region.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e exposes a crucial contradiction in the relationship between institutional resources and student need. For-profit institutions, despite having the smallest aid budgets, serve the neediest student populations with median federal aid coverage around 60%. In contrast, both public and private institutions serve similar proportions of high-need students (~\u0026thinsp;30%), yet deploy vastly different aid strategies. This pattern suggests that prestige considerations, rather than need-based targeting, drive private institution behavior.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e quantifies these strategic differences through distributional analysis. Private institutions provide aid amounts nearly four times larger than public institutions (median ~\u003cspan\u003e$\u003c/span\u003e22,000 vs ~\u003cspan\u003e$\u003c/span\u003e6,000) while also achieving superior coverage rates, with most private institutions exceeding 85% coverage compared to more variable coverage at public and for-profit institutions. Together, these patterns demonstrate that private institutions operate with fundamentally different resource constraints and optimization objectives than their counterparts.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Methodology: Inverse Optimization Framework","content":"\u003ch3\u003e4.1 Theoretical Model\u003c/h3\u003e\n\u003cp\u003eWe model institutional financial aid allocation as a constrained optimization problem where each institution maximizes a weighted combination of competing objectives subject to budget and operational constraints. The institutional decision-maker\u0026apos;s problem can be formalized as:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eForward Optimization Problem\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003emax \u0026Sigma;ₖ wₖ \u0026middot; fₖ(a, c, p)\u003cbr\u003e\u0026nbsp;a,c\u003c/p\u003e\n\u003cp\u003esubject to:\u003cbr\u003e\u0026nbsp;a \u0026middot; c \u0026middot; N \u0026le; B (budget constraint)\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;0 \u0026le; c \u0026le; 1 (coverage constraint)\u003c/p\u003e\n\u003cp\u003ea \u0026ge; 0 (non-negativity constraint)\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u0026Omega;(a, c) (institution-specific constraints)\u003c/p\u003e\n\u003cp\u003ewhere the decision variables are average aid amount per recipient (a) and coverage rate (c), defined as the proportion of enrolled students receiving institutional aid. The parameter vector includes the proportion of high-need students (p), measured as federal grant recipients; total undergraduate enrollment (N); and the institutional aid budget (B). The functions fₖ represent distinct institutional objectives, with weights wₖ capturing the relative importance assigned to each objective.\u003c/p\u003e\n\u003ch3\u003e4.2 Objective Function Specification\u003c/h3\u003e\n\u003cp\u003eDrawing from institutional mission statements and higher education finance theory, we specify four potential objectives that capture the primary dimensions of institutional aid allocation:\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCoverage Maximization:\u003c/strong\u003e f₁(c) = c\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e\u003cem\u003eEconomic interpretation:\u003c/em\u003e Maximize the proportion of students receiving aid\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eTheoretical foundation:\u003c/em\u003e Equity-oriented institutional missions emphasizing broad access\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003eNeed-Based Targeting:\u003c/strong\u003e f₂(c,p) = c \u0026middot; p\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e\u003cem\u003eEconomic interpretation:\u003c/em\u003e Maximize aid allocation to financially disadvantaged students\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eTheoretical foundation:\u003c/em\u003e Social mobility and human capital development objectives\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003ePrestige Signaling:\u003c/strong\u003e f₃(a) = a/ā, where ā = $30,000\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e\u003cem\u003eEconomic interpretation:\u003c/em\u003e Signal institutional quality through generous award amounts\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eTheoretical foundation:\u003c/em\u003e Positional competition theory and reputation-building strategies\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eNormalization rationale:\u003c/em\u003e The normalization constant ensures scale comparability across objectives while preserving relative differences in award generosity\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003eCost Efficiency:\u003c/strong\u003e f₄(a,c,N,B) = 1 - (a\u0026middot;c\u0026middot;N)/B\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003e\u003cem\u003eEconomic interpretation:\u003c/em\u003e Minimize aid expenditure relative to available budget\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eTheoretical foundation:\u003c/em\u003e Resource constraint optimization and financial sustainability\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3\u003e4.3 Inverse Optimization Approach\u003c/h3\u003e\n\u003cp\u003eGiven observed aid allocation decisions (a\u003cem\u003e, c\u003c/em\u003e) for each institution, we employ inverse optimization techniques to recover the implicit objective function weights w = [w₁, w₂, w₃, w₄] that rationalize these choices. Our approach integrates revealed preference theory with variance-based weight identification.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStep 1: Calculate Objective Values\u003c/strong\u003e\u003cbr\u003eFor each institution i, we calculate realized objective values:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003eCoverage: f₁ᵢ = cᵢ\u003c/li\u003e\n \u003cli\u003eNeed-based targeting: f₂ᵢ = cᵢ \u0026middot; pᵢ\u003c/li\u003e\n \u003cli\u003ePrestige signaling: f₃ᵢ = aᵢ / 30,000\u003c/li\u003e\n \u003cli\u003eCost efficiency: f₄ᵢ = 1 - (aᵢ\u0026middot;cᵢ\u0026middot;Nᵢ)/Bᵢ\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003eStep 2: Revealed Preference Analysis\u003c/strong\u003e\u003cbr\u003eWithin each institutional control type g, we apply revealed preference logic: objectives receiving higher optimization weight should exhibit greater cross-institutional variation as institutions make different trade-offs along these dimensions. We compute the sample variance \u0026sigma;ₖ\u0026sup2; and mean \u0026mu;ₖ for each objective k.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStep 3: Weight Recovery\u003c/strong\u003e\u003cbr\u003eFollowing the variance-proportionality principle, we recover institutional weights as:\u003c/p\u003e\n\u003cp\u003ewₖ = (\u0026sigma;ₖ\u0026sup2; \u0026middot; \u0026mu;ₖ) / \u0026Sigma;ⱼ(\u0026sigma;ⱼ\u0026sup2; \u0026middot; \u0026mu;ⱼ)\u003c/p\u003e\n\u003cp\u003eWhere \u0026sigma;ₖ\u0026sup2; is variance and \u0026mu;ₖ is mean of objective k. This specification ensures that objectives showing both high variation (indicating active optimization) and high mean achievement receive proportionally higher weights.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStep 4: Validation\u003c/strong\u003e\u003cbr\u003eWe validate recovered weights through multiple empirical tests: (i) cross-sectional regression analysis examining whether institution type predicts aid generosity after controlling for enrollment size, (ii) within-type heterogeneity analysis testing whether private institutions with different aid strategies serve systematically different student populations, (iii) bootstrap resampling with 1,000 replications to assess statistical precision and generate confidence intervals for key institutional metrics, and (iv) for each institution we compute the coefficient of complementarity (\u0026rho;; Chan et al., 2019) to quantify how closely observed allocations satisfy the KKT conditions: \u0026rho; ranges from 0 (no fit) to 1 (perfect optimality).\u003c/p\u003e"},{"header":"5. Results","content":"\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Revealed Institutional Objectives\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003epresents the recovered objective weights by institution type:\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObjective\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eFor-Profit\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCoverage\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.597 (59.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.298 (29.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.434 (43.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNeed Focus\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.201 (20.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.150 (15.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.277 (27.7%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePrestige\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.179 (17.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.548 (54.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.105 (10.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCost Efficiency\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.023 (2.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.003 (0.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.185 (18.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eInterpretation\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCoverage-focused\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrestige-focused\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBalanced\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePrimary Trade-off\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCoverage vs. Budget\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrestige vs. Access\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNeed vs. Cost\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e \u003cb\u003epresents the recovered objective weights from the inverse optimization analysis.\u003c/b\u003e The results demonstrate significant variation in optimization strategies across institution types, with each sector exhibiting distinct priority structures.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePrivate institutions exhibit the highest weight on prestige signaling (54.8%).\u003c/b\u003e This objective receives substantially greater emphasis than coverage (29.8%) or need-based targeting (15.0%). Cost efficiency receives minimal weight (0.3%), indicating limited consideration of spending constraints in aid allocation decisions.\u003c/p\u003e \u003cp\u003e \u003cb\u003ePublic institutions prioritize coverage above other objectives (59.7%).\u003c/b\u003e This emphasis on broad student access is accompanied by moderate attention to need-based targeting (20.1%) and prestige signaling (17.9%). The relatively balanced approach across coverage and need-based objectives suggests optimization under resource constraints.\u003c/p\u003e \u003cp\u003e \u003cb\u003eFor-profit institutions demonstrate a more balanced objective structure.\u003c/b\u003e Coverage (43.4%) and need-based targeting (27.7%) receive comparable emphasis, while this sector shows the highest weight on cost efficiency (18.5%) among all institution types. This pattern indicates greater attention to operational efficiency in aid allocation.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe recovered weights reveal systematic differences in institutional priorities. Private institutions optimize primarily for prestige outcomes, public institutions emphasize access maximization, and for-profit institutions incorporate efficiency considerations more prominently than other sectors. These differences suggest that institutional type serves as a strong predictor of aid allocation strategy.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Statistical Validation\u003c/h2\u003e \u003cp\u003eTo validate our inverse optimization findings, we conduct two empirical tests examining whether the recovered objectives predict institutional behavior.\u003c/p\u003e \u003cdiv id=\"Sec19\" class=\"Section3\"\u003e \u003ch2\u003e5.2.1 Cross-Sectional Analysis\u003c/h2\u003e \u003cp\u003eWe first test whether institution type predicts aid generosity, controlling for institutional size. Estimating the model:\u003c/p\u003e \u003cp\u003elog(aid_amount) = β₀ + β₁Private\u0026thinsp;+\u0026thinsp;β₂ForProfit\u0026thinsp;+\u0026thinsp;β₃log(enrollment) + ε\u003c/p\u003e \u003cp\u003eyields the following results:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003ePrivate institution coefficient: β₁ = 1.323 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eFor-profit coefficient: β₂ = -0.046 (p\u0026thinsp;\u0026gt;\u0026thinsp;0.05)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eModel R\u0026sup2; = 0.665\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe private institution coefficient of 1.323 indicates that private institutions provide 275% more aid per recipient than public institutions (e^1.323\u0026thinsp;\u0026asymp;\u0026thinsp;3.75) after controlling for enrollment size. This substantial difference, explaining 66.5% of variation in aid amounts, confirms that institution type fundamentally determines aid allocation strategy. The non-significant for-profit coefficient suggests these institutions behave similarly to publics in terms of award size, consistent with their balanced objective weights.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e \u003ch2\u003e5.2.2. Within-Type Heterogeneity\u003c/h2\u003e \u003cp\u003eTo examine whether prestige optimization varies within private institutions, we partition them into quartiles based on average aid amounts and analyze their characteristics:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAid Quartile\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAverage Award\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCoverage Rate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eFederal Aid Recipients\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1 (Lowest)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e9,887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e49%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e19,056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e87%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e34%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e25,283\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e90%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e31%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4 (Highest)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e38,401\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eA striking pattern emerges: institutions providing the largest awards serve substantially fewer high-need students. The percentage of federal aid recipients decreases monotonically from 49% in the lowest quartile to 23% in the highest quartile\u0026mdash;a 26 percentage point decline. This inverse relationship between award generosity and student need provides additional evidence that high awards serve prestige rather than access objectives. Institutions in the top quartile effectively purchase prestige by concentrating resources on students who require the least financial assistance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e visualizes these patterns across the three key dimensions of aid allocation. The top panel demonstrates the dramatic escalation in average awards, with the most generous institutions providing nearly four times larger awards than the least generous. The middle panel reveals coverage rates that peak in the third quartile before declining as the highest-award institutions become more selective about recipients. Most critically, the bottom panel confirms the systematic decline in federal aid recipients as award generosity increases. This within-type variation demonstrates that even among private institutions sharing similar organizational structures and missions, those that weight prestige objectives more heavily systematically serve different student populations than those prioritizing access or coverage. The heterogeneity in revealed preferences within the private sector thus provides additional validation of our inverse optimization approach by showing that objective weights vary meaningfully even within institutional types.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section3\"\u003e \u003ch2\u003e5.2.3. Goodness-of-Fit via Complementarity\u003c/h2\u003e \u003cp\u003eThe coefficient of complementarity confirms that the recovered objectives explain observed behavior. Median ρ across all 3,130 institutions equals 0.65, indicating allocations lie within 35% of perfect optimality under the recovered weights. ρ varies systematically by sector \u0026mdash;private\u0026thinsp;=\u0026thinsp;0.726, public\u0026thinsp;=\u0026thinsp;0.647, for-profit\u0026thinsp;=\u0026thinsp;0.608 \u0026mdash; with private institutions exhibiting the tightest fit to their prestige-focused objective. This level of fit is well within the range typically reported in inverse-optimisation studies involving real-world human or institutional behavior, where values between 0.6 and 0.7 are considered strong (Chan, Lee \u0026amp; Terekhov, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInstitution Type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian ρ\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInterpretation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.726\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e73% optimality\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.647\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e65% optimality\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFor-Profit\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.608\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e61% optimality\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOverall\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e0.650\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKey model-fit statistic\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Heterogeneity Analysis\u003c/h2\u003e \u003cp\u003eOur aggregate findings may mask important variation within institution types. We examine heterogeneity along two dimensions: geographic region for private institutions and institutional size for public institutions.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003e5.3.1. Regional Patterns in Private Institution Behavior\u003c/h2\u003e \u003cp\u003eAnalysis of private institutions by region reveals systematic geographic variation in aid strategies:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabc\" border=\"1\"\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAverage Award\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInst/Fed Ratio\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNew England\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e30,642\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6.19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e89\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFar West\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e25,944\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMid East\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e25,408\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e207\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGreat Lakes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e23,512\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.92\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e156\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePlains\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e22,037\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e104\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSoutheast\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e20,381\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e222\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRocky Mountains\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e19,568\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSouthwest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e18,244\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e58\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe geographic gradient is pronounced: New England institutions provide average awards 68% higher than those in the Southwest (\u003cspan\u003e$\u003c/span\u003e30,642 vs. \u003cspan\u003e$\u003c/span\u003e18,244). The institutional-to-federal aid ratio similarly varies from 6.19 in New England to 3.73 in the Southwest, indicating that competitive dynamics shape optimization strategies.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e visualizes this geographic gradient through a dual-axis presentation showing that average awards (bars) and institutional-to-federal aid ratios (line) follow remarkably similar patterns across regions. Both metrics peak in New England and decline systematically toward the Southwest. This correlation between award generosity and relative emphasis on institutional versus federal aid suggests systematic regional differences in aid allocation strategies. The geographic clustering of similar behaviors is consistent with positional goods theory's prediction that competitive dynamics influence institutional strategy, though the underlying mechanisms driving these regional patterns warrant further investigation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section3\"\u003e \u003ch2\u003e5.3.2. Size and Public Institution Objectives\u003c/h2\u003e \u003cp\u003eAmong public institutions, we examine how institutional scale affects revealed objectives:\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabd\" border=\"1\"\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSize Category\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInstitutions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCoverage Rate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAverage Award\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmall (\u0026lt;\u0026thinsp;5,000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1,504\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e68.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e6,184\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedium (5,000\u0026ndash;20,000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e70.4%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cspan\u003e$\u003c/span\u003e11,340\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLarge (\u0026gt;\u0026thinsp;20,000)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe data reveal that medium-sized public institutions achieve modestly higher coverage rates (70.4% vs. 68.2%) while providing substantially larger awards (\u003cspan\u003e$\u003c/span\u003e11,340 vs. \u003cspan\u003e$\u003c/span\u003e6,184). This pattern suggests that economies of scale in aid administration allow larger institutions to pursue coverage objectives more effectively. However, the limited number of medium-sized institutions (n\u0026thinsp;=\u0026thinsp;48) and complete absence of large public institutions (\u0026gt;\u0026thinsp;20,000 students) in our sample significantly constrains analysis of size effects. The directional relationship observed between the two size categories supports our interpretation that operational capacity influences objective weights, though definitive conclusions require broader institutional representation.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Robustness Analysis\u003c/h2\u003e \u003cp\u003eWe conduct three robustness tests to assess the stability and statistical reliability of our findings.\u003c/p\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e5.4.1 Statistical Inference Through Bootstrap Resampling\u003c/h2\u003e \u003cp\u003eTo quantify uncertainty in our revealed objective estimates, we implement bootstrap resampling with 1,000 replications. For each iteration, we resample institutions with replacement and recalculate key metrics:\u003c/p\u003e \u003cp\u003e \u003cb\u003ePrivate Institutions\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eInstitutional/Federal aid ratio: 4.734 [95% CI: 4.591, 4.865]\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eStandard error: 0.070\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003ePublic Institutions\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eCoverage rate: 0.682 [95% CI: 0.672, 0.693]\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eStandard error: 0.005\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe narrow confidence intervals indicate precise estimation of institutional behavior. The private institution confidence interval for the inst/fed ratio (4.591, 4.865) lies entirely above the public institution mean of 1.40, confirming that the differential optimization strategies are not artifacts of sampling variation. Bootstrap distributions show no evidence of multimodality, suggesting our revealed preference approach identifies unique, stable objective functions rather than averaging across multiple equilibria.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003e5.4.2 Hypothesis Testing for Differential Objectives\u003c/h2\u003e \u003cp\u003eTo formally test whether private and public institutions pursue different objectives, we compare their institutional-to-federal aid ratios using a two-sample t-test:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eTest statistic: t\u0026thinsp;=\u0026thinsp;56.262\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDegrees of freedom: 2,493\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ep-value\u0026thinsp;\u0026lt;\u0026thinsp;0.000001\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe extreme test statistic provides overwhelming evidence against the null hypothesis of equal objectives. The probability of observing such different aid strategies if institutions truly shared common objectives is effectively zero. This statistical significance, combined with the economic magnitude of the differences (private ratio is 3.38 times the public ratio), confirms that institution types fundamentally differ in their optimization strategies.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e6\u003c/span\u003e provides intuitive visual evidence for the statistical differences documented in our hypothesis testing between private and public institutions. The dramatic separation between these two institution types is immediately apparent: public institutions (blue) cluster tightly around a median ratio of 1.4, with most institutions falling between 1.0 and 2.0. In stark contrast, private institutions (orange) show both higher central tendency and greater dispersion, with the majority of institutions operating at ratios between 3.0 and 6.0\u0026mdash;entirely above the public distribution. This complete separation of distributions explains the extreme t-statistic (56.262) in our formal hypothesis test and demonstrates that the differential optimization strategies between private and public institutions are not marginal differences but fundamental distinctions in institutional behavior.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"6. Policy Implications","content":"\u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e6.1 Transparency Through Mathematical Clarity\u003c/h2\u003e \u003cp\u003ePerhaps the most significant contribution of this analysis is providing mathematical clarity to previously opaque institutional decisions. By recovering objective weights through inverse optimization, we transform abstract discussions of \"institutional priorities\" into concrete, quantifiable metrics.\u003c/p\u003e \u003cp\u003eInstitutional leaders can now answer specific questions: What percentage of our aid optimization focuses on prestige versus need? How does our revealed objective function compare to peer institutions? What would be the concrete impact of shifting our allocation weights?\u003c/p\u003e \u003cp\u003eThis transparency extends to other stakeholders as well. Students and families can better understand the forces shaping aid offers. Donors can assess whether their contributions support stated institutional missions. Policymakers can evaluate whether current incentive structures promote desired social outcomes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e6.2 A Tool for Institutional Analysis\u003c/h2\u003e \u003cp\u003eThe methodology developed here offers institutions a tool for self-examination. By applying inverse optimization to their own historical aid data, institutions can uncover their revealed priorities and assess alignment with stated missions. This approach moves beyond anecdotal evidence or good intentions to provide rigorous, quantitative insight into actual institutional behavior.\u003c/p\u003e \u003cp\u003eFor institutions genuinely committed to expanding access, our findings highlight the magnitude of change required. Moving from prestige-dominant optimization (54.8% weight) to need-focused allocation would represent fundamental transformation rather than incremental adjustment.\u003c/p\u003e \u003c/div\u003e"},{"header":"7. Discussion","content":"\u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003e7.1 Theoretical Implications\u003c/h2\u003e \u003cp\u003eOur empirical findings provide strong evidence for the applicability of Positional Goods Theory to understand financial aid allocation in higher education. The recovered objective weights reveal that private institutions allocate 54.8% of their optimization effort toward prestige enhancement, mathematically confirming what Hirsch (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e1976\u003c/span\u003e) theorized about competition for relative position. This finding challenges the prevailing assumption that educational institutions, particularly those with substantial endowments and expressed commitments to access, operate according to Human Capital Theory's prescription of maximizing social returns.\u003c/p\u003e \u003cp\u003eThe contrast between institution types proves particularly illuminating. Public institutions, with 59.7% weight on coverage and only 17.9% on prestige, demonstrate behavior more aligned with their statutory missions and public accountability structures. This differential supports Resource Dependence Theory's prediction that institutional behavior reflects stakeholder constraints. For-profit institutions present an unexpected finding: despite their profit motive, they show greater balance between coverage (43.4%) and need-focus (27.7%) than private non-profits, suggesting that market pressures may sometimes align better with social objectives than philanthropic governance structures.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec33\" class=\"Section2\"\u003e \u003ch2\u003e7.2 Methodological Contributions\u003c/h2\u003e \u003cp\u003eThis study advances the application of inverse optimization methods to social science questions. While inverse optimization has proven valuable in engineering and operations research contexts, its application to institutional behavior in education opens new methodological possibilities. Our approach differs fundamentally from traditional econometric methods that require researchers to specify behavioral models a priori. Instead, we allow the data to reveal the objective function, providing a more inductive path to understanding institutional priorities.\u003c/p\u003e \u003cp\u003eThe revealed preference framework adapted here addresses a persistent challenge in studying organizational behavior: the gap between stated and actual objectives. By mathematically recovering the weights institutions place on competing goals, we move beyond reliance on mission statements, strategic plans, or administrator surveys. This methodology could extend to numerous educational contexts where institutions face multi-objective optimization problems, from admissions decisions to resource allocation across academic programs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec34\" class=\"Section2\"\u003e \u003ch2\u003e7.3 Limitations\u003c/h2\u003e \u003cp\u003eSeveral limitations warrant consideration when interpreting our findings. First, the aggregated nature of IPEDS data prevents observation of individual student aid packages, potentially masking within-institution heterogeneity in allocation strategies. Institutions may apply different objectives to different student segments, a nuance our institution-level analysis cannot capture.\u003c/p\u003e \u003cp\u003eSecond, our model assumes institutions freely optimize their stated objectives, but unobserved constraints may influence allocation decisions. Restricted donations, state-mandated aid programs, and historical commitments create path dependencies not reflected in our optimization framework. While our robustness checks suggest these constraints do not fundamentally alter the revealed objectives, they may explain some variation in institutional behavior.\u003c/p\u003e \u003cp\u003eThird, our analysis focuses on absolute dollar amounts rather than aid as a proportion of institutional costs. While institutional grants represent different proportions of total costs across institution types, our focus on revealed optimization priorities through aid allocation patterns captures fundamental institutional behavior regardless of underlying cost structures.\u003c/p\u003e \u003cp\u003eFourth, our cross-sectional analysis captures a snapshot of institutional behavior rather than dynamic evolution. Market pressures, leadership changes, and policy interventions may alter institutional priorities over time scales our data cannot observe.\u003c/p\u003e \u003cp\u003eFinally, the functional form of objective specifications influences results, though our robustness tests using alternative formulations yield qualitatively similar findings. The choice to normalize prestige by \u003cspan\u003e$\u003c/span\u003e30,000 and cost efficiency by budget size reflects reasonable but ultimately subjective scaling decisions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec35\" class=\"Section2\"\u003e \u003ch2\u003e7.4 Future Research\u003c/h2\u003e \u003cp\u003eThis work establishes a foundation for multiple research trajectories. Longitudinal analysis could trace how institutional objectives evolve in response to market conditions, policy changes, or competitive pressures. The COVID-19 pandemic's disruption to higher education financing provides a natural experiment for studying objective stability under stress.\u003c/p\u003e \u003cp\u003eAccess to student-level data would enable more nuanced inverse optimization, potentially revealing how institutions apply different objectives to different populations. Such analysis could test whether the aggregate patterns we observe result from uniform policies or averaged heterogeneous strategies.\u003c/p\u003e \u003cp\u003eComparative international research could examine whether the prestige optimization we document reflects uniquely American competitive dynamics or broader patterns in marketized higher education systems. Countries with different financing models and governance structures provide natural contrasts for testing the generalizability of our findings.\u003c/p\u003e \u003cp\u003eExperimental or quasi-experimental studies of institutions that have reformed their aid policies could validate whether stated changes in priorities translate to shifted objective weights in practice. Such research would move beyond revealing current behavior to understanding the conditions under which institutions can successfully transform their allocation strategies.\u003c/p\u003e \u003cp\u003eThe methodological framework developed here invites application to other domains where institutions balance multiple objectives. From hospital resource allocation to public school funding formulas, inverse optimization offers a tool for revealing the true priorities governing institutional decisions that affect social welfare.\u003c/p\u003e \u003c/div\u003e"},{"header":"8. Conclusion","content":"\u003cp\u003eThis study provides the first mathematical evidence that private US universities optimize financial aid for prestige rather than access or need. Our analysis of 3,130 institutions reveals that private universities allocate 54.8% weight on prestige signaling while dedicating only 15.0% to need-based targeting\u0026mdash;a nearly 4:1 ratio that stands in stark contrast to public institutions, which prioritize coverage (59.7%) over prestige (17.9%). The inverse relationship we document between award generosity and student need among private institutions demonstrates that prestige optimization fundamentally conflicts with access objectives, quantifying patterns that prior research suggested but never measured precisely.\u003c/p\u003e \u003cp\u003eThe significance extends beyond higher education to organizational behavior more broadly. By demonstrating that institutional actions can reveal true priorities regardless of stated missions, this research offers a methodological foundation for understanding decision-making in any context where organizations balance multiple competing objectives.\u003c/p\u003e \u003cp\u003eAs higher education faces increasing scrutiny over affordability and equity, this research contributes quantitative evidence about institutional aid priorities that can inform policy discussions and institutional self-reflection about the \u003cspan\u003e$\u003c/span\u003e185\u0026nbsp;billion distributed annually in financial aid.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data used in this study are derived from publicly available IPEDS datasets for the 2021\u0026ndash;22 academic year. The processed dataset\u0026mdash;including recovered objective weights and validation metrics for 3,130 institutions\u0026mdash;along with full replication code and figures, is archived at:\u003c/p\u003e\n\u003cp\u003ehttps://doi.org/10.5281/zenodo.15496894\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAhuja, R. K., \u0026amp; Orlin, J. B. (2001). Inverse optimization. Operations Research, 49(5), 771-783.\u003c/li\u003e\n\u003cli\u003eAswani, A., Shen, Z. J. M., \u0026amp; Siddiq, A. (2018). Inverse optimization with noisy data. Operations Research, 66(3), 870-892.\u003c/li\u003e\n\u003cli\u003eAvery, C., \u0026amp; Hoxby, C. M. (2004). Do and should financial aid packages affect students\u0026apos; college choices? In College choices: The economics of where to go, when to go, and how to pay for it (pp. 239-302). University of Chicago Press.\u003c/li\u003e\n\u003cli\u003eBecker, G. S. (1964). Human capital: A theoretical and empirical analysis, with special reference to education. University of Chicago Press.\u003c/li\u003e\n\u003cli\u003eBertsimas, D., Gupta, V., \u0026amp; Paschalidis, I. C. (2015). Data-driven estimation in equilibrium using inverse optimization. Mathematical Programming, 153(2), 595-633.\u003c/li\u003e\n\u003cli\u003eBettinger, E. (2015). Need-based aid and college persistence: The effects of the Ohio College Opportunity Grant. Educational Evaluation and Policy Analysis, 37(1_suppl), 102S-119S.\u003c/li\u003e\n\u003cli\u003eBourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education (pp. 241-258). Greenwood.\u003c/li\u003e\n\u003cli\u003eBowman, N. A., \u0026amp; Bastedo, M. N. (2009). Getting on the front page: Organizational reputation, status signals, and the impact of US News and World Report on student decisions. Research in Higher Education, 50(5), 415-436.\u003c/li\u003e\n\u003cli\u003eChan, T. C., Lee, T., \u0026amp; Terekhov, D. (2019). Inverse optimization: Closed-form solutions, geometry, and goodness of fit. Management Science, 65(3), 1115-1135.\u003c/li\u003e\n\u003cli\u003eDeming, D., \u0026amp; Dynarski, S. (2010). College aid. In Targeting investments in children: Fighting poverty when resources are limited (pp. 283-302). University of Chicago Press.\u003c/li\u003e\n\u003cli\u003eDynarski, S., \u0026amp; Scott-Clayton, J. (2013). Financial aid policy: Lessons from research. The Future of Children, 23(1), 67-91.\u003c/li\u003e\n\u003cli\u003eEhrenberg, R. G. (2000). Tuition rising: Why college costs so much. Harvard University Press.\u003c/li\u003e\n\u003cli\u003eEsfahani, P. M., Shafieezadeh-Abadeh, S., Hanasusanto, G. A., \u0026amp; Kuhn, D. (2018). Data-driven inverse optimization with imperfect information. Mathematical Programming, 167(1), 191-234.\u003c/li\u003e\n\u003cli\u003eFrank, R. H. (1985). Choosing the right pond: Human behavior and the quest for status. Oxford University Press.\u003c/li\u003e\n\u003cli\u003eGoldrick-Rab, S. (2016). Paying the price: College costs, financial aid, and the betrayal of the American dream. University of Chicago Press.\u003c/li\u003e\n\u003cli\u003eHeuberger, C. (2004). Inverse combinatorial optimization: A survey on problems, methods, and results. Journal of Combinatorial Optimization, 8(3), 329-361.\u003c/li\u003e\n\u003cli\u003eHirsch, F. (1976). Social limits to growth. Harvard University Press.\u003c/li\u003e\n\u003cli\u003eKeshavarz, A., Wang, Y., \u0026amp; Boyd, S. (2011). Imputing a convex objective function. In 2011 IEEE International Symposium on Intelligent Control (pp. 613-619).\u003c/li\u003e\n\u003cli\u003eMa, J., Pender, M., \u0026amp; Welch, M. (2022). \u003cem\u003eTrends in college pricing and student aid 2022\u003c/em\u003e. College Board. https://research.collegeboard.org/media/pdf/trends-in-college-pricing-student-aid-2022.pdf\u003c/li\u003e\n\u003cli\u003eMcPherson, M. S., \u0026amp; Schapiro, M. O. (1998). The student aid game: Meeting need and rewarding talent in American higher education. Princeton University Press.\u003c/li\u003e\n\u003cli\u003eMonks, J., \u0026amp; Ehrenberg, R. G. (1999). The impact of US News and World Report college rankings on admission outcomes and pricing decisions at selective private institutions (No. w7227). National Bureau of Economic Research.\u003c/li\u003e\n\u003cli\u003ePfeffer, J., \u0026amp; Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. Harper \u0026amp; Row.\u003c/li\u003e\n\u003cli\u003ePhan, Vinhthuy; Wright, Laura; \u0026amp; Decent, Bridgette. (2022). Optimizing Financial Aid Allocation to Improve Access and Affordability to Higher Education. \u003cem\u003eJournal of Educational Data Mining\u003c/em\u003e, 14(3), 26\u0026ndash;51.\u003c/li\u003e\n\u003cli\u003eRoland, J., De Smet, Y., \u0026amp; Figueira, J. R. (2013). Inverse multi-objective combinatorial optimization. Discrete Applied Mathematics, 161(16-17), 2764-2771.\u003c/li\u003e\n\u003cli\u003eSaez-Gallego, J., Morales, J. M., Zugno, M., \u0026amp; Madsen, H. (2016). A data-driven bidding model for a cluster of price-responsive consumers of electricity. IEEE Transactions on Power Systems, 31(6), 5001-5011.\u003c/li\u003e\n\u003cli\u003eSugrue, P. K., Mehrotra, A., \u0026amp; Orehrovec, P. M. (2006). Financial aid management: an optimisation approach. International Journal of Operational Research, 1(3), 347-362.\u003c/li\u003e\n\u003cli\u003eWinston, G. C. (1999). Subsidies, hierarchy and peers: The awkward economics of higher education. Journal of Economic Perspectives, 13(1), 13-36.\u003c/li\u003e\n\u003cli\u003eZhang, L. (2013). Effects of college educational debt on graduate school attendance and early career and lifestyle choices. Education Economics, 21(2), 154-175.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"financial aid, higher education policy, institutional behavior, equity in access, inverse optimization","lastPublishedDoi":"10.21203/rs.3.rs-6778626/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6778626/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study uses inverse optimization to uncover the implicit objectives driving financial aid allocation decisions across U.S. higher education institutions. Analyzing data from 3,130 institutions in the 2021\u0026ndash;22 IPEDS dataset, we estimate the strategic weight each institution places on four competing goals: access (coverage), equity (need-based targeting), prestige, and cost control. Private nonprofit universities prioritize prestige (54.8%), awarding an average of \u003cspan\u003e$\u003c/span\u003e23,155 per student; public universities emphasize broad coverage (59.7%) with more moderate awards; and for-profit institutions adopt mixed strategies, balancing access and need. By recovering these revealed institutional priorities from observed behavior, we provide the first large-scale quantitative evidence that financial aid is frequently used to enhance institutional reputation rather than expand access. These findings highlight the equity implications of strategic aid allocation and offer a methodological framework to evaluate whether institutional priorities align with access goals.\u003c/p\u003e","manuscriptTitle":"Revealing Objectives in University Financial Aid: An Inverse Optimization Approach","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-30 10:01:00","doi":"10.21203/rs.3.rs-6778626/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"166c5a4a-edce-4a54-b2c0-c1fd9b08bba2","owner":[],"postedDate":"May 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":49247481,"name":"Decision Sciences"}],"tags":[],"updatedAt":"2025-05-30T10:01:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-30 10:01:00","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6778626","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6778626","identity":"rs-6778626","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}