Credit Constraints & Human Capital

The paper begins by establishing the longstanding recognition of credit constraints as a significant barrier to human capital investment, particularly in higher education. Because human capital (education and skills) is not easily repossessed as collateral, securing loans to finance education is challenging, especially for young people without established credit histories. While government intervention through student loan programs has helped, these programs often provide limited credit, leaving credit constraints a persistent influence on educational decisions. The authors note that existing models often make simplistic assumptions about credit constraints (either rising interest rates with loan size or fixed maximum loan amounts), ignoring the crucial link between borrowing and investment decisions. This link, however, is central to both government student loan programs (GSL) and private lending practices. The paper aims to address these limitations by examining the interaction between the design of student loan programs, the incentives of private lenders, and student investment behavior to better explain observed patterns in human capital investment and college attendance.

This section presents empirical evidence illustrating the relationship between family income, ability (measured by AFQT scores), and college attendance, using data from the National Longitudinal Survey of Youth (NLSY79 and NLSY97). The authors reference previous research (Belley and Lochner 2007) showing a weak correlation between family income and college attendance in the early 1980s but a significantly stronger positive relationship in the early 2000s. This shift is attributed to increasingly binding credit constraints due to rising tuition costs and returns to schooling. Supporting this observation, data from the U.S. Department of Education indicate a substantial increase in the percentage of undergraduate borrowers reaching maximum loan limits under federal student loan programs (from 18% in 1989-90 to 52% in 1999-2000). Concurrently, private student loan borrowing dramatically increased, reaching almost $14 billion in the 2004-05 academic year. This growth in private lending, while seemingly contradictory, is analyzed within the paper's framework as a market response to heightened demand for educational financing driven by the rising financial rewards of higher education.

The core of the paper lies in comparing two models of credit constraints: exogenous and endogenous. The standard exogenous model assumes either increasing interest rates or a fixed borrowing limit, failing to capture the interconnectedness between credit availability and investment decisions seen in GSL programs and private lending. The authors demonstrate that the exogenous model predicts a counterfactual negative relationship between ability and human capital investment for low-income youth when the consumption intertemporal elasticity of substitution (IES) is less than one (a common empirical finding). The paper introduces a framework that incorporates endogenous credit constraints, originating from the structure of both GSL programs and private lenders' responses to default risk. GSL programs directly link credit to investment (borrowing is limited to educational expenses), and private lenders, considering repayment incentives, similarly link credit limits to investment and individual characteristics that influence earnings potential. This endogenous model generates more realistic relationships between ability, investment, and family resources, overcoming the limitations of the exogenous approach.

To test the empirical validity of their model with endogenous credit constraints, the authors extend their framework to a multi-period lifecycle model, incorporating government education subsidies. They calibrate the model using US data on schooling, ability, government subsidies, and post-school earnings. This allows them to simulate the effects of rising returns to schooling and increasing tuition costs observed during the period under study. By comparing model predictions to empirical data from the NLSY79 and NLSY97 cohorts, they demonstrate that the model accurately reproduces the observed cross-sectional patterns of investment, ability, and wealth. In contrast, a model with exogenous credit constraints fails to achieve this fit. This quantitative analysis strongly supports the paper's central argument that endogenous credit constraints are essential for understanding the complex interplay between family income, ability, and college attendance in the face of changing economic conditions and evolving credit markets.

The paper details the key features of US Government Student Loan Programs (GSL), emphasizing their limitations as a source of educational financing. A crucial characteristic is the direct link between loan availability and educational investment; students can borrow only to cover college-related expenses, preventing the use of GSL funds for non-educational purposes. This direct tie between credit and investment is a pivotal aspect of the model. Furthermore, GSL programs establish both pre-specified maximum loan amounts (a crucial parameter affecting affordability) and endogenous limits tied to actual spending on education. The existence of these upper limits contributes to credit constraints even within the GSL system. The study points out that GSL loans usually have more extensive enforcement provisions than typical unsecured private loans, a factor that impacts default rates and, consequently, the willingness of private lenders to extend credit. The specific programs mentioned are the Stafford and PLUS loan programs, along with the Perkins Loan Program, which although providing funds to needy students, only constitutes a small fraction of total GSL disbursements. The PLUS loan program, noteworthy for its lack of a fixed maximum borrowing limit after 1993-94 (although limited by total college cost less other financial aid), highlights the evolving structure of GSL lending over time.

A significant portion of the paper focuses on the dramatic expansion of the private student loan market. The paper documents the remarkable growth of non-federal student loans, from $1.3 billion in 1995-96 to almost $14 billion in 2004-05, representing nearly 20% of all student loans disbursed. This surge is interpreted as evidence that GSL limits are no longer sufficient to meet the growing demand for educational funding. Private loans, while often mirroring the basic structure of GSL loans (e.g., limits tied to school costs or overall borrowing caps), differ significantly in interest rates (typically higher than GSL loans) and eligibility requirements. The expansion in private lending is attributed to several factors: rising returns to schooling, increasing tuition costs, and relatively stable real borrowing limits under GSL programs. Private loans are often used after available GSL funds are exhausted, indicating that GSL programs are frequently insufficient. The paper notes that private lending is particularly prevalent among graduate students (especially in professional schools) and undergraduates attending expensive private universities, providing insights into the specific segments of the student population most impacted by constraints.

The analysis delves into the mechanisms by which private lenders determine student credit levels. While human capital itself cannot be directly seized by lenders, various methods exist to punish borrowers who default (lowering credit scores, seizing assets, garnishing wages). The efficacy of these penalties is tied to the borrower's post-school earnings, which are, in turn, influenced by ability and educational investment. This observation is used to explain the increasing reliance of private lenders on both investment and ability to assess creditworthiness and determine credit limits. Higher ability students, who invest more in education, will be offered more credit due to their increased ability to repay given the consequences of default. This is significant because the link between credit limits, ability, and investment is essential to generating a positive relationship between ability and investment, a pattern frequently seen empirically but not easily explained using traditional exogenous constraint models. The authors position their model of private lending within the broader literature on endogenous credit constraints, highlighting the similarities with models focused on risk-sharing, asset prices, and firm dynamics.

The study utilizes data from the National Longitudinal Survey of Youth (NLSY79 and NLSY97) to examine the relationship between family income, ability (measured by AFQT scores), and college attendance. A key finding highlights the increased importance of family income in determining college attendance in the early 2000s (NLSY97) compared to the early 1980s (NLSY79). This observation supports the hypothesis that credit constraints have become more binding over time, as rising tuition costs and returns to schooling exacerbate the financial burden of college. The authors note that, controlling for ability and family background, youth from high-income families in the NLSY97 were sixteen percentage points more likely to attend college than those from low-income families—nearly double the effect observed in the NLSY79. This difference, which rises to almost 30 percentage points when considering both income and wealth, underscores the growing influence of family resources on college access. The analysis further explores how ability affects college attendance among different family income groups, consistently revealing a strong positive correlation between ability and college attendance across all income levels in both datasets, a pattern that the paper's model aims to explain.

This section contrasts the predictions of models employing exogenous versus endogenous credit constraints. The standard exogenous model, which assumes either rising interest rates with borrowing or a fixed maximum loan amount, is shown to yield a problematic prediction: a negative relationship between ability and human capital investment for constrained borrowers when the intertemporal elasticity of substitution (IES) is less than one—a condition supported by most empirical estimates. This counterfactual result contrasts sharply with the robust empirical observation of a positive relationship between ability and college attendance across all income levels. In contrast, the endogenous model, which incorporates constraints derived from both GSL program design and private lenders’ responses to default risk, avoids this problematic prediction. By linking credit limits to both investment levels and observable individual characteristics influencing returns, this model generates a more realistic positive relationship between ability and investment, even with an IES less than one. This difference in predictions is central to the paper's argument for a more nuanced understanding of credit constraints in human capital investment.

The paper briefly surveys the existing literature on credit constraints and human capital investment, highlighting some of the key findings and limitations of previous studies. The authors cite research by Becker (1975) as foundational, emphasizing the underinvestment in human capital by youth lacking family resources and adequate credit. They also mention several studies that have examined the relationship between family income and college attendance (Manski and Wise 1983; Cameron and Heckman 1998, 2001; Ellwood and Kane 2000; Carneiro and Heckman 2002; Belley and Lochner 2007), but the authors note that some of these prior works either found a small effect of family income after controlling for other factors (Cameron and Heckman 1998, 1999; Carneiro and Heckman 2002) or found little evidence of borrowing constraints affecting educational attainment (Cameron and Taber 2004; Keane and Wolpin 2001). This review of existing research provides context for the present study, motivating the need for a new approach that addresses the limitations of previous models and data analyses.

To rigorously test the implications of their endogenous credit constraint model, the authors extend their framework to a multi-period lifecycle model. This more sophisticated model incorporates government education subsidies, providing a more realistic representation of the educational financing landscape. The model's calibration utilizes US data on schooling, ability, government subsidies, and post-school earnings, enabling a direct comparison between model predictions and observed patterns in the economy. A key simplification is the assumption of frictionless credit markets after the 'youth' investment period, allowing for a clearer focus on the effects of credit constraints during the schooling years. Initial resources (w) are interpreted as the present value of family transfers, while ability (a) encompasses genetic traits and prior investments affecting returns on educational investment. The model accounts for both direct costs (tuition and government expenditures) and indirect costs (foregone earnings) of education, with a government subsidy rate (s) calibrated based on data from the Digest of Education Statistics (2003). This calibration, along with the incorporation of private and public lending, creates a robust framework for testing the model's ability to accurately reflect real-world investment patterns.

The calibrated model is used to simulate human capital investment under various credit market assumptions: both GSL and private lending, private lending only, GSL only, and an exogenous credit constraint model. The simulations incorporate an increase in the wage returns to education and a decrease in the government subsidy rate (representing rising tuition costs) while keeping maximum GSL loan limits constant. The results show that investment is significantly higher when both GSL and private lending are available, compared to scenarios where either source of credit is absent. This emphasizes the synergistic effects of public and private student financing. The model with both credit sources accurately reflects the increased importance of family resources in determining educational investment, especially in the latter period, mirroring the trends observed in the NLSY97 data. In contrast, the model with exogenous constraints fails to reproduce the observed patterns, underscoring the superiority of the endogenous constraint model in explaining the empirical evidence.

The simulations also analyze consumption patterns during the investment period under the different credit market scenarios. A key observation is that consumption is much higher when both GSL and private lending are available, compared to the scenarios with only one or no credit source. This highlights the importance of access to multiple credit sources for consumption smoothing, particularly for low-income students. The model demonstrates the potential cost associated with the GSL program's restriction against using borrowed funds for non-educational expenses; for those with low initial assets, consumption during school is very low. The comparison across various credit market structures reveals the positive relationship between ability and investment in the models incorporating private lending, but this positive relationship is not observed in the GSL-only or exogenous models. This contrast underscores the importance of including both GSL programs and private lenders for a more complete understanding of investment patterns and their relationship with ability and family resources.

The study concludes that incorporating endogenous credit constraints and the interplay between GSL programs and private lending is crucial for accurately modeling human capital investment decisions. The authors' model demonstrates that the direct link between credit and investment inherent in GSL programs alters the trade-off between income maximization and consumption smoothing for some borrowers. This is an important finding, because it means that some students, even while credit constrained by GSL limits, may invest at their unconstrained optimal level. This subtlety is often overlooked in empirical studies that focus solely on educational attainment. The model's success in replicating the observed increase in the effect of family income on college attendance, the rise in students borrowing the maximum GSL amount, and the growth of private student lending strongly supports the model's validity and highlights the importance of considering both public and private credit markets when analyzing educational investment decisions. The study's findings challenge the adequacy of traditional exogenous constraint models, offering a more nuanced understanding of the factors influencing human capital investment.

The authors acknowledge that their model, while successfully explaining several key trends, has limitations. For example, the model does not fully incorporate uncertainty in labor market outcomes, which could influence investment and default behavior. The authors suggest that future research should address this gap by incorporating uncertainty into the model, investigating the role of default as insurance against adverse labor market outcomes, and exploring optimal lending and enforcement policies. The inclusion of uncertainty introduces an interesting trade-off for lenders and policymakers between providing insurance and enforcing repayment, complicated by students' private information about their abilities or willingness to study. Additionally, future empirical work should aim to improve data collection, including information on schooling, borrowing, earnings, and loan repayment to enable more robust structural estimations of student loan markets and the effects of government policies on human capital investment.