Borrowing Constraints, Portfolio Choice, and Precautionary Motives: Theoretical Predictions and Empirical Complications

This paper studies effects of two classes of borrowing constraints, collateral- and income-based, on wealth accumulation, portfolio behavior and on precautionary motives. We examine the sensitivity of solutions to tightness of constraints, education level, and preference parameters. The models are calibrated using the 1992 Survey of Consumer Finances. The idea that constrained households engage in less borrowing and less holding of risky assets than desired is borne out for income-based constraints but not necessarily for constraints where assets also serve as collateral. The commonly used nonnegativity constraint on wealth turns out to be a very special case among collateral constraints: not only is constrained consumption equal to income but precautionary wealth holding is zero. Income-based constraints reverse the sign of precautionary effects on holdings of risky assets, and so do relatively tight collateral constraints. The latter reverse the sign of precautionary effects on borrowing, as well. Precautionary effects on wealth holding and on borrowing are smaller when income-based constraints are binding, though not necessarily so for collateral constraints. Results suggest that inclusion of constrained households in a sample of unconstrained ones is quite likely when using standard wealth-level cutoffs for sample splitting, and that it tends to bias empirically observed precautionary effects on wealth downwards. Estimated precautionary effects on risky assets and on borrowing may even be biased towards zero. These findings may help explain the failure of recent empirical studies to uncover sizeable precautionary effects on wealth and on portfolio composition.