³ÉÈË´óƬ

Prior to an increase in the guaranty cap offered to mortgage lenders through the VA Home Loan program, I document nearly complete bunching of loan amounts at this cap. When the cap rises, I show that borrowers were more positively selected: they were more likely to repay their loans despite taking out substantially larger loans. Analyzing the economics behind the guaranty cap reveals that these ndings are inconsistent with rational expectations of borrowers and protmaximizing lenders. An unintended benefit of this bunching behavior is that the increase in the cap induces large, discrete improvements in home and neighborhood quality even though lenders and borrowers have not changed. Leveraging this allows me to evaluate the causal effect of living in a home that is more expensive partly because it is larger and partly because it is associated with a better neighborhood. I find this causal effect is substantial; borrowers' incomes rise by three percent due in part to greater proximity to high-paying jobs.