Drawing from the largest existing database of U.S. home equity lines of credit, they use county-level aggregates of these loans to estimate panel regressions on the characteristics of the borrowers and their loans, and competing risk hazard regressions on the outcomes of the loans. The authors show that when the expected unemployment risk of households increases, riskier households tend to borrow more. As a consequence, the pool of households that borrow on home equity lines of credit worsens along both observable and unobservable dimensions. This is an interesting example of a type of dynamic adverse selection that can worsen the risk characteristics of new lending, and suggests another avenue by which the precautionary demand for liquidity may affect borrowing.