The application of information technology to finance, or "fintech," is expected to revolutionize many aspects of borrowing and lending in the future, but technology has been reshaping consumer and mortgage lending for many years. During the 1990s, computerization allowed mortgage lenders to reduce loanprocessing times and largely replace human-based assessments of credit risk with default predictions generated by sophisticated empirical models. Debt-to-income ratios at origination add little to the predictive power of these models, so the new automated underwriting systems allowed higher debt-toincome ratios than previous underwriting guidelines would have allowed. In this way, technology brought about an exogenous change in lending standards that was especially relevant for borrowers with low current incomes relative to their expected future incomes-in particular, young college graduates. By contrast, the data suggest that the credit expansion during the 2000s housing boom was an endogenous response to widespread expectations of higher future house prices, as average mortgage sizes rose for borrowers across the entire income distribution.