Essays on consumption, borrowing, and behavioral economics

This thesis consists of three essays on consumption, saving, and consumer credit. The first essay estimates time preferences using data on wealth accumulation, credit card borrowing, and consumption-income comovement. The second essay measures individual-level causal impacts of access to credit. The third essay tests for sophistication and naivete using a structural model of payday borrowing.; Intertemporal preferences are difficult to measure. The first essay attempts to estimate time preferences using a structural buffer stock consumption model and the Method of Simulated Moments. The model includes stochastic labor income, liquidity constraints, child and adult dependents, liquid and illiquid assets, revolving credit, retirement, and discount functions that allow short-run and long-run discount rates to differ. Field data on retirement wealth accumulation, credit card borrowing, and consumption-income comovement identify the model. The benchmark estimates imply a 40% short-term annualized discount rate and a 4.3% long-term annualized discount rate. Almost all specifications reject the restriction to a constant discount rate. Quantitative results are sensitive to assumptions about the return on illiquid assets and the coefficient of relative risk aversion.; The second and third essays study payday loans, on which an estimated ten million American households borrow each year. Using a regression-discontinuity framework, the second essay measures the causal impact of access to payday loans. First-time payday loan applicants who are approved for a loan apply for additional payday loans, take out fewer pawn loans, and file for Chapter 13 bankruptcy at higher rates. No evidence is found that access to payday loan cash affects the incidence of crime.; The third essay accounts for the prevalance of payday borrowing by estimating a structural dynamic programming model of consumption, saving, payday-loan borrowing, and default on payday loans. The model's key parameters are identified by evidence of repeated borrowing and evidence that the average borrower who eventually defaults first repays or services five payday loans, paying 90% of the original loan's principal in interest. Procrastinating on default in this way is most consistent with partially naive quasi-hyperbolic discounting. The exponential and sophisticated quasi-hyperbolic null hypotheses are statistically rejected.