The three chapters of my dissertation study household finance, with a particular interest in financial institutions and poor households. Specifically, I study constraints households have that are unobservable to institutions yet impact their interaction and the effectiveness of government interventions that aim to ease or facilitate these exchanges. Chapter 1 studies the behavior of payday loan borrowers by using bank-transaction level data on household spending, income, and loan activity to document three novel stylized facts. They suggest that a payday borrower is poor, has volatile income, and uses payday loans to smooth baseline consumption needs after an adverse idiosyncratic income shock. Chapter 2 builds on these findings to motivate a short-term lending model. The equilibrium contract under realistic frictions matches the observed payday loan contract on multiple dimensions, specifically when borrowers have low expected income and high-income volatility. I then calibrate my model using my bank-transaction dataset and find welfare increases between 5% and 28.7% when rollover fees decrease and initial fees increase. Chapter 3 studies the efficiency of government healthcare subsidization schemes, with a specific interest in the underlying mechanisms that financially motivate hospitals to serve the uninsured. We study a Disproportionate Care Hospital (DSH) payment scheme that supports hospitals treating a disproportionately higher fraction of uninsured patients. We demonstrate that DSH payments lead to social welfare loss, especially in a system with large and small hospitals, compared to the second-best mechanism. We then use the setting of the Global Payment Program (GPP) program, which compliments DSH by providing primary care, to show that direct and assured payment for primary care improves social welfare. Overall, my dissertation seeks to understand the development of an equilibrium contract structure between financial institutions and the poor, examine methods of greater efficiency, and evaluate the ...