Essays on the Evolution of Housing and Mortgage Markets

Date of Completion

January 2011


Economics, General




Theoretical models and empirical analyses argue that mortgage underwriting is a dynamic process in which previous mortgage and housing market conditions affect current mortgage approvals. Neighborhoods likely differ in important ways and over important events or shocks that influence both housing prices and mortgage underwriting decisions. This potential endogeneity complicates the causal analyses in a model of mortgage underwriting and failure to control for neighborhood heterogeneity risks confounding spurious and true state dependence. Most empirical studies control for neighborhood time-invariant heterogeneity with inclusion of neighborhood fixed effects. Fixed effects estimator in specifications using typical panel data often suffers from incidental parameters bias. While incidental parameters bias is most often discussed in the context of a panel, it can also arise in estimations that use repeated clustered cross-sectional data and control for neighborhood fixed effects as long as time-varying neighborhood common shocks exist. In the context of mortgage underwriting, these time-varying neighborhood heterogeneity could include expectation shifts of equity risk or compositional changes in a community over time. ^ My dissertation attempts to examine the housing dynamics and distinguish between the sources of time persistence on neighborhood mortgage underwriting. Specifically, I extend traditional and recently developed dynamic panel data techniques for use of repeated clustered cross-sectional individual mortgage applications linearly and nonlinearly, respectively. Essay one specifies a linear probability model to examine whether mortgage market activities creates information and lowers the costs of underwriting mortgages. Essay two proposes an analytical approach to reduce incidental parameters bias associated with fixed effects estimators in panel nonlinear fractional response models. Essay three examines the impact of local housing market conditions on mortgage underwriting using a panel of mortgage approvals collapsed from repeated clustered cross-sectional individual applications. It also extends the bias correction approach typically used in panel analyses to studies using repeated clustered cross-sections. ^