Skip to main content
eScholarship
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Essays on Mortgage Risk

  • Author(s): Neale, Alan J.
  • Advisor(s): Craine, Roger
  • et al.
Abstract

This dissertation contributes three essays in areas of mortgage risk that are rapidly growing in importance. The first essay develops a fully dynamic optimization model for a borrower's redefault decision on a modified mortgage incorporating real-world frictions relevant for default decisions. Solutions to the model reveal large differences across modification structures and a basic pecking order for redefault performance controlling for resulting mortgage present value. Further, empirical tests utilizing unique and extensive data on modified loans offer broad agreement with the predictions of the model.

The second essay provides one of the most complete studies for termination behavior of non-U.S. mortgages to date, jointly estimating the competing risks of prepayment and default in a grouped duration mixed proportional hazard framework applied to Singapore mortgages. The study tests option-theoretic motivations for prepayments and defaults as well as "trigger event" explanations, explores comparative results to U.S. mortgage studies, examines unique institutional characteristics of this market impacting option-theoretic motivations for loan termination, and documents that variation in sources of borrower equity matter for the exercise of default options.

The final essay argues that the estimation of tail credit risk in residential mortgage portfolios remains relatively poorly understood, and that many common approaches to the problem have been incomplete or inadequate. In addition to laying out the fundamental components of sound portfolio credit risk assessment, the essay develops competing models for realistic dynamics of underlying risk factors, such as home prices. Particular attention is paid to identifying the properties of these models most consequential for the estimated distribution of losses, and to measures of implied sensitivity to geographic diversification.

Main Content
Current View