The existing literature on mortgage default and foreclosure largely examines the development of mortgage problems (the first stage) and the possibility of recovery from such problems (the second stage) separately. Such studies may provide an incomplete picture of the development and possible resolution of mortgage problems. This paper uses the proprietary geo-coded version of the Panel Study of Income Dynamics from 2009 through 2013 to implement a different type of probability model. Three transitional probability equations are estimated: the likelihood of homeowners developing mortgage payment problems, the likelihood that those who develop such problems move to resolve their problems, and the conditional probability that those who develop such problems and don’t move will recover from them by other means. The coefficients from all three are combined to calculate the joint (state) probability of households having mortgage problems from which they will not recover. The analysis contains a number of unique results: changes in the mortgage front-end ratio are shown to have a substantial influence on both developing a mortgage problem and its resolution for those who do not move, 2) prior foreclosures are determined to affect the development of current mortgage problems, and 3) out-of-pocket medical expenses are found to influence both mortgage problem development and recovery. Also, the impact of policy actions are considered across several at-risk demographic groups. This experiment demonstrates that by improving the financial and mortgage characteristics, and the education level of households that actually develop mortgage problems, their predicted probability of developing mortgage problems that cannot be resolved could be reduced substantially.