This paper proposes two measures of credit risk for the population of outstanding mortgages. The first uses an average ex ante default probability to characterize risk, the second uses the unexpected loss generated by the asymptotic single factor risk (ASFR) model, a probabilistic model of portfolio risk. Both approaches show that average market–wide expected default rate and the unexpected loss per dollar of outstanding mortgage balances were roughly constant during the 2002-2006 boom in US house prices.

The future state of the housing finance system in the United States is the subject of ACTIVE debate. Central to this discussion is the question of who should hold the exposure to mortgage credit risk in the future. This raises the question of how to measure the total mortgage credit risk. Existing measures of market-wide mortgage credit risk have focused only on newly originated loans rather than the stock of outstanding loans. Yet, from a policy perspective understanding the risk of the entire US mortgage market is of even greater importance, as it represents the total risk that is to be borne by society and the decisions about the future of housing finance will largely determine how this risk will be allocated across individuals, financial institutions and the government. This paper proposes two related measures of market-wide mortgage risk. The first is based on an ex ante assessment of default probability and extends the methodology of Li and Goodman (2014) beyond new mortgage originations. This approach assesses ex ante default rate for all outstanding mortgages at any point in time through a weighted average of the ex post default experience of a ‘normal’ and a ‘stress’ cohort of loans. This measure can be thought of as a ‘through the cycle’ notion of default risk, as changing housing market conditions do not influence the assessment of default risk. The second measure of risk is based on an estimate of the unexpected loss of the population of mortgages.