This paper studies the impact of housing market cycles on loss given default (LGD). Previous studies have shown that the current loan-to-value ratio (CLTV) is the most important determinant of LGD. This paper establishes another linkage which is between the house price cycles before the time of mortgage origination and LGD. The empirical analysis is based on a large loan-level sub-prime residential mortgage loss dataset from 1998 to 2009. Results show that house price history has a long memory in explaining LGD. Its explanatory power far exceeds the original LTV and other loan characteristics. This paper offers a countercyclical view of LGD risk. The model can be combined with a default probability model to serve as a regulatory prudential tool. Such a tool provides a solution to the inherent procyclical bias in BASEL II capital requirements, and can contribute to the safety and soundness of banking institutions.