In this paper we argue that loans collateralized by new construction are more likely to go into default relative to purchase loans for existing homes because of non-linear depreciation schedules and appraisal complications. Using loan-level mortgage records for about 4 million loans originated between 2004 and 2009, we provide strong empirical evidence in support of this hypothesis. The unconditional default rate for mortgages used to purchase new construction was 5 percentage points higher than the default rates for other purchase loans in our sample. In models that include extensive controls for borrower and loan characteristics as well as Census-tract-origination-year fixed effects, we find that loans for new homes were roughly 1.7 percentage points more likely to default, while our instrumental variables analysis suggests that new-home loans are 4.5 percentage points more likely to default.

This study develops estimates of expected loss severities on mortgage exposures using data from Florida during the Great Recession. This paper marks the first attempt at addressing sample selectivity in the context of loss models. We also construct measures of home equity that are more accurate than those employed in previous studies. We find that failing to address sample selection and the use of noisy equity measures in loss models can bias loss estimates significantly. We also find significantly higher loss severities and a greater sensitivity of loss severity to equity than what previous studies report.