Working Paper: NBER ID: w21320
Authors: Manuel Adelino; Antoinette Schoar; Felipe Severino
Abstract: This paper addresses two critiques by Mian and Sufi (2015a, 2015b) that were released in response to the results documented in Adelino, Schoar and Severino (2015). We confirm that none of the results in our previous paper are affected by the issues put forward in these critiques; in particular income overstatement does not drive any of our results. Our analysis shows that the origination of purchase mortgages increased across the whole income distribution during the 2002-2006 housing boom, and did not flow disproportionately to low-income borrowers. In addition, middle- and high-income, as well as middle- and high-credit-score borrowers (not the poor), represent a larger fraction of delinquencies in the crisis relative to earlier periods. The results are inconsistent with the idea that distortions in the origination of credit caused the housing boom and the crisis and are more consistent with an expectations-based view where both home buyers and lenders were buying into increasing housing values and defaulted once prices dropped.
Keywords: mortgage crisis; loan originations; defaults; income distribution
JEL Codes: D30; G21; R30
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| mortgage origination increased across the income distribution (G51) | increase in delinquencies among middle and high-income borrowers (G51) |
| defaults rose significantly among borrowers with credit scores above 660 (G21) | higher-income and higher-credit-quality borrowers were more likely to default (G51) |
| buyer income growth (F61) | mortgage origination growth (G21) |
| increase in delinquencies among middle and high-income borrowers (G51) | contradicts the notion that credit distortions led to the housing crisis (G21) |
| share of defaults from low-income borrowers decreased relative to earlier periods (G51) | supports the argument that higher-income and higher-credit-quality borrowers were more likely to default (G51) |