Working Paper: NBER ID: w27499
Authors: Manuel Adelino; William B. McCartney; Antoinette Schoar
Abstract: We show that the distribution of combined loan-to-value ratios (CLTVs) for purchase mortgages in the U.S. has been remarkably stable over the last 25 years. But there was a dramatic shift during the housing boom of the 2000s in the provision of high- CLTV loans through private sources, which replaced almost one-for-one the share of high-CLTV loans directly guaranteed by the government, via FHA and VA. Post 2008, FHA/VA loans increased back to 30% of all purchase mortgages. This substitution between government and privately backed high-CLTV loans holds within ZIP codes, properties and borrower types over the full sample period. We also show that the increase in private high-CLTV lending follows local house price increases rather than preceding them. These findings suggest that the housing boom was not accompanied by a shift towards more high-CLTV loans, and instead favor models that rely on changes in collateral values or broad changes in house price expectations.
Keywords: mortgage market; credit cycles; loan-to-value ratios
JEL Codes: E03; G21; G28; G30; R30
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| local house price increases (R31) | increase in private high CLTV lending (G21) |
| house price increases (R31) | increase in private high CLTV lending (G21) |
| increase in private high CLTV lending (G21) | substitution of loan sources (G21) |
| substitution of loan sources (G21) | cyclical relationship between government and private lending sources (F34) |
| debt-to-income ratios of FHA and private high CLTV borrowers are similar (G51) | types of loans were used interchangeably (G21) |
| changes in collateral values and house price expectations (G59) | dynamics of the US housing market (R31) |