---

Abstract: This paper estimates the jumbo-conforming spread using data from the Federal Housing Finance Board’s Monthly Interest Rate Survey from January 1993 to June 2007. Importantly, this paper augments the typical parametric approach by adding state-level foreclosure laws and ZIP-level demographic variables to the model, estimating the effects of loan size and loan-to-value ratio on mortgage rates nonparametrically, and including geographic location as a control for some potentially unobserved borrower and market characteristics that might vary over geography, such as credit scores, debt-to-income ratios, and house price volatility. A partial locallinear regression approach is used to estimate the jumbo-conforming spread, on the premise that loans similar to each other in terms of loan size, loan-to-value ratio, or geographic location might also be similar in other, unobservable borrower and market characteristics. I find estimates of the jumbo-conforming spread of 13 to 24 basis points—50 to 24 percent smaller since about 1996, when credit scores became widely used in mortgage underwriting, than estimates from a commonly used parametric model. I therefore attribute the difference in estimates to credit quality and other unobserved characteristics, among other potential explanations, making these controls an important issue in estimating the jumbo-conforming spread.

Keywords: Mortgages, jumbo-conforming spread, partial-linear regression, locallinear regression

Full paper (403 KB PDF) | Full paper (Screen Reader Version)

---