Essays on mortgage broker, borrower, and lender behavior in the subprime mortgage market

This dissertation presents three essays on the behavior of mortgage brokers, borrowers, and lenders in the subprime mortgage market. In the first essay, I examine the relationship between the type of broker-borrower interaction in the origination process and subsequent mortgage performance. I show that face-to-face interaction between a mortgage broker and borrower before the loan funds is associated with lower levels of ex post default. This relation holds only for loans that have certain characteristics associated with low levels of financial literacy. Specifically, face-to-face interaction is negatively related to default for minorities, borrowers located in areas with low levels of education, low-income borrowers, and borrowers with low FICO scores. Additionally, a face-to-face meeting between the mortgage broker and the borrower is associated with lower default rates on cash-out loans, which brokers often originated as credit-repair" loans. Moreover, the relation between default and broker-borrower interaction is significant only for certain loan products (ARMs, low-doc) that have been linked to low levels of financial literacy. Taken together, these results suggest that face-to-face interaction between the mortgage broker and borrower may reduce problems associated with financial illiteracy.;In the second essay I explore differences in low-doc mortgages across employment types. Evidence suggests that stated income documentation mortgages, or liars' loans," are used to falsify income so that borrowers can obtain mortgages that would be deemed unaffordable by traditional underwriting standards (Jiang et al. (2014a) and LaCour-Little & Yang (2013)). By examining dfferences in stated income loans across employment types (e.g. self-employed versus W2 borrowers), I show that not all lies are created equal. My results demonstrate that high default rates for stated income loans are driven by W2 borrowers. For self-employed borrowers, default is not significantly related to the level of income documentation. Also, I find that income falsification is problematic only on stated income loans to W2 borrowers. In addition, I provide evidence that selection effects before the loan funds, at both the lender and borrower level, exist on stated income loans. Taken together, the results indicate that important differences exist between stated income loans to W2 borrowers versus low-doc loans to self-employed borrowers.;The third essay investigates real earnings management in the context of the residential mortgage market. Specifically, I try to determine if the lender originates lower quality loans to meet earnings benchmarks. My results are consistent with the lender lowering origination standards to manipulate earnings. Depending on the specification, loans originated in the final month of income constrained quarters are 27% - 34% more likely to default, ceteris paribus. Furthermore, expected default rates, based on ex ante risk characteristics observable at origination, are higher in the final month of income constrained quarters. In addition, I try to detect how the lender manipulates its activities in the final month of quarters where it meets or just beats earnings expectations. The results indicate the lender is less likely to reject loan applications when it is in jeopardy of meeting earnings expectations.