Third party originators and mortgage prepayment risk: An agency problem?

Journal of Real Estate Research, The, 1999 by LaCour-Little, Michael, Chun, Georgory H

Abstract. We focus on an agency problem encountered by mortgage lenders and investors in mortgage-backed securities when the underlying collateral is originated by third parties. Third parties, such as mortgage brokers, have economic incentives to encourage borrowers to refinance and, accordingly, their actions may affect asset values. We sketch the principal-agent problem and examine two sets of data. Results support the argument: loans originated by third parties are significantly more likely to prepay after controlling for other known determinants of termination risk. Moreover, third party loans are about three times as sensitive to refinancing incentives, compared to retail loans.

Introduction

A basic tenant of finance is that the value of an asset is the present value of its future cash flows. In the case of mortgages and mortgage-backed securities, prepayment estimation is essential in forecasting expected mortgage cash flow patterns. Accordingly, security prices are highly dependent on prepayment assumptions. With roughly 1.7 trillion dollars in mortgage loans securitized, many of which re-packaged into highly volatile derivative instruments, accurate forecasting of mortgage loan prepayments is of increasing importance to a wide variety of players in the capital markets.

In this article, we focus on an agency problem encountered by mortgage lenders and investors in mortgage-backed securities when the underlying collateral is originated by third parties (TPOs), such as mortgage brokers. We show how the contractual relationship between mortgage lender and third party originator may be cast in terms of the classic principal-agent problem. Since prepayment risk is not shared between lender and originator, TPOs can "churn" the customer, earning additional fees every time the mortgage is refinanced. Regression results using loan level data on mortgage terminations over the period 1992-1997 supports the argument. Results show that loans originated through third parties are significantly more likely to prepay, after controlling for re-financing incentive, loan type and size, and age. While it is difficult, if not impossible, to control for all borrower characteristics that may affect prepayment probabilities, the magnitude of these effects strongly supports the agency problem hypothesis developed. Additional market level data is consistent with loan level regression results, although the magnitude of the difference appears much smaller.

Institutional Background

The secondary mortgage market represents a large segment of the bond market. As of 1998, about 60% of single-family mortgage debt has been securitized, with the bulk of these loans pooled into mortgage pass-through securities issued by the Federal Home Loan Mortgage Corporation or the Federal National Mortgage Corporation (collectively called the government-sponsored enterprises or GSEs). Loans larger than the congressionally set loan limits, currently $240,000, are often securitized by major mortgage banking firms, these issues are known as "non-agency" or "private label" mortgage-backed securities (MBS). Agency guaranties protect investors from default risk; in the case of non-agency securities, subordinate bond structures provide this protection. But since, in general, borrowers may prepay their mortgage at any time without penalty, investors in MBS are exposed to prepayment risk, which increases as interest rates fall and refinancing becomes more attractive. Even in a stable interest rate environment, investors are exposed to prepayment risk, since borrowers who pay off their mortgage early to move to another house produce unscheduled cash flows to the mortgage pool. The rate at which these unscheduled cash flows arrive is known in the industry as the single-monthly-mortality rate (SMM) or, on an annualized basis, the conditional prepayment rate (CPR).I While investors in MBS bear this risk directly, firms that sell these mortgage pools to the GSEs, or to the capital markets directly in non-agency issues, retain exposure. Those firms typically service the loans they sell on behalf of the investor, retaining a servicing fee, usually 25 basis points for a conventional fixed rate product, that is a function of the principal balance of loan. This cash flow stream, known as the "servicing asset" in the industry, behaves much like an interest-only (IO) strip: when interest rates fall, principal balances runoff faster due to prepayments and servicing fee income is reduced. Since mortgage firms are required by GAAP rules to capitalize their servicing assets, unanticipated prepayments cause large accounting losses. For example, the FDIC recently reported that the 48 largest commercial banks saw the value of their servicing rights fall by $990 million during the third quarter of 1998, due to a surge of mortgage refinancing during that period.2 Furthermore, those same institutions were able to successfully hedge only $746 million of that decline in value, producing net losses of $244 million due to unanticipated prepayments.