|
Article Excerpt
Empirical research on mortgage default in the single-family market has focused on the value of the borrower's put option using house price indices to estimate contemporaneous loan-to-value ratio or the probability of negative equity. But since the borrower possesses the option to increase leverage by taking on additional debt secured by junior liens subsequent to loan origination (a phenomenon termed here equity dilution), even a perfect house price adjustment cannot be expected to accurately measure changes in borrower equity over time. Since junior liens are generally unobservable to senior debt holders, proxies are required in empirical applications. This paper employs an independent estimate of junior lien probability developed from the 1998 Survey of Consumer Finances combined with loan level mortgage performance data to examine the role junior liens play in increasing default risk.
**********
Fostering home ownership has long been an important public policy objective in the United States, yet the majority of home purchases require use of mortgage debt and not all mortgages are repaid as contracted. Indeed, Elmer and Seelig (1999b) document a long-term secular trend of increasing rates of default. Delinquency and foreclosure rates are currently high, notwithstanding robust housing conditions. (1) Likewise, personal bankruptcies continue to reach all-time highs. (2) Accordingly, lenders, researchers, policy analysts and governmental officials continue to seek to better understand mortgage default. Figure 1 depicts the paradoxical pattern of increasing house prices together with increasing rates of mortgage default. Since 1986, house prices in the United States have approximately doubled, while foreclosure rates on government-insured loans have tripled and foreclosure rates on conventional loans have increased by 50%.
[FIGURE 1 OMITTED]
Researchers have recognized that contingent claims methodologies can provide important insights into the mortgage market. Under this view, a mortgage loan may be treated as a fixed income instrument combined with American put and call options held by the borrower and written by the lender. The right to prepay the mortgage at any time is a call option at par; the ability to default on the mortgage at any time is a put option in which the mortgage is "sold" to the lender for the market value of the property. (3) Despite the focus on option-based models, an important option possessed by mortgagors has been overlooked: the option to add mortgage debt secured by junior liens subsequent to loan origination. As a general rule, except in the case of some commercial loans, acceleration provisions of mortgage notes are not triggered by further encumbrance of the collateral property; hence, an implicit option to further encumber is available to the borrower.
How do junior liens affect the risk borne by the senior mortgage holder? This is a complex question. On the one hand, a reduction in borrower equity and an increase in debt service requirements would seem likely to increase default probability on the first mortgage. But since junior liens are by definition subordinate, there would seem to be no effect on loss severity and hence credit losses. On the other hand, if debt service requirements on the junior lien consume funds that would have otherwise gone to maintenance, the condition (and hence the value) of the collateral may be impaired, producing an increase in loss severity upon default. (4) Alternatively, if funds from junior lien borrowing are used to improve the condition and thereby increase the value of the property, then, if the incremental value is sufficient to offset the incremental debt, the combined loan-to-value ratio (LTV; including both senior and subordinate financing) could remain constant or actually decline.
It is worth noting that in segments of the mortgage market in which default is more frequent and its consequences more severe, namely, commercial property, subordinate financing is often expressly prohibited. (5) By contrast, under the general theory of contract, the option to further encumber residential property with loans secured by junior liens is implicitly allowed, since it is not expressly prohibited.
Collin-Dufresne and Goldstein (2001) have recently analyzed an analogous issue in the corporate finance literature, focusing on the determinants of credit spreads paid by firms with different levels of leverage. They argue that credit spreads on bond debt issued by firms should account for "the firm's option to issue additional debt in the future." (Collin-Dufresne and Goldstein 2001, p. 1930). Results imply higher required spreads for low leverage firms. The analogy is imperfect, of course, since household assets and liabilities differ in some significant ways from firm capital structure. Moreover, in terms of contract structure, household mortgage debt has a priority ranking based on lien status and is generally continuously callable, while firm bond debt has neither feature.
Two important phenomena have promoted expansion of mortgage debt usage in recent years. First, a new industry segment of subprime lenders has emerged. These firms specialize in making mortgage loans that do not meet agency guidelines, yet much of this debt is ultimately securitized in the non-agency market. A second major development has been the growth of home equity lending, fueled by generally increasing housing prices during the late 1990s. This segment, particularly home equity lines of credit, has grown rapidly in recent years, particularly since 1986 tax law changes eliminated the deductibility of interest on consumer loans. According to accounts in the trade press (American Banker 2003), home equity lending has continued to expand with the average size of new home equity lines as of June 30, 2003, increasing 26% over 2002 to $69,513, while the average size of new home equity loan commitments increased 44% to $58,054. About 16% of these loans were to subprime borrowers, defined as those with credit scores less than or equal to 620.
How big is the home equity market? Using data from the first half of the 1990s, Weicher (1997) estimated that home equity lending accounted for between 5 and 10% of total mortgage origination in the United States. During the mid-1990s, this would have represented $50-$100 billion in annual volume; today, holding the estimated percentage constant, that figure would be $100-$200 billion. Using more current data, SMD Research (2001) estimated that the total home equity market ranges from $525 to $700 billion. From the 1998 Federal Reserve's Survey of Consumer Finances (SCF) data used for part of the empirical work reported here, one may calculate that about 18% of homeowners with first mortgage loans also held a second mortgage at that time. Accounts in the trade press reported that as of year-end 2002, commercial banks and thrifts held $256.4 billion in home equity lines alone on their balance sheets, up 39% over year-end 2001 (Davenport 2003).
What is the likely effect of these junior liens on the senior debt holder? As previously discussed, the a priori relationship is unclear. Second mortgages may dilute borrower equity and increase leverage, raising the risk of mortgage default; alternatively, if funds are reinvested in the property in such a way as to increase its value, there may be no such effect. Existing research on home equity debt usage, discussed in the next section, has not addressed the effect of junior liens on first mortgage loan performance. But since information on the existence of junior liens is generally not available to the senior debt holder, proxies are required in empirical applications. This paper develops and tests one such proxy, an independent estimate of the probability of a second mortgage estimated from SCF data.
The organization of the balance of the paper is as follows. In the next section, I review the existing literature on mortgage default and the limited research on junior liens. In the third section, I briefly discuss theories of mortgage default and identify the impact of equity dilution. The fourth section describes the two data sources used in the empirical analysis. In the fifth section, results of a model that estimates the probability of a junior lien using the SCF data are reported. In the sixth section, I turn to loan performance models, modeling both default and prepayment, and introducing the probability measure for the presence of a junior lien. Graphs illustrate how default risk appears to be amplified by the likelihood of subordinate debt. The final section offers conclusions and open research questions.
Literature Review
The literature on residential mortgage loan performance is vast, so the review here is necessarily limited. Important early empirical research includes Von Furstenberg and Green (1974), Cambell and Dietrich (1983), Foster and Van Order (1984, 1985), Evans, Maris and Weinstein (1985), Vandell and Thibodeau (1985) and Cunningham and Capone (1990). Quercia and Stegman (1992) and Vandell (1993) provide surveys focusing on default. Virtually all researchers conclude that borrower equity, or loan-to-value ratio, are critical determinants of default probability. The role of trigger events, such as loss of job, divorce and health problems, is also thought important, though more difficult to formally model (Vandell 1995). Ambrose and Capone (1998) decompose the process still further, separating default from foreclosure and modeling the process in a two-stage context. Extending the work on defaults and cures still further, Ambrose and Buttimer (2000) develop simulation models in which delinquency may result from trigger events, but reinstatement is more likely where sufficient equity is present. Recently, Elmer and Seelig (1999a) have developed a broader theoretical framework in which mortgage default can arise from...
|
