Delinquency Risk in Residential ARMs: A Hazard Function Approach

Journal of Real Estate Portfolio Management, Sep-Dec 2004 by Teo, Alan H L

Executive Summary. Delinquency risk is a major area of concern to mortgage lenders and underwriters of Mortgage-backed Securities (MBS). Utilizing the hazard function approach, this paper provides an analysis of the characteristics and duration of delinquency risk of residential adjustable-rate mortgages (ARMs). Results show that mortgages with average ages of around ten and twelve years have the greatest likelihood of going into delinquency. Results also show that lenders' emphasis on borrower characteristics has been misplaced. Instead, uncontrollable mortgage-specific and macroeconomic characteristics are more significant in affecting delinquency incidence. This critically impairs lenders' ability to mitigate delinquency risk in their mortgage portfolio and hinders MBS underwriters' attempts to package good quality mortgages for securitization.

Introduction

Mortgage lending is an imperative component of the businesses of financial institutions worldwide. This is accelerated by the growth of the private residential property markets in the respective countries. With the increase in the number of mortgage loans originated, it is expected that the issuance of Mortgage-backed Securities (MBS) would continue to accelerate.

Similar trends can be found in Singapore (see Exhibit 1). In 1998, the Monetary Authority of Singapore, the de facto central bank, recommended the securitization of real estate (Sing and Ong, 2004). Together with the provision of favorable tax treatment (Ong, Ooi and Sing, 2000), there is enormous potential for the securitization of real estate and real estate-related debt in Singapore.

According to Campbell and Dietrich (1983), at any point of time, possible borrower action can be categorized into one of four groups:

1. Delinquent (payment delay);

2. Default;

3. Prepay the mortgage (through the sale of property or refinancing); and

4. Continue to service the mortgage.

Quercia and Stegman (1992) described the assumption in borrower payment models that the utility derived from each of the action is assessed and compared separately by the borrowers. Subsequently, the borrowers will select the action that yields the highest utility. Thus, a comprehension of the factors influencing each possible borrower action is essential towards understanding overall mortgage risk. Prior research studies predominantly focus on mortgage defaults or prepayments. On the other hand, there is generally a lack of studies on mortgage delinquency. This represents a gap in existing knowledge of mortgage risk. Thus, the first contribution of this paper is to attempt to fill this gap by examining the nature of delinquency risk and the factors affecting their intensity.

It is also imperative to note that the essential preceding step of a defaulting mortgage is delinquency.1 By providing insights into delinquency risk, this paper serves to expand the current perspectives and understanding of mortgage defaults. On the other hand, when a loan becomes delinquent, default might not necessarily follow (Waller, 1988). With fundamental differences in the motivations for default and delinquency (Campbell and Dietrich, 1983), this suggests that studies on delinquency risk can be an imperative area of research that contribute significantly to the current knowledge of overall mortgage risk.

The approach of this study is to utilize the hazard function to model delinquency. Two different duration models-the non-parametric and parametric models-will be undertaken to test the influence of the proposed regressors and also various duration implications.

The purpose of this paper is to provide insights into duration since originations have an exceptionally high risk of delinquency, and also the intensity and direction of relationships of various determinants postulated to affect delinquency. The knowledge of such critical periods provides a forewarning for mortgage lenders to be more careful about the possible disruption in cash flow caused by of mortgage delinquency. It is also useful for underwriters of MBS and Collateralized Mortgage Obligations (CMOs). The implications for these are discussed later. A comprehension of the critical factors that influence delinquency would aid lenders in setting their mortgage-lending criterion and affect the pricing and packaging of MBS.

The key results are that mortgages that originated ten and twelve years ago have the greatest likelihood of going into delinquency. Furthermore, various mortgage-specific and macroeconomic variables are found to have significant relationships with delinquency incidence. Specifically, the important covariates include the use of CPF funds, change in mortgage rate, premium of mortgage rate over investment returns, change in market sentiment and change in residential property price index.

The next section provides a review of related literature followed by a description of the research methodology for the intended duration analysis. The data and its descriptive statistics are then explained before the findings of the results are presented.