Borrower age differences in residential mortgage lending by commercial banks *.(Statistical Data Included)

Introduction

Empirical studies on the patterns of residential mortgage lending activity relative to the borrowers' age are almost nonexistent. Reliable lender data that contain sufficient detail on the mortgage terms, including the applicant's age, are virtually nonexistent. As the population grows older, the composition and terms of the mortgage bundle become more critical to every older potential borrower. Further, every loan officer is interested in the borrowing characteristics of its customers in order to effectively package and market future mortgages. Each bank also has a keen interest in loan-approval trends by type of customer to assure that all regulatory requirements on unfair treatment are satisfied.

The exact terms and composition of the loan offered, called the "mortgage bundle," have been investigated by other researchers primarily to uncover potential violations of the Equal Credit Opportunity Act (ECOA). The lender may be offering loans that contain terms and features that are significantly different and more punitive to one or more groups of applicants. Typically, these studies use the loan interest rate as the dependent variable. The loan amount, the term to maturity, and the loan costs are the minimum essential independent variables over which the lender may have some control. If the lender perceives that the information on the loan application warrants a significantly different credit score because a higher probability of default exists, the risk requires a slightly higher interest rate, higher loan costs, a shorter maturity, a smaller loan, or some appropriate combination.

The regression techniques shown in this paper could be used for a further, more intensive explanatory investigation by the individual institutions interested in regulatory compliance and by academic researchers. A more intensive investigation would require a detailed study of all loans in the lender's portfolio in a predetermined time period, combined with a study of the socioeconomic characteristics of the community, to assess the risk profiles of various applicant groups as suggested by Ferguson and Peters (1995). Obviously, the level of analysis which could be necessary to assure regulatory compliance is costly and time-consuming.

The task of the researcher is twofold. First, the level of appropriate data aggregation must be addressed. Although earlier researchers such as Shlay, Goldstein, and Bartelt (1992) suggest that a national aggregation of data makes little sense because housing market considerations are local in nature, this paper relies on statistical results from large, cross-sectional surveys for two reasons. The local market is influenced by national trends which are best seen in more representative data, as suggested by Hula (1992). Evidence of this is found in a study by Dunson and Reed (1991), one of the few empirical studies on age discrimination, that relied on a national survey of finance companies, and in the investigation of Texas consumer finance companies by Elliehausen and Lawrence (1990). The use of a larger data set with aggregation of local lenders is useful to estimate the impact of legislation, to evaluate the need for regulation without a specific consumer complaint, and to determine the need for further, m ore costly lender-specific studies.

Second, local lender data with sufficient detail to derive statistically meaningful results is not available to academic researchers in the current environment of increased regulation of residential loan origination markets. Local lenders are not responsive to requests for data to support needed academic research, for fear of the results, however innocent the intentions. This explains the void in recent years of local lender studies on loan origination patterns. Until this data becomes available in data sets or consumer complaints arise, the opportunity for academic research will rest in aggregate information provided by surveys, associations, or government institutions.

This paper takes the first step in assessing the loan pattern between two age groups. It is among the first to investigate the …