Chapter 4

Leveling the Playing Field

After years of disappointing attempts, Dawn Clark of Laurel, Maryland, finally managed to give her 10-year-old daughter a place they could call their own. The single mother prepared diligently, going to homeownership counseling classes, paying off overdue bills and stitching together a 5 percent down payment of $5,600—“more money than I have ever saved before.” With a stable annual income of $40,000, Ms. Clark believed she was set.
Unfortunately, delays and requests for information left the 42-year-old African-American wondering whether she was victimized by racial discrimination. “I do not know if race was the issue, but I cannot see them treating me like this if I were White,” she said.
While Ms. Clark is quick to add that other factors besides discrimination may account for the treatment she received, doubts linger. “I am just suspicious. I can’t pinpoint anything, but this comes from a lifetime of hearing minorities say they have had trouble getting homes.”

Automated underwriting, by evaluating loan applications solely on the basis of objective financial information, can help dispel lingering perceptions like Ms. Clark’s. In every case, Loan Prospector ignores any information about a loan applicant that is not directly relevant to assessing a borrower’s likelihood of default.

Because of its statistical basis, Loan Prospector is proving to be an accurate predictor of default, not just overall, but for borrowers across demographic and economic groups. The objectivity and consistency of Loan Prospector gives every family applying for a loan a fair shake.

Limitations of Traditional Underwriting

All too often, homebuying is an intimidating process, particularly for minority families. In part, this wariness is a legacy of misguided policies of the past. “Redlining” and other discriminatory lending practices occurred with some frequency in the housing market in earlier decades.

At least since the passage of the Fair Housing Act of 1968, however, the mortgage industry and the public have fought to stamp out lending discrimination. While industry observers continue to debate the effectiveness of these efforts, one fact is beyond dispute: Perceptions of unfair treatment in the mortgage market continue today.

A 1994 Gallup Poll commissioned by the Mortgage Bankers Association of America (MBA) documented a widespread view among minority renters that they would fare poorly when seeking a mortgage. Among those who had never applied for a mortgage, 32 percent of African-Americans and 24 percent of Hispanics believed they would encounter discrimination because of their race or ethnic background.1

During the past several years, the mortgage industry has eliminated significant barriers to home financing. At the same time, it has taken action to reach out to communities historically underrepresented in the mortgage market.

Freddie Mac, for example, has brought industry and community groups together to eliminate any unnecessary barriers in secondary market underwriting guidelines and to ensure that the application of the guidelines is consistent and fair. We also have formed alliances with community groups to draw more minority borrowers to lenders’ doors.

These efforts, while essential, cannot change the fact that traditional underwriting remains dependent on subjective human judgment. Each of the thousands of mortgage originators in the country employs many individual loan underwriters with different backgrounds and skills. Many of these lenders also purchase a substantial portion of their loans from mortgage brokers or other third-party originators.

Inevitably, different people make different decisions. These inconsistencies stem from the complex and multifaceted task of tallying up all the factors that go into an underwriting decision—including assessing the voluminous material contained within applicants’ credit files.

The result is an uneven process whereby loan applicants are treated differently from case to case. Some families who are ready to become homeowners get turned away. Others who are not ready for the financial responsibilities of homeownership obtain mortgages only to suffer foreclosure. The cumulative effect of each of these mistakes weighs heaviest on households whose applications fall in the gray area between acceptance and denial.

Loan Prospector Is Predictive for All Borrower Groups

By consistently applying uniform standards of creditworthiness, Loan Prospector provides the same objective treatment to all borrowers. Every attribute entered into the system is evaluated the same way for every borrower every time. Moreover, automated risk assessment is blind to the demographic or cultural characteristics of a loan applicant. The most compelling evidence of Loan Prospector’s objectivity and fairness is found in the accuracy with which it predicts default risk for all groups of borrowers.

This predictive power is illustrated in Exhibit 8, which is based on 1994 Freddie Mac purchases. The comparison looks at the foreclosure experience of borrowers in each of Loan Prospector’s risk classifications, across racial and ethnic groups. Whether a borrower is African-American, Hispanic or White, loans rated caution performed far worse than those rated refer, which in turn performed far worse than those rated accept. 2

The same strong pattern of predictiveness holds true across income groups, as depicted in Exhibit 9. Regardless of income, borrowers in this sample who received a caution classification faced far higher foreclosure rates than borrowers classified as refer or accept.

Increasing Homeownership for Minority Families

Automated underwriting marks an important milestone for the housing finance industry. By promoting system-wide fairness, automated underwriting has the potential to increase dramatically the number of minority families who own their homes.

As more minorities approach lenders with the belief they will receive fair treatment—and as the treatment they are accorded confirms those beliefs—the gap in homeownership rates that currently exists between minority and nonminority households should begin to dissolve. The potential impact is enormous. By leveling the playing field, automated underwriting could bring an additional 400,000 of today’s African-American and Hispanic renters into the ranks of homeowners. 3

Automated underwriting represents a significant breakthrough in mortgage lending. By evaluating all applications accurately and consistently, Loan Prospector will bring new families into the housing finance system.

1.Calculated as the percentage of African-American and Hispanic renters who said they were somewhat or very likely to face discrimination, multiplied by the percentage of each group that attributed that discrimination to their race or ethnicity rather than to other reasons. Study on Barriers to Homeownership and Perceptions of Discrimination in Mortgage Lending, Executive Summary, Mortgage Bankers of America, March 22, 1994.
2. The relatively high foreclosure rates shown for Hispanic borrowers in these data likely reflect the recent poor house-price appreciation rates in California and the fact that Hispanic borrowers in this sample disproportionately live in this state.
3. Susan M. Wachter and Isaac F. Megbolugbe, “Racial and Ethnic Disparities in Homeownership,” Housing Policy Debate, Spring 1992. This research estimates that, after accounting for income, housing prices, age, marital status and other characteristics, about a 20 percent differential in homeownership rates remains between nonminority and minority households—a difference of approximately one million households. Similarly, see George C. Galster, Laudon Aron and William J. Reeder, “Estimating the Number, Characteristics, and Risk Profile of Potential Homeowners,” March 1996. These researchers estimate that between 500,000 and 600,000 families would become homeowners “if the lending practices found in white suburban areas were applied uniformly across the nation.” Freddie Mac’s estimate of 400,000 families adjusts for the fact that these studies did not consider factors such as credit history.