Freddie Mac Exec Details Evolution of Credit Scoring

More than any other player, the Federal Home Loan Mortgage Co. has propelled the use of credit scores in the home loan business.

Freddie Mac set the ball rolling in July 1995 by asking lenders to supplement their traditional review of borrower credit histories with bureau credit scores. Applicants with poor scores were to be targeted for the most rigorous scrutiny, the agency said.

Freddie's own automated underwriting system, also introduced in 1995, uses the same statistical analysis to score not just the borrower's credit history, but several other borrower and loan attributes as well. By yearend, Freddie estimates that half of all loans it buys will be handled by Loan Prospector, its electronic underwriting system.

The agency also has worked with Standard & Poor's to develop scoring models for jumbo and subprime loans, and is testing a model for government- insured loans.

Long used by credit card and auto lenders, credit scores give a snapshot of a borrower's willingness to repay debt by statistically analyzing the borrower's track record in paying off previous debtors - credit card companies, department stores, banks and finance companies.

In a recent interview with American Banker, Ann Schnare, senior vice president of corporate relations at Freddie Mac, explained why the agency believes credit scores belong in the home loan business, and how their use is changing the mortgage industry.

How did Freddie Mac decide to go with credit scoring?

SCHNARE: The underwriting guidelines that have been used in the mortgage market have evolved over time and had some real truth in them.

One of the things you don't get is the ability to trade off against compensating factors or to identify when too many risk factors combined put an otherwise sound application over the edge.

This is what you really get from a more statistical approach to underwriting mortgages.

In the early 1990s, we began quite extensive research based on the performance of actual mortgages - looking at the relationship between loan performance and a wide variety of underwriting factors.

We discovered the predictive power of generic credit scores, which is really a way of combining all the information in a borrower's credit file into one objective and highly predictive measure of risk. We were the first major player to advocate their use as an aid to manual underwriting.

At the same time, we've introduced Loan Prospector, (which) takes credit history as just one of several things that contribute to loan performance. These new tools are bringing more consistency to the underwriting process, more fairness, and they're also reducing the cost.

In the future, you'll see the use of this technology in a much broader way. We just developed a model for the servicing and handling of delinquent loans. Most delinquent loans cure. Right now servicers have to make the same effort to go after all of them. What these technologies do is enable you to target resources, much like the front end of the business.

One of the things we hope to do is (see) whether you can use automated underwriting to focus counseling (by) identifying the people who have the greatest probability of being risky loans.

So you feel pretty confident that the use of scoring is well established in the mortgage industry?

SCHNARE: I think it's very well established. I think a year ago there were a lot of concerns over consumer issues. Those have largely been resolved.

First and foremost, (scoring is) going to lower costs. These savings are going to be passed on to consumers, and even small savings in up-front costs can generate not-insignificant increases in the number of potential homeowners that would qualify for a mortgage.

Even a $400 decrease would lead to an 8% increase in the number of African-Americans that qualify for a mortgage.

What about the concerns that if a consumer doesn't have an extensive credit history, they're not likely to get a fair shake.

SCHNARE: Under the traditional system, if you don't have a credit history, we encourage our underwriters to look at rent payments and utility payments. That's not changed under automated underwriting.

I think the biggest issue is whether credit scoring or a system like Loan Prospector would treat minority and low-income people fairly. Loan Prospector is equally predictive of loan performance, regardless of whether the borrower is African-American or Hispanic or white, and regardless of borrower income.

There was a great concern about the use of finance companies, (which) are a small part of the scores. We got data from FHA and found that the use of finance companies was no more common among African-Americans than it was among whites.

Is there anything you'd like to add?

SCHNARE: There has been a lot of fear about technology. The mortgage industry was one of the last to embrace technology, and I think a lot of these initial concerns were quite understandable.

I think that they are going to die down over time, because we have a very powerful set of tools that are going to improve the efficiency of the business (and) the quality of the lending decisions, but also have the potential of bringing households that are either excluded from the current system or relegated to the more expensive subprime market into a lower-cost mortgage.

This is good for consumers, good for lenders, and also good for investors.

For reprint and licensing requests for this article, click here.
MORE FROM AMERICAN BANKER