Printer Friendly



Last month, we discussed the potential importance of artificial intelligence-based systems for the mortgage industry. The fundamental conceptual advantages include leveraging the experience of the most effective employees, employing the consensus judgment of departmental management, standardizing decision-making across specific business areas (i.e. underwriting), increasing the volume of "good business," and improving customer service through faster and more accurate decision-making.

Despite the compelling advantages of enhanced systems intelligence, no one in the mortgage industry has successfully employed expert systems in a production environment (although several are close to coming on stream).

What causes the delay in adopting a new technology with so much promise?

The relative advantage of expert systems over conventional operational approaches has not been demonstrated to the industry.

Peter Carrol of the management consulting firm of Oliver, Wyman and Company of New York observes, "There appears to be a cultural block to the acceptance of artificial intelligence in the mortgage industry. We have done a good deal of work on ways to integrate artificial intelligence-based underwriting into the flow of production. The benefits are real; faster and more consistent underwriting decisions; better resolution in discriminating between a good and a bad loan; greater volumes of high quality product being closed."

Carroll continues, "There is confusion between how statistical models, rigid in their application, and how expert systems, adaptable to changing underwriting conditions, perform it. It takes a year or more to build AI into operations. Until model demonstration sites are up and running, the industry will be slow in adopting something this new."

Expert systems have yet to be assimilated into the operational and cultural norms of the industry. The hard part in introducing new techniques is in figuring out how the users and technology will work together.

A case in point is the authorizer's assistant (AA), an expert system aid to credit authorizers at the credit card business of American Express. AA extends the ability of a credit authorizer to make decisions on referred credit transactions (not automatically approved). AA has been integrated into the authorizers work station. The expert system does not replace the human operator, but provides a consistent screening tool for improving decisions that authorizers had always made. By increasing the effectiveness of the authorization staff, AA has resulted in reducing the declination rate by about 30 percent and a 50 percent reduction in the loss rate from those approved transactions. The effectiveness of the AA greatly exceeded the gains in productivity from the authorization staff (the original objective of developing AA).

The use of expert systems had to elbow its way into the operations flow at American Express. In the beginning, it was the commitment and funding from top management that enabled the work to be done. Senior management was determined to support experimentation with "risky" technology in research and development. American Express created a challenge fund to encourage high profile projects that promised high financial and organizational payback. Enculturation of innovative technology had to be initiated by senior management.

Vendors of artificial intelligence are just beginning to demonstrate how intelligent systems can reduce operational complexity.

Buzz Burt, founder and chairman of COGENSYS, La Jolla, California-based supplier of judgment software, states, "Acceptance (of judgment software) depends on the existence of visionary leadership in the industry. AI will be a vital component of lenders' re-conceptualizing their marketing relationships with borrowers and investors. A number of point-of-sale experiments (most unannounced) utilizing judgment software are underway. Aggressive firms are redefining service levels to their clients."

Burt forecasts, "The mortgage insurers may be the first to simplify client relationships using judgment software. The MIs are pressing to deliver on-line, almost instantaneous commitments. Within the next 12 months, we will see several loan processing systems in production with judgment software acting as the decision-making engines. A fully automated processing and underwriting operation will enable the innovating lenders to make reliable commitments almost at the time a loan application is received. The only contingency will be the appraisal.

"The conventional approach to underwriting has actually been an inhibitor to growth. Data acquisition and underwriting combine to become the longest duration tasks in originating a loan. Judgment software, integrated into a network that gathers information and decisions from credit agencies, MIs and appraisers will convert the longest duration tasks to nearly simultaneous transactions. Customer service will be greatly enhanced. Further, with electronic underwriting, the decision-making capability of the best underwriters will be behind every originated loan. Cleaner loan packages and portfolios will be offered to the secondary market."

The ability of AI firms to communicate their inventions is limited by a lack of marketing and advertising budgets.

Both Peter Carroll and Buzz Burt expressed the hope that the secondary market would play a major role in disseminating the benefits of judgment software to the mortgage industry.

"By holding to conventional document and information standards," observes Burt, "Fannie Mae and Freddie Mac may be inhibiting the acceptance of AI-based systems. Freddie Mac, however, has invested significant resources in developing and working with judgment software. We believe that the views of the secondary market will be vital to the future of the modernization of the mortgage decision process."

The mortgage industry is structurally more fragmented that other industries have adopted and found good use for AI-based systems. AI development and technological innovation, in general, have become industry issues that should be more actively discussed and supported by the agencies and the MBA than is now the case.
COPYRIGHT 1990 Mortgage Bankers Association of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:artificial intelligence at mortgage companies
Author:Lebowitz, Jeffrey A.
Publication:Mortgage Banking
Article Type:column
Date:Apr 1, 1990
Previous Article:Complexity is thriving.
Next Article:Who's who in wholesale: the big players are pulling in large amounts of new business.

Related Articles
New technologies emerge in medical AI.
Paper is down, but not out.
The MI technology race.
What's hot in technology.
The brave new world of artificial intelligence.
Artificial intelligence on a budget.
The risks of mortgage automation.
Time to invest in (the right) automation technology.
LogicEase receives initial approval for automated loan compliance patent.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters