Origination Views

Lend to the Person, Not the Score

By Michael A. Covino

Mr. Covino, president of Tarrytown, New York-based jumbo loan specialist LUXMAC Home Mortgage, has written an opinion piece on why he believes a "dehumanization" of the underwriting process is the fundamental cause of the mortgage market's current crisis and why that needs to be changed. The following are his opinions on these topics.

The fundamental cause of the credit crisis in this writer's opinion is the relatively recent (last 10 years) complete reliance on the credit scoring system in rendering credit decisions on everything from home mortgages to credit cards, car loans and even insurance policies.

We now have all the proof we need that software programs, based on logarithms that are unknown for the most part by anyone associated with the lending decision, are not accurate predictors of loan performance. Is there anyone out there that still believes they are other than the Federal Trade Commission, which ruled that credit scoring is "a reliable means of credit evaluation?" Certainly I do not believe to any degree that relying on a credit score in making a decision to lend is prudent. I do however believe it can have significant adverse consequences for lenders and borrowers alike.

A credit score attempts to condense a borrower's credit history into a single number. Credit scores are calculated by using scoring models and mathematical tables that assign "points" to bits of information about an individuals credit profile which some analyst (from another credit environment much different than the one we have currently) determined would best predict future credit performance. While these models legitimately involved the study of how millions of people have used credit in the past, for them to be useful today, logic would demand that the economic environment be identical to that environment from which the data was compiled. Without a consistent environment how could the historical data be predictive? Model developers found "predictive factors" in the data that they analyzed and anticipated that it would give what they thought to be an indication of how future credit performance could be anticipated. Models were then developed from different sources of data. Credit bureau models were developed from information in consumer credit bureau reports, a score was assigned, and the rating agencies, Wall Street firms, bond insurers and investors jumped on these digitalized summations like bums on a bologna sandwich. The fixed income traders and loan product designers had their own unique interpretations of how to utilize the scores and sold a bill of goods to the bond holders, rating agencies, and bond insurers that with these new predictive credit scores, creditworthiness could now be quantified.

Some lenders required three "scores" from the three different credit bureaus. Some lenders relied on just one bureau's "score;" some lenders took the "middle score" as gospel. Some took the highest of the three. Programs were established around these "scores." Loan-to-values were governed by "scores." Extending credit to a particular borrower and the conditions under which that credit would be extended, were score driven. Whether or not income documentation was necessary, or what amount of liquid reserves would be required to approve the loan were all a function of the credit score. If you had a high "score" sometimes (usually above 700) it often was not necessary to check income because the score (they claimed) indicated that the borrower in all likelihood would pay his bills on time. High 700 scores also afforded a borrower the privilege of 90% to 100% financing. The credit "score" had become the foundation that dictated and continues to dictate the rest of the loan application parameters that lenders are willing to offer. The interest rate, loan to value, income documentation, debit ratio, reserve requirements, etc. are all determined by the credit score of the borrower. "Scores" today still determine the requirements and the terms and risk appetite for the whole mortgage loan transaction.

Credit scores now more than ever are determining whether or not people can refinance, get a car loan or even insurance. This scoring system is bogus and is only compounding the tightening of credit during this credit crisis. It is making it more difficult for creditworthy people to obtain credit at this time, to access lower rates on refinances or to acquire properties now at the reduced market prices. I suggest that this corruption of tried and true conventional underwriting policies and procedures evolved out of convenience. The Wall Street firms had a voracious appetite for volume. The rating agencies loved the fees that the Street paid for the ratings on the securitizations the Street was issuing. In an effort to facilitate volume a mechanism was needed to streamline or automate the cumbersome and archaic traditional mortgage lending system. Credit scoring seemed to fit the bill. With just a few bits of information entered in a PC or laptop, an immediate response pops up with the credit viability of a potential borrower. A score they predicted would be much like a report card. The lender/originator or underwriter could look at it and trust it as a predictor of future loan performance.

Underwriters loved it because it relieved them in large part from being solely responsible for the loan decision. Interpreting a borrower's creditworthiness was historically a skill that took years and thousands of files to develop. The secondary market loved this scoring system because they could establish risk-based pricing and calculate weighted average credit scores, along with their weighted average coupon (WAC), weighted average maturity (WAM), weighted average loan-to-value, etc. As volume increased, more reliance was placed on credit scores as automated underwriting came into popularity.

Now all an originator had to do was to enter in data on a borrower, like credit score, loan-to-value, loan amount, and presto, a loan decision, in some cases up to $2 million.

Loan originators needed to fulfill a few conditions that the automated artificial intelligence programs came up with and your loan was guaranteed to fund, or be bought in the secondary market.

What a fantastic evolution this was. Billions of dollars put into the hands of waiting homebuyers and those looking to cash out equity via a refinance loan, all at the push of a button, without the cumbersome process of interacting with a human being. No longer did prospective borrowers have to answer underwriter's silly questions. With automated underwriting approval (accepted by every loan buyer, including Fannie Mae and Freddie Mac), the result was an unprecedented ease of access to liquidity via the secondary market. The borrowers got fast decisions with minimum due diligence. Loan originators made fast commissions. Realtors, title and settlement agents all experienced increased revenue as a result of the improved mechanics of home lending. The real estate finance industry had modernized. We had met the challenge and we had changed our procedures with the times.

There was a major part of the underwriting process, however, that credit scoring and automated underwriting could never be programmed to do and that may have resulted in the economic meltdown that is the mortgage business today. Our computers, their software programs with logarithms based on predictive analysis and all its innovative capabilities could not "smell" the file for inconsistencies, tendencies, trends, misrepresentations, personalities, character flaws, or borrower's motivations. I am referring to that sixth sense that a good underwriter develops in evaluating a transaction.

Profiling the borrower, assessing the collateral, and the circumstances surrounding the loan transaction are all integral to making a prudent loan decision. Having an experienced underwriter's input on a file trumps that of an automated process that could and often did rely on faulty data. I will just summarize by saying many bad loans resulted from automated underwriting.

Let's get back to our credit scoring system that has become the cornerstone of the underwriting process. What if credit scoring is not a reliable means of credit evaluation? What if economic conditions change rapidly or real estate market conditions change sharply, resulting in a borrower's motivations and priorities changing? What if borrowers and their loan originators or third-party credit counselors discover ways to legally manipulate the credit score through software programs like "What If," which currently is available through several credit companies and offers tips and instructions to increase one's credit score? These are software programs that coach you in how to increase your score by modifying your credit profile. If, in this example, a borrower paid down his credit card balances and gets a higher score should he be afforded more credit as a direct result of manipulating his/her score with the aid of a software program? Will the automated underwriting engine be able to pick up this manipulation in time to stop itself from issuing an approval for a $1 million dollar loan because the borrower went from a middle score of 669 to a new revised score of 680? Is the same borrower several hours later more creditworthy because he/she manipulated their credit score by 11 points?

There are thousands of examples where relying solely on scoring has resulted in credit being extended imprudently or denied unfairly. Yet, since the mortgage meltdown, more than ever greater reliance has been placed on credit scoring. This is the equivalent to trying to extinguish a fire by dousing it with gasoline.

We are headed for an unprecedented devaluation in housing prices the likes of which this country has not seen since the Great Depression. This, in my opinion, is brought on in large part by the dehumanization of the underwriting process in the last 10 years in this country. Customers and their loan applications became numbers that made them worthy or unworthy of obtaining finance and the numbers very well may have been a wrong predictor.

I have been in the finance business and have been underwriting loans for 32 years. I learned the business of credit, and how and when to extend it, from my friend Bob.

The year was 1977. I was in my mid-20s and Bob was in his 50s. Bob had a sour wrinkled face and was as bitter as a lemon until you got to know him. He had been underwriting for 20 years at the time he and I met, and he had seen it all. To say Bob was negative would be an understatement. Getting to know him, I began to see why he always expected the worst. He got his start in the lending business by making what was called "pay day loans" for a small loan company. It was the late '50s when he first got into the business. Pay day loans were unsecured loans usually $500 to $2,000 in principal and payments were due on the day you got paid. So if the borrower got paid every Friday, you expected him to make some small payment each Friday. When Friday came at Bob's company and a borrower did not show up with his or her payment, a good loan officer was expected to go to the borrower's house and knock on his door to collect the payment. They called it having to do "door knocks." Bob told me stories of getting beat up, having guns pulled on him, of being chased, and bitten by dogs. Bob said he quickly became an expert on figuring out during his loan interview process with potential borrowers, just who was most likely to make timely payments and who was not. I learned a lot from Bob. The single most important lesson I learned was that, as an underwriter, in order to evaluate borrowers, you must talk to each and every one and profile them. Profiling the borrower interestingly enough requires a human touch and human interaction. It cannot be accomplished by interpreting a score or a number derived from a logarithm.

I am afraid that in our industry's attempt to bring efficiency to the extremely inefficient mortgage underwriting process we have strayed too far from our roots. The American banking system has long been the envy of the world, a banking system that had a long successful history of lending to people who paid back the loans that were made to them. Lending in America was always a "know your borrower" ideology that required face-to-face interaction.

How do we fix these delinquency problems of today? Start by immediately abandoning automated underwriting engines.

Traditional lenders like Bank of America, Chase and Wells Fargo have the lowest delinquencies and these are the lenders that refused to relinquish their lending authority to a software program. They underwrite the old fashioned way, one loan at a time, manually. We as an industry must stop issuing guidelines based on credit scores that are clearly based on historical credit data that has no significance in this current market environment of rising delinquencies, unavailable credit and depreciating home values. When profiling a borrowers creditworthiness, if the history is short in nature (two or three) years, and the borrower's score is 720, one must question whether that borrower is more creditworthy than a borrower with a 670 credit score and 10 pages of timely credit going back over 25 years. I personally would think not, and yet with a mortgage system and guidelines based on credit scores, more attractive rates and more aggressive loan amounts and terms will be afforded the 720 borrower with a track record that is no where near as extensive as the borrower with a 25 year history. This just doesn't make sense. Additionally we should reconsider making loans to individuals with high scores (680 and up) without looking at credit depth and history. A college student can have a score of 780 and it is not a measure of his/her creditworthiness. It's probably more indicative of his/her parent's credit worthiness if he or she is still a student. We have to ask ourselves when we look at the credit mess we are experiencing industry wide, have we allowed our dependence on a credit scoring system to dehumanize the underwriting process to such a point that loan transactions were actually made without a final human determination? In the volume frenzy that followed the reliance on credit scoring and automated underwriting we are faced with unprecedented defaults, no sound investor confidence in the securities that are collateralized by these loans, and bond ratings that are in question as well as the creditworthiness of the insurers that are insuring the bonds themselves. The subprime crisis has been the phraseology for the mortgage meltdown, yet the dehumanization of the underwriting process and the reliance upon credit scoring spans subprime, alt A and prime mortgages and is the one constant denominator that all delinquencies can be traced to.

Consumer groups, regulators, the investors all want to find blame with the originators. Loan officers and mortgage brokers that were handed matrixes and rate sheets that provided for loan products based primarily on a credit scoring system that may been flawed. These loan officers and brokers are not necessarily to blame. If I leave my teenage children at home with the key to the liquor cabinet and tell them I am going away for the weekend, I would deserve the consequences of my poor judgment. The consequences of relying solely on a credit scoring and automated underwriting system to the degree in which we as a mortgage industry did, across all sectors - subprime, alt A and prime -was an enormous mistake. A disaster in which the cause lies not with the originators and brokers, but with the system - loan product designers, loan aggregators, investment banks and their traders. The willingness to accept information and extend credit without a personalized human touch approach at the decision making level by experienced and competent personnel, is - in my opinion - the primary cause of the poor performance of the loans originated in the last four years. This resulting market meltdown is the byproduct. The guilty are the senior credit officers and operation managers that bought into the credit scoring hype and authorized guidelines pegged solely to scores and automated approvals. They are the ones who authorized the guidelines that relied too heavily on scoring and not enough on traditional underwriting protocols. While unscrupulous brokers and originators may have committed fraud in certain instances, the scope and sheer volume of the problem stems from a system wide failure to underwrite risk, one loan at a time, one borrower at a time, one transaction at a time.

Mortgage originators, loan officers and brokers are not the pariahs that the media, regulators, politicians and the lenders would have you believe they are. The responsibility lies with decision makers at every level in the secondary mortgage process - senior credit officers, rating agencies, analysts, traders, and investment managers for the bondholders -that all said in their own way, "Bring it on." They all loved the volume.

I recently attended a mortgage servicing summit held by one of the larger investment banks in response to the collapses of the securitization markets and the impact delinquencies are having on the day-to-day operations of some of the largest servicers in the country.

One large servicer there, who services over 500 million loans, was telling us about a new initiative in servicing delinquent loans. His company had been analyzing their loan modification efforts for nonperforming subprime loans. They determined that the hardest part of the modification process was getting the delinquent borrower on the phone to offer the borrower a loan modification. The modification process would lower the loan rate and sometimes forgive a portion of the principal on the loan. Among other traditional attempts to contact the borrower by phone or overnight courier, he said they had begun a campaign that was proving to be the most effective tool in their workout effort. He called it the "Door Knocks" program. He said they actually send a representative out to the delinquent homeowner's house to knock on the door and speak with the homeowner. He said they have learned a great deal about the borrowers they have loans with. The company often learns about the borrower's current needs, their capability, their circumstances, and a little about their character. Their representatives are trained to talk about modifying their customer's mortgage and find out how much of a mortgage the borrower can manage. They qualify the borrower about their employment picture and other income sources as well as any other obligations both on and off the credit report.

Let's see if I got this right. After we lend the money, we talk to the borrower when he doesn't pay it back to see how much he can afford and we modify the amount we lent him so that his payment can fit into his budget. We establish a face-to-face relationship with the borrower in an attempt to collect his debt. We build a relationship and understanding of our customer and we do so without any regard for their credit score only after he stops paying, and not before we make him the loan. The loan representative becomes the borrower's contact at the company, and no longer are his calls placed into a pool. Each representative is assigned 150 accounts only and the borrower and the representative establish a constant two-way communication for the rest of the life of the loan.

I happened to agree that this type of grass roots effort is what it will take to keep millions of homeowners in their homes. I think it is a good program and what is necessary in this time of unprecedented delinquencies and foreclosures. I guess we will all be taking a page out of my friend Bob's book. When our clients stop paying, go see them and work it out.

What I don't understand is why at this juncture we, as lenders, still want to remove ourselves from the borrower during the underwriting process by continuing to rely on credit scoring as the foundation of the underwriting decision and most of our industry's current loan product matrixes. Why do we still utilize automated underwriting engines? Why don't we want to talk with the borrowers before we lend them money they may not be able to pay back? Why do we still seek to establish a procedural method of qualifying and quantifying someone's creditworthiness based on a "score" that could very possibly be an invalid predictor of loan performance.

There are thousands of unemployed underwriters out there. Let's put them to work.

Let's be humane to our mortgage customers. Let's treat them as individuals with credit profiles that merit study and interpretation. Let's profile them with expert underwriters who are trained to analyze not only the credit history as it appears on the credit bureau but the income stream, their asset profile, their career expertise, their responsibility level and character either on the phone or face-to-face. Let's stop relying on the credit scoring and automated underwriting systems, and get to "know our borrowers."

What do they plan to do with your money after you lend it to them? Let's fix this problem one loan at a time. If we approach it one borrower at a time, we will return our financial system to the tenets of our forefathers who made the American banking system the envy of the world. Make loans to borrowers that you personally know will have the ability, willingness, and motivation to make timely payments after you have extended them credit. Make your next loan to the person, not the score.