Grappling with Loan Quality

LENDERS DON'T USUALLY WINCE IN PAIN OVER visions of liquidity woes. But a quick glance at 2004's headlines left them troubled as they read about federal investigators busting a Charlotte, North Carolina, fraud ring swindling lenders out of up to $50 million. Another headline talked about a bust of

Other scams feeding lenders' nightmares are: balloon-flipping scams targeting emerging markets; falsified Social Security numbers cloaking criminal histories to secure mortgage licenses; and an FBI probe of well-organized fraud rings comprised of brokers, lawyers and myriad loan stakeholders. All these misdeeds had lenders flinching to uncover what the FBI thinks will be billions in cost.

With parasitic fraud schemes flaring up and mutating cleverly ahead of authorities into burgeoning markets nationwide, mortgage fraud begs a wait-a-minute, Will Rogers philosophy of horse sense: When it comes to loan quality, would you rather perform preventative medicine-or an autopsy?

Now visualize a volume wholesale lender originating thousands of loans nationwide each month through a large broker network. This lender audits loans randomly for quality control (QC), mostly post-closing, to ensure adequate due diligence. So where is the disconnect between news headlines and the routine QC grind?

Most troublesome about the practice of conducting blind, random audits after origination is that they typify the industry's current tactics against fraud: steering while looking through the rearview mirror. This is still common practice, even though these tactics can take lenders two years or more to discover that loans closed today won't be attractive or convertible to cash on the secondary market later.

Whether borrowers inadvertently misstated income, or outright fraud for profit has unraveled investor confidence on a percentage of the lender's loan volume, when loans trickle into defaults well after closing, it's much too late for lenders to contemplate fraud prevention.

In a post-boom environment where increasingly sophisticated and virulent fraud is a rising tide that will affect everything in the water, only lenders that maintain a reputation for loan quality in the secondary market will survive. "Will it sell?" is no longer just a chief concern of secondary marketing officers and other executives. Portfolio marketability has become a lender mantra for 2005. Welcome to the new dilemma facing lenders-how to accurately quantify, define and screen fraud within existing workflows. How does a lender implement compatible fraud-filter technology while incurring minimal costs yet producing maximum loan quality? Increasingly problematic is the need to decipher which indexes and tools are most effective at intercepting fraud.

A competitive mortgage market prompts lenders to innovate new tactics to increase loan volume and balance loan quality. But grappling with loan quality means not only uncovering the causes of fraud, but also identifying the market's barriers to loan quality, and then crafting solutions that effectively integrate into legacy lending practices and nomenclature.

Lenders have only to canvass their increasingly vital broker networks-a primary source of sustaining volume, including alternative-A and nonprime products-to glean threats of fraud. Twenty years ago, most people secured home loans face-to-face through their local banker, someone who usually knew the borrower and the borrower's circumstances. Thanks to the onset of cheaper browser-based origination technology and wider adoption of the wholesale lending model, 60 percent to 70 percent of all mortgage loans today originate through a third party, typically a broker.

The increasing decentralization of loan processing coupled with accelerating speed-to-market of loan products that capitalize on rapid point-of-purchase decisioning and cross-promotion have led to many more hands touching each loan.

With the advent of "no-doc" and "stated-income" loans-sometimes dubbed the "liar's loan"-versus full-income disclosure, lenders increasingly face potential exposure to a smorgasbord of what the FBI calls fraud for profit. In October 2004, Chris Swecker, assistant director, criminal investigative division of the FBI, gave testimony before the House of Representatives about the cost of fraud to lenders.

"A significant portion of the mortgage industry is void of any mandatory fraud report," said Swecker in a written statement. "Based upon existing investigations and mortgage fraud reporting, 80 percent of all reported fraud losses involve collaboration or collusion by industry insiders. The FBI defines industry insiders as appraisers, accountants, attorneys, real estate brokers, mortgage underwriters and processors, settlement/title company employees, mortgage brokers, loan originators and other mortgage professionals engaged in the mortgage industry," Swecker adds.

Swecker also pointed out that if fraud continues unrestrained, it might cause investors to lose faith and require high returns from mortgage-backed securities. During the refi era, less-scrupulous brokers have generated a large number of poor-quality loans to churn up volume. The lending climate was more like a game of Little League with brokers waving in every borrower without reservation-where everyone gets to play and touch home base and no loan is tagged out.

Short-term, aggressive lending at low rates creates application costs; credit reports and appraisals, for example. At stake long-term are the liquidity levels and capitalization ratios of the secondary market itself, if fraud is not identified and denied on the front end in loan origination.

Rising foreclosure rates will soon caution investors to choose loan quality first, not volume. Reducing fraud requires that lenders identify at-risk elements in their loan volume, identify ways to adjust workflow for maximum efficiency to improve loan quality, and determine what tools to use to ensure loan quality.

The Mortgage Bankers Association's (MBA's) Fraud Subcommittee, the FBI and independent think tanks have all tried to produce reliable fraud statistics, but fraud has proven amorphous and elusive to calculate.

Ellen Rauch, vice president of compliance and quality assurance for Franklin, Tennessee-based Franklin American Mortgage Co., with assets of more than $4.7 billion and almost 35,000 loans in 2003, calculates that the price of fraud can take lenders two years or more to uncover and quantify against the bottom line.

"It takes so long to accumulate a number that, before we implemented fraud software, we would not know our fraud losses for 2004 until 2006; unless it's an outright fraud ring, which [produces loans that] generally default in the first six months," says Rauch. But Rauch knows that borrowers who commit fraud to get a house by lying about employment or income often struggle along. "Then you will get defaults and repurchases in year two, three or even five; by then, the damage is done."

All of these factors make it difficult for lenders to quantify their fraud-related losses.

Traditionally, quality control is performed, according to Rauch, after a loan closes. "This is fine for employee education. But if you've closed a loan with fraud, you own it," notes Rauch.

"If your investor can prove that the loan information was fraudulent-and investors can and will pull all old borrower bankruptcy filings-then the lender is responsible for portions of that loan and shortfall," she adds.

Nationwide, home mortgage originations are projected to reach $2.8 trillion in total volume in 2004, according to MBA. As the FBI has recently uncovered, hot spots for fraud rings are not only in high-volume states such as California and Florida, but also in places such as Nevada and Georgia. Fraud perpetrated on just a fraction of that amount of overall lending volume would cost the industry billions.

Other factors also contribute to opportunities for fraud to seep into the lending process. As emerging markets lending business adds to loan volume, income tax identification numbers (ITIN) numbers are being considered to address borrower documentation issues-but they also fuel the need to ensure proper identification.

Another factor to consider when assessing the financial risk from fraud is what type of production channels a lender employs. Rauch says Franklin American's volume is about 60 percent wholesale, while correspondent lending accounts for approximately 30 percent to 40 percent of volume.

"With correspondent loans, many of [which] are [originated by] banks, they will buy the loan back if something goes wrong or we find misrepresentation. Many brokers don't have that net worth or option-so we were exposed to risk," she says. This was a factor behind Franklin American's decision to integrate a software system that scrubs loans on the front end, saving workflow administrative costs and allowing the right group of loans to receive special attention.

Now Franklin American identifies and declines loan applications that involve property flips, says Rauch. "We can now say we don't even want one flip making it through the process. When our underwriter assistants receive broker packages, they enter loan information. A report is generated in two minutes as a PDF [portable document format]. Just in the last quarter of 2003, we had 5,500 fraud-alert instances [where] the Social Security number on the loan reported with multiple names-and that was within one month. Approximately 1,200 times the Social Security number was reported with multiple names, where names didn't match the borrower," says Rauch.

Also, within that period, Rauch says Franklin American received 266 alerts that the property value was possibly inflated. "When we get alerts like these, we immediately review the appraisal," she says.

Rauch says the two biggest categories of fraud that now usually result in alerts are invalid Social Security numbers and misstated income.

"As for misstated income, anyone today can generate false tax returns out of commercial tax software, so lenders are at risk if they don't verify income," says Rauch.

Other resources that can be employed to help combat various types of fraud are the Department of Housing and Urban Development's (HUD's) Limited Denial Participation Database (LDP) and Freddie Mac's exclusionary list. Rauch says the latter list contains names of people Freddie Mac doesn't want to do business with anymore.

Detecting property flips

One of the more costly types of mortgage fraud involves property flipping. Here is an area where good fraud detection tools can pay for themselves. "If you consider, for example, that the lender cost is $25,000 for a flipped property involved in double escrow, where the house isn't worth $75,000-only $50,000-then the price of fraud software pays for itself quickly [if] a lender identifies even just a few potential property flips," says Rauch.

In June 2004, the median home price was $191,800, according to the National Association of Realtors (NAR), Chicago. As house prices go steadily higher in 2005, so will the stakes of getting burned by a property flipping scheme.

Lenders are faced with the challenge of how to integrate fraud tools within the property assessment function. Mike Ela, president of Sacramento, California-based C&S Marketing, a provider of property valuation and collateral assessment data for the mortgage lending industry, says, "Lenders should seek holistic analysis of collateral, which also includes market condition, foreclosure statistics and market volatility rather than just property value."

Ela adds, "Within collateral assessment, geographic area market conditions suggest that AVMs [automated valuation models] can pose a greater risk than more traditional valuations. For example, property values are artificially inflated in areas where flipping activity is present. If 'flip' values are used in AVM estimates, values will likely be inflated because flips are not true reflections of market value."

So the best practice is to screen out flipped properties prior to collateral assessment. Lenders that manage a large volume of loans benefit exponentially when they scrub applications on the front end of origination and workflow.

"The real challenge is to know your customer, from the borrower to the broker to the appraiser-everyone," says Linda Johnson, national wholesale operations manager for RBC Mortgage Co., Houston. RBC integrated fraud software within its system to comply with the USA PATRIOT Act and Office of Foreign Assets Control (OFAC) requirements, and also to reap the rewards of fraud software.

"The Social Security issues like the PATRIOT Act requirements are totally seamless now," says Johnson. "It's vital to know our customers and to verify the identity of a borrower on our application and the property title. Lenders should look for software [that] generates reports [that] are easy to interpret, [that is simple to operate and maintain, and [is] enforceable."

Johnson says no one today can totally verify all identities, but that RBC's software picks up broker and loan officer identities and validates them, while also tracking names on HUD's LDP list. She receives multiple alerts in different categories as key indicators.

"The age calculation and verification of Social Security numbers on loans recently highlighted and stopped about 15 percent of attempts at loans on our system containing misinformation within one month," says Johnson.

This occurs, for instance, when the borrower lists one age on the application, but a check of the Social Security number reveals a different age attached to that number. Identity misrepresentation is much more ubiquitous than the industry ever imagined prior to fraud-intercept detection tools.

Fraud prevention with teeth

"There are almost never consequences for committing fraud," says Rauch. "Those engaging in fraud rarely get prosecuted; it has to be massive before prosecution occurs through the government. In order for brokers to offer FHA [Federal Housing Administration] loans, for instance, brokers have to have a minimum net worth, which guarantees repurchasing capability], if required. But there are no specific net worth requirements for conventional loans.

"Every lender sets [its] own requirements for broker profiles and net worth requirements, says Rauch. "Yet a broker's net worth can include his or her house, and many would be unable to repurchase a $100,000 loan. So there is almost no standard way to establish accountability. If you try to do something against brokers, they often threaten to take their loan volume elsewhere."

The most practical available solution to reducing mortgage fraud is to institute standardized safeguards. What many have called for is a nationwide registry of mortgage originators with uniform licensing and registrations of loan officers, brokers and appraisers.

Shrewd lenders will identify cost points throughout their origination process, to determine where they might best place fraud checks to abort unneeded administrative costs if a loan proves fraudulent along the way. Lenders should also institute appraisal technology that can filter out false-positives. These are cases where loans seem risky only because they got caught in fraud filters because of data-entry error, according to Ela. Look for predatory warning signs such as loan amounts that are higher than home values (loan-to-value ratios [LTVs]), for example.

Finally, here are a few more tips for lenders hoping to ward off fraud before it does damage to the bottom line. Establish personal identification standards and risk alerts for everyone involved in the loan process, through cross-referenced index evaluation models to verify identity and credibility. Then wield internal quality-control audits to analyze a much richer pool of higher-risk loans.

If lenders don't begin to standardize filters to monitor the sprawling maze of third-party originations in the next year or two, complicated fraud schemes and fraudsters simply exploiting unchecked workflows will not only make lenders suffer losses, they will jeopardize secondary market liquidity and investor confidence.

In time, however, the mortgage industry can innovate new mechanisms by working closely with the government to instill incentives in third-party originators to deliver superior loan quality and produce disincentives for all parties involved in fraud.