Making Quality Control Automatic: Every Step to the Electronic Mortgage Counts

John tapped his fingers impatiently as the clock rolled past 7 pm and his Outlook Calendar fastidiously reminded him of a nearing dinner date. As president of a thriving mortgage bank in reach of the billion dollar mark, John couldn’t let himself think about food until he resolved whether or not to extend the about-to-expire locks on the handful of loans for which he had just faxed final docs to clear conditions. There had to be a better way.

John’s predicament puts in stark relief a number of things that are still very broken in the origination process. He was late submitting one of the condition-clearing docs not because his borrowers were remiss but because (as learned in a last-minute phone call) guidelines were misstated on the investor’s Web site. In another case, a processor had mistyped an asset value in keying an application into another investor’s site; she caught the error and tried to fix it but was stymied by days of phone tag. In a third case, an underwriter knew she could place the loan with a particular investor, but since it was just at the edge of guidelines, she couldn’t do a smooth electronic submit. Again, manual keying and phone-based exception-handling used up the lock period runway.

Proponents of the electronic mortgage have been promising to revolutionize this process for several years. But those expecting to wake up one day to fully-automated, end-to-end origination and fulfillment have likely been disappointed thus far. Progress has been made, particularly at the point of sale, but it has been more evolutionary than revolutionary. As a result, most lenders must cope with a system that is only partially automated and still susceptible to the vagaries of human error.

Grow and Merge Islands of Automation

The good news is that even before the end-to-end paperless mortgage becomes a reality, there are numerous incremental benefits in the steps towards that ideal. A lender can continually grow and merge smaller islands of electronic process into larger islands, iterating until these islands merge into the long-sought continent of e-Mortgage.

Within a particular island of automation, it is intuitive that work is more efficient and quality is higher: data is moved and manipulated electronically, and software rules validate data integrity. The breakdowns occur where the lending process and its associated loan data must traverse from island to island. Data must be translated or re-keyed. Assumptions on the sending island don’t match the rules on the receiving island. The benefits are in merging the adjacent islands.

Two such opportunities are immediately apparent. The first is between the originator’s point of sale and the wholesaler/investor’s underwriting. The second is the final delivery from the originator’s processing system to the wholesaler/investor’s processing system.

Quality Starts at the Point of Sale

Revisiting John’s situation, several of the problems were the result of a disconnect between John’s automated point of sale island and his investors’ automated underwriting islands. The keys to merging these islands are:

• Electronic upload from the originator’s point-of-sale system
• Automated data validation
• Automated product eligibility, price adjustment, and lock policy administration
• Automated condition setting by applying guideline rules to loan data
• Automated condition clearing by document upload or fax
• Support for exception handling

Emerging technologies and standards have made some of this possible today, and the situation continues to improve. Widening support for the MISMO XML standard means that more point-of-sale systems can upload into more underwriting systems. Systems based on a service-oriented architecture (SOA) can more easily integrate with other systems and can also more easily accommodate changes or upgrades to their own constituent components without disrupting the process.

Electronic upload itself significantly improves quality by reducing introduction of human error in re-keying. The next three items in the list take quality to a higher level and significantly reduce cost. Validating data—making sure all required fields are present, in the right format, and consistent (state and zip code match, income is not a trillion dollars, etc.) is the beginning. Filtering the product selection and adjusting the prices based on this data is next. And upon selecting and locking the rate on a particular product, setting conditions based on guidelines is a jump in quality control sophistication. These activities involve rules, and the technology to support the entering, managing, and updating of these rules has improved dramatically in the past few years. No longer is it necessary to engage programmers every time a lender needs to introduce a new loan product—the secondary marketing department can enter the rules directly.

Condition clearing typically happens shortly after the actual point of sale but is tightly related to this part of the process. Couriering or faxing documents to clear conditions means that a person must receive the documents, associate them with a loan file, verify that they in fact satisfy the conditions, and indicate to the originator that the conditions are cleared. Though completely automated verification is beyond affordable technology at the moment, an important improvement to John’s situation can be achieved by automatically associating and storing an imaged fax with the electronic loan file and then instantly indicating to the sender that it was received.

The last item in our point-of-sale list is the most difficult to realize. Certainly, just by virtue of having reduced manual data entry and error correction, the automation we’ve described thus far will free up more human time for exception handling where it’s truly needed. But a notch better is to have the system actually support the human exception handling rather than simply force the exception outside the system, as in John’s example. Enabling such exception support involves a clever combination of rules and user interface: rules that identify the exception in the first place while categorizing and quantifying the exception’s parameters, and a user interface that presents the parameters along with sufficient ability act upon them (e.g. overriding an eligibility rule).

Quality in Follow-through

Once the point of sale and underwriting islands are merged, the lives of correspondents and other originators like John are significantly improved. But let’s fast-forward John’s vignette a week or two, and now there’s a different opportunity. For each loan John places with an investor, he ultimately couriers a fairly thick paper loan file to complete the transaction. The e-Mortgage vision replaces this with the click of a mouse-button that sends a MISMO-based electronic package of SMARTDocs, replete with tamper-proof e-signatures, change histories, and easily parsable data for upload to other systems. Between now and the realization of that vision are a lot of county recorders, compliance departments, servicers, and even investors, all of whom still require paper.

But it’s not all-or-none. By keeping the mortgage as electronic as possible, for as much of the process as possible, all parties in the transaction benefit from higher quality and lower costs. Only the documents that absolutely must be rendered to paper, such as the note, should be printed, and printing should occur as far along in the process as is practical. This allows for a single electronic representation of the loan file to be managed and maintained in a consistent manner, and in particular maximizes the ability to carry out rules-based automated quality control rather than the laborious paper-based equivalent.

A particular opportunity for merging automation islands in this area involves the interdependency between loan product, loan data, and documents. In a common scenario, an investor introduces a new loan product that has unique document requirements. The correspondent uses his own doc prep service. If that service hasn’t (or couldn’t have) retooled to accommodate the new product, the document-set that is generated will not be correct, and this will cause problems after the investor receives the loan file. If instead the correspondent’s and investor’s systems are integrated in such a way that either the correct document rules are automatically downloaded with the product definition—or, better yet, the correspondent simply generates the documents directly on the investor’s system—the process will be smoother. Again, rendering to paper only what is necessary improves the transaction even more.

Finally, there is the bulk scenario. John’s examples so far have been for single loans, but there are even more synergies in the multiple loan context. Bulk sales are commonly supported by the emailing of spreadsheets. Excel is powerful, but John’s spreadsheet might not map cleanly to his investor’s spreadsheet, and this means quality control effort and potential for human error. Now replay the scenario with John selecting the desired loans for his pool within his processing system (even better: the pool was automatically constructed based on rules). John clicks a button to upload the pool to his investor. The investor’s system reads and parses the pool, runs its validation checks, and instantly gives John a status.

Build the Habit into the System

Aristotle said, “We are what we repeatedly do. Excellence, then, is not an act but a habit.” While quality assurance has always been critical in the mortgage business, it has too often been viewed as located outside the process itself, with quality checks done retrospectively. Electronic mortgages promise tighter integration from point of sale to loan close, creating the opportunity to reinvent quality control by building it into the fabric of the system. The way to get there is by merging islands of automation. The net result will be faster close times, fewer errors, and higher productivity up and down the chain.