Remember the car from the movie Back to the Future? From an aesthetics standpoint, the DMC DeLorean was the perfect centerpiece for one of the most successful franchises in Hollywood history. For true car fanatics, however, it was kind of a dud.
Despite its stunning lines and overall “coolness,” the DeLorean was massively overweight for a sports car. It also housed a modest 130-horsepower V6 engine capable of doing 0 to 60 in rather pedestrian 9.5 seconds. In fact, its top speed was just 110 mph—only slightly faster than the 88 mph Marty McFly needed to travel back to 1955.
The mortgage industry is a little like the DeLorean these days. Much of the technological sophistication in recent years has taken place on the front end of transactions and in eClosing technologies, both of which are highly visible within the borrower experience.
If you lift up the hood, however, much of our industry is still being powered by technology built more than a decade ago—creating downstream impacts to the secondary market, especially when it comes to the trading of mortgage servicing rights (MSRs). Thankfully, however, this is starting to change.
Rife With Bottlenecks
Currently, originators that sell servicing rights have extraordinary trouble meeting the multiplicity of buyer delivery requirements. As a result, buyers often have to key in information and process documents manually when transferring loans into their servicing systems or delivering loan file packages to subservicers. This creates extra costs and delays, which ultimately can make it challenging to comply with the required timely communications to borrowers about who is servicing their loan.
The source of friction is a lack of standard processes for exchanging MSRs. Documents and data are being received from sellers in so many different ways that buyers can’t keep up with the incongruities. For example, document images are classified using different taxonomies, stacking orders aren’t consistent, and image quality varies, so no two loans are alike in terms of structure or legibility. Buyers receive loan packages from different sellers that look completely different from each other. All of this means buyers must manually go through every loan and restack documents in their preferred order and determine which things are missing from individual loan files.
There are also nuances that cause data discrepancies in the loan files that lead to extra costs. These can slip by when work is done by staff with insufficient expertise to identify these issues. As an example, for a loan with co-borrowers, the fact that a credit report exists isn’t enough. Is that credit report joint or single? Do you have a loan app for each borrower? For any loan, is hyphenated content consistent throughout the loan file?
Fixing these inconsistencies across documents and data files using manual procedures is rife with human error. Rules-driven technology can capture these nuances much more efficiently. As opposed to reviewing all loan files to find the few loans that have discrepancies, staff can be focused on clearing conditions for only those loans that have issues.
The lack of standardization in MSR trading also limits business opportunities. Banks, lenders, and investors that buy MSRs want to conduct trades quickly and go back to market and do more—but every time they buy loans, they get a big blob of due diligence and document processing work on their hands, which means they can only do so many trades. Banks are particularly hamstrung, as they have special compliance burdens that make loan scrutiny more costly and time-consuming.
There are MSR trading platforms that make negotiating the deal easier, of course. But the delivery of the loans hasn’t been substantially improved on in years. Sellers are challenged in meeting buyer requirements, which means buyers must still process documents and address discrepancies through a combination of manual work and piecemeal automation that is still not robust enough.
Recently, however, a leap forward has taken place in the area of MSR transfers that could remove friction for all parties, including the consumer.
Taking the Leap
Over the past couple of years, some pieces have fallen into place that will help streamline MSR trades. The Industry Loan Application Dataset (ILAD), a loan application data “superset” based on MISMO v3.4, may make it easier for buyers and sellers to exchange loan information digitally. Until recently, however, the biggest piece of the puzzle—the development of technology that can normalize and standardize both MSR document and data sets regardless of loan origination date—has been missing. Now that piece has arrived.
By applying automated, repeatable processes, new technologies are streamlining MSR transactions by normalizing document naming and stacking, regardless of individual loan seller procedures or protocols. Technologies today can extract data from hundreds of data fields on dozens of the most common loan documents, enabling buyers and sellers to identify and resolve data inconsistencies more quickly.
The key to these technologies is machine learning-based automated document recognition (ADR) technology and automated data extraction (ADE) tools. Together, they can perform multiple automated tasks and power automated rules for any type of document check, document-to-document comparison, and document-to-data comparison. Configurations can then be used to define a buyer-specific naming convention and stacking order for onboarding into a servicing system. This is significantly faster than sending large files to an overseas partner and waiting days for the files to come back.
Essentially, these technologies consolidate the many manual steps in MSR transfers into highly automated steps, which are both configurable and capable of being universally applied to all loan files. They are also much more powerful than standard optical character recognition (OCR) tools. For example, on loans involving two borrowers, most current OCR systems often fail to detect whether there are credit reports for both borrowers or whether all the pages of the credit report are in the file. New technologies, however, can.
These technologies can be implemented at extraordinary speed, often in as little as two days. They also allow sellers to drop and drag loan files into a secure, online portal used by both buyers and sellers. If there is something wrong with one or more loans, such as a missing document, both parties can see the errors and address them through one communication platform.
Even if a buyer outsources its due diligence to a third-party provider, that third party can log into the portal and perform the required work in a secure, web-based environment. Plus, using APIs, data and documents can be easily integrated into a buyer’s servicing system.
This “leap” in MSR trading innovation has already been achieved with great success by Freddie Mac through its Freddie Automated Servicing Transfer (FASTSM). The technology behind this tool has greatly streamlined MSR transactions, saving time and lowering costs by standardizing and simplifying the exchange of documents and data during servicing transfers. FAST uses the same type of machine learning document processing automation and data extraction technologies mentioned earlier, which can create a verified and validated stream of data out of digital images and scanned documents.
The Real Prize—Lower Costs
Making MSR trades faster by applying automation to normalize and standardize the process creates enormous potential for cost savings, so much so that it might catch buyers off guard. If buyers can automate 80% of the tasks involved with processing documents and identifying and resolving data discrepancies, they only have to apply their full-time employees on the 20% that are the exceptions with conditions that need to be cleared.
This means they can significantly lower staff costs or simply deploy their teams to other work. For many buyers, this could also surface an opportunity to reduce the army of business process outsource (BPO) staff needed to work on large trades and the corresponding FTE expense. By automating MSR transfer tasks, buyers may find themselves with adequate in-house staff to perform due diligence work on their own.
There are other benefits, too. Because MSRs being bought require fewer manual touches, there’s fewer chances of errors. By shrinking the amount of time it takes for transfers to be completed and loans absorbed into their servicing system, buyers can move on to the next MSR acquisition much faster. That’s good news, as we have seen a spike in trades over the last 60 to 90 days.
These benefits seem to be why the number of buyers that are automating MSR transfer work has soared over the past year. In fact, our own platform, LoanLogics IDEA, provides 95% coverage of the top co-issuance buyers for MSRs. Just this past year, we have also ramped up support for the bulk market, assisting over 50% of these large buyers.
When these technologies are combined with data standards, they can revolutionize the secondary and servicing markets. By standardizing MSR trading tasks, bottlenecks created by the wide disparities in loan file delivery simply disappear.
Currently, the only barrier that remains is the willingness to change. As a matter of fact, for all its faults, there are still many people who prefer the DeLorean—even though today’s hybrid cars would leave it in the dust. But for most drivers, and for most mortgage participants, knowing what’s under the hood is what really counts.