According to data compiled by Mortgage Bankers Association, the total number of loans in forbearance as of April 19 was just shy of seven percent. The high number of homeowners looking for mortgage assistance has raised the alarm of a wave of delinquencies following in the wake of COVID-19. Mortgage servicers who attended a recent roundtable held by MBA and STRATMOR estimated that delinquencies could reach as high as 30 percent.
According to Neil Fraser, director of US operations for Paradatec, servicers who have not completed a thorough auditing of their MSR data have ample reason to fear that wave of delinquent loans: The inaccurate fee calculations that result when a prior servicer’s MSR data doesn’t match up with what’s in the actual loan documents can impact their clients’ long-term finances and expose the servicer to potential compliance infractions.
“What we’re seeing is, the regulatory bodies have not given up on going out to servicing organizations and checking things like late fees,” Fraser says. “We’ve seen situations where lenders are asked to go back five years through their archives and double-check every late-fee calculation that they perform, and every late-fee charge they think they should apply.”
That is a big ask, particularly in a corner of the mortgage space where the margins are low and resources are generally spread thin.
“It’s very difficult for a new servicer, the one that’s taking in that MSR transfer, to justify the cost of comparing carefully all of the information on the data tape against the constituent documents that make up the legal loan itself. In many cases, what tends to happen is there can potentially be a disconnect between the paperwork and what’s in the database, which is effectively your servicing system. So really, the servicing system is out of date or inaccurate with respect to the actual paperwork that pertains to the loans,” Fraser explains.
Fraser says servicers have a growing number of AI document recognition platforms to choose from, but many of them still require human verification for any information the system can’t confirm.
“It’s expensive to pay for two things,” says Fraser. “You’re paying for the software license and then you’re paying for some humans, often offshore, to go verify the exceptions in that process.”
In the last 12 months, companies like Fraser’s have been able to eliminate that human component, speeding up the document verification process considerably while also bringing down the overall cost. He says a servicer leveraging such an AI-powered platform could potentially verify a million loans in only 40 days and “reindex every single page of every document in your system” without the need for outside verification.
“Effectively, this then cleans up the problem of data quality inside your servicing system,” says Fraser.
Servicers never really know what they’re receiving when they accept a new stack of MSR data. Maybe it’s correct, maybe it isn’t. Maybe there’s a final note stuck to page 45 of a PDF the new servicer – and possibly the last one – doesn’t know is there. If a wave of delinquent loans hits servicers later this year, those knowledge gaps could prove costly. Servicers now have the tools to start filling them in.