The standard term for mortgages is 30 years, but it's rare for a loan to remain outstanding for that long.
Prepayments are an unavoidable part of the business for the mortgage industry. The median age of mortgages refinanced during the fourth quarter of 2015 was 6.7 years, according to Freddie Mac. That's just shy of the record high of 7.3 years during the first half of 2014, going back to 1994.
When a borrower refinances or sells a home and pays off the mortgage, it cuts off the flow of interest payments to investors and fees to servicers. But prepayments aren't all bad news for mortgage companies. They can actually be an opportunity for servicers to funnel borrowers back to their colleagues in the origination side of the business to keep their customers from going to a competitor for their next loans.
Servicers' portfolio retention strategies have typically used basic data like loan-to-value ratio and property value to identify customers who can benefit from refinancing, and proactively approaching them about the opportunity.
Years of historically low interest rates fueled a lengthy refinance boom, and borrowers who were able to take advantage of those low rates are unlikely to refinance. That's good for keep those borrowers in servicers' portfolios, but it comes at the expense of new origination business.
At the same time, rising home prices offer more options for homeowners who have delayed decisions on selling their homes or refinancing. Now, servicers don't just have to be mindful of refi-driven prepayments, but purchase related ones as well.
As a result, servicers are increasingly looking for ways they can identify borrowers who may be on the cusp of buying a new home by using enhanced analytics to further segment their customer base and adapt their marketing to meet homebuyers' needs.
Since existing borrowers are already versed in getting a mortgage, they tend to be a bit more comfortable in terms of shopping around. "They tend to be a little less sticky…a second-time customer is a little bit savvier and tends to be a bit more of a rate shopper," said Roelof Slump, a managing director at Fitch Ratings. "They may be more likely to use another entity if the rates are better elsewhere."
Just recapturing a small percentage of those borrowers can make the difference in a tight margin business.
"With the customers that we work with, if they can get a 1%, 2%, 3% lift in their retention rates, those are pretty significant impacts on their business. It's a benefit on the margin for sure," said Jerry Halbrook, president of the origination technologies division at Black Knight Financial Services.
The increase in prepayment speeds makes for a good news/bad news conundrum for mortgage bankers. "It's wonderful for originators but on the servicing side, they're taking it on the chin. We've been sending out our second-quarter valuations and the response is not all that wonderful; we're not real popular fellows these days because of this risk of prepayment," said Tom Healy, president of Level1 Analytics.
Interest rates are only one factor of whether a consumer can buy a house, but it's often what draws people to the market. Home prices are rising, driven by a
Servicers can look at the borrower's prior behavior when rates drop a certain level, such as 25 basis points, and notify their origination unit to say this person is likely to prepay and leave unless he or she is contacted.
For a refinance that is a fairly easy thing to do, Slump said. "Some servicer/originators who have that strong relationship with the borrower are able to defend their portfolio in the case of a move-up situation. But it is hard to do on a very large scale, that is the reality."
"Regardless of who contacts them, they want to get the best price. So I think that even if you were to be proactively reaching out to these borrowers who you think are about to purchase a home, you're ability to capture them may not be that successful unless you got the best rate in the market," he added.
Improved data analysis help servicers identify borrowers who need a new loan, but more can be done. "The industry needs to get better at predicting who is going to prepay and when. The models to date have gotten much better than they were a decade ago. But at best, we're becoming less and less wrong," Healy said.
Today's models now take into account things like credit scores and loan-to-value ratios. But more can still be done to improve accuracy.
"I think we need to recognize that prepays is a behavioral science and we're trying to predict how mortgagors are going to behave. And they [can prepay] for a lot of reasons other than just interest rate arbitrage," Healy said.
Level1 can look at a portfolio down to the loan level, but the demographic data about the borrower, like family size, is often lost when the loan moves from origination to servicing systems.
"It's a real shame because that kind of information would be really, really helpful, but by and large it's lost," Healy said. "I would strongly argue that that marketing information you pick up in the underwriting process should be automatically populated into the servicing system."
Since it is very difficult and costly to get updated broker price opinion valuations for an entire portfolio, Level1 looks at Federal Housing Finance Agency data and other housing indices to identify markets where "the probability is going to be higher that you are going to have people who are refinancing or selling and buying a new house," Healy said.
Given the low returns of the business, any information edge a mortgage banker can gain through analytics is a plus, he explained. So among the things Level1 clients are asking for now is for it to assign a prepay propensity at the loan level so they know which loans to target for a refi campaign. Plus, because the Great Recession is not completely out of servicers' minds, it is getting requests to look for the likelihood of loans becoming delinquency.
Just 10 years ago, servicers used analytics as a financial exercise to find where prepayments would come from; this was because of the FAS 157 accounting standard requiring a mark-to-market analysis of MSR values, he said. Now, they are starting to use analytics as an operational tool — which loans have a higher propensity to prepay; what's going to happen to servicing costs over time; what loans are going to become delinquent or stay current; and what 30-day delinquencies are going to go 60 days late or become current again.
No matter whether the customer is recaptured or not, any time a loan prepays it is a financial loss for the servicer, Healy said.
"We look at the prepayment rate and we look at the recapture rate. If a loan pays off, even if you recapture it, there is a negative economic impact to your organization. But it is much less of a negative than if it pays off and your competitor gets it," he said.
But there is still more data that can be brought into the models that can help servicers identify at a loan level basis those borrowers who might be likely to become move-up customers.
Black Knight launched its LoanSphere technology late last year, which includes a "Data Hub" that provides data and analytic products from within LoanSphere for originations, servicing and default. Then in June, the company
Users can access public records and multiple listing systems information through the Data Hub, said Halbrook. Because of this, Black Knight is "able to identify when a customer has listed their house for sale, which could be quite a while before they would actually tell the servicer they were going to move," he said.
Plus, Black Knight is working with its clients to build predictive models that use data sets combining home value appreciation, the borrower's equity position and the property's equity position in the market to develop a propensity score that allows the servicer to target people who are in a position to make a move up.
"I don't think many servicers have done this kind of analytics historically because it was very difficult to get the data," he said. "But we're starting to see a lot more of that and are working with several clients on those kinds of activities," Halbrook said.
Even so, using analytics is a best guess attempt at determining likely outcomes. "I don't think we're ever going to be 100% right, but we've become less and less wrong trying to predict who or what group, what cohort is going to have a higher propensity to prepay," said Healy.
For example, Level1 recently found that borrowers who have monthly payment automatically debited from their bank accounts are 50% more likely to prepay than those who still pay using coupon cards, he said.
While data can identify consumers who are in a financial situation where a move makes sense, there are many non-financial reasons that influence that decision as well, Halbrook said.
"There's a lot of knowledge and predictive capabilities in these analytics. It's an art, not a science, but it gives the servicer another set of data points about that customer and the behavior of that customer that they can use to try to retain them," he said.
While Motivity's clients have been using business intelligence on the origination side, it has only become of recent interest to those customers for their servicing portfolio.
Many of its midsize mortgage banking customers have started retaining their servicing rights "and so that was a request that was becoming more and more frequent was to get the servicing data to be able to do the types of things that Black Knight has done in their Data Hub. Our existing client base really wanted and needed that data set," said Motivity Solutions CEO Tyler Sherman.
And while the volatility of MSR values is a concern for lenders, they also wanted to become more predictive about the performance of the loans they hold the servicing rights on and not just react to events like persistent low interest rates that affect portfolio performance, he said.
Mortgage bankers want an overall view of their business because "using spreadsheets and siloed data sets are not acceptable tools in today's environment," Sherman said.
However, there are limits to the capabilities of analytics technology.
"We have found over time that models work very well within fairly narrow bands of reality. And once you get outside of those bands of reality, it's more difficult to predict mortgagor behavior," Healy said.
So if things go into uncharted territory, like negative rates? "All bets are off," he said. "You just don't know what's going to happen."
