Automation Attracts Investors Who Love Homogenous Loan Pools

Figure was founded in 2018 with a singularly ambitious goal: use technology to change the way that loans are originated and traded – automate underwriting to securitize homogenous loan pools, eliminating inefficiencies and origination risks to increase liquidity in the private-label secondary. If 2018 was the year that Figure pitched its premise, 2023 was the year that Figure proved it – investors have fallen in love with Figure’s loans. 

Last April, the company closed its first rated HELOC securitization ever – a $236.7 million loan pool of Class A and B notes, rated AAA and A by the rating agency, DBRS Morningstar. By December of 2023, Figure had closed its fourth rated securitization of the year. 

The latest transaction was backed by more than 2,600 fixed-rate open HELOCs with an aggregate principal balance of over $195 million and collective credit limits of over $204 million, with 75% of the loans having a 30-year term. The loans had a weighted average FICO score of 744, a combined loan-to-value (CLTV) ratio of 65%, and a debt-to-income (DTI) ratio of 38%. The notes were rated AAA, AA, A-, and BBB- by Kroll Bond Rating Agency (KBRA) – a first for KBRA.

More notably, the deal included 20 unique class A-D investors, including alternative asset managers, insurance companies, private equity funds, and hedge funds, 15 of whom were new to the Figure shelf of securities. To that end, it was the largest order book of any Figure deal and significantly oversubscribed: Class A was 5.25x oversubscribed; Class B was 2.95x oversubscribed; Class C was 4x oversubscribed; and Class D was 4.25x oversubscribed.

Watch it on The Interest: Expanding Automatic Underwriting

These numbers demonstrate the increasing number of price points for investors to buy into Figure’s loan pools and that there is more investor demand than supply of loans at each price point. As a result, the investor spread has tightened. AAA notes priced at 240 basis points and A- notes priced at 330 basis points are the tightest for any Figure transactions to date.

With the home equity puzzle seemingly solved and an initial public offering (IPO) rumored for 2024, programmers at Figure have been playing in a new sandbox. “We’re going to start to focus on using this exact same technology for jumbos and Non-QM,” says Frommer. “We’re actually talking to someone about potentially a DSCR product as well.” 

Without the guarantees of the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac, non-conforming mortgage borrowers – those with too much bread for the government’s spread – find themselves with costlier financing and fewer loan options. Similar issues plaguing home equity lending also plague Non-QM lending – high costs of origination, cumbersome underwriting, and worst of all, illiquidity.

“The ultimate goal is to use our technology to drive more liquidity for everybody in the capital markets and be able to have more people offer non-traditional, non-agency products,” Frommer explains. In non-agency markets, lenders need more certainty that investors will buy the loans they fund; investors need more certainty the loans they buy will perform. 

Figure can make that happen, at least on a pilot basis, by the end of 2024, says Frommer. And yet, the company’s long-term plans are even grander.

“We’re hoping,” she says, “long run, and we’re using HELOC as a use case, that really we can create a market that is very much like a TBA [to be announced] market where the homogeneity of the product that’s originated on the tech enables a lot more efficient purchase and ultimate trading of the loan because it’s 100% understood by everybody who is in the process.”

However, Frommer adds, “it’s a very, very difficult task."

Automate To Underwrite Homogeneity 

With manual underwriting, humans assessing loan files make subjective judgments based on a set of rules. “What we are doing,” Frommer explains, “is we’re taking rules that we’ve predefined over the six years we’ve been originating HELOCs – how we’ve been able to understand how certain characteristics drive performance of our loans.” 

Those rules are tweaked over time in accordance with data analyses – not artificial intelligence (AI) – that assess the performance of different cohorts of loans. When performance deteriorates or improves, the rules are refined, “but the rules at any given point are all exactly the same,” says Frommer. Though the underwriting rules are written by humans, all subjectivity has been removed from the decision-making process of qualification, inherently eliminating human error. 

The only time humans have their hands on the machine is when certain pieces of borrower information cannot be verified digitally. For example, if Plaid, a data transfer network used for cross-checking applicants’ financial data, can not verify an applicant’s income, the applicant can upload paystubs to be fed through Figure’s rules engine.

Fully automated underwriting lowers barriers to entry for both borrowers and originators by reducing the costs and complications of origination. But, Figure also employs automation outside of underwriting to trim other inefficiencies and costs in the loan process. Automated valuation models (AVMs) replace appraisals, lien searches replace full title and insurance, customer service agents replace loan assistants for contacting borrowers when information can not be gleaned from digital sources.

That being said, if an applicant does not fit Figure’s underwriting rules, “we’re not going to change anything in our system to allow you to somehow contort yourself to fit into the process,” Frommer continues. “There’s not anyone you can call here to make a discretionary decision about doing the loan for you. The process is the process on any given day.” 

Figure implemented that “no exceptions” policy so as to differentiate themselves from other lenders – and differentiate their rules engine from other automated underwriting programs – all in service to investors’ preferences. Fully automated, no-exceptions underwriting produces homogenization in securitized loan pools, which reduces uncertainty for investors – and investors hate uncertainty.

“There are other lenders who will allow more of an exception process, and we’re fine with that because the benefit of using our technology ultimately is to create this marketplace that can trade very efficiently because of the nature of the fact that the loans have been originated in a consistent way,” Frommer says. While she acknowledges that Figure cannot serve every borrower seeking a home equity loan, refining the rules on the back end is a way to serve more.

“In my mind,” Frommer shrugs, “I’m not sure why any investor wouldn’t like that process better because you know exactly what you’re getting.”

Overcoming Fears Of The Unfamiliar

Frommer says that many originators and lenders fail to grasp that they can make more money – and build their borrower base – by originating Figure’s HELOCs instead of purchase mortgages.

The abundance of use cases for Figure’s five-days-to-close HELOCs greatly expands the potential borrower base. The potential borrower base expands in a higher interest rate, higher home price environment where people cannot afford to move. Not only does Figure’s HELOC stand in for the cash-out refinance product, but also replaces the “huge” personal loan market.

The speed, ease, and lower costs making HELOCs an attractive alternative to personal loans for borrowers are precisely what make HELOCs a compelling alternative for originators.

Though loan balances are smaller – the minimum at Figure is $15,000, though their average is $90,000 –  Frommer says originators can do “20 times as many, 30 times as many” HELOCs in the same amount of time it takes to complete one purchase mortgage. 

One purchase mortgage carries the hope of one refinance, maybe a few referrals, maybe a home equity loan, somewhere down the line. Originating 20 or 30 times more HELOCs in the time it takes to originate one purchase mortgages means working with 20 or 30 times more borrowers, multiplying the future business to be gleaned from those relationships. 

“The smart lending officers, they get that,” Frommer shrugs. “They get the fact that this is just really fast and requires little work on their end . . . We definitely let all of our partners know, and all the lending officers know, that it is as simple as sending a link. It is very low-touch for them.” 

Educating partners and originators about the HELOC opportunity remains a challenge, the narrative that HELOCs are time-intensive, costly, and cumbersome to underwrite still difficult to quash. Originators need to understand the effort and time trade-off, Frommer says. It is a brand-new way of originating loans and people fear what they do not understand.

“I think that scares some of them,” Frommer levels, “because it’s so fast and we give them transparency into where the loan is. But, it’s not that same manual process that they may have with a loan that’s underwritten by a guy in their shop that they can call and try to push through the platform.”

Investors, seeing the opportunity, have quickly overcome the fear of the unfamiliar. 

Besides being homogeneously underwritten and originated, Figure can amortize the draw period over 30 years, lowering the monthly payments for borrowers. Though Figure offers various terms, the 30-year is most popular because of the lower payments. The longer, lower payment schedule makes Figure’s loan pools that much more attractive to investors.

Homogeneity Lowers Origination Risk

Frommer attributes the growing attractiveness of Figure’s loan pools to investors’ growing familiarity with the broader asset class of home equity loans – which have not been originated at scale since the Great Financial Crisis – and Figure’s fully automated rules engine.

“People have gotten comfortable that the process actually, in a lot of ways, is better than a process with any sort of human intervention,” Frommer says. The strong performance of Figure’s securitized loan pools proves the potential for automated underwriting to eliminate origination risk Figure’s hands-free approach also carries over to partners – the loans they originate with Figure’s tech “look exactly like” the loans Figure originates.

White-labeled loans are originated under partners’ names – like Movement Mortgage, the Loan Store, CMG Financial, or Guaranteed Rate – who white-label Figure’s technology and take the funding risk for those loans. “We tell them how much they need to fund,” Frommer explains, “then they put the loan up for sale to us and we’ll purchase it. If for whatever reason they want to sell it to somebody else, there’s an agreement where we potentially allow them to do that.” 

Figure has a much broader base of partners through their wholesale platform, Frommer says, which was launched in June 2023. Brokered loans close in Figure’s name. Brokers need to walk borrowers through some of the attributes of the loans “so that they can be doing the work that they need to do to truly be a broker,” Frommer explains, but then it is as easy as sending a link.

“If you get a loan with us, you have met those rules,” Frommer continues, “and there’s nothing that someone can do in terms of human intervention that can change those rules. We don’t allow for exceptions, so you have to have met those rules.”

Homogeneity means it does not matter who originates the loan; to investors, it all looks the same. An investor who buys Movement’s loans or Guaranteed Rate’s loans does not need to understand new underwriting teams or processes because their loans are underwritten in the exact same way – through Figure’s rules engine.

What does change from partner to partner, Frommer explains, is average FICO credit scores, combined loan-to-value ratios (CLTVs), or debt-to-income ratios (DTIs), on account of their customer base. These differences blend into different loan pools for securitizations.

A New Sandbox, The Same Toys

Frommer says that attracting more investors to the secondary market and reducing the cost to originate home equity loans has helped expand access to second-lien financing for borrowers across the mortgage industry.

But, the company is not angling for home equity lending dominance. “We actually saw a huge gap in the market,” she explains, “and nobody actually offering those loans in a way that worked for the customer and, quite frankly, worked for the originator, either.” Unlike the agency space, “originators aren’t necessarily comfortable originating loans if they don’t know the takeout’s going to be there,” explains Frommer. 

Despite traditional underwriting for Non-QM loans being slightly more complex than for home equity, including more regulation due to their closed-end nature, Frommer says Figure has already developed many of the rules necessary for exporting their HELOC rules-engine to other non-agency spaces. 

“There’s got to be some sort of market,” Frommer continues, “that develops where people have confidence in how these loans are underwritten.” Bringing the certainty of homogeneity to Non-QM originations will lower costs and risks, helping to attract more investors to Non-QM’s private-label secondary, improving the cost and access to financing for borrowers. 

“We’re going to focus on using our technology in the closed-end first lien space to originate sort of in the exact same way with our tech, in a homogenous way, where we can get investors comfortable that the loans all perform in the same way,” Frommer explains, the goal being, “no matter who originates them, the tech and the automation will drive the underwriting process and create that same homogeneity in the non-agency space.” 

The fact that banks have pulled even further away from originating non-conforming loans only amplifies the opportunity for Figure to bring its tech to bear on Non-QM lending, originating and securitizing homogeneity in order to build liquidity into the market. Because Figure’s partners already use Figure’s technology for HELOCs, the comfort with Figure’s process is already there.

“They allowed us to put new technology right into their organizations, and so they now have access to our technology, which can be easily modified to do a Non-QM loan,” Frommer says. 

Here, too, the performance of Figure’s loans will serve as the basis for building investors’ confidence in non-agency loans originated with fully automated underwriting. With some fine-tuning over the next couple of months, Frommer expects to roll out a Non-QM product with fully automated underwriting “in the back half of the year, at least on a pilot basis.”

Competing In An Automated Future

Differentiation can be difficult in the technology industry. It can be even more difficult in the lending industry. Everyone wants something cheaper. But, cheaper usually comes at someone else’s expense, be it the borrower, originator, or lender. 

Never does cheaper come at the cost to investors. A lack of liquidity in the secondary market only drives up costs in the primary market. Frommer says the long-term benefit of Figure’s efforts lies in making the non-agency market work more like the agency market.

“We’re creating a loan that is very easy for any investor to understand as long as it’s been originated with our tech because it looks like every other loan,” Frommer explains. “Similar to what exists in the TBA market with Fannie and Freddie, people can really get comfortable, hopefully, in the long run, buying a loan before it’s even originated.” 

The strong emphasis on process now allows Figure to replicate that process in other markets, like Non-QM and DSCR. Knowing what the loan will look like and how it will perform reduces many of the due diligence factors for investors and rating agencies. The impact of lowering these barriers is even greater in non-agency markets given the lack of government guarantees.

“We have a multi-stage process that we’re working on to create the TBA,” says Frommer. “The first one is getting a group of buyers comfortable buying all of our loans regardless of who our originators are and our common loan purchase agreement. We’ll probably have that done pretty soon. Then, ultimately, we’re going to introduce a guarantor structure and create something that looks much more like a traditional TBA that exists in the agency space.”

Where increasingly more companies are leveraging automation, few are leveraging blockchain. Even fewer are leveraging both, but Figure is one of them. The launch of Figure Markets – independent of Figure Technology Solutions – in mid-March came with the announcement of a new decentralized custody crypto exchange and blockchain-native security marketplace for trading in various digital assets.

Not only do investors trust Figure’s technology independent of the originator, “the fact that it was put on blockchain in an immutable fashion,” Frommer explains, “you can track all of the transactions and understand what has changed and what hasn’t changed” for individual loans and loan pools. 

Storing loan transactions on the blockchain gives investors full transparency into loan performance, enhancing the ability to efficiently monitor and trade these assets in the secondary market. Frommer calls this “a big benefit” given Figures’ ongoing goal of creating a permanent capital structure for non-agency securitizations.

“It’s not something that’s going to happen overnight,” Frommer admits, “but the steps to it, I think, will happen relatively quickly.” 

To read more on this subject read Raising An Ai Brainchild In The Mortgage Industry