The Israeli fintech Pagaya partners with lenders, financial institutions, and fintechs to improve their assessment of underwriting and lending opportunities. Pagaya’s backbone is an AI-driven data network that simultaneously helps banks and fintechs originate more loans while creating investment opportunities for institutional capital investors. For this episode of Talking Banking Matters, we spoke with Pagaya cofounder and CEO Gal Krubiner. The following edited transcript presents highlights of the conversation, which took place in September 2022.
Matt Cooke, McKinsey: It’s not surprising that we should see a change in how loans are made, with the rapid increase in available data, easy-to-use APIs, cloud-based computing power, and AI capabilities. A number of companies have emerged to target the space, including “buy now, pay later” players, peer-to-peer and direct lenders, and lending marketplaces.
Pagaya is a six-year-old Israeli fintech that has taken a different approach. It’s a B2B2C player that operates at the nexus of two sets of customers. On the one hand, banks and other lenders use its AI-based technology to decision on and originate lending products to their customers. On the other, institutional investors deploy capital to acquire loans that originators aren’t able or willing to fund themselves—in effect benefiting lenders, investors, and consumers.
Since we spoke with Gal in September, Pagaya—like much of the fintech world—has faced headwinds, and its valuation in the public markets has dropped significantly from earlier this year. But as you’ll hear through the session, Gal remains an optimist and takes a long-term view to shareholder value creation at Pagaya.
Gal was just 26 when he started the company with two others. He was already a serial entrepreneur who had also worked in investment banking. We talked with him about why he and his colleagues started Pagaya, why they chose the model they did, and how he views the ongoing turbulence in the market and the company’s stock. But first, here’s Gal describing what Pagaya does.
Gal Krubiner, Pagaya: I founded Pagaya with Avital Pardo, whose background is hard-core data science, and Yahav Yulzari, whose background is capital raising. Our vision for the US consumer credit world was connecting technology capabilities—specifically AI and research to assess risk—with financial capabilities to help institutional clients manage money.
There is an endless amount of debt assets and a very high level of integrity in the data on the US consumer. A credit file of a person in the US is typically 17 years. So you can run models and build infrastructure, bringing millions of different trades with millions of different loans that have been provided in the past.
Even though hundreds of banks in the US did that for three or four more decades than we did, we saw the ability to create a better outcome fairly quickly. The opportunity was that the infrastructure was behind. Looking from this infrastructure perspective, we set out to create something to enable lenders to serve more consumers. We built an ecosystem that is basically bringing more efficiency to the US consumer credit world, and we’re able to drive it in a very complementary way into how banks work.
The opportunity was that the infrastructure was behind … we set out to create something to enable lenders to serve more consumers.
So Pagaya’s role is to build a B2B2C model that enables our mission of making life-changing financial products and services available to more people. But we decided to look at it from an infrastructure perspective—to be able to create something that enables others to be able to serve more consumers.
Matt Cooke: Pagaya’s model provides lenders an underwriting capability to help them risk-assess and price their loans. Second, for those loans that fall outside these lenders’ credit box, the model created a new institutional funding capability to hold those loans, so fewer customers are turned away. Here’s Gal to explain it further.
Gal Krubiner: The lenders can be fintechs, banks, or different financial institutions. What’s interesting is that these types of organizations still don’t know how to underwrite all the flow that’s coming their way. One reason is constraints caused by the cost of capital; think of banks that are not very effective below a 640 FICO, or the subprime. Another reason is lack of technology.
We thought if we could combine two capabilities—AI’s ability to connect to these organizations and to assess—and be able to match those with liquidity less dependent on the actual cost of the different types of bank regulations, we could provide a meaningful lift to the amount of bookings these departments can do from the flow coming through their door.
That actually has two outcomes. One, it’s creating a unique asset flow, so that includes creation of consumer credit, auto loans, and others. And two, we created more customers that actually went to the bank. The bank got an enhancement of its ability to market itself to more customers and not actually take the weighted average cost of capital on its books because the risk is not sitting with them.
Conceptually, we decoupled the relationship from the asset. While the asset is traveling in our network and sitting in the asset investor space, the relationship always stayed with the lender the consumer came with. This approach and the fact that we don’t have a B2C brand created full alignment between us, the banks that want to book more loans and have more relationships with consumers, and the asset investors who don’t care about the relationships and just want to create return.
Conceptually, we decoupled the relationship from the asset. While the asset is traveling in our network and sitting in the asset investor space, the relationship always stayed with the lender the consumer came with.
That was built in a very forward-looking tech capability that we created an API plug for, which connects into the loan origination systems of all the banks and the fintechs that we have and proposes in real time an automated process to be able to identify which type of borrowers we actually want to provide credit to. And they can become customers of those partners while we take their risk and place that with the asset investors.
So it’s a situation where, at the same time as consumers enjoy getting more credit, the partners win because they have more relationships and actually get revenue from us. The investors also get a diversified exposure to unique asset classes and the ability to deploy capital at scale, which is important for them. Because we focus only on the underwriting and the capital, we free up a lot of resources for them to do marketing and servicing.
And that has allowed us to travel in between markets. We started in personal loans, then moved into auto loans, point of sale, and then credit cards.
Matt Cooke: It’s starting to look like Pagaya is working to position itself as a network rather than simply a service provider. While on one hand, it hopes to serve a growing group of lenders with customers looking for loans, on the other, it hopes to find an increasing number of institutional investors looking for yield.
Gal Krubiner: Because we’re sitting in the back end of the ecosystem and connecting to many lenders, we’re experiencing a unique situation. When the market is opening up or closing, as it is today, the banks become even more conservative. So the flow and the need for Pagaya become bigger. Those types of changes stabilize our ability to operate over cycles as much as possible in a very rapidly changing world.
Matt Cooke: Pagaya’s model raises an interesting question of risk sharing. With multiple parties involved, there is a potential misalignment of incentives that can arise between the entity pricing or servicing a loan and the entity holding the credit risk on that loan. In Pagaya’s case, it has a set of investors, on one hand, who are the risk takers and an originating bank, on the other, servicing the loan, including collections, with Pagaya in the middle of each transaction. We asked Gal to talk about how he balances the interests of investors, the banks and servicers, and the end consumer.
Gal Krubiner: All pricing is set by us—the full set of pricing, maturity, credit box, and AI. The bank is adopting our AI model and pricing as a recommendation; they are the originator based on our recommendation.
These things are updated every quarter or so. We brought in a chief risk officer from Bank of America, where he was the head of risk modeling in the consumer bank, and built a process with the triple-A-like standards the banks require.
The two main issues are typically fair lending and model validation, and we cracked the code on both. You do a certain level of due diligence if you think an organization doesn’t understand the consequences of reputational risk, which is very crucial for its ability to fund generally. As we are tapping into the banks, this is the less common theme but something that they still used from the world of ABS [asset-backed securities], etcetera. So that has already been standardized. Having said that, oversight, a very rigid understanding, and taking actions are very important.
Let me give you one example. During the COVID shutdowns, one of the biggest imperatives was to be able to determine whether and how to help borrowers who had lost their source of income and needed a deferment, especially in the first 45 days of the shutdowns. We were able to quickly develop some best practices for supporting these different types of deferment programs. Some banks and platforms chose to wait for their customers to call and then assess whether to offer a deferment. Others took a more active approach. What they all learned quickly was that their call centers got overwhelmed. So the big push was to convince them to develop the tech capabilities to do all the deferment payments online without a long validation time, so that they could keep the phone lines open for customers who needed anything else. They were monitoring the phone queues and were able to see how effective the technology was in helping reduce that burden on their call centers.
Matt Cooke: With this much automation, lenders often worry that working with AI and machine learning makes it more difficult to identify and eliminate biases in the models. We asked Gal to talk about how Pagaya addresses this issue so that the banks it works with comply with regulatory requirements like the FDIC’s fair lending laws and the Consumer Financial Protection Bureau rules.
Gal Krubiner: In the US, there are good practices already, showing that a superior model does not affect any minority population, for example. Theoretically, you can do that check on every model. That ability to have an explanatory piece is becoming more mainstream.
You can have processes such as identifying the top ten or 20 parameters. One thing we felt the banks were struggling with is the fact that in an AI model, all the parameters are relevant because you don’t have regression—the reduction of the effectiveness-of-information ratio. There is really no reason to limit them. Our first models were built on over 1,600 parameters.
One of the projects internally, Pathfinder, which is to help banks get more comfortable with automation, was to reduce the amount of attributes in a model from 1,600 to 266. Sometimes the compliance team will come and say, “We want 15 or so parameters,” but we help them understand very quickly that this doesn’t work well with AI. A good compromise is in the low hundreds of parameters. It takes a lot of phone calls and a lot of explanation.
Efficiency in the world is coming through technology, and the more we can get these tools into the different parts of the system, the more we can get a result that is a better outcome for all of us as a society.
Matt Cooke: Most large banks tend to want to develop their credit risk models themselves, especially since they have more data than smaller banks, given their larger customer bases and range of products. But once you get past the four largest US banks, each with over a trillion dollars in assets, there are another 4,000 or so banks in the United States and slightly more credit unions. We wondered whether Gal sees these smaller institutions as a source of future growth.
Gal Krubiner: There are two sides to that: funding and sourcing. From a sourcing perspective, the answer is no. The small banks in the US are in a tough spot. Scale is ten times more important in today’s world, and that disadvantage is happening over and over. Not to say there aren’t good midsize banks with specialization that will be very strong. But in the long run, I’m not sure there is a lot from a sourcing perspective.
On the pure bank funding model, definitely. I think that’s actually an advantage for them because today, through players like us, they can have simple access to assets to generate revenue and returns. There is a question of how much that model will allow you to become a major player rather than just a service provider, to keep asset flow coming through the door. But that’s the place where I think it will be more meaningful for them and for us.
Matt Cooke: With economic indicators seeming to point toward a slowdown or even a recession, Pagaya faces some uncertain times. Rates have been rising, though economic indicators at the end of 2022 pointed to a potential slowing of increases. And we could start to see delinquencies rise, too. According to one analysis, while consumers still have $1.3 trillion in excess household savings left over from government COVID support programs, those balances are already winding down. These are times when lenders separate into winners and losers. Gal talked about current macroconditions and how Pagaya’s model will respond to changes in the environment.
Gal Krubiner: I think it’s becoming all about relative value. US consumer credit comes with a lot of competition. Typically a borrower has between five and seven offers. So thinking that a lender is only as good as how aggressive it is on different days is not that simple. You have two options: to increase or decrease the price. If you increase the price too much, only the ones who didn’t find the right outcome will end up with you.
Matt Cooke: What Gal is referring to is the age-old problem of adverse selection. At any point in time, lenders who price too high are left with desperate borrowers who have been turned down from cheaper alternatives, and lenders who price too low have depressed margins and could lose money. This balance is constantly shifting as the credit environment evolves. Those lenders who can anticipate changes early and adapt pricing and credit criteria accordingly will have an edge.
Gal Krubiner: The understanding of the macroeconomic environment and the availability of funding is a crucial part of being able to be a good lender and show consistent returns. One of the advantages we have is that now we are as good as how we react and how we train the new models. There are more steps to that, but our unique vantage point allows us to react more quickly. And if you will notice our Q2 earning reports, our peak was already in October [2021] because of the deterioration that started in April.
Next time, we’re going to be better, and we’ll recognize that hopefully in August or September. But the bottom line is that the ability to reduce that until March by 30 or 40 percent is a meaningful outcome. That is what you would expect from your lender or asset owner or manager.
There is so much you can do with relative value, and we are all living in the world of macro, where the Fed is deciding the cost of capital is four or six, and all the markets follow. But with unique tech capabilities, you can take a better approach.
Matt Cooke: Since its IPO in June 2022, Pagaya’s stock price fluctuated, ranging from below $3 to above $34 over its first three months, which some analysts attributed to its low public float. In late September, a secondary offering of 46 million shares and a lockup expiration for early stakeholders seemed to trigger additional concerns that sellers would be flooding the market. At the time of our conversation, Pagaya’s stock had hit $2.20.
Gal Krubiner: Let’s clear the easy answers and the facts. Lockup came up two days ago [September 2022]. Most of the investors are from the early days, and in that environment, it’s hard to build a book of buyers. It takes time for people to see and learn and to be able to pull the trigger. That’s something we think will evolve into a solution in the next few quarters.
The most shocking thing for me is explaining what is happening in the world of investment. We all grew up believing there are two types of investors: long term and short term. A force like the Fed has a year of impact. Let’s call it two quarters on the short side, two years on the long side, more or less. That’s it: no more long or short investors, only short-term investors. It doesn’t matter which company; no one is able to keep positions because the world has shifted from growth to value. Even long-term investors are having a hard time following that type of precision.
When long-term players cannot be long-term players anymore, it takes time to readjust, reprice, and regroup. We experienced that a bit in 2008 after the shock. The big question was who are going to be the largest players in that industry afterward. What’s happening now is this reshifting of who the next long-term growth investors for the next ten years are going to be. And sometimes it’s as good as luck as choosing the five good names. Sometimes it’s someone that was so persistent throughout the time and just did the right thing.
The world of investing is in a freeze. People are not making decisions. That’s why we’re seeing all these valuations and discussions happening. As soon as the Fed marks the end, it’s going to push everyone to decide which position they want to take. Today no one is taking positions.
Matt Cooke: Gal was talking, of course, about the current investor focus on earnings versus the promise of long-term growth. While some factors driving Pagaya’s stock are unique to it, the current market turmoil for fintechs is enough to test even hardened leaders. Before starting Pagaya, Gal worked at UBS in London and Zurich, first in electronic currency trading and then with ultra-high-net-worth banking clients. We asked him to talk about what he learned in those previous roles when he started Pagaya. But he surprised us and instead talked about why the fact that he and his cofounders were friends will really be the basis for their long-term success.
Gal Krubiner: Let’s separate the vision that we spoke about from the decision to be entrepreneur. When I came back to Israel [from working in the United States], I talked to my two good friends, who have a lot of strong capabilities, about doing something together. It started there. Then I realized that a company is a way to spend 12 to 16 hours of your day with people who share the same values, have the same beliefs, and have an influence on who you are becoming.
Having people around you who are good-hearted, always challenging you, learning and trying to be the best—that’s influencing who you are and how you grow. That was the biggest motivation for me and for my two cofounders and how we could create the right ecosystem that would continue to develop us for many years to come.
Having people around you who are good-hearted, always challenging you, learning and trying to be the best—that’s influencing who you are and how you grow.
Matt Cooke: Since we spoke with Gal, the unsettled economic trends affecting so many fintechs had also touched Pagaya. In its recent third-quarter results released in November, Pagaya’s network volume had grown by 26 percent versus a year earlier, and revenues grew by 49 percent. But the slower credit environment was also taking its toll; volume and net revenues were actually lower than in the second quarter. And EBITDA swung from a profit of $5 million to a loss of the same amount in the latest quarter. Even before earnings were announced, Pagaya’s stock had slid further since our conversation. And as of mid-April, it was trading at just under a dollar per share. According to Eugene Simuni, an industry analyst at MoffettNathanson, investors in fintech companies will need to see the resilience of the credit and business models under stress scenarios before they jump back into these stocks. At the end of the day, businesses succeed based on individual choices. As the credit environment and interest rates continue to pressure the market, the shakeout between winners and losers will now be driven by a combination of lending experience, technology, adaptability, and sheer grit. In Pagaya’s case, the outcome will also depend on how well the company can convince investors that it has all of these attributes and more.
