DSC Meridian Capital LP is an opportunistic, event-driven credit investment firm founded by Sheru Chowdhry, the former Head of Credit Research & Co-Portfolio Manager of the Paulson Credit Fund (2004-2017). The firm seeks to generate absolute returns through the credit cycle. DSC Meridian will prioritize principal protection, liquidity and shorting at the top of the credit cycle, and then shift portfolio focus to capital appreciation and equity-like upside at the bottom of the cycle. DSC Meridian will invest both long and short, across small, mid, and large-sized capital structures, and across the full spectrum of credit opportunities from performing, stressed and distressed credit (bankruptcy reorganizations, litigations, liquidations) to post re-organization, event-driven equities and capital structure arbitrage. The firm integrates material sustainability factors throughout the investment process and pursues an active corporate engagement strategy to help shape positive sustainability-related outcomes.
DSC Meridian, Sustainability Spotlight Series:
AI for Good – Expanding Access to Consumer Credit Creates Value for Lenders and Borrowers
Featuring Dave Girouard, Co-Founder & CEO of Upstart
This interview with Upstart Co-Founder and CEO Dave Girouard was conducted by Paula Luff, Director of ESG Research & Engagement at DSC Meridian Capital. Opinions expressed herein are the own opinions of Mr. Girouard and Ms. Luff and are not necessarily the opinions of DSC Meridian Capital LP or Upstart or their respective partners, employees or affiliates. The interview has been edited for clarity.
Paula Luff: Dave, thanks so much for joining me today and sharing a bit about Upstart. We started the Spotlight series to highlight companies in our portfolio where sustainability considerations are central to value creation. Please give us your background and discuss your company, Upstart.
Dave Girouard: I grew up on the East Coast outside of Boston, but I’ve spent most of my adult life in the Bay Area. And, I’ve been in the technology industry for a long time with roles at Apple and Google. I left Google thirteen years ago to found Upstart.
As for why? Maybe the simplest explanation is that I was very interested in credit as this large part of our economy. I didn’t come from banking, credit, or lending, but I had a very intuitive sense that it was a system that had not changed in decades. It had not benefited from any of the advances in technology since the dawn of the Internet or even the PC.
We thought we could apply some of the technology we had developed at Google to improve access to capital. Credit is a foundational and central part of our economy; it is critical for families. If you can make the system work more efficiently and better for everybody, consumers and lenders, there’s so much potential to create value.
So, essentially, you sought to harness the power of what we now call “AI” to expand access to credit. I think I read somewhere that you have described this as “better math.”
Well, effectively, that’s what AI is. It’s better math. When we started, we always had the notion that more data and more sophisticated software were a way to build better risk models, which would lead to better outcomes. More data, more sophisticated algorithms. The term “AI” wasn’t really used a lot back in 2017, which is when we started using it on our website.
There were many internal discussions about whether to use the phrase “AI lending platform,” given the industry and the people we served. It’s a scary term to many people. But in the end, we felt strongly that it would be important to be out in front, advocate for the technology, and explain that AI is nothing more than sophisticated math. It’s lots of data, and in this particular case, it leads to very good outcomes.
You also identified a disconnect between default risk and the sort of traditional data that banks rely on, notably FICO scores, right?
Absolutely. At its first adoption, the FICO score was revolutionary. Prior to FICO scores, some thirty-five years ago or so, a prospective borrower would sit down with a loan officer and answer a bunch of questions, and it was up to the intuition of that loan officer to determine whether a client was creditworthy. FICO was pretty revolutionary in its day since it meant that we had a centralized system that gathered data on credit performance, etc. Not much has changed in the process since. If you think about building sophisticated risk models, trying to compress everything you could know about the risk of an individual in a particular type of loan into a three-digit number means so much information is lost.
A FICO score is very useful, particularly for consumers to understand where they stand in the industry. To build more sophisticated models, you need to move beyond FICO and understand as much as you can about an individual. In the early days, we added information about employment and education. That continued to evolve into a more sophisticated model that continues to learn more and more about borrowers.
Many companies talk about partnering with NGOs, the public sector, or other companies. What struck me is that you sought strategic engagement early on with several stakeholders, including the Consumer Financial Protection Bureau (CFPB). How did that relationship help ensure that your innovation would be fair and unbiased?
Well, I think many startups, particularly in Silicon Valley, take the “just build it first, fly under the radar, and address the regulatory issues later” approach when they are on the edge of a very regulated area.
Lending is heavily regulated, so we felt it was not a strategy that would ultimately be successful for us. Fair lending laws go back to the 1970s, and while technology has evolved, the rules are still fairly cogent and useful today.
We wanted to get feedback early in the game on how we planned to comply with fair lending laws in particular. Is the product fair, is it biased, and will it run afoul of these laws? So even before we launched our product, we naively marched ourselves up to the CFPB’s San Francisco office—not the policy people, by the way, but the enforcement team. We showed them what we were doing and asked for their feedback. I think the CFPB was a little surprised that we sought their input, but in the end, that upfront investment was very useful. We don’t have that many more people left to convince that what we are doing is a giant leap forward—it’s very pro-consumer and pro-lender—and we have built some very sophisticated tools working alongside the CFPB. Flying under the radar may have been faster in many ways, but the outcome we got to was great in the end.
One of the things that really struck us when we started looking at Upstart through the sustainability lens was the accolades the model received for actively avoiding bias, which is not what we normally hear about AI models. How did you solve for this?
The hardest part was actually defining what bias is since the laws are a bit obtuse in their definition. The law is definitely protective of certain demographics, but if somebody has a much higher income than somebody else, that person with a higher income is probably going to qualify for a bigger loan. So, the first thing we worked on with the CFPB was to develop a common understanding of how to measure bias. We decided that the right way forward was to measure outcomes. You can try to analyze inputs and which variables the model is using, but in the end, what really matters, particularly in a sophisticated system like this one, is what comes out the other side.
We built a rigorous system where we test every single application and every single loan using a waterfall of tests, and we can assess whether there is bias or not. If we detect bias, then we can take a step back and correct for it. A lot of bias sampling in lending is currently done by pulling a few loans off the shelf and seeing if they look OK. We knew we could do better.
So, Upstart has clearly brought a lot of value to consumers who might have to resort to payday lenders or other costly sources of capital. What about the other side of the equation? What value has been created for lenders?
At a very basic level, it allows them to offer a better product to more people. It is better in the sense that you can approve more people for the types of loans they are looking for—a car loan, a personal loan, or something else. It removes virtually all the friction from the process. About 90 percent of our loans are approved in a moment with no documentation to upload and no phone call. So, the burden on the consumer to get through the process is shockingly easy.
That also benefits the lender, who saves a lot on the cost of loan origination. It’s also a much more rigorous, modern system that provides a better experience and a better product for their customers. Lenders see better performance and reliability; a handful of large institutions have thousands of engineers who can build bespoke systems. But many banks and credit unions in our communities are not in a position to build these kinds of things themselves. They need partners to build a modern product that will appeal to the consumer. Being a technology provider to banks and credit unions across the country enables them to compete with larger financial institutions as well as other FinTechs. We provide a valuable service to lenders and the communities they serve.
AI is often described as a black box; What techniques has Upstart developed to ensure that what you’re doing is both explainable and accountable?
That’s a very common perception. When there are so many variables, it is easy to wonder why, for example, ChatGPT says what it says in response to a query. In our world, we have to justify every loan we make and every loan we decline. Again, we operate in a highly regulated environment. If someone applies for credit and you decline them, you have to explain why. The legal term is an adverse action notice. Back in the day, you could simply tell a prospective borrower that their FICO score was too low, or they couldn’t pay back this loan because they didn’t have a job.
But our system will generate easily understandable responses that look at all the various inputs to the decision, identifying the most important ones. What it really does is identify what the individual could do differently so that the outcome would be different next time. For example, your debt burden is already too high, or your income is either too low or a little unpredictable, which means that there is too much risk associated with this loan. These are technologies that we have developed but are also in the market, which are sometimes referred to as SHAP. They look at how the decision is made; they cluster the different parts of the algorithm in order to develop an easily understood response as to why the decision was made.
Which borrowers must appreciate since the reasons behind the decision are not always clear, right?
I mean, we certainly like to do better than “your FICO score was too low,” though a lot of our lenders just have a minimum FICO score that ultimately is the answer, and there is not much we can do about that!
Obviously, this is a field that is rapidly evolving. It’s also not very well understood by the average layperson. There’s a lot of talk about AI’s potential impacts—how it might displace workers and how it may also create new fields and employment opportunities. What kind of regulations should be in place, generally, to ensure that we fully mine technology’s potential for good while protecting us from bad actors. How do we balance smart regulation that enables us to deliver great products, minimize the potential for harm, and maintain our leadership position in technology?
Sure. I think my answer to the question is quite simple. Laws should be made that prohibit illegal activity rather than to prevent building products that have the potential to be used illegally. Even e-mail could be used for illicit purposes. The thing that has worked for us is that there were already laws on the books clearly prohibiting practices like redlining, for example. AI doesn’t change that—you have to comply regardless. AI doesn’t necessarily need its own set of regulations. If the application of the technology is illegal, it’s illegal. That should be the focus. A lot of the pieces are already in place. Laws against money laundering, for example, are already on the books. AI may change how you need to protect against things happening, but the simple answer is that we need to prohibit illegal activity.
Any final thoughts or lessons learned that you’d like to share with our readers?
We recently had an annual kick-off meeting with employees, during which we touched on some big-picture sustainability questions. Whether you like what a company is doing or not, how do you take a big step back and think about its success? Is its success good for humanity or not?
I like to think that if Upstart is as big and successful as possible, it will only be because we dramatically reduced the cost of borrowing money for millions, if not tens of millions, of people. Only good can come from Upstart’s success.
That’s not true of all companies, although I wish it were. I’ll just point out a few examples. Take social media. I’m not sure that if we all spend more time on our phones scrolling through social media, it’s actually a good thing. Or, food companies promoting unhealthy products. What about credit card companies encouraging us to spend beyond our means? And so, when I think about sustainability at Upstart, my mental model is to picture our success and ask myself, should people see our success as a positive or a cause for concern? I think with Upstart, the answer is a resounding positive.
IMPORTANT DISCLOSURES
THIS DOCUMENT IS NOT AN OFFER TO SELL OR THE SOLICITATION OF AN OFFER TO BUY INTERESTS AND/OR SHARES IN A FUND (THE “FUND”) MANAGED BY DSC MERIDIAN CAPITAL LP (“DSC MERIDIAN” OR THE “FIRM”). AN OFFERING OF INTERESTS OR SHARES WILL BE MADE ONLY BY MEANS OF A CONFIDENTIAL PRIVATE OFFERING MEMORANDUM (A “MEMORANDUM”) AND ONLY TO QUALIFIED INVESTORS IN JURISDICTIONS WHERE PERMITTED BY LAW.
An investment in the Fund is speculative and involves a high degree of risk. The Fund is not intended to be a complete investment program. DSC Meridian is a recently formed entity with a limited operating history and intends to employ certain strategies and techniques, such as investing in distressed securities, illiquid securities, short selling and the use of leverage and derivatives, which may increase the risk of investment loss. DSC Meridian’s investment program involves substantial risk, including the loss of principal, and no assurance can be given that the Fund’s investment objectives will be achieved. As a result, the Fund’s performance may be volatile, and an investor could lose all or a substantial amount of his or her investment. The Fund’s fees and expenses may offset trading profits. There can be no assurances that the Fund will have a return on invested capital similar to the returns of other funds or accounts with which Sheru Chowdhry was associated, due to, among other reasons, differences in investment policies, risk parameters, economic conditions, regulatory climate, portfolio size, leverage, fee structure and expenses. Past performance is not a guarantee of, and is not necessarily indicative of, future results.
The Fund also has substantial limitations on investors’ ability to withdraw or transfer their shares or interests therein. No secondary market for the Fund’s shares or interests exists or is expected to develop, so investors may not have access to capital when needed. All of these risks, and other important risks, are described in detail the Fund’s Memorandum. Prospective investors are strongly urged to review the Fund’s Memorandum carefully and consult with their own financial, legal and tax advisors before investing.
FOR EEA INVESTORS: This document does not constitute an offer of interests or shares in the Fund to investors domiciled or with a registered office in the European Economic Area (“EEA”). None of the Fund, DSC Meridian or any of their respective affiliates currently intends to engage in any marketing (as defined in the Alternative Investment Fund Managers Directive) in the EEA with respect to interests or shares in the Fund. Receipt of this document by an EEA investor is solely in response to a request for information about the Fund which was initialed by such investor. Any other receipt of this document is in error and the recipient thereof shall immediately return to the Fund, or destroy, this document without any use, dissemination, distribution or copying of the information set forth in this document.
FOR SWISS INVESTORS: The Fund does not need to be licensed for distribution to non-qualified investors with the Swiss Financial Market Supervisory Authority (FINMA) as a foreign collective investment scheme pursuant to Article 120 paragraph 1 of the Swiss Federal Act on collective investment schemes of June 23 2006 as amended (CISA). Accordingly, pursuant to Article 120 paragraph 1 of CISA the shares may only be offered, and an offering memorandum may only be distributed into or from Switzerland by way of distribution to qualified investors as defined in Article 10 paragraph 3 of CISA and its implementing ordinance, if the Fund has appointed a representative and a paying agent in Switzerland. Investors in the shares do not benefit from the specific investor protection provided by CISA and the supervision by FINMA in connection with the licensing for distribution.
FOR UK INVESTORS: This is neither an offer to sell nor a solicitation of any offer to buy any securities in any Fund managed by DSC Meridian. Any offering is made only pursuant to the relevant offering document, together with the current financial statements of the relevant Fund, if available, and the relevant subscription application, all of which must be read in their entirety. No offer to purchase securities will be made or accepted prior to receipt by the offeree of these documents and the completion of all appropriate documentation. This document is not an advertisement and is not intended for public use or distribution. It has not been approved under Section 21 of the Financial Services and Markets Act 2000 (FSMA). In the United Kingdom, this document is being distributed only to and is directed at (i) persons who have professional experience in matters relating to investments falling within Article 19 of the Financial Services and Markets Act 2000 (financial promotion) order 2005 (as amended) (the financial promotion order) or within Article 14 of the Financial Services and Markets Act 2000 (promotion of collective investment schemes) (exemptions) order 2001 (the CIS promotion order) or (ii) high net worth entities and other persons to whom it may otherwise lawfully be communicated, falling within Article 49 of the financial promotion order or within Article 22 of the CIS promotion order or (iii) certified sophisticated investors falling within Article 50 of the financial promotion order or within Article 23 of the CIS promotion order or (iv) other persons to whom it may lawfully be directed under an exemption contained in the financial promotion order or the CIS promotion order (all such persons together being referred to as relevant persons). Accordingly, this document is exempt from the general restriction in Section 21 of FSMA and from the restriction on the promotion of unregulated collective investment schemes in Section 238 FSMA on the grounds that it will be communicated only to relevant persons. Persons who are not relevant persons must not act on or rely on this document or any of its contents. Any investment or investment activity to which this document relates is available only to relevant persons and will be engaged in only with relevant persons. Relevant persons in receipt of this document must not distribute, publish, reproduce, or disclose this document (in whole or in part) to any person who is not a relevant person. “certified sophisticated investor”, for this purpose, means a person: (a) who has a current certificate in writing or other legible form signed by an authorized person to the effect that he is sufficiently knowledgeable to understand the risks associated with participating in unregulated collective investment schemes and of investments of the type described in this document; and (b) who has signed, within the period of twelve months ending with the day on which the communication is made, a statement in the prescribed form.