This thought piece provides a rundown of some of the primary misconceptions about the marketplace lending industry. Investors in this asset class should not only be aware of these issues, but able to understand them at a bit more depth. Investing in marketplace lending is not without risk, but savvy investors will come to understand what information should be included in a rigorous due diligence process and what should be disregarded as unfounded.
- “Marketplace Lending hasn’t been tested through a credit cycle.”
- “Marketplace Lending platforms are incentivized to originate.”
- “Impossible to add alpha.”
- “Investors have platform bankruptcy risk.”
- “Default rates have been trending higher, driven by looser underwriting.”
- “Early returns are indicative of the return an investor can expect to make over the life of a portfolio.”
- “Borrowers taking out online personal loans must be risky.”
- “There is no collateral, so borrowers have no incentive to pay.”
“Marketplace Lending hasn’t been tested through a credit cycle.”
Quite possibly the most common message coming from conservative investors regarding this asset class, this misconception deserves an entire whitepaper by itself. However, for the sake of brevity, this paper will focus on two main points: a) consumer credit as an asset class, and b) marketplace lending from 2007 through 2010.
Consumer credit has existed in various forms for generations, and in a very familiar form as unsecured credit card debt for several decades in the United States. It has survived multiple credit cycles and a large variety of recessions, inflationary and deflationary economies, and dynamic consumer credit markets. Yet, over time, the spread between interest rates and default rates on credit cards as a whole has always remained positive.
Thus far, most marketplace lenders have opted for term installment loans as opposed to lines of credit. This product, too, has existed for decades from major banks, albeit in lessor volume than credit cards for the most recent few decades. That said, marketplace lending unsecured consumer loans have their own unique aspects, so one can always argue they are different and could be vulnerable to unforeseen risks.
Even so, Prosper and Lending Club were founded in 2007 in the US, and Zopa in 2006 in the UK, thus originating loans for almost a decade now. In that decade, we have seen both falling and, now, rising interest rate environments, and a huge swing in unemployment and macroeconomic trends. Vintages from 2006-2010, while smaller than today, still contain thousands of loans that can be studied closely. A quick glimpse at Lending Club’s 2007 and 2008 vintages combined show positive performance overall. Considering these were primarily FICO-driven, unsophisticated version 1.0 underwriting models, the results provide a strong floor for what could be expected in future similar situations.
“Marketplace Lending platforms are incentivized to originate.”
Also referred to as the “platforms have no skin in the game” argument, the general concern is that platforms generate a majority of their revenue via origination, either in the form of an origination fee to the borrower or a loan sale at a premium to a loan buyer. This misconception has a varying degree of responses by platforms. Some platforms have found ways to have “skin in the game” – as alternative finance traditionalists refer to it – by holding whole loans or portions of loans they originate. While a sound solution, concerns around fair allocations and the proper amount of “skin” will always exist. Others have aligned fees in accordance with the performance of the portfolios. This is primarily done through servicing fees, but some platforms have charged performance fees on fund structures they have set up.
However, the most important aspect of all of these marketplaces is that buyers of loans will disappear if the performance does not support the investment thesis of buying the loans. Yes, there is a delay between origination and the initial underperformance of a vintage before loan buyers can react. In the very short term, platforms are incentivized to originate. However, a great thing about these assets is their short duration and thus fast feedback loop. Buyers can often see how a vintage is performing within just a couple of quarters. Any platform wishing to survive longer than a few quarters cannot sacrifice on underwriting for short term opportunity and incentives.
This type of situation has happened with several platforms already. Some platforms have been able to avoid it – or even hide it – by taking more balance sheet risk. However, in fact, the more of a true marketplace the platform is – meaning, the less balance sheet risk it is taking and the larger percent that is sold off – the more reliance it has upon buyers of loans and thus less incentive it has to loosen standards and take risks.
“Impossible to add alpha.”
Before arguing for alpha, one must set a baseline for the beta. Generally, the beta of this asset class should be referred to as the average return of all assets in the asset class, meaning all platforms and products available. If we focus solely on US unsecured consumer credit, we can generalize by setting the beta as the average return of Lending Club, Prosper, Avant, SoFi (consumer unsecured), and Marlette, as they originate more than 90% of the total volume combined.
Next, we must determine a way to quantify alpha. The return on a constantly revolving portfolio, especially if growing or shrinking as new capital is infused or removed, is an inaccurate measure. The most accurate measure is to review the final IRR on fully realized vintages. Since term loans have 5-year, and now 7-year (SoFi), durations, it’s a long wait to determine outperformance. However, one must be careful to compare vintages in-flight, as these loans have non-linear conditional default and pre-payment curves. Simply put, the IRR of a higher interest rate loan portfolio against a lower interest rate portfolio may look better at the beginning, may look worse in the middle, and may look better at the end just due to the nuances of default curves and expected cash flows. If expected conditional default rates are relatively accurate and portfolios are adjusted for credit quality and interest rate differential, quarterly vintages can be fairly accurately compared.
Rather than providing example portfolios that have proven alpha, let’s examine how this is possible. First, at a high level, if one of those platforms returns outperforms another, an astute investor can provide alpha simply by picking the right platform. Next, if one product within each platform outperforms another product, simple product selection can provide another layer of alpha. Third, if the platform allows for grade level selection, properly selecting the best-performing grades against the underperforming grades will provide the next layer of alpha. And finally, selecting the best X% of loans within each grade will provide the most refined layer of alpha. It is important not to take for granted each one of these layers, as significant value can be added at each level, providing cumulative alpha that significantly outperforms the industry beta.
“Investors have platform bankruptcy risk.”
It is easy to mentally link the loans originated by a platform with the performance of the equity of the platform itself. When Lending Club went through some corporate turmoil in early 2016, many investors were quick to assume the loans themselves were bad or at risk as well. If not properly structured, this could, in theory, be the case. However, most platforms and investment managers have taken extensive measures to remove any platform corporate risk from the underlying loans themselves.
The first method of doing so includes removing custody of the loans from the platform’s balance sheet. This is primarily done by institutional investors who transfer custody of the loans to a third party custodian, place loans within a bankruptcy remote vehicle, or a combination of the two. In a simplified explanation, this protection is done with the intent of removing risk of the loans being seized by a bankruptcy court.
The next step would be to protect against servicing issues. Almost all marketplace lending platforms maintain master servicing rights on the loans they originate, especially with consumer unsecured loans. The risk with a platform bankruptcy is that the primary servicer would cease being able to collect on payments. Here, most platforms and most managers engage backup servicers that are ready to step in and service the portfolio were anything to happen with the primary servicer.
“Default rates have been trending higher, driven by looser underwriting.”
Most platforms, including Lending Club, have acknowledged that pockets of borrowers have been underperforming against expectations, beginning in late 2015 and running into mid-2016. The general assumption has been that these recent vintages were underperforming because platforms have been primarily incentivized to originate and have loosened their underwriting standards.
There are two methods for disproving this assertion. The first is to rigorously analyze the underwriting methodology used in recent vintages. The second is to review vintage performance accounting for vintage age.
With the first method, analysis must be done independently, as no marketplace lending platforms will provide their full underwriting methodology, publicly or privately, outside of general high-level limits. These limits may include a FICO floor, debt-to-income ceilings, a cap on recent credit report inquiries, and minimum years of credit history. Beyond that, external analysis must be done to monitor trends in stratification of platform-identified borrower risk (interest rates). Most analysis completed thus far has yet to identify any areas where the major platforms loosened underwriting standards significantly through 2015 vintages on a risk-adjusted basis.
Secondly, to review actual vintage performance, we can compare each vintage at its present month’s age versus historical vintages when they were the same age. But, what to compare? A cumulative or absolute default rate, on which most analysis has been done, taints the data because of what has happened previously on that vintage. A conditional default rate gets closer, but it is still slightly tainted by prior vintage performance. However, comparing percentages of non-performing loans can get an analysis quite accurate, as long as roll rates – or the percent of loans moving from one late stage to another – are holding steady. It also provides a larger sample set than only comparing charged-off loans, as there are usually 2-3x the amount of non-performing loans than loans charging off in a given month.
Thus far, analysis completed using the above strategy has been able to identify a clear wave of worse performance across all vintages relative to where each vintage should be. This decline in performance began around October of 2015. The exact buckets of borrowers driving this performance decline has not yet been identified absolutely, but in general it seems to be borrowers that have over-levered themselves due to the increase in available credit globally. However, performance began to improve beginning around May 2016. The improvement may be due to more efficient collections, a lessening of available credit to consumers in general, or tighter current underwriting, but it is most likely a result of these changes combined.
“Early returns are indicative of the return an investor can expect to make over the life of a portfolio.”
One of the most important aspects of the return profile of this asset class is the non-linearity of defaults. Simply put, defaults do not occur at a constant rate over the life of a loan. A quick example of this is to think about how many borrowers you would expect to default in month 1 after taking out a loan, versus, say month 12. Clearly, most borrowers are in a good position to make their payments just a month after taking out a loan, or at least the underwriter thought this was the case. But, a lot can happen in 12 months, so there is a greater chance that the borrower may lose their job, have another life event, or run into some other issue that pushes them into default.
Cash on cash returns will be higher in the first few months of the life of a vintage of loans, and will generally continue to decline over time. Below is a model of all Lending Club loans, showing their cash-on-cash returns on a monthly basis. It is the calculation of the amount of interest received minus the amount of charged-off (defaulted) loans in each month, divided by the start of month outstanding principal.
There are a variety of methods for adjusting for this steady increase in defaults in order to predict a more accurate monthly return based on default expectations and vintage progress. However, if just using a cash-on-cash analysis, or even a linear expected default rate, to guide your performance expectations, one must be shrewdly aware of the effects of an aging portfolio.
“Borrowers taking out online personal loans must be risky.”
No good or sophisticated borrower would go to a website to borrow money, right? Any smart borrower would go to their bank, right?
There are both untrue and unfortunate answers to both of these. To start with the former, many smart borrowers actually have taken personal financial management into their own hands, tracking expenses and income on sites like Mint.com, tracking their credit scores on Credit Karma, and intelligently shopping for loans on sites like Lending Club and Prosper, or for credit cards on aggregator sites. On the flipside, these online platforms have gone offline, using direct mail (and even TV/radio ads) as their primary point of contact with many potential borrowers. While the application is still completed online, completing a form online is much less of a turnoff to borrowers than being solicited online.
And, unfortunately, personal bank loans have almost disappeared from the opportunity set for most borrowers since the late 2000s. The cost to a bank to underwrite a small dollar personal loan in a branch office has become extremely expensive. It takes many man-hours and forms to complete the process and is simply a loss-leading business. Banks are also heavily regulated on holding these types of assets. From a capital adequacy and regulatory burden standpoint, it is really only beneficial to hold the highest end of the prime and super prime debt on their balance sheet. To that end, banks have been actually shifting towards partnerships with these marketplaces as opposed to competition. Banks have been happy to buy only super prime loans from the platforms that have been originated within banking guidelines. Some banks are even directing their walk-in branch customers to marketplace lending platforms when they request to take out a personal loan, which the bank then may re-purchase from the platform later.
All in all, the online borrower is much higher quality than most believe, and in actually is becoming the mainstream borrower as it becomes the best or only choice for installment loans.
“There is no collateral, so borrowers have no incentive to pay.”
Unsecured lending is not new. It has existed on the back of creditors’ trust based on a borrower’s reputation and borrowing history since the dawn of trade. Today, the United States has the most robust credit bureau, credit reporting, and credit scoring system in the world. Fortunately, marketplace lending platforms are a major part of that system.
All proper marketplace lending platforms report both positive and negative reports to one or multiple of the three major credit bureaus: Experian, Equifax, and TransUnion. Positive reports may mean that a borrower is rewarded within the FICO scoring system for making payments on time. However, negative reports have the opposite effect.
If a borrower misses a payment, multiple payments, or eventually defaults on a loan, negative reports to credit bureaus can do extensive damage to a FICO score that may be very difficult to repair. The impact of a default could almost singlehandedly prevent a borrower from being able to get additional credit. This carries through to getting a mortgage, getting a new car, or even getting a new phone. This fear of losing credit is by far the most overwhelming factor toward maintaining the over $1 trillion unsecured consumer lending market in the United States.
Controversy does surround the priority put on unsecured credit in a crisis. If funds were depleted but not entirely absent, would a borrower chose to make a mortgage payment over a car payment? Or a credit card payment over an installment loan payment? Does the borrower even have a choice? There are many theories around what decisions would be made in a recession or even what individual borrowers do at present. However, very few facts support each theory. In reality, there are many factors that go into each borrower’s decision on what to pay first, when, or how. Many of these theories may be right and many wrong, but most likely it is a confluence of all of them. But, whatever situation may arise, one thing continues to hold true; all things considered, a vast majority of borrowers prefer to make payments rather than suffer the implosion of their credit score.