How to rate the risks of peer-to-peer lending, the newest bubble

BY Doug Dachille  January 1, 2014 at 12:45 PM EST

The careless lending practices in peer-to-peer lending may be a harbinger of greater problems to come in our financial system. Photo by Chris Hondros/Getty Images.

The stock market had an incredibly good run in 2013: The Standard and Poor’s 500 Index closed Tuesday up 30 percent for the year, far surpassing the 7.3 percent it was expected to climb.

But as we examined in our Making Sense segment on the market’s recent highs, there are plenty on Wall Street who are leery of their apparent good fortune. Wall Street money manager Doug Dachille wasn’t shy about calling the market overstimulated. And he identified a new bubble, whose risky lending practices, he wrote on the Business Desk, could be symptomatic of greater troubles for the financial system. Dachille, trained as a medical doctor before founding First Principles Capital Management, warned about investors carelessly lending to borrowers over the Internet in so-called peer-to-peer lending. But the feedback Dachille received on that first post troubled him; not enough people were aware of the loan risks.

Doug Dachille: Last month on the Business Desk, I wrote about the newest bubble: peer-to-peer lending over the Internet. Although still relatively small, the peer-to-peer lending bubble shares many of the hallmark signs of the bubbles that undermined our financial system in 2008. I’m worried that the careless lending in this Internet phenomenon may be a harbinger of bigger problems to come in our financial system. The post received a number of comments which merit responses so that folks can be better educated about the potential loan risks and be aware of additional factors they should consider when making these loans. I was particularly concerned with one commenter who indicated that 25 percent of his retirement savings were invested in peer-to-peer loans.

If we learned anything from the financial crisis it’s that investors need to take full accountability for performing appropriate due diligence when making investments. At a minimum, they should carefully read all documentation and offering prospectuses and independently assess the validity of any third-party data, such as credit ratings, which may be used in the investment decision process.

Investors also forgot during the period leading up to the financial crisis that the utility of historical loan performance and asset price data is limited; only looking in the rear view mirror is as unsafe for driving an investment decision as it is for driving an automobile.

Performing this kind of due diligence is not always easy. Increasingly, the risk factors outlined in financial prospectuses resemble the daunting list of risk factors described on medication labels. Given the length of the list of risks disclosed, most consumers have become completely desensitized to them, with many no longer reading the risks at all. Since it is difficult for the typical consumer of medications and financial products to distinguish which risks are the most likely and require serious consideration from those that are remote but simply included to reduce legal liability, the entire risk factor section becomes white noise. Only highly trained physicians can read a package insert for a medication and quickly determine the relevant and important risk factors. The same can be said for financial prospectuses.

The LendingClub was the first peer-to-peer lender to register its offerings with the U.S. Securities and Exchange Commission. A careful review of the LendingClub prospectus should cause investors to question the reliability of the credit ratings it designates for the borrowers. Investors should wonder, is offering a specific-rated borrower a lower interest rate really justified?

According to the April 30, 2013, LendingClub Prospectus, in order for a borrower to qualify for a loan, the prospective borrower must have a minimum FICO score of 660. (This credit score is obtained from a standard third-party consumer reporting agency, and a report from any of the major consumer credit reporting agencies should be reliable.) Additional qualifications for the borrower include having a debt-to-income ratio (excluding mortgage) below 35 percent. The calculation of the debt-to-income ratio is based upon the prospective borrower’s debt reported by a consumer reporting agency and income that is stated by the borrower, which is unverified by LendingClub unless otherwise noted.

It appears that only a limited number of borrowers have their stated income actually verified. Here is an excerpt of the relevant section from page 15 of the LendingClub prospectus describing its verification process for borrower-supplied information:

Information supplied by borrower members may be inaccurate or intentionally false and should generally not be relied upon.

Borrower members supply a variety of information that is included in the borrower Member Loan listings on our website and in the posting reports and sales reports we file with the SEC. Other than as described below, we do not verify this information, and it may be inaccurate or incomplete. For example, we do not verify a borrower member’s stated tenure, job title, home ownership status or intention for the use of loan proceeds, and the information borrower member’s [sic] supply may be inaccurate or intentionally false.

Unless we have specifically indicated otherwise in a loan listing, we do not verify a borrower member’s stated income. For example, we do not verify borrower member paystubs, IRS Forms W-2, federal or state income tax returns, bank and savings account balances, retirement account balances, letters from employers, home ownership or rental records, car ownership records or any records related to past bankruptcy and legal proceedings.

In the limited cases in which we have selected borrower members for income or employment verification, for the nine months ended December 31, 2012, approximately 60.1% of requested borrower members provided us with satisfactory responses to verify their income or employment; approximately 10.2% of requested borrower members withdrew their applications for loans, and approximately 29.7% of requested borrower members either failed to respond to our request in full or provided information that failed to verify their stated information, and we therefore removed those borrower Members’ Loan postings.

LendingClub assigns 35 loan grades to each borrower loan request based upon the borrower’s FICO score and LendingClub’s “proprietary scoring model.” How accurate can this model be to generate 35 unique credit scores for borrowers when, with the exception of FICO, LendingClub doesn’t verify the accuracy of borrower attributes important to prospective loan performance, such as current income?

While LendingClub doesn’t provide many specifics about what factors into this model, it appears that results are primarily driven by the historical loan performance of the borrowers. How reliable a predictor of future loan performance is a model that does not adjust historical loan performance for verified changes in current employment status and income levels? Stability of employment and income are critical to prospective loan performance. Therefore, verified employment status, employment tenor and income are important borrower attributes to consider when assessing the likelihood of historical borrower loan performance as a forecast of future performance. Yet, these attributes are not always verified for LendingClub member loans.

LendingClub admits that when it attempts to verify stated borrower information, it cannot do so for approximately 40 percent of the sample borrowers. Is it possible that 40 percent of the proprietary credit scores are inaccurate? Since it is rare for a borrower to provide stated information that makes him appear less creditworthy (especially when he believes the information will not be verified), is it not possible that 40 percent of the credit scores may overstate the true creditworthiness of the borrowers, resulting in the offering of a loan interest rate that is too low?

It’s not the concept of peer-to-peer lending to which I object. My main concern is the execution of the lending process, as compared to other types of lending such as credit card debt. The higher interest charged on credit card debt is based upon the fact that credit card loans are unsecured with limited credit underwriting. Generally, the underwriting is limited to credit bureau information and FICO scores. Since the credit card companies don’t know much about the borrower, there is minimal fine-tuning of the interest rate charged on traditional credit cards based upon borrower specifics.

Certainly, as more information about borrowers is obtained, the risk premium attributable to the unknown should adjust accordingly.

Probably the most important piece of borrower-specific information needed to fine tune the interest rate charged on an unsecured loan is the debt-to-income ratio. Obtaining independently verified borrower income, in combination with credit bureau reports and other independently verified borrower attributes, should result in improved credit underwriting with a reduction in the risk premium implicit in the interest rate on a credit card. Therefore, a business model where lenders offer unsecured loans at reduced interest rates to borrowers who provide verifiable income and other relevant attributes makes sense. But the lower rates peer-to-peer lenders offer don’t seem to be based on verified data.

The concern I have with the current execution of peer-to-peer lending is that the intermediary companies are giving investors the impression that they are conducting greater underwriting on the borrowers than they actually are. It’s this impression of stringent underwriting that can lead lenders to offer reduced rates to these borrowers.

The sections highlighted above from the LendingClub prospectus should raise doubts in the mind of investors. If the underwriting process used by peer-to-peer lending companies provides no greater and more reliable borrower credit information than that obtained by traditional credit card companies, then lenders need to question why they are offering reduced risk premiums to these borrowers. If you, as a lender, are going to offer a borrower a lower rate than another unsecured loan alternative, you need to be certain there is sufficient verifiable credit information to justify the rate reduction.



Is the stock market a bubble in the making?


This entry is cross-posted on the Making Sen$e page, where correspondent Paul Solman answers your economic and business questions