Federal Reserve of Cleveland Makes Some Dubious Claims in a Report on P2P Lending

Last week the Federal Reserve of Cleveland released a report titled Three Myths About Peer-to-Peer Loans that had some pretty damning things to say about the P2P lending industry. They analyzed 90,000 P2P loans issued between 2007 and 2012 and compared these borrowers to similar people who did not take out a P2P loan. The biggest issues I had with the study was that I don’t trust the data set they used and they then made an inflammatory claim that P2P lending has predatory practices without backing up that claim with rigorous analysis.

[Side note: This report seems to be using P2P lending to mean the broader online lending industry beyond just Lending Club and Prosper so keep that in mind when you see me using the P2P lending term in this article].

I was surprised by the outcomes of this report because it contradicted a report from July by researchers at the Philadelphia and Chicago Federal Reserve Banks. That report was based on an analysis of Lending Club data and found that borrowers were getting lower priced credit and that Lending Club was expanding access to credit in underserved areas. The Cleveland Fed report said the opposite.

One of the things I disliked most about this new report was the how opaque their analysis was. They just said they used data provided by TransUnion “in which we observe about 90,000 distinct individuals who received their first P2P loan between 2007 and 2012.”. They provide no indication as to which platforms this data is based upon and then we see this very strange chart below showing supposed delinquency rates by year. This is where they lost me.

This chart shows that the lowest delinquencies from P2P loans occurred in 2006. Really? I am sorry but this is just plain wrong and I challenge the authors to show me the actual data this is based upon. In 2006, the only consumer P2P lending (or any significant online lending) platform in existence in this country was Prosper and their 2006 vintage was terrible. Several years ago I wrote about Prosper 1.0 (the period from 2006-2008 where Prosper allowed borrowers with FICO scores down to 520). This data was all publicly available at one point and I reported that of the 28,936 loans issued during those initial two and a half years 10,456 loans defaulted, a 36% default rate. Subsequent vintages at Prosper performed much better. When I see the table above showing those numbers I have to question the entire data set of the report because obviously their 2006 data is wrong.

Another area I was concerned about was their statements on regulation. Here is a paragraph in the report related to regulation of P2P lending:

On the borrower side, there is no specific regulatory body dedicated to overseeing P2P marketplace lending practices. Arguably, many of the major consumer protection laws, such as the Truth-in-Lending Act or the Equal Credit Opportunity Act, still apply to both P2P lenders and investors. Enforcement is delegated to local attorney general offices and is triggered by repeat violations, leaving P2P borrowers potentially vulnerable to predatory lending practices.

Much has been written about the regulation of P2P lending. Law firm Chapman and Cutler have produced a white paper every year for several years providing the state of regulation for P2P (now marketplace) lending. Here they go into detail on the various consumer protection laws that P2P lending platforms must adhere to and they specifically mention the Truth in Lending Act and the Equal Credit Opportunity Act.

The reality is that most P2P lending platforms partner with an issuing bank such as WebBank or Cross River Bank and these banks ensure that platforms are in compliance with all fair lending laws. I have never heard of any platform receiving repeat violations by any state attorney general so I am not sure where the authors of this report make the jump to predatory lending practices.

[Update: It was pointed out to me after I published this that there was one other chart that makes it even more clear that this report had nothing directly to do with P2P lending. The chart below shows the supposed growth of P2P lending. In 2009 they are saying the industry had outstanding balances of just under $60 billion. At the end of 2009, Lending Club and Prosper had issued less than $300 million in total loans in their history. Clearly, the Cleveland Fed is talking about something other than P2P lending here.]

To be fair the report does bring up some valid concerns. Here are the three questions the report is referring to that they consider as industry myths.

  1. Are P2P loans used to refinance previous loans?

They claim that a typical P2P lending borrower does not in reality pay down their credit card balances, but the opposite is actually true – they show an increase of 47% of these balances after taking out a P2P loan. This is something many of us have wondered about and it would be useful for platforms to share these statistics to dispel or confirm this myth.

  1. Do P2P loans help in building a better credit history?

The industry has often claimed that P2P borrowers typically see a bump in their credit score because they are paying down high interest credit card balances with a lower interest rate thereby freeing up cash flow and leading to better financial health. Again, this may or may not be true but the platforms should all have excellent aggregate data they could share in this area.

  1. Do P2P lenders serve individuals or markets underserved by the traditional banking system??

I have always felt that this point was not central to the whole online lending revolution. The industry, for the most part, has not focused on the underserved. While access to credit has expanded somewhat due to the easy ability to apply for a loan from any geographic location most borrowers have other options. Of course, in the subprime space this would not be as true but again most online lenders are focused on the prime and near prime space.

In reality I am not necessarily disputing their points on the three myths. I think they are very interesting topics for discussion and I would like to see some more exploration around all three topics. What I am disputing is the entire methodology used to produce this report and the authors’ grasp of their subject matter. It is a shame because a rigorous analysis of this data would have been very useful.

Notify of
Newest Most Voted
Inline Feedbacks
View all comments
Homero Garza
Homero Garza
Nov. 13, 2017 4:51 am

They lie, funded by banks

Nov. 13, 2017 10:53 am

Definitely some questions around methodology, but for me the central question would be something like:
Do borrowers of unsec. consumer installment loans from these platforms generally use the proceeds to re-fi and reduce debt? Or do they pay off credit cards just to free up headroom to then run the balance back up over the next 12-24 months?
That is, is the typical borrower improving his financial situation vis-a-vis household debt, or just adding more debt through a new channel?
Would need a time series analysis that reliably tracks borrowers over a fairly long window with enough sample size to control for various confounding factors as well as macro economic trends.

Nov. 13, 2017 1:52 pm

If one looks at “P2P Lending” strictly from the borrower side, it is simply what banks called, for a hundred years, “a personal loan” or a “signature loan.” So, that’s point #1 – “de-jargonization.”

Next, the periods selected for analysis were a) during the spin-up of the credit models, when they were new and inoptimal – one doesn’t go to a restaurant during the soft-open and write up a critique for the newspaper for this very reason, and, b) the selected period for analysis includes the collapse of the housing bubble and markets, themselves, during 2008 – a full blown recession striking me as something that might need to be factored into the analysis.

This is not “news” to any of us, since we bore many of the losses associated with loans made during that timeframe, as it was our personal funds at risk (we didn’t get a federal bailout), so we remember.

There are other methodological “questions”, but, as an academic who used to study these things, my first, blunt reaction was the word “bunk.” I am flummoxed as to how the author makes conclusions re: regulation – or lack thereof – as well, given the huge number of compliance acronyms involved, today.

Additionally, whether or not a borrower behaves “responsibly” (“rationally”, in the econ speak) seems something somewhat beyond either investor or platform/issuer control. I suppose if one made that a requirement, not too many people would be borrowing money at legal rates and banks would die out, though millions of government jobs would be created so that politicians could “supervise” consumers.

Without getting too partisan – or too personal – I’m really not a fan of this research, but, luckily, some other work already exists which is a bit more careful about the data used and the conclusions drawn.

As to the third question, about the “underserved”, and whether personal loans fill a need, I think the market speaks for itself. If today’s credit models mature, they will expand. For the moment, they’re, largely, just re-introducing that old, bank-invented “personal loan”; it’s online, sure, but it ain’t ‘new’.

“If it ain’t broken, and if it’s getting better, and if you didn’t even look at the last 5 years – don’t fix it.”


Nov. 13, 2017 3:09 pm

I recommend reading the source paper not the report. It does a better job distinguishing online lending vs P2P


Diogenes Lamp
Diogenes Lamp
Nov. 14, 2017 5:53 pm
Reply to  Peter Renton

Reading the study did nothing but raise more questions in my mind about the commissioning and goals of the study, the timing of its release during LC’s earnings week and the drumbeat alarmist headlines from @ClevelandFed over the last week, using those flawed charts based upon the flawed study, which prompted the financial press to repeat the claims, at face value.

I’ll note that LC put-buying, prior to earnings, was relatively heavy; I’m sure it’s a coincidence.

Brian Wolfe
Nov. 13, 2017 9:33 pm

All three credit bureaus sell a product that is data on individuals stratified across the US and over time. I’ve looked into purchasing it from one of the bureaus for similar research. The actual identity of the individual is removed but relevant information such as zip code, salary, credit score, etc… comes with the individual. The data depends a little on which vendor you buy it from but gives information on all “trade lines” which are credit accounts of the individual at that point in time. It looks like the authors are using one from TransUnion that would list out all of the credit lines/loans for an individual. That’s why they identify in Table 1 of the full paper (see DSC’s comment) which lenders they considered P2P. It would probably be more appropriate to call these FinTech lenders, but this explains the difference between Prosper’s first few vintages of loan default rate and what they show in the figure above.

With regard to the legislative environment for P2P platforms – it is incredibly complex. The Chapman and Cutler summary has grown every year and the one released last spring was over 100 pages. I have a working paper (academic) looking at the amount of P2P volume that is coming out of the traditional banking system and we discuss the legal environment on both the borrower and the investor. On top of complying with the Truth-in-Lending Act and ECOA both the issuing bank (WebBank/Cross River) and the platforms have to comply with state-level loan licensing provisions. Sometimes the platforms are considered loan brokers. Sometimes the platforms don’t have to do anything besides let the issuing bank license. There are other states (Iowa) that have origination fee rebates that make it difficult for a P2P platform to make money.

I would agree with Peter that I haven’t seen any evidence of a platform being fined by the same state twice for issues with borrower compliance – but I’ve seen plenty of evidence of the platforms being fined across multiple states on the investor side for security registration issues. I would agree with the authors that there is not one consistent set of banking laws or a single regulatory body that P2P platforms have to satisfy on the borrower side.

Nov. 15, 2017 4:37 am

There is a difference between being subject to laws and being subject to the regulators who enforce those laws.

It sounds like the report is focused on the latter while people here seem focused on the former. A good example is the CFPB, who is now regulating entities that previously were subject to laws but had no one checking compliance. I have heard that the peer lending companies have refused to give data to regulators when asked. I have been waiting for the CFPB to show up at some point and uncover some naughty practices.

Nov. 15, 2017 6:36 pm

Following up on this, I think additional regulation is only justified if one can show that there is a meaningful information asymmetry to address, or a real externality that needs to be assumed by the public. P2P loans, or whatever we want to call them, tend to be very straightforward and far easier to understand for the typical borrower than revolving credit card debt, helocs, or other types of consumer credit. They also represent a fairly small segment of the total market for consumer credit in the US, and so it is unlikely that there are major externalities that we have to worry about at the societal level.

If the originators or the lenders are mispricing credit risk in this market, I think we can be comfortable that those same parties will bear the consequences and are able to do so. If the borrowers end up over-leveraging themselves to the detriment of their financial health, I fail to see how this is any different than if they ended up taking out more credit card debt, etc. We are too far down the road of being an economy dependent on consumer credit at this point to turn back, and anyway that is a question outside the scope of any regulatory agency.

Nov. 16, 2017 7:23 pm

This is hysterical. Lending Club and Prosper suck at credit and both companies are trainwrecks.