P2P Lending is mostly anonymous and loans are unsecured. To make the risks of lending to a stranger acceptable for lenders, p2p lending services had to provide models for the lenders to judge the dimension of the risk of not getting paid back.
The initial estimation of the risk-level could not come from the platform itself as it had no track record and could not build a model that “calculated” the level of risk involved for the lender. The consistent consequence was that nearly all p2p lenders relied on established third party providers for credit history data and credit scores. Prosper for example showed Experian data on default levels to be expected depending on credit grade.
Over the time it became obvious that the actual default levels at Prosper were much higher than the expected default levels based on Experian data. We don’t actually need to argue here what led to this (be it financial development of the economy, be it that p2p lending attracted bad risks, be it a poor validation process), but the result was that since defaults were much higher than expected, lender ROIs were much lower than expected at the time of the investment.
And this is not Prosper specific. Several other p2p lending services show clear signs that default levels will (or have) surpassed the initially published percentages of defaults to be expected based on external data.
Boober failed due to default levels, on Smava levels are higher than the Schufa percentages fore-casted, same is likely for Auxmoney defaults which will be higher then Schufa and Arvato Infoscore data suggested. The one exception from the rule is Zopa UK, which successfully manages to keep defaults low, as CEO Giles Andrews rightly points out.
So, is this “conventional” data, that banks rely on too, useless for p2p lending and especially for a lender trying to assess the risks?
Not at all, because used on a relative base it still works well. Smava borrowers with a credit grade of A proved to be a lower risk than borrowers with a credit grade of D which in turn defaulted less than credit grade G borrowers.
But what more can be done? Obviously stringent vetting of borrower applications by the p2p lending service. But that’s not the topic on this article (if you want to digress for a moment read former Zopa US’s CEO Rajesh Jayaraman recent blog post).
When Prosper started, it introduced groups – based on the assumption that the ties in these groups would add social responsibility to repay the loan. A concept that works well in microfinance, but it failed at Prosper. The groups formed, rarely had previously formed offline ties between the members. Due to anonymity non-payers did not need to fear social pressure in case of non-payment. And Prosper gave the wrong incentives by paying group-leaders to recruit new borrowers.
So far groups are useless in p2p lending (not p2p microfinance).
Let’s look at social networks like Linkedin or Facebook. They are booming. Nearly everyone is a member in at least one. And many of the contacts there are contacts that go beyond the “pure” online world (be it family, friends, work or business contacts).
I currently have 90 contacts at XING, my most actively used social network.
What do I know (or could via the social network easily verify) about these contacts that might be useful in p2p lending (from a lender’s perspective):
- Their current financial situation? no
- Their level of income? no
- Their employment status and the company they work for? In most cases yes
- Their education and past work history? In many cases yes
- Can I assess if they did honor their financial agreements in the past? Only in rare cases where I have a very long time business relationship and the person is an entrepreneur
- Could I say I am convinced they would repay a loan, or at least do everything possible? In a few cases
And on a more emotional level, it would boil down to: Do I trust them?
Upon login, Xing not only shows me how many direct contacts I have but also displays the number of second and third level connections.
For these second and third level connections my answers to the above questions would by no nearly all of the time, as I won’t even know most of them.
So how could a p2p lending service make use of social network data and social network ties? For the sake of the debate, I picture the following model:
First borrowers would need to be ready to waive anonymity in their loan request. That’s a big step, but it doesn’t mean the borrower’s name needs to be written on the p2p loan listing. And as discussed earlier the risks for lenders so far are much higher than for borrowers.
I actually have observed many listings, where borrowers voluntarily disclosed there identity. Not by giving their name or phone number, but rather by disclosing information like the website of the business they run, that makes their identity researchable. And many borrowers at Prosper used personal photos. Listings of borrowers who disclosed much information in my observation were more attractive to lenders then those that lacked personal information.
So let’s assume that there is a small, but relevant share (e.g. 15%) of borrowers that is willing to forego complete anonymity. For them an optional additional process during bidding could be offered.
Most social lending networks allow applications to access data via interfaces (APIs). If a borrower A and a lender B are using the same social network the p2p lending service could use the API to crawl and fetch the connections between them.
It is very unlikely that there will be a direct connection between A and B but chances for a second or third level connection are better.
What could then happen?
The service could allow B to query C, if he trusts A in financial matters and if he thinks he can judge, if A is in a financial stable situation.
Would that be an information I would rely on, if I am the lender? Not as a single factor. But it might be one of many points influencing my decision, if all other “conventional” data (e.g. credit score) of the loan requests to choose from does not differentiate. But it would certainly influence my decision, if C says A cannot be trusted because… . So it might block bad risks in some cases.
And a loan is not funded by one lender but by several. Even better the decision is made at a point in time when several hundred lenders are reviewing the same loan request.
So if I query C about A and get a neutral feedback – that does not tell me anything. But suppose there are 20 other lenders with second level connections to A and they query M, H, O, R, Z and G in between. If the p2p lending service automatically aggregates the feedback and 5 “voted” positive, 1 neutral (plus the C neutral vote) than I might get a positive picture.
The signal in third level connections would be much less trustable and weaker but would still add to the overall picture.
Would the intermediaries be annoyed? I don’t know. Depends on how good they know the lender/the borrower. I feel that friends and family contacts would be more than happy to help out dedicating a few mouse clicks.
What about data privacy? The social network connections and the data in the profiles is public anyway. The new information is the one disclosed in the process. The impact is different if C was asked “Do you think A is financially trustworthy” compared to “A wants a 20,000 US$ loan. Do you think he can afford this?”.
Should p2p lending services pursue this route? I don’t know. Would it work? I don’t know. Could it be manipulated, if implemented as suggested above? Sure. Do I think it would be transparent? No.
But I feel that any p2p lending service, that succeeds in adding an additional, useful “trust” parameter not derived from credit scores or credit history, but rather based on social connections, will have a valuable competitive advantage and might lower risks in mid-term.
This “trust” parameter most likely would not work in the way suggested above. If it is doable without waiving borrower anonymity – even better. So I invite you to share your ideas and opinions on if and how p2p lending services could make use of social network data here or in the forum.