P2P Loan Servicing
John Paul Lewicke
I've recently been reading about the early history of Prosper.com, and
I found that some of the issues they encountered seem to be general
issues for P2P loan servicing. Their fundamental problem is that they
didn't have a scalable model for collections, which resulted in
shockingly high default rates(up to 44%) and negative returns for the
average lender. A lot has been written about how Prosper was
constrained by US securities laws and how alternative underwriting
models might reduce the overall loss rate, but there's one simple
reason why proper loan servicing matters: even if a P2P/distributed
lending system can provide superior underwriting vs. existing lenders,
below average loan servicing can more than erase that edge. The
centralized nature of loan servicing in the current US P2P lending
services is one of the largest drags on their cash flow and loan
performance.
There are several components to loan servicing, many of which can
create perverse incentives for the loan servicer.
http://www.calculatedriskblog.com/2007/02/tanta-mortgage-servicing-for-ubernerds.html
has a good description of how loan servicing works in an established
market. The root problem is that lenders have much more at stake than
servicers, and so it's a lot easier for a servicer to just collect
money for performing loans while doing very little for non-performing
loan. By all signs, this is what happened at Prosper -- they tried a
couple of fairly ineffective collections agencies, didn't do timely
loan sales, and only have initiated legal actions against a handful of
delinquent borrowers ( See
http://www.prospers.org/blogs/media/blogs/Fred93/open-letter-number-2.pdf
, http://fred93blog.blogspot.com/ , and http://www.prospers.org for
details).
While part of the problem is that incentives are mis-aligned and
Prosper is mainly focused on their loan origination activities, most
of it stems from the fact that it's very hard to let the lenders have
greater control over the collections process. If lenders
systematically collect and share too much personal information about a
prospective borrower, various state and federal privacy laws can
apply. When borrowers default or are in arrears, the Fair Debt
Collection Practices Act applies and the loan servicer could be held
liable for lenders' overly vigorous collections efforts.
Another issue is that the cash flow implications of having to service
loans are not very healthy for a fledgling P2P lending startup.
Having to directly pay for attorney fees and similar issues is a
backloaded cost that makes it harder to establish a positive cashflow
loop of customer acquisition -> transaction -> fulfillment.
In typical agile-banking style, I think the solution to these issues
is not one perfect monolithic startup that happens to get everything
right, but instead an ecosystem of partnering services working on a
common standard. Thus, you could see something like the following
breakdown of services:
- Sites that originate P2P loans between lenders and borrowers. It's
best if a way can be found for the lenders to keep title.
- Know Your Customer/anti-identity theft providers who can
authenticate people. While still a cost center, this would help
maintain loan quality and wouldn't just be a regulatory requirement.
- Payments processing for repayments.
- A platform for secondary trading of loans and notes.
- Various collections providers.
- Legal services that can enforce the terms of loans.
I'm also not sure that traditional collections agencies would be the
best fit. If there were some way for the community of lenders to work
closely with a financially distressed borrower and to try to help them
solve their issues, that could be a much more positive approach than
making their phone ring all day. The Fair Debt Collection Practices
Act would get in the way, but maybe something where borrower's
information is carefully protected and not necessarily known to all
lenders could work. They could also use some sort of telephony API to
proxy the borrower's real phone number and only allow creditor calls
in during appropriate hours, and monitor the transcripts to ensure
that lenders are not being abusive.
Just some random thoughts -- hope someone finds it interesting. I'm
personally not very excited about distributed loan sharking, but found
a lot of the story of Prosper fascinating when I started digging in.
Best,
JP