Channel Code for Prosper
There are two kinds of acquisition methods for consumer lending institutions to bring on borrowers – active and passive. Passive acquisitions are borrowers who apply for credit without responding to a direct offer – although the applicant could be responding to an advertisement on television, radio, online or print. In addition, virtually all consumer lending institutions have programs to actively acquire some of their borrowers. These active methods include direct mail, telemarketing, emails, and online advertising. Most lenders use a combination of active methods in order to acquire many of their borrowers (although telemarketing is not used much), the purpose being two-fold:
- Increase general brand presence: The more often potential borrowers see links and receive offers from a lender, the more likely the potential borrower will be to think about that lender when credit is needed.
- To be in the right place at the right time: On the day that a consumer decides to borrow funds, the lender wants to be in the mailbox/inbox/etc. of that consumer. Otherwise, the lender runs the risk that another lender is there instead.
Currently, the Prosper data lists 5 different acquisition channels by the following codes:
40000, 50000, 70000, 80000, and 90000
At this time, we don’t know what the difference is between the codes, but we are able to see the patterns and behaviors related to the codes.
From 2010 to mid-2012, Prosper only used channels 40000, 80000, and 90000 and was fairly evenly distributed between the three. 50000 was introduced in July 2012 and has grown slowly since that time. 70000 was introduced in September 2012 and has grown quickly – becoming the largest channel fairly quickly.
Given that we don’t know what the channels mean, it is hard to draw much of a conclusion about this change and what it might mean from a marketing perspective. It is possible that these new codes are completely new acquisition channels, or simply variations on existing ones.
However, when reviewing the relative credit grades for each of these channel codes to better understand their characteristics, we see a few things:
- 70000 comprises a larger portion of the lower risk credit grades
- 40000 and 90000 comprise a larger portion of the higher risk credit grades
- 50000 and 80000 are similar among the different credit grades
This would indicate that 70000 is likely a prescreened type of acquisition offer. Lenders have the option to send potential borrowers marketing materials with or without actual offers. If a lender sends a potential borrower a marketing piece that includes an actual offer, they are bound to that offer in most circumstances. This means that the lenders are generally prescreening the credit bureau of these individuals to confirm that the credit profile meets some minimum standard.
Looking back at 2012 and early 2013 data, it is clear that certain Channel Codes are trending better than others. One point to keep in mind is that 50000 is quite small (< 5% of the loans) so this default rate is based on small, and more recent volume.
Given that we know that channels are not distributed evenly for the different credit grades, looking at the default rate by credit grade will help to understand if the difference in performance is simply due to a difference in credit grade distribution.
The overall performance for the different channel codes, for the most part, mirrors the performance within each credit grade.
Channel 90000 has the highest default rates. As we saw earlier, the relative volume of these loans have been shrinking, so that could be due to this negative selection and changes in the way Prosper solicits and approves loans.
Channel 70000 is generally better within in each credit grade, and it has been growing over time – which could also be due to a change in how Prosper solicits customers.
Ultimately, it is important for investors on Prosper to know what these different channels mean, however it is also interesting to conduct the analysis without knowing and being able to assess the relative performance without adding subjective thoughts based on preconceived opinions about the channels.
In terms of using channel code in a risk strategy, if it works within an investment strategy to bring it closer to risk/return objectives, it might be sufficient, despite the lack of definitions for the different codes. However, it is always important to continue to track variables that are so closely related to an originator’s underwriting strategy. As one could see in the first graph, Prosper changes the codes used frequently enough that a model or strategy with this as an input variable could certainly be off if Prosper changed the way the channel was used.