Oct 21, 2013

Roll Rates for Lending Club

What are roll rates and why are they important?

Roll Rates are the percentage of outstanding balances that go from being moderately past due to significantly past due.   At any given time in a portfolio of loans, there are a portion of outstanding balances that are 30 days past due – meaning that the customer has missed more than one payment.  If you understand the roll rates of that population, you can extrapolate what the charge off rates will be in several months.


Understanding the basic roll rates of a consumer lending portfolio is important.  With Online Lending today, given that notes are held to maturity, roll rates are mostly used to inform underwriting decisions.  Underwriting models are generally more predictive and applicable if built on reasonably recent data.  The issue is that, if the data is recent, charge offs cannot be modeled well because recently issued loans haven’t matured to the point of charge off.  In this case, you can model to predict a less severe measure of default (e.g. 30 days past due date) and then apply roll rate assumptions to those predictions.


How do you calculate roll rates?

First, we will focus on portfolio or “coincidental” roll rates.  This takes the entire loan population at any given time and looks at how many dollars are moderately past due (in this case, 30 days) and then compares that number to how many dollars are charged off 120 days later.  Lending Club’s data is a little difficult to navigate, because they only provide the status of “Late (31-120 days)”, which is pretty broad.  By performing some data manipulation, we were able to identify all the loans at a given time that were just 31-60 days late.


For Lending Club originations in 2010 and 2011, we saw a significant amount of fluctuation in the 30 day-charge off roll rates, due to the smaller size of the Lending Club portfolio at that time.  Based on a weighted average, the percentage of 30 day past due balances that ended up charging off was 69%.

On more recent data (Lending Club 2012 – 2013), we see the roll rates increasing a bit, but also stabilizing significantly.  They now hover around 83%.

The next view is by origination date, or “vintage” roll rates.  This tracks a group of loans that were booked during a specific time period (in this case, Lending Club 2010 and 2011 originations) to see how the roll rates change over the life of the loan.   The calculation is the same – moderately past due dollars in one month (30 days) to the charge off dollars in later months to see how the dollars “rolled”.  We see that, as the portfolio becomes more seasoned, the roll rates become lower and less stable.


The lower roll rates at higher tenures are likely due to lower outstanding loan balances to an individual borrower. It is easier for the borrower with a loan near maturity to pay off in full, and they are incentivized to do so in order to protect their credit bureau history (i.e. why charge off now and negatively impact your credit history when you’ve already paid back 75% of the loan?).  The stability of the roll rates has to do with the small balances in general, given that many loans will be paid off by now, and the outstanding balances are smaller.

Conclusion

Roll Rates are an important input for underwriting model strategy and management of a consumer loan portfolio.  For underwriting, models can be developed using recent data to predict credit losses based on an earlier default definition.  Roll Rates also help to give an early indication as to what portfolio credit losses will be in the future.  This is particularly important for large institutions setting aside capital to cover future losses.  While retail investors do not do this, it is important to understand the performance of your portfolio and any adjustments that should be made to make sure you are hitting your target returns.  While roll rate analysis is certainly not an exact science, and there are many different ways to track them, it is important to keep an eye on roll rates so that you are a step ahead of losses in your portfolio.

Angela Ceresnie