Today, we’re excited to publish our first Insight Report, which provides marketplace loan investors an in-depth look at modeling techniques and findings from our data science team. This is the first report in our newly launched Insights series, and we’re excited to continue sharing some of the exciting findings we’ve been working on at dv01.
For the first Insight Report, we take a look at roll rate analysis. While this type of modeling is prevalent in the broader consumer credit space, its applicability to unsecured consumer loans in MPL portfolios has never been studied or proven. We use data from more than 1.6 million marketplace loans to investigate the applicability and nuances of applying roll rate analysis to MPL portfolios composed of loans of different ages.
First, we review the data to see whether there is a relationship/pattern between the default behavior of seasoned loans and the default behavior of newly originated loans, and whether this pattern is similar to what we see in consumer lending as a whole.
Then, we examine the differences between the default curves of seasoned and newly issued loans quantitatively to see if they resemble consumer credit markets, and therefore support roll rate analysis as an accurate modeling technique for this asset class.
After reviewing default curves of seasoned and new loans of varying ages, loan grades, loan terms, and even different marketplace lending platforms, we find a common and relatively simple pattern in the curve differences.
The pattern is quite similar to what we see across consumer credit, suggesting that using roll rate analysis is as accurate for MPL portfolios as it is for other consumer credit product portfolios.