We’ve written in the past about borrower income and debt levels and how these variables predict risk of default. In those analyses we assessed whether a borrower with a larger debt to income (“DTI”) ratio is more likely to default on a loan payment. We found that DTI and disposable income are both predictive.
These analyses are useful when developing an investment strategy, as they help identify borrowers that are more likely to default. Last week, a friend shared some interesting data from Federal Reserve Economic Data (FRED), a robust database maintained by the St. Louis Federal Reserve Bank; the database includes almost 150,000 economic time series data from nearly 60 sources. The specific data measures Household Debt Service Payments as a Percent of Disposable Personal Income. In other words, the trends in the United States over time of debt payments to disposable income (i.e. income after taxes). The graph below shows this metric over time, which has been dramatically decreasing in the recent years.
Over the past 6 years, this metric has gone from a 30-year high of 13.2% to a 30-year low of 9.9%. This volatility is result of the credit bubble leading up to the 2008 recession and the subsequent household de-leveraging during the 2010 to 2014 gradual economic recovery.
Given that marketplace lending has grown during this time period, we wanted to assess what trends exist with the borrowers on Prosper and how Prosper borrowers compared to the nation as a whole.
In order to conduct this analysis, we first had to identify the inputs:
Debt: Given that all of the borrowers applying for loans on Proper have a credit bureau pulled, assessing the monthly debt payments is easy. In fact, Prosper provides a variable called "monthly debt" through their API; this is exactly what we need.
Disposable Personal Income: Prosper provides "stated monthly income". However, this is the borrower’s gross income. In order to calculate FRED’s definition we must take 72% of that income to account for taxes.
Some important pieces of information to keep in mind when analyzing this data are:
1. The inputs may not be apples to apples. FRED data is pretty clear, as is Prosper’s, but ensuring that they are measuring the same things in the same way is impossible.
2. FRED data is looking at the country overall at a snapshot in time, while Prosper’s data show the loans originated at that time. Prosper does not provide periodic updates as the borrower’s income and debt change over time, so it would be difficult to measure the DTI for the platform as a whole without making many assumptions.
With those caveats in mind, we will compare DTI for Prosper v. the FRED data over the past 3 years. We find that the median Prosper borrower has a higher DTI than the overall country. This is not surprising given that the majority of borrowers are applying for a loan in order to consolidate debt. The average Prosper borrower DTI remained in the ~15% over the last few years.
To further analyze this data, we reviewed an additional time series published by FRED: the real disposable income per capita. Income per capita has been around $37,000 for the past year, meaning the average per capita monthly household debt service payment is $306.
Next we compare the per capita income from FRED to Prosper for the same time period we used to assess DTI.
The average income for Prosper is $56,000, putting the monthly debt service payment at $827. The average Prosper borrower’s income is 51% higher than the average American household. This difference is significant and is likely one of the drivers for Prosper’s low observed and predicted defaults in the most recent vintages, despite the higher debt amounts.
There are many sources of economic data out there, and while some might be more useful than others, it is valuable to analyze and assess how marketplace lending borrowers compare to the general population. Such data can be used to compare and assess various marketplace originators’ platforms and underwriting strategies. It can also be used to develop benchmarks for an investment strategy. If nothing else, it helps to understand economic trends that affect all asset classes.