By Sam Griffiths on 29th August 2014
Last week we published an article outlining the analysis that AltFi Data has recently carried out on Funding Circle’s loan book. The results of the analysis and the conclusions are very interesting. This week we are going to go a little deeper into the analysis, revealing the methodology and taking a closer look at the results. The analysis is made possible by the fact that Funding Circle provides investors with a loan by loan breakdown of its loan book, offering maximum transparency.
As discussed last week, we found that Funding Circle investors, when selecting their loans, only seem to take into account 2 (Funding Circle credit rating and loan size) of the 15 risk factors that we identified and examined. Of these two factors, one could be a ‘technical’ factor rather than being caused by investor choice – Funding Circle investors are advised to minimise concentration risk in their portfolio, this limits the size that most investors will lend to any one borrower. With a finite number of investors investing at any one time, this may mean that competition for the larger loans is less and therefore the average rate at which they are funded is slightly higher. The perverse consequence of this being that as people try to minimize risk by seeking diversification they may in fact be limiting their return – and indeed reducing their risk adjusted return.
Having analysed loss rates (the principal lost in defaults net of any recovery) across the entire Funding Circle loan book, looking at the same 15 risk factors, we found that risk factors other than the two mentioned above had statistically significant impacts on loss rates. In other words, when selecting loans, lenders should take a lot more into account than just the Funding Circle credit rating.
The 15 risk factors that we looked at were:
In examining these factors, we normalised each subset within the category for three factors:
In adjusting the actual loss rate for the above factors, we were able to compare groups of loans with effectively the same interest rates at the same point in their life. We were then able to pick out certain loan characteristics within these risk factors that exhibited statistically significant higher or lower normalised loss rates than the average. Below are a selection of our findings:
The reasons for some of the patterns we observed seem pretty obvious – companies that make more profit should find loans easier to pay back and likewise, companies with a large ‘buffer’ (shareholder funds), should still be able to make repayments even if unforeseen events occur. Expansion and growth is always a risky time for a small business, so it makes sense to see that as the riskiest reason to borrow and companies with a lower debt burden may be more likely to be able to service additional debt (although this metric could be taken a step further by examining the financial leverage of the company – EBITDA/Net Debt). We were surprised to see such a variation across the UK’s regions with the data suggesting that the north-south divide is very much in existence with respect to SME default rates (for Funding Circle loans at least).
One finding that has us scratching our heads is the peak in defaults for loans with an External Credit Score of between 70 and 80. We’ve broken it down and it is definitely statistically significant, not caused by idiosyncratic defaults. We’ve bounced around many theories about what could be causing this phenomenon, but have been unable to confirm any of them from the data. We’d love to hear any reader’s ideas on why this effect may be present.
If all of these risk factors are observable, then why do Funding Circle’s investors not take them into account when selecting loans? There could be many reasons:
Obviously, as with any analysis, there are limitations to this research. Whilst Funding Circle’s loan book is certainly the most comprehensive and oldest SME loan book in the UK, it is still relatively small and only has 4 years of lending history with the vast majority (over 75%) of loans having been made in the last 18 months. This means that the analysis is not carried out on full life cycle loans and also there is still potential for idiosyncratic risks to distort findings. Likewise, this analysis is backward looking and as we all know, ‘past performance is no guarantee of future returns’!
However, the conclusion is clear: to get the best return from Funding Circle loans, investors should invest actively (i.e. not use Auto-bid); consider factors other than just the Funding Circle credit rating when choosing their loans and use the unique wealth of data that Funding Circle provides to inform their investment decisions.
Next week we combine the findings from our loss rate analysis detailed above with an IRR analysis to see just how much difference active loan selection can make.
For any questions or comments, please contact: email@example.com
Insurance AI & Analytics USA (June 27-28, Chicago) is the only forum bridging the gap between the analytical and data minds and the business transformation leaders. As carriers rush to meet customer demands and deliver continuous business growth without dramatically increasing costs, deploying innovative technologies such as AI, machine learning and advanced analytics can be the only way to remain competitive. But in order to deliver real value to the organization, these innovations must have a real application in the core business areas and directly improve operational efficiency and deliver a seamless customer experience