In our last post in this series we discussed the idea of risk analysis at the level of the Seller, on the assumption that a probability of default might be usefully considered in terms of a Seller-based metric.

That actually brings up an interesting distinction between the TRE model and that of traditional factoring relationships.

In the traditional full-line factoring relationship the factor will have funded a significant portion of the total receivables portfolio of the client. Not necessarily the entire portfolio, but perhaps all of the invoices due from specified debtors or all invoices due from specified debtors within a specified age limit, or something of the sort. And the factor will have a buffer against loss equal to the aggregate dollars held-back from all invoices funded plus (in many cases) additional value from invoices not actually funded.

In such a case, a debtor’s default on one invoice, or the default by one debtor among many, will not necessarily put the factor's advance position at risk; there might be enough value in the unfunded positions to cover the defaults.

In the spot-factoring model, particularly in the case where the lien is specific to the invoices purchased and there is no other security provided, the buyer of the invoice has fewer options to effect a cure of any default.

That is typically the case in a TRE transaction.

So, some might argue that the risk of loss might be better viewed at the level of the individual transaction rather than at the level of the Seller. I don’t necessarily agree but it does provide another interesting approach to risk analysis.

Let's say that we’re looking at a $50,000 single-invoice auction and that the terms of sale are: an 85% advance; a monthly discount fee of 1.5%; and, an expected duration of 30 days. The net earnings on an auction with those parameters will be approximately 1/79th of the initial advance.

In that case, just for illustration, if the buyer thought there was 1 chance in 79 of suffering a complete loss on an auction with similar characteristics, the net expected return after credit losses would be zero.

Inverting the analysis, if the buyer thought that a credit loss equal to 10% of the expected earnings was acceptable (just to keep the numbers round), he would have to attach a 1 in 790 probability of a total loss to this auction in order for it to meet his loss tolerance.

Using that approach a Buyer can fairly easily construct a loss tolerance distribution using variables for size, duration, rate and targeted loss levels. This won’t answer the question of what the probability of loss might actually be. But it will provide a measure against which the Buyer can test the reasonableness of various assumptions.

Let’s keep size, rate and targeted loss levels constant and test for the effects of duration change. If the duration were 15 days instead of 30 days, the net earnings expectation would be about 1/212th of the initial advance and, in order to hold losses to a 10% level, it would take about 2,120 successful auctions for each 1 that was a total loss.

If the duration were to be double the initial case i.e. 60 days, the expected earnings would be 1/35th of the advance amount and 1 auction in 350 could be allowed as a total loss while maintaining a 10% loss ratio.

We’ve noted that there is not enough data yet to reliably attach probabilities of loss to TRE auctions. And the variations among Seller and Debtor strength, experience, and other metrics are so wide that any analysis of the TRE market as a whole is perilous.

However, a Buyer CAN approach a single auction armed with the calculation of implied loss tolerance given the auction parameters and his own appetite for risk.

Loss tolerance calculation is not loss probability calculation. But it’s something.

It seems that we’ll soon be able to discuss a significant new risk-mitigation step being taken by TRE. When we CAN discuss it we will.

But the fact that mitigation actions are taken doesn’t relieve us of the need to assess the risk itself. It just imposes the additional requirement of analyzing the extent to which the mitigation measures actually affect the net loss probability.

It's About Time!

6 years ago

## No comments:

## Post a Comment