Wednesday, August 31, 2011

Appropriate Compensation #11: Keep it Simple(r)

Here in post-Irene New Jersey many of us still have some recovery to attend to but, as the month ends, I want to quickly clarify the idea put forward in our last post.

This is what I mean by a Seller-based approach to an allowance for credit losses:

1. Let’s say that all active Sellers on TRE represent the same share of total auction activity. (Clearly, that’s unrealistic, but it’s just for illustration.) And let’s say that our analysis suggests that 3% of active Sellers will default in any given year. If we assume that every default results in a total loss, we might say that overall TRE pricing has to provide for an annual loss of 3% of capital (before recovery expenses).

That 3% loss provision is not based on the gross income of auctions purchased. It’s not a top-down calculation. It is an absolute loss provision that has to be included in auction pricing to compensate for perceived Seller-default risk.

That’s an extremely simplified example of a platform-level approach but, hopefully, it conveys the idea.

It’s not possible at this point, for a Buyer to “buy the Exchange” i.e. there is no single transaction that will expose the Buyer to a pro-rata position in the entire TRE portfolio. So an Exchange-level analysis, even if it were really possible in a statistically meaningful way, wouldn’t go far enough for the individual Buyer.

The individual Buyer will have to account for his own strategic portfolio and auction purchase decisions in creating an allowance.

2. Let’s say that a given Buyer has a portfolio diversification rule that limits his exposure to any one Seller to an amount equal to 10% of the Buyer’s capital. And let’s say that the Buyer believes that his Seller-qualification criteria allow an annual probability of default of 1 in 20. If a total loss on default were assumed, and the Buyer’s investment at the time of default were at its 10% maximum, the simplistic loss allowance should be about 5% of capital per year.

3. Let’s take the same situation as in #2 and assume that the Buyer actually expects a net loss recovery of 50% of defaulted amounts. In that case a net loss allowance of 2.5% of capital per year would be indicated.

Obviously, a Buyer projecting a loss of 2.5% of capital per year will think and act differently than one projecting a 5% loss. (And those making no conscious allowance for loss will be unpleasantly surprised sooner or later!)

There is not yet enough history of TRE operations to reasonably estimate the probability of a “typical” Seller defaulting. Nor is the information about actual default experience publicly available.

What we ARE able to say is that the experience of TRE to-date has caused it to make a number of meaningful changes to procedures, staffing and operations that affect both risk of default and likelihood of recovery post-default. And that those changes have been constructive.

However, it remains true that:

• There is a wide range of (reported) financial capacity among TRE Sellers.

• There is a wide range of financial capacity among TRE Account Debtors.

• There is a wide range of past experience among TRE Seller/Debtor pairings.

• There is a wide range of documentation strength provided in TRE transactions.

• There is a wide range of actual pricing in TRE transactions.

Because each Buyer’s actual and perceived risk/return profile is dependent on many individual portfolio construction and auction purchase decisions; and many assessments of risk are still largely subjective in the context of limited TRE history; the question of an appropriate loss reserve will also necessarily be both individual and largely subjective.

I might think, based on my own portfolio construction and auction purchase criteria, that an annual Seller default probability of 1 in 20 is reasonable. Another Buyer might find some other number to be more reasonable.

I might think that a net recovery expectation of 25% is reasonable. Another Buyer might think differently.

I’ll develop this idea further in subsequent posts but I hope this clarifies the distinction between a Seller-based analytical process and one that is Debtor-based.

I should note that I’m not ignoring the differential strength of Account Debtors. Regardless of Debtor capacity, it is the Seller that is ultimately responsible to make good on invoices sold. And, in the context of TRE notification/verification procedures, I think that the quality of the Debtor is best considered as one element in the assessment of Seller risk.

Thursday, August 18, 2011

Appropriate Compensation #10 : The View from Below

In this series of posts we’ve identified a number of risks assumed by Buyers of TRE auctions that we’ve suggested deserve incremental compensation beyond that provided for the risk in typical factoring transactions.

From the start we’ve acknowledged that quantifying an appropriate level of incremental compensation is going to be difficult.

There is no evidence (that I am aware of) to support a rigorous quantitative analysis of the incremental risk associated with reliance on unaudited financial statements as opposed to audited ones. There is no evidence (that I am aware of) to support a rigorous quantitative analysis of having a limited lien position versus an “all asset” lien, or a junior lien as opposed to a first lien. And so on.

We all know that the additional risks assumed in TRE transactions DO exist and that they are real and that they are not trivial.

But how can we approach assigning a value to them?

In our post of May 11, 2011 I wrote, having provided some supporting data and analysis:

“It seems to me that an average credit loss allowance in the range of 10% to 12.5% of gross income in typical factoring transactions, over the course of a credit cycle, is not unreasonable.”

My expectation at that time was that, after examining the major factors that add risk to TRE transactions, I’d arrive at a suggested analogous number that might be used in analyzing TRE transactions.

The more I’ve thought about it, though, the more I’m drawn to a different approach.

Here’s why.

When we look at the factoring industry as a whole we’re looking at a highly competitive environment in which competition constrains the participants’ pricing power.

Average gross revenue is largely a market-determined number and the participants are challenged to operate their businesses within that revenue environment in a way that generates an adequate profit after costs.

A market-wide average credit loss experience will affect top-line, market-wide pricing only over the longer term. That is, increasing (or decreasing) credit losses will tend to affect FUTURE pricing to the extent that they appear to be a reflection of a structural change in overall risk. They are analyzed, calculated and reported in terms of their relationship to the market-driven gross fee environment.

They occupy one line in a top-down analysis of profitability.

The question we’re really asking in THIS analysis, it seems to me, is NOT one of the relative size of a number in a top-down profitability analysis but rather one of the ABSOLUTE size of a number in a bottom-up pricing structure.

The ultimate question is NOT how much of a relatively fixed top line should be reserved for losses but rather what level of loss expectation does the top line need to INCLUDE, as a compensation for risk, in the unique environment that TRE represents. The top line, rather than being relatively fixed, has to be flexible enough to expand to accommodate the risk assumed.

So, rather than stating the conclusion in terms of a percentage of income to be subtracted from the top line, we would state the conclusion in terms of a percentage of capital that must be included in the pricing of auctions to compensate for losses.

This is an approach that implicitly assumes that much more of the TRE Buyer’s risk is Seller-based than Debtor-based.

The question becomes not how many Account Debtors fail to pay invoices posted for sale but rather how many Sellers, for whatever reason, default on THEIR obligations.

That would include simple inability to make good on invoices not paid by their Debtors as well as Seller insolvency or default arising from one of the various forms of potential Seller fraud.

So we would ask:

a) What is the percentage of the total TRE Seller universe that is likely to default in any given period of time?

b) What is the percentage of total TRE auction volume that is likely to be represented by those Seller defaults?

c) What is the likely NET recovery rate (gross recoveries less costs of recovery) in cases of Seller default?

The answer to those questions gets us to a projected loss-of-capital allowance for the Exchange as a whole. Each Buyer is, of course, able to construct diversification and Seller-qualification strategies that he believes would mitigate either the absolute or risk-adjusted loss potential.

In the next post I’ll explore the impacts on pricing of a range of assumptions on these issues.

I invite anyone interested to share their own views of this approach and their thoughts about reasonable values to assign to the variables.

Monday, August 8, 2011

Appropriate Compensation #9: Upon Mature Reflection

I worked with a guy once who had a wonderful way of reversing his expressed opinion on a matter, which was particularly useful when he found that it conflicted with the boss’s opinion.

When it was clear that he’d dug himself into a hole, he’d invariably preface his remarks with: “Upon more MATURE reflection….”

We’ve identified a number of significant issues that we believe add risk to TRE transactions when compared to the typical factoring transaction. The financial statement issue; the personal guarantee issue; the lien priority issue; for example, are all significant and all clearly differentiate the TRE transaction.

Before moving to the question of what kind of pricing response might be appropriate I want to bring up another element that is perhaps the most difficult to analyze and probably impossible to quantify. It is, nevertheless, quite real.

It is the fact that a real-time, competitive auction environment creates a unique potential for self-defeating behavior: behavior that is influenced by the trading environment in a way that does not affect the typical factoring transaction.

There is a great deal of literature about the impact of bidder motivation and psychology on behavior in various auction environments.

The TRE auction environment is one in which, unlike some others, the Buyer does not have the practical option to exit the trade to correct a buying or bidding error. There is no opportunity to say: “Upon more MATURE reflection…..get me out of this thing!” And then to sell what might have been purchased in error, or in haste, or under pressure to put money to work or even just in a momentary fit of pique at having lost a number of auctions that day.

All of these things, and more, can be the cause of a Buyer looking up from an Awarded Auction Report and saying “I can’t believe I really did that.” The only thing to be done at that point is own up to the lapse and hope the thing actually gets paid.

That is not to say that the decisions of the typical factoring company are not subject to the motivations of competitive pressure and internal goal setting. Of course they are.

But they are not subject to the pressures of a real-time auction environment, which requires that decisions be made very quickly, usually without prior warning, frequently in the face of new information that has not been previously analyzed, often in an environment of excess market liquidity, and often in competition with others with differing motivations for participation.

One of the dynamics proven true in competitive auctions is the increase in the perceived “need” to win as the number of “lost” auctions increases. That is, a buyer’s propensity to act against his own bidding rules will tend to increase with the number of unsuccessful bids made. That’s a dangerous but very real temptation when being the successful bidder is defined in terms of “winning” the auction.

In fact, the successful bidder is the one who only bids according to his pre-established bidding rules.

And it might be that a consistently successful bidder is NOT particularly successful at keeping funds deployed in some market environments. But that’s a different question.

I can tell you that, based on my own experience, there is a dynamic in the competitive, real-time auction environment that has clear error-making potential. Some Buyers will be more susceptible to it and some less. So the actual level of incremental risk is as much Buyer-driven as process-driven.

Whatever the actual MEASURE of incremental risk, the fact of incremental risk in real-time, competitive auctions is well documented in academic literature and clearly felt in practice.

I confess that I have found myself (and I will bet that many other Buyers have as well), after placing a bid or winning an auction, wishing that “upon more mature reflection” there was an un-do function on the TRE platform!

It is no coincidence that the competitive nature of the TRE buying environment is stressed in TRE's marketing to SELLERS rather than to Buyers!