Borders' DIP Financing and Risk Weights

At the UCLA Big-Bankruptcy Empirical Research conference last week, an interesting question was raised regarding debtor-in-possession ("DIP") financing. Why do DIP facilities offer such high returns even though the historical default rates are so low? A study by Moody's shows that, of 297 cases of DIP facilities extended to large public companies which defaulted between 1998 and 2008, there were only 2 defaults.

Even though the probability of default is low, an argument may be made that DIP facilities pose risks which are "low probability, high severity". After all, one of the defaults in the Moody's study resulted in a low recovery rate of between 20-30%. Given the scarcity of default data, a common technique used in the industry is the "distance from loss" method. The super-priority status enjoyed by DIP lenders in the bankruptcy waterfall usually implies a high "distance from loss" for the debt. Indeed, the $505 million DIP facility in the Borders bankruptcy appears to provide a good illustration as to the low-risk nature of this asset class, as it enjoys a huge debt cushion of more junior debt.

Here's a quick and dirty analysis (caveat emptor!) from my days as the LossCalc Model Manager in KMV (i.e., the person responsible for modeling and estimating loss severity or loss given default). The "distance from loss" method is based on a probability distribution of the value attributable to assets representing the sources of repayment to the lender:

1) Calculate estimated market value of assets: In its bankruptcy filings, Borders reported its total assets to be $1.275 billion dollars. A haircut of 27% is taken (this is based on the stalking horse bid for the right to conduct the store-closing sales which guarantees 73% of the cost value of all merchandise). A more conservative haircut can be taken but then again, if we expect that Borders may emerge as a going concern, one may also argue a price/book closer to that of Barnes & Noble (which is about 1.1 based on yesterday's closing price).
 
2) Calculate asset volatility: The volatility of Borders' stock returns for the past year is calculated in order to compute estimates for implied asset volatility, using the Black-Scholes option pricing model (using total assets and liabilities numbers as well). This comes to 12%.

3) Estimate the size of the tail region for the distribution of Borders' possible asset values: Using the above numbers, a normal distribution (this could be skewed, etc, but again, this is a 10-minute exercise) of asset value is constructed, with an examination of the "tail" at the 95% confidence level.

We can say that under my assumptions, there is only a 5% probability that the firm's realized value will be less than $750 million, i.e., there is a vanishingly small probability that the firm's assets will be worth less than the $505 million DIP facility. Doubling the volatility gets us $562 million – which would still be pretty good!

Why, then, do DIP facilities offer such high returns to lenders? The usual reasons include: the labor-intensive nature of the transactional work (lots of deal-making and potential inter-lender litigation) and the difficulty of winning priming lien fights (which create a barrier to entry). There is, however, a potential other reason which is seldom explored - capital requirements and risk weights (the higher the risk weights, the higher capital requirements).

Under the Basel II framework, risk weights for corporate exposures and specialized lending exposures are different. DIP facilities fall under specialized lending. For example, under the standardized weighting system, risk weights can be as low as 20% for high quality corporate exposures and are capped at 150% for corporate exposures generally. A high quality specialized finance exposure attracts 50% risk weight and the cap is 250%.

The higher capital requirements required for such facilities are then fed into banks' calculation of risk-adjusted return on capital ("RAROC"). To compensate for the high denominator, it would be logical to increase the numerator – there is usually only one way – by increasing returns!

Comments

[Ginger Yellow]

So why don't non-bank lenders step up and offer DIP finance at much lower rates commensurate with the economic risk?

[mt]

That was very interesting and well beyond my skill level so thanks to you and Prof Lawless for arranging this contribution.

I endorse the "difficulty of winning priming lien fights / barriers to entry" argument. I would add one other factor: a risk premium for the risk that the bankruptcy judge will not let the DIP lender exercise remedies if s/he believes that will result in harm to "the little guys". That is, the bell curve may not be as symmetrical as you show and the tail may be longer and fatter because you may not be allowed to act at all points on the downward slope toward the loss region.

For ginger yellow, many such potential lenders are constrained by regulatory or contractual limitations. There are some nonbanks who do it, but they often have much more restrictive terms on cash management and collateral review (and a reputation for being difficult in default situations) than the winning DIP lender, and management nd stakeholders may be willing to pay up for greater flexibility. Also there can be a continuity benefit of staying with the same institution(s) from which the debtor has been borrowing from and expects to borrow from post emergence.

[Sarah Woo]

mt, good point about the risk that the DIP lender may not be able to exercise its remedies under certain circumstances. Anyway, as I mentioned in the post, this is a quick and dirty analysis which I did on the fly. The asset value distribution may not actually be a normal distribution. Nonetheless, with sufficient data on loss severity in bankruptcy cases (a recoveries database which I am currently collecting), including factors such as debt structure, etc etc, we can try to parametrize the distribution empirically (as best as we can).