Lehman Fees

The FT has a front page story today about the total fees incurred by the debtor's professionals in the London part of the Lehman case. So far the accountants (who play the lead role in the UK) and the attorneys are due about $363 million. This got me thinking about a recent WSJ article, which noted that the US part of the case had topped $400 million, further noting that "[f]ee experts have estimated that the bankruptcy could cost between $800 million and $1.4 billion."  The Journal does not say who these unnamed experts are, but I suspect they might be myself and Lynn LoPucki, since the numbers are very close to those we estimated at the outset of the case.

But I estimated total Lehman fees based on its most recent pre-bankruptcy 10-K, and I expect Professor LoPucki did something similar. That means that the London assets were likely included in the estimate (to the extent they were Lehman assets and not trust assets for customers), since the 10-K reports a consolidated balance sheet. That means the "$800 million and $1.4 billion" includes at least part of the UK case, albeit probably with a substantial margin of error, because we are treating the thing as one big chapter 11 case. I know I've never done any testing to see how my model performs on a foreign bankruptcy case.

At this point, the careful reader is probably saying, "wait a second --  once I sum up the total fees in the two jurisdictions, we are already at your estimate.  And the case is not over yet."  True enough, but at least with regard to my fee estimate, this illustrates the simplified nature of the media stories on this issue.

Typically when a reporter calls me and wants an estimate of how much a case might cost, I run my model with the best available data (using 10-K numbers in place of the schedules, for example) and tell them the predicted cost and the upper value of the prediction interval. The prediction interval is like the "margin of error" you often hear about in political polling, and it acknowledges that there is some degree of imprecision in predicting the future based on what happened in the past.

In this particular case, there are two key sources for of "play" in the prediction of fees in Lehman, beyond the cross-border issue.  First, Lehman is by far the biggest chapter 11 debtor ever.  As such, it is out on the right end of the graph, with very few datapoints even near it, and the degree of certainty is reduced.  And in all cases, big or small, Professor LoPucki and I both measure professional fees as the log of fees -- which means the prediction interval is also in log units and once converted back into dollars there is often a pretty big range of possible outcomes, because of the nature of the underlying scale.

In short, for Lehman my model predicted costs of about $700 billion million, if the case lasts two years, but it could be as high as $5 billion.  The press typically only reports the first number.  I doubt it will get as high as $5 billion, but it shows what is possible.  And again, this is likely subject to some substantial degree of error, because the model assumes that the UK proceeding is just like chapter 11 -- the exact thing I warn against in a recent paper.

Comments

[jim ryan]

Suggested correction -
From:
"...$700 billion...as high as $5 billion..."
To:
"...$700 million...as high as $5 billion..."

[Stephen Lubben]

Indeed. Thanks.