There is no Evidence that Unsecured Recoveries in Chapter 11 are Diminishing (Part 4: Valuation Differences)

This is the 4th in a series of blog posts demonstrating flaws in an argument that the Bankruptcy Code needs to be reformed to restore some purportedly lost balance between secured creditors and unsecured creditors.  In the prior posts, I demonstrated that the thesis that unsecured creditor recovery in chapter 11s has declined in recent years is based on incredibly shoddy data that is far too unreliable to support the reform agenda, and that the purported increase in second lien debt cannot be the cause of the purported decline. In this post, I lay out two valuation-related points:  1) that the two sample sets that were compared to make the argument of a decline in unsecured recoveries are not properly comparable because they cover different lengths of time and different economic environments;  and 2) any decline in recoveries is driven much more by valuation and intrinsic business merit than by the mix of secured versus unsecured debt.

Mismatched Sample Periods

Wood has chosen to analyze reorganizations that took place over a shorter period of time – 2 years – vs. the 6 years in the LoPucki study.  This makes for a highly misleading comparison. This misleading comparison is partly responsible for the purported differences in recoveries.

First, the earlier period had much more economic growth than the 2009-2010 time period. In nominal GDP terms (nominal because the debt of a chapter 11 debtors is quantified in nominal dollars, not inflation-adjusted dollars),  nominal GDP was more or less the same at the end of 2010 as it was at the end of 2008, the period Wood studies, while it rose at an average rate of more than 5% per annum over the 1991-96 period.

More significantly, the Lopucki study measures recoveries during a 6-year bull market, but Wood only studies 2 such years.  The S&P 500 index stood at 330.22 at January 1, 1991, the outset of the period covered in the “Lopucki study”, and rose to 740.74 by the end of that period, December 31, 1996  -- a 124% increase.   In contrast, from its January 1, 2009 level of 903.25 to its December 31, 2010 close at 1257.64, the same index only rose 39%, slightly less than one-third the increase in value that occurred over the period of the Lopucki study (which is understandable given that the Lopucki study period was three times as long). Both periods had roughly the same IRR – 18%.  The only difference is that the 18% IRR runs for 6 years in the sample with the higher recoveries, in contrast to only 2 years in the one with lower recoveries. Since valuations performed by investment bankers for disclosure statements reference comparable public companies explicitly (and also rely on analyses such as discounted cash flow and comparisons to acquisitions of similar companies that implicitly reflect public market valuations), the 1991-96 period is going to have higher average valuations in it than the 2009-10 period.  That’s all. There was no imbalance in the Bankruptcy Code at work.  Recovery differences were just a function of compounding over a longer period.

I can understand why a student graduating in 2011 would not be in a position to wait four more years to publish research.  But there is no excuse for his professor, or the editors of the American Bankruptcy Law Journal, or Gotbaum, to disseminate such obviously half-baked work as remotely reliable, let alone a basis for fundamental legislative changes.  It’s ridiculous to draw any inferences from comparing a six-year period of 18% IRR to a two-year period of 18% IRR.

Wood is not blameless, though. He conspicuously ignores recoveries during intervening periods, for example, the 2001-02 wave of restructurings.  But there was some data readily available to him for that period, right in the Lopucki study. In footnote 70 of that 2003 article, LoPucki wrote that a preliminary study showed unsecured creditors were experiencing substantial declines in recoveries in 2001 and 2002, compared to the 1991-96 data set.  In the 90’s, he stated, unsecured creditors received full recoveries in 59% of the cases, and only 27% of the time did they recover less than half of their claim.  But in 2001-02, he stated, the proportions were essentially reversed, with unsecured creditors recovering less than half their claim 62% of the time, and getting par only 25% of the time.  Wood does not make any use of the 2001-02 data.  Had he done so, he would have recognized that recoveries for unsecured creditors in his 2009-10 dataset were actually better than those in the 2001-02 cases, with unsecureds recovering less than half their claim only about 53% of the time (adjusting for errors Wood made in several cases).  That completely vitiates his thesis that recent recoveries are being affected by recent increases in the use of secured debt.  But the fact that both 2-year samples showed lower recoveries than the 6-year sample obviously confirms that the difference in recovery is in part attributable to the length of the sample. 

It’s Valuation,  Not the Mix of Secured Debt.

I think most practitioners familiar with the cases in Wood’s sample can tell that the recoveries would not have changed much had there been less secured debt.  Every one of the 25 companies with below-par recoveries for unsecureds, save possibly Spansion, had a debt/EBITDA ratio of over 10:1 when it filed chapter 11.   That’s going to affect recoveries.  In fact, in contrast to the typical bell-curve distribution one would typically see in a sample, the cases he cites tend to fall into a barbell pattern of two groups with extreme results:  100% recoveries and very low recoveries.  That shows the proportion of secured debt was largely irrelevant to unsecureds’ recoveries.  Only a small minority of the cases wind up with sizable but still below par recoveries to unsecureds. 

Several of the companies come from industries that bore the brunt of the “Great Recession” like auto (Hayes Lemmerz, Lear and Visteon, each of which also had large pension and OPEB liabilities that I didn’t include in the debt/EBITDA ratio), and housing (WCI, Building Materials, LandAmerica and Luminent, the last two of which were basically out of business even though they had little secured debt).  Valuations of those companies were dramatically affected by the depressed condition of their respective industries. 

Secular changes in the media industry drove ION Media, Idearc, Readers’ Digest, Young Broadcasting and Journal Register into reorganization, and pre-petition secured claims were badly haircut in all of them. Some companies in the sample had badly shrunken or even negative cash flow and, again, the secured lenders were badly haircut in each of those (VeraSun, Linens, MagnaChip, Aleris). 

A surprising number of the filings either were chapter 22s (Bally’s, Hayes Lemmerz, Pliant) or, since emergence, the debtors have filed again (Idearc, Reader’s Digest, Buffets, Journal Register), indicating that there were fundamental problems with their businesses that impaired their value even with de-levered capital structures, i.e., independent of capital structure.

Ignoring Secured Creditor Recoveries. 

Neither the LoPucki study nor the Wood article tabulates secured creditor recoveries.  By ignoring them, Wood and LoPucki deprive the reader of essential data with which to evaluate the hypotheses that secured creditors are somehow taking more value from unsecureds than in past eras. There’s no way one can determine that declines in valuation are not responsible for declines in recoveries for unsecureds and equity, without examining what happened to recoveries for the secureds in those cases.  It is clear from Wood’s data, taking it at face value, that the recoveries for every constituency that he chooses to look at have declined.  So what happened to the secured creditor recoveries that he chooses not to examine? An analyst needs to know that to figure out if the decline was valuation-driven, or if the secureds were simply capturing more of the value. The only correct way to decide which is at work is to include an analysis of  secured debt recoveries in the 1991-96 time period and then compare the results to secured debt recoveries in the more recent time period.

Still, the declines in equity recoveries are instructive if one is trying to get a handle on causation of declines in recoveries.  Even if one subscribes to the dogma that secured debt takes away from unsecureds’ recoveries and blames increases in the former for declines in the latter, one must recognize that secured debt does not take away from equity any more than unsecured debt does, so shifts in the mix of secured debt vs. unsecured debt should not change equity recoveries (yes, if you are a good soldier in the war against secured creditors, you can conjure up Rube Goldbergian sequences of causation, where being secured makes a lender act differently and that changes the case dynamics and that causes losses, but in the real world, no: it’s valuation that determines the recoveries.). Since Wood’s data show declines in equity recoveries, so it seems fairly obvious that there is another explanation for pervasive declines throughout the capital structure, i.e., valuations were different.

In sum, even if declines in unsecured recoveries have happened, and are not the result of bad data, or artifacts of methodological decisions or mistakes, one cannot explain them with confidence as the result of developments in secured debt, if one has not analyzed the recoveries of secured debt!  This seems incredibly obvious but is completely missing from the Wood analysis.  However, an understanding of the specific business dynamics of the cases in his sample informs one that the quantity of secured debt was irrelevant in many cases to unsecureds’ recoveries.