|Shares Out. (in M):||98||P/E||n/a||n/a|
|Market Cap (in $M):||1,535||P/FCF||n/a||n/a|
|Net Debt (in $M):||0||EBIT||0||0|
I believe they are wrong. Even assuming MIP’s original valuation was highly erroneous, my base-case estimate of expected returns is currently $52 per unit, above the original write-up’s $50 estimate (plus my number excludes the $2.65 in intervening distributions). The original write-up may have greatly overestimated one important factor (the % losses incurred on the defendants’ loans), but it greatly underestimated an equally important factor (the amount of loans on which ResCap will ultimately be suing). MIP has been faithfully updating us in the comments as the loan total has increased. The original write-up also omitted several significant sources of recovery that MIP later detailed in the comment thread. I have recast the original valuation into a different analytical framework to account for these conceptual changes. I tracked down MIP, who still knows far more than I do about many details, and have discussed these issues with him to help minimize new errors on my part. (Given how complicated this situation is, I would be surprised if I have not still made some sort of error, but I believe I have the big picture right and this supplement will be helpful.)
There also remains a chance that the originally estimated loss ratio is roughly correct, in which case all these other changes boost the base-case valuation above $100, with over 500% upside.
I also believe that the legal posture is as good as I have ever seen. At the risk of setting myself up to get embarrassed by “unknown unknowns,” I see low odds that ResCap could lose these cases, lower than in any other litigation situation I have seen. I also believe that many people are overestimating the length of time it will take before settlements are reached; if I am right that the cash will come sooner, then the IRR is higher. Based on these conclusions, we have made ResCap a large long holding.
This write-up assumes the reader has carefully read MIP’s original write-up; reading all the message board posts is also worthwhile. It proceeds as follows:
REFRAMING THE STORY
Despite 89 separate lawsuits and all the complexity of its details, the basic story here is simple: The banks sold bad (= poorly underwritten) loans to ResCap while promising (through reps & warranties) that the loans were good. ResCap resold the loans to the trusts while promising they were good. For breaking its promise, ResCap was forced to pay $12.2 billion in awarded bankruptcy claims. (Plus bankruptcy costs, plus other costs that I will discuss later.) For breaking their functionally identical promise, the banks are liable for their share of the $12.2 billion (plus other costs).
To calculate the defendant banks’ share, MIP took a detour into how much the actual loan losses were and how big the loan pool was that created those losses. It is clear from the 220+ message board comments that, because of the detour, many people got lost in the detail, despite MIP trying to correct them at least a dozen times. The $12.2 billion is not merely a figure that can be used to calculate ratios for estimating damages; it literally is the damages, the primary harm that ResCap suffered due to the banks’ breach of contract. (ResCap suffered other harm via liability to other parties beside the trusts and monolines in bankruptcy, plus via payments made prior to bankruptcy.) So, for example, when someone asks, can’t the banks argue that some of their loans went bad not because of poor underwriting but because the economy and home prices tanked, the first-level answer is “no.” ResCap, too, resold loans that went bad without poor underwriting. That is largely why it only had to pay a $12.2 billion bankruptcy claim on total loan losses that were estimated at $47.1 billion.
Each bank sold some well-underwritten loans and some poorly-underwritten loans that went into the trusts. Their collective poor underwriting resulted in a $12.2 billion loss for ResCap (plus bankruptcy costs). To calculate each bank’s share of the $12.2 billion, the lawsuits need to determine each bank’s share of the poorly-underwritten loans. That share can be derived from two numbers: (1) the principal balance each bank sold to ResCap, put into the trusts, and still remaining in the trusts as of the bankruptcy, and (2) the underwriting defect rate for each bank’s loans, as compared to the average defect rate. In other words, if one bank did a better job of underwriting and has a lower defect rate, it will be liable for a smaller share of the $12.2 billion, relative to its total loans sold, and vice versa. Put differently, a bank can’t argue merely that some of their loans went bad for reasons other than poor underwriting, but it can argue that more of them than average failed for other reasons, because it was a better-than-average underwriter.
One of the only uncertainties in the lawsuits is what method the courts will require the parties to use to determine each bank’s share of the $12.2 billion. The banks want to force ResCap to prove a contract breach for every single loan. But ResCap is likely to persuade the judges to allow some version of what was done in ResCap’s bankruptcy: re-underwrite only a sample of the loans, calculate an underwriting defect rate in that sample, and use the calculated rate to determine liability.
The legal significance of the $12.2 billion cuts both ways, however. I do not agree with MIP and some others that there is potential additional upside from the breach of contract claims on top of the indemnity claims. Rule #1 of contract law – almost literally, you learn it in your first day of Contract Law class, just like in the famous scene in The Paper Chase with Professor Kingsfield – is that damages equal the difference in value between what you were supposed to get and what you actually got because of the breach. As I have said before, for the loans in the trusts, the $12.2 billion in awarded bankruptcy claims (plus costs) is the damages, the reified harm to ResCap. If no bank had breached its contract on the trust loans in bankruptcy, ResCap would have been only $12.2 billion better off today (plus costs). It does not matter that the total loan losses on the trust loans are $47.1 billion or that the total loan losses on the loans with underwriting defects was $22.6 billion ($18.3 billion for the trusts and $4.3 billion for the monolines). Thanks to the bankruptcy process’s extinguishment of claims against ResCap, any losses above the $12.2 billion (plus costs) were suffered by other parties, not ResCap, and cannot be part of ResCap’s breach of contract award.
Moving to the next level of detail: Some people are thinking that, because the contract allows ResCap to put loans back to the banks, ResCap can put them back now, which would effectively allow it to collect 100% of the loan losses. Unless the banks are being run by idiots, that is not going to happen. Putting back the loans is a contractual right. The banks will say “we’re not taking back the loans, we are breaching that contractual provision, go ahead and sue us for breach of contract – oh yeah, you already did.” And then we are right back to the usual measure of damages. ResCap may try to argue that, because it still has the right to put back loans now, post-bankruptcy, a breach of that provision still creates damages equal to 100% of the loan losses, notwithstanding the bankruptcy. That argument has some Jesuit validity, but I would be very surprised if courts would allow ResCap to lop 46% off its liability for these loans in the bankruptcy (paying only $12.2 billion on $22.6 billion of losses) and then turn around and get 100% of the losses from the banks.
WHY THIS LEGAL SET-UP IS SO GOOD
Two separate factors contribute to unusually low legal risk in this situation: the conceptual strength of ResCap’s claims, and the fact that the precise courts that will decide the claims have already ruled on most of the key issues.
The claim is so conceptually strong in part because it is so simple: The banks sold loans with reps & warranties, and ResCap resold them with essentially the same reps & warranties. ResCap has already been held liable for breaches of the reps & warranties; apart from procedural defenses like statute of limitations, it is not conceptually possible for the banks to escape liability for the same reps & warranties on the same loans. On top of that, MIP fully laid out the incredibly generous terms of ResCap’s contracts with the banks, via the Client Guide. In case you missed it, I want to highlight a transcript snippet that Chuplin posted from one of the Minnesota judges in these cases: “[ResCap’s] contract is a pretty remarkably generous contract. If I ever become a movie star or something, I’m going to hire the lawyers that wrote this contract to write my agreements for me, because I would get everything and the other side would get nothing. They are pretty one-sided contracts.”
Even so, litigants lose strong cases all the time, because juries or judges get things “wrong” all the time. They may believe the other side’s witnesses when they should have believed yours. They may simply mis-analyze. They may have some sort of bias that causes them to intentionally lean the other way. The more complicated the legal claim, and the more facts in dispute, the higher the risk. A typical risky litigation situation with great complexity and many factual disputes is a jury trial of a patent infringement claim. See, e.g., VirnetX and Parkervision, each of which managed to persuade a jury to find for them only to have the trial judges reverse the verdicts recently because – this is the actual legal standard for reversal – no reasonable jury could have found for them.
In ResCap’s case (or rather 89 cases), there is almost nothing a judge could get wrong. The liability claims are simple. And there are no disputed facts, apart from the underwriting defect rate, which will merely move the needle one way or the other in a given individual case. (And based on the one-sided Client Guide, ResCap may be legally entitled to declare what the defect rate is.) The statute of limitations issues are a bit more complex, but not badly so.
Even those strengths would not be enough to eliminate material risk entirely, because of the chance for judicial error or bias. I can contrast with two other situations: In the Gyrodyne case prior to the appellate oral argument, I used a 20% chance of losing, even though they had already won the trial, the other side’s lawyer had completely botched the defense at trial, the law was simple, and the trial court’s factual rulings were theoretically not open to appellate review. The primary risk was that the judges would be politically biased to favor the state government. In the current Fannie/Freddie litigation, last month I read all ~600 pages of pleadings and briefs and spoke with one of the nation’s top DC appellate litigators, concluded that the plaintiffs “should” win on the legal merits, but didn’t invest because (1) the hedge fund plaintiffs made unsympathetic plaintiffs, (2) a federal government attempting to implement national policy to “save” and shore up the entire U.S. housing market was a very sympathetic defendant, (3) the plaintiffs had a very unsympathetic claim, because the government committed hundreds of $billions and promised to commit almost unlimited additional amounts to save the GSEs from certain liquidation absent the government help, (4) the plaintiffs had a trial judge who tends not to favor the government but would then face a court of appeals dominated by judges who would favor the government, and (5) the legal issues were complex enough and debatable enough that a judge who wanted to find for the government could easily do so on several different grounds.
In these cases, I do not see much risk of bias. The defendants are banks who sold mountains of crappy loans, and ResCap has a garden-variety contract claim with no policy implications. Just as importantly, as MIP laid out up-front, the very trial court hearing these cases, (the District of Minnesota) and the court that would hear any appeal (the Eighth Circuit) have essentially already ruled for ResCap on the key liability issues in another case, Terrace Mortgage. MIP laid that out clearly, but some people have expressed concern that he is not a lawyer, so I will be the ex-litigator telling you, to quote Vice President Biden, this is a big @#$ing deal. The courts are not going to reverse themselves on those issues. Even for the cases in New York, those courts have a precedent involving the same contract with the same plaintiff; they are highly likely to follow that precedent. The Terrace decisions significantly lower my estimate of the litigation risk. That lowering does not apply to the statute of limitations issue, but even on that issue, three magistrate judges have already ruled for ResCap, and I agree that they have ruled correctly.
WHY SETTLEMENTS MAY COME SOONER RATHER THAN LATER
Five words: ZIRP, prejudgment interest, and attorneys’ fees.
In a normal litigation world, defendants want to drag the case out and make the plaintiff do everything the hard way. Even if they think they are probably going to lose, every day of delay is a day that they get to keep and use the money and the plaintiff doesn’t. If they lose they have to pay prejudgment interest for those days of delay, but they might win and pay nothing, and normally the interest isn’t such a big deal. Furthermore, every dollar they force the plaintiff to spend on attorneys and other litigation costs eats into the plaintiffs’ net recovery. The defendant can use the prospect of that spending as leverage in its settlement negotiations.
Almost none of that applies here. The contract allows ResCap to recover all its litigation costs from the banks, so it is the banks who are hurt by delay-induced spending. Tthe banks must know that they have little hope of winning – if they don’t know now, they will once all the preliminary motions are decided – so they should not meaningfully discount the prospect of paying prejudgment interest and costs.
Just as important, probably, is the fact that the Fed Funds rate is zero and these defendants are banks. In the old-normal world of 5% Fed Funds rates and 7% 10-year Treasury rates, paying 9% or 10% prejudgment interest isn’t so bad. It looks very different when risk-free rates are 0-3% today and perhaps 2-4% 18 months from now. And nobody is more acutely focused on its cost of capital than a bank. Today the average U.S. bank’s net interest margin is 3.0% (I just looked it up), composed of a gross return on assets of maybe 5-6% and a cost of liabilities of 2-3% (I am making an educated guess on those numbers). The ResCap litigation liability is an unquantified but real liability, effectively a source of capital just like the banks’ other ones, except this one is costing them not 2-3%, but rather 10-11% (including the attorneys’ fees etc. that are accruing month after month). The management teams may have some incentive to delay if they have under-reserved for this liability, but my guess is that the double-digit cost of doing so creates a stronger incentive to settle more quickly.
Based on the strength of ResCap’s claims and the interest and costs that are accruing, I also believe MIP is reasonable to assume that settlement amounts will be for a high percentage of claimed damages.
ADJUSTMENTS TO THE ORIGINAL LITIGATION ASSET VALUATION
MIP’s original analysis started with the Original Principal Balance (OPB) of loans claimed in all the current lawsuits. He then calculated a “loss ratio” of the loan losses that had been estimated in the bankruptcy case, divided by the OPB in the bankruptcy case. That loss ratio is not necessary for calculating the banks’ liability in the current cases, because as explained above, the $12.2 billion bankruptcy claim award makes the original loss amount almost irrelevant. However, MIP did not want to start with the $12.2 billion because he thought it would require him to make an estimate that he did not think he could make accurately (the % of correspondent banks that have survived since 2007). The method he did use was a perfectly sound analytical framework.
Unfortunately, MIP may have mis-estimated his loss ratio. He used losses and OPB from only a portion of the trusts (the ones without a monoline insurance wrapper) and calculated a loss ratio of 63.5%, and then applied that ratio to the entirety of the bank defendants’ OPB. The number may be correct for the unwrapped trusts, which presumably were unwrapped because they contained the lower-quality loans. But as Wolverine noted (comment #214), as a loss ratio for all loans sold to ResCap, it does not pass a basic sanity check; there is no way that 2/3 of all loans went bad with 100% losses on each loan.
We don’t need to guess at the actual loss ratio for all the loans; the Sillman Examination provides data to make a better ratio. Annex C, the defect rate analysis, lists OPBs for each trust that sum to $292.8 billion. The total losses estimated by Sillman were $47.1 billion, $42.3-43.2 billion for the trusts and $4.2-4.4 billion for the monolines. Therefore the total loss ratio is 16.1%, only ¼ as large as MIP’s ratio.
MIP’s response is that, in the lawsuits against the banks, ResCap has only included the dodgy categories of loans (e.g., subprime and Alt-A) that have much higher defect rates and severity and therefore much higher loss ratios. That argument has some strengths and some weaknesses such that the truth is probably somewhere between the two extremes; the loss ratio on the defendants’ OPB is probably greater than the 16.1% for the total trust OPB but less than the 63.5% on the unwrapped trust OPB. Working against MIP, in any given lawsuit against a specific bank, ResCap has no incentive to exclude OPB from the non-dodgy categories. In every category of loans, surely some of them had underwriting defects that caused losses, for which ResCap could recover. Also, we know for a fact that the defect rate on the entire universe of loans in the trusts was quite high. Remember, Annex C, which includes the full $292.8 billion of trust loans, was the defect rate analysis, and Sillman found underwriting defects in 39% of them. Therefore many loans with defects didn’t have much losses.
In MIP’s favor is the following argument/question: With a universe of $292.8 billion in trust OPB, why has ResCap sued (so far) based on only $43 billion, 15% of it? What happened to the other $250 billion? Assuming that only correspondent banks can be sued now and those banks sold 61% of the OPB, that still leaves 46% unaccounted for. Some banks failed and were liquidated and cannot be sued, but the assets of all liquidated banks, relative to total bank assets, were tiny. (The banks that were liquidated, although numerous, were tiny banks. I had thought that WaMu was an exception, but I was reminded that its operating entity was sold to JPM with liabilities intact; only WaMu’s holdco was zeroed out.) The mere fact that the complaints are missing 46% of the OPB strongly implies that ResCap has in fact excluded a lot of low-loss OPB. One way this could have happened is if a number of banks sold only non-dodgy (prime) loans to ResCap. The losses and potential recovery from those banks could be low enough that ResCap would decide it was not worth suing them. Or perhaps I am wrong and there was some valid reason for ResCap not to include all the OPB for some of the banks it did sue. Another plausible explanation is that ResCap will keep adding much more OPB to the lawsuits.
If the loss ratio were the only adjustment to be made to the original valuation, and if I and others are correct that a 63.5% loss ratio for the $51 billion of currently claimed OPB is far too high, the litigation asset value estimate would need to be cut by as much as 75%. But new events and new analysis (mostly by MIP himself) have pushed the number all the way back up to where it started. Those factors are:
1. Higher trust OPB from the defendant banks
Back in May, the OPB in the securitized trusts identified in the federal-court lawsuits was only $31.5 billion. Today it is $43 billion, 37% higher. The increase comes from three sources: A few new cases were filed. Some cases were moved from state court, where the complaints were not easily accessible to see the OPB, to federal court, where the complaints are accessible. And some complaints were amended to add more trust loans with more OPB. Based on the increase in OPB in the lawsuits so far and the large amount of “missing” OPB, I believe there is a decent chance the claimed OPB continues to increase, potentially by a large amount.2. New non-trust OPB from the defendant banks
ResCap amended the complaints in the Southern District of New York to add, for the first time, $7.8 billion of non-securitized loans that were outside the trusts and were directly resold by ResCap. (Thus the grand total of OPB is now $51 billion.) Prior to amendment, the total OPB in the New York cases was $6.4 billion. The new non-securitized OPB of $7.8 billion more than doubles the total. I have no idea if proportionately similar amounts, or any amounts, can be added to the Minnesota complaints. If so, the total OPB may double from here…
…but with a catch. By definition, the non-securitized loans were not contained in the trusts and therefore were not part of the $12.2 billion bankruptcy awards to the trusts and monolines. It is likely that most of the non-securitized loans were in the bankruptcy in the form of other claims by other parties, and that ResCap has some other claim award covering them. It is also possible that some non-securitized loans created damages to ResCap before the bankruptcy, in the form of ResCap making forced repurchases of bad loans or making restitution payments. Assuming that most of these non-securitized loans were in the bankruptcy, ResCap’s losses on them, per dollar of OPB, is likely to be lower than for the trust loans, because the claims allowed to other parties were lower.
3. Pre-bankruptcy losses
The other major sources of additional value were flagged by MIP in comment #216. One, similar to what was just discussed, is pre-bankruptcy losses: “From January 1 2006 through March 31, 2012, [ResCap] repurchased mortgage loans or otherwise made payments with respect to representations and warranty claims of approximately $2.8 billion.” These amounts were not part of the bankruptcy process and are in addition to the $12.2 billion in bankruptcy awards. The defect rate on these loans is necessarily 100%. If, say half the $2.8 billion in outlays were repurchases, the loss severity on defective loans was 54% similar to the trust loans, and the other half of the payments was straight compensation to the buyers (100% damages), then ResCap has $2.2 billion of potential legal claims from this bucket.
4. Pre-judgment interest
Although the breach of contract claims shouldn’t get ResCap more damages, they should get ResCap more prejudgment interest. I agree with MIP that the prejudgment interest clock did not begin ticking for indemnity claims until the bankruptcy plan was effective (Dec.2013), while for breach of contract claims it began when the loans were purchased. Contract claims for loans purchased prior to May 2006 will probably be time-barred by the statute of limitations. MIP says that 40% of the OPB was purchased after that date and should have a valid contract claim; I’ll use his number. Assuming these loans were mostly purchased in 2007, and that settlements will come in 2 years in late 2016, that is a full 9 years of prejudgment interest. Add 3 years of interest on the other 60% of damages, and at 9-10% non-compounded the total prejudgment bill is going to come to 52% of the damages amount.
5. Bankruptcy costs and attorneys’ fees
The indemnification claim covers all costs and expenses resulting from events of default, which would include the bankruptcy costs. Those costs are in the hundreds of millions, and the defendants are liable for their share.
6. Preference cases / DOJ/AG settlement
As these amounts are in the tens of millions, which do not move the needle, I am going to ignore them.
MY NEW NET ASSET VALUATION
I take a different approach to sizing the bankruptcy claim recovery. I start with the $43 billion in trust OPB in the complaints, as before. I then assume that more OPB is going to be added, 10% more in the bear case and 30% in the bull case. That yields $48-56 billion of trust OPB. I then assume that, at a minimum, the defendant banks will be liable for their share of the $12.2 billion in awarded claims, proportional to their share of the total $292.8 billion in trust OPB. Again, the better underwriters will pay less, the worse ones more, but it should average out. $48-56 billion would be 16.2-19.2% of the total trust OPB, and those shares of the $12.2 billion would be $2.0-2.3 billion.
This first pass implicitly assumes that the loss ratio on the OPB in the complaints is only the 16.1% that was estimated for all the trust OPB. I leave that unchanged in the bear case, but for the reasons described above, I increase the loss ratio to 24% in the base case and 32% in the bull case – still only half as high in the bull case as the original write-up had assumed. Using these higher loss ratios as compared to the 16.1% total loss ratio, I increase the banks’ share of the $12.2 billion settlement to match their assumed share of the losses instead of their share of the OPB. That yields damages of $2.0, 3.2, and 4.7 billion in the bear/base/bull cases, still well below the theoretical $12.2 billion of total damages.
(Note that, mathematically, I have now duplicated MIP’s original framework via a different formula. I prefer mine because it intuitively follows the legal reality of the $12.2 billion as the damages.)
Next I add a recovery for non-securitized OPB. I assume that more new non-securitized OPB will be added than new trust OPB, $10-30 billion. However, I assume that the damages per dollar of OPB will be only about 1/8th as much as for the trust OPB, 0.5-1.0% versus 4.2-8.3% for the trust OPB. Those assumptions yield $89-378m of additional damages:
Next I add the pre-bankruptcy losses. Barring any data on how much of the outlays were repurchases versus payments, but noting that the defect rate on these loans is 100% and much of the payments could be compensation (100% loss) rather than repurchases, 50% seems like a good floor for how much of the $2.8 billion could be damages. However, all those damages did not come from loans sold by the currently-sued banks, and I assume that ResCap has sued nearly everyone it intends to by now. If we assume that the defendant banks are responsible for the same share of these losses as of the trust losses (16-38%), that yields additional damage claims of $227-747m.
The total base damages from trust, non-securitized, and non-bankruptcy losses is $2.3-$5.8 billion.
Next I add the prejudgment interest, fees, and costs. The interest calculation assumptions are as described above: 40% of damages are from breach of contract with 9 years’ duration, the other 60% from indemnification with 3 years’ duration. I add $100-300m for bankruptcy attorneys’ fees and costs. The total interest, fees, and costs is $1.3-3.3 billion. The base case of $2.2 billion for these add-on recoveries is 144% of ResCap’s current market cap.
Note that ResCap should also recover attorneys’ fees for its litigation against the banks, but because that will be reimbursement for outlays that are not included in its current net asset value, the recovery is a wash for valuation purposes.
The total litigation claims against the banks are worth $3.6-$9.1 billion if litigated to judgment. I next apply a settlement percentage of 70-80%. If I were ResCap, with a case this strong and with prejudgment interest ticking at 9-10% and my attorneys’ fees ultimately reimbursed, I would reject settlement offers of less than that, take the case through trial and appeal, and collect my 100% plus interest.
Thus the total litigation settlement proceeds should be $2.5-$7.3 billion, or $26-$74 per ResCap unit.
Finally I add the non-litigation net assets. I have done nothing to the original write-up’s numbers other than subtract the June and September distributions of $115m and $150m.
That yields a total value of $2.8-$8.0 billion or $29-81 per unit, with a base case of $52 and 228% upside.
Finally, what if MIP’s original 63.5% loss ratio for the OPB in these lawsuits is correct? His original estimate of solely the trust claims damages, back when there was only $31.5 billion of OPB, was $3.5 billion in the bear case and $5.2 billion in the base and bull cases. If I raise those values for the increase in OPB to $43 billion, the estimated damages increase to $4.9-7.1 billion, compared to my $2.0-4.7 billion. Plugging those numbers into my model, so that I add the non-securitized settlement value, pre-bankruptcy settlement value, interest (which goes up as the damages go up), fees, and costs, I get the following total net asset values and upsides:
This outcome seems unlikely to me, but as the Beach Boys said, wouldn’t it be nice….