AMBAC FINANCIAL GROUP INC AMBC
September 28, 2019 - 6:43pm EST by
katana
2019 2020
Price: 19.14 EPS 0 0
Shares Out. (in M): 49 P/E 0 0
Market Cap (in $M): 940 P/FCF 0 0
Net Debt (in $M): 0 EBIT 0 0
TEV ($): 0 TEV/EBIT 0 0

Sign up for free guest access to view investment idea with a 45 days delay.

Description

INTRODUCTION

We have been following Ambac for more than six years; now is the time to buy.  Ambac has become a full-blown event-driven special situation with possible 100-200% upside to be realized within a year, and possibly much sooner, when it (probably) announces a large lawsuit settlement that should be well in excess of what most investors are expecting.  If we are wrong about the settlement size, the likely stock downside is zero because the settlement will still likely meet most investors’ expectations. We see only a small risk of a market-disappointing settlement or a trial loss.

For several years, Ambac has steadily become more and more of a pure bet on the value of its residential mortgage-backed securities (RMBS) representations-and-warranties claims against some big banks.  The other potential negative and positive factors that existed several years ago have largely played out. The RMBS claims’ true value has grown every quarter while their recorded book value has not, due to a quirk in GAAP.  The true value’s excess above its recorded book value is likely more than Ambac’s entire market capitalization. The largest hidden value is Ambac’s entitlement after a trial to 12 years’ worth of 9% prejudgment interest on its RMBS damages (= 108%), which GAAP requires Ambac to exclude from its balance sheet.  Ambac currently values its expected RMBS recoveries, excluding prejudgment interest, at $1.8 billion.  Thus as a first rough cut, the unbooked value of prejudgment interest alone could be $1.9 billion, 2x Ambac’s current equity market value (i.e., 200% upside for the stock price).  More refined cuts are below.

All of the settlement’s excess value will benefit Ambac’s shareholders, thanks to $billions in NOLs that can offset the realized gain.  Most of it should be realized when Ambac settles its litigation with Bank of America (BAC). Last week the primary BAC lawsuit cleared the final major pre-trial hurdle and should be scheduled for trial by mid-2020.  Settlement should occur at or before the trial date and could occur at any time; the likelihood of a settlement before trial has risen now that all major pretrial rulings and appeals are complete. 

The settlement upside is higher than it was for Assured Guaranty’s settlement with BAC in 2011, MBIA’s with BAC in 2013, or Ambac’s with JPMorgan in January 2016, because Ambac’s BAC litigation is much closer to trial, has already survived key legal challenges on appeal, and has been accruing 9% prejudgment interest for up to four times as long.  These factors are probably not obvious to most investors, so the settlement’s upside surprise will also probably be larger. We can use the RMBS claims’ book value amount, previous settlement amounts, prejudgment interest calculations, and some assumed settlement discount ranges to estimate that Ambac’s recoveries should exceed their book valuations by at least 100% of Ambac’s market capitalization.

 

CONTEXTUAL BACKGROUND

Ambac remains a fairly complicated and opaque beast, and this write-up focuses almost entirely on the RMBS litigation upside.  For a background on where Ambac has come from, you can review the 2013 write-up from mip14, the 2017 write-up from lvampa1070, and the 2019 write-up from Den1200. Lvampatook a unique and unusually helpful look at Ambac’s equity value, starting with the most certain and concrete assets and working down to the more speculative upsides.  Lvampa found considerable total upside but listed upside of “only” $675m = $15 per share in book value from the RMBS litigation. That number is probably too low and potentially much too low.  Den1200 briefly covered the RMBS claims and their hidden value but did not try to size them. It is the sizing, and last week’s appellate decision clearing the way for a trial date, that makes this investment so compelling today.

 

THE AMBAC STORY HAS BEEN CLEANED UP

Compared to 2013 or 2017, Ambac has materially reduced several important uncertainties and downside risks.

First, Ambac’s outstanding insurance obligations have steadily and rapidly declined, thanks to a combination of maturities, payments, settlements, and active risk-reduction efforts by management.  Net par insurance exposure has fallen from $200 billion in 2013 to $60 billion in 2017 to $42 billion today. Adversely classified exposure has fallen from $32 billion in 2013 to $15 billion in 2017 to $9 billion today.  Ambac’s adjusted book value has increased from $4 in mid-2013 to $24 in 2017 to $30 today.

Second, Ambac successfully exited its rehabilitation in early 2018 and positively restructured its obligations as part of that process.  The exit gives Ambac a freer hand to pursue a number of activities without risk of the Wisconsin regulator disallowing them. Most notably, Ambac has $1.3 billion of NOLs at the holding company level and $2.2 billion at the opco level, with no holdco income against which it can use the holdco NOLs.  Now Ambac can pursue corporate transactions to use them. We assume Ambac is waiting for the BAC settlement before acting on this opportunity, because the settlement’s size could materially change the amount of available NOLs.

Third, Ambac’s largest insurance risk, Puerto Rico, has largely played out and crystalized itself in Ambac’s stated book value.  Ambac insures $1.3 billion par value of Puerto Rico bonds, all of which went into default. Much of the original Puerto Rico par value came from the COFINA senior bonds; at a previous value of $805m, COFINA was Ambac’s single largest insurance liability.   Last year Ambac hit a home run with those bonds. In early 2018 when investor fears over Puerto Rico were at their worst, the bonds were trading at 50-60 cents. Ambac bought up every bond it could in the open market, 58% of the total. In August 2018, COFINA became one of the first Puerto Rico bonds to reach a settlement between bondholders and the debtor.  Senior bondholders received a new package worth 93 cents on the dollar. Ambac realized a nice gain on its purchases and eliminated most of the downside risk on its largest insurance liability. The COFINA insured par value is now down to $178m. On the remaining Puerto Rico liabilities, Ambac has been increasing its reserves by modest amounts for the last two years, which has caused its adjusted book value to stop rising during those years but also, probably, means the downside is largely covered.  Ambac continues to post a frequently-updated summary of its Puerto Rico exposure, which can be found here.

 

RMBS LITIGATION BACKGROUND

Ambac became insolvent and entered rehabilitation because it had ensured tens of $billions worth of RMBS that were backed by toxic mortgage loans.  Many of these loans defaulted in part because the U.S. housing market imploded but in part because they violated the banks’ stated underwriting standards and violated other representations and warranties in the relevant RMBS loan-sales contracts.  The banks’ liability to the monoline insurers for these loan sales is unusually clear. Most of these lawsuits between the monolines and big banks were settled years ago, with the monolines receiving most of their claimed damages amounts even at early stages of litigation.  Ambac settled its second-largest set of claims, with JPMorgan Chase, in January 2016.

The sole remaining large holdout is Bank of America, via its acquired subprime poster-child, Countrywide.  BAC settled with MBIA and Assured Guaranty years ago but has not yet settled with Ambac. The reasons for this outlier large-ticket delay are unclear.  It is possible that Ambac is simply holding out for a higher settlement relative to its claim size than what MBIA or Assured Guaranty were willing to accept.  It is also possible that BAC thought Ambac’s rehabilitation put it in a weaker negotiating position than MBIA or Assured Guaranty had. Whatever the reason, the procedural posture is this:  The primary case against BAC is in New York state court. The parties conducted discovery and filed a number of summary judgment motions back and forth, leading to two rounds of appeals. The second round of appeals was decided in June 2018, and trial was scheduled for February 2019.  BAC then filed a new set of pretrial motions, distinct from the earlier summary judgment motions. The trial judge denied all of them last December, and BAC appealed those rulings.  

The appellate court heard oral arguments in May and issued its decision last week.  It affirmed the rulings for Ambac on each of the substantive issues – upholding the fraud claims, declining to limit which loans can create liability, and allowing statistical sampling to establish liability and damage amounts.  It reversed and held for BAC on the procedural issues of when to determine whether liability extends from Countrywide to BAC (in a separate trial following the primary one) and who will decide that issue (the judge rather than a jury).  The litigation value of each side’s appellate win roughly balances that side’s loss, and the stock price did not move when the decision was published.  

Because each side lost part of the appeal, each has 30 days to request leave to appeal the new decision to New York’s highest court, and at least one side is likely to make the request.  If the request were granted, the trial judge would likely delay the trial date further to await the high court’s decision, which would add another six months or so to the schedule.   However, the request is highly likely to be denied, because the appeal was pre-trial rather than on a final judgment, not dispositive to the case on any issue, and decided unanimously by the appellate judges.  The trial court’s next scheduled conference with the parties is in November, right around the time a motion for leave to appeal is (probably) denied. The trial court is likely to schedule a trial date during the November conference or soon thereafter.  We can infer from the previous lower-court schedule – trial was scheduled for five months after the judge heard these pretrial motions – that little pretrial activity remains to be done. Thus, if the high court doesn’t get involved, the new trial date should be six to nine months from now.  (With the caveat that litigation frequently takes longer than the estimates, even though the estimates are made by people who know that litigation frequently takes longer than the estimates.)

Ambac has several other RMBS lawsuits in process.  Just as it regularly updates a slide deck detailing its Puerto Rico exposure, it regularly updates a slide deck detailing its RMBS lawsuits’ status, which can be found here just below the Puerto Rico deck.  To summarize Ambac’s summary, those lawsuits are:

  • Several others against BAC/Countrywide, which would presumably be packaged with the primary litigation into a global settlement.

  • A lawsuit against Nomura in New York state court.  Fact discovery is complete, expert discovery is ongoing.

  • A lawsuit against U.S. Bank in New York federal court and Minnesota state court.  This one is tiny and is qualitatively different from the others. U.S. Bank, as RMBS trustee, sued Countrywide, the mortgage originator, and later settled for $94 million.  Ambac sued U.S. Bank for allegedly poor litigation choices and for accepting too low a settlement, which increased the amount of insurance claims Ambac had to pay. The trial court recently dismissed four of Ambac’s five causes of action; the core claim for breach of contract remains.

  • A lawsuit against U.S. Bank in New York federal court.  Fact discovery is ongoing.

 

THE RMBS LITIGATION UPSIDE

The earlier settled RMBS cases resulted in payments that were much larger than the amounts recorded on the monolines’ financial statements.  MBIA settled with BAC in May 2013 for $1.7 billion. On the day of the announcement, MBIA’s stock rose 45%, which was about $1 billion of market cap.  Assured Guaranty settled with BAC for $1.6 billion in April 2011; its stock rose 24% that day. In January 2016, Ambac settled with JPMorgan Chase for $995m.  Management does not disclose the recovery estimates for specific cases, but in that quarter’s financial statements, Ambac increased the litigation recovery asset’s book value by $278m.  Normally this number doesn’t change much quarter to quarter, so we can safely assume that the JPMorgan settlement was about 40% higher than the JPMorgan proceeds valuation on Ambac’s books prior to settlement.  ($995m less $278m = $717m book value; $278m is 39% of $717m.) Ambac’s stock rose “only” 15% or $83m on announcement day, well below the book value gain, because by then investors had come to expect some excess gains and had already priced them into the stock.

This systematic undervaluation on the financial statements occurs for three reasons, two of which are required by GAAP.  First, as detailed in each 10Q and 10K, the litigation asset amount represents the lawyers’ recovery estimates, discounted for the risk of losing at trial, the risk of a jury awarding lower damages than expected, or – the most likely outcome – settling for less than the expected jury award in order to eliminate those risks.  Litigators tend to be very conservative when making such estimates; they have strong incentives to under-promise and over-deliver for their clients.

Second, Ambac has, and the other monolines have had, legal causes of action for fraud in addition to contract-law claims for breach of reps and warranties.  The trial and appeals courts rejected BAC’s motion to strike Ambac’s fraud claims specifically because those claims can result in compensatory damages measured differently from, and in addition to, the contract claim damages.  More importantly, fraud claims can result in punitive damages on top of the usual compensatory damages. The fraud claims here seem fairly strong in terms of both legal theory and how they will play emotionally in front of a jury.  (Big banks knowingly pumped out hundreds of billions in toxic loans, destroyed the economy and bankrupted countless firms and people, got bailed out by the government, no one went to jail, then the banks refused to take back the mortgages as required by the contracts and refused to pay what they owed to the plaintiffs under clear contracts.)  The risk of punitive damages will influence how much BAC and the other banks are willing to pay in settlement to eliminate those claims. But GAAP requires Ambac to exclude punitive damages from its book value.

Third, and most important, is the 9% prejudgment interest.

New York statutes dictate that a winning plaintiff in a contract case receives from the defendant both the damages award and 9% simple (non-compounded) interest on the award.  This amount is non-discretionary, i.e., a judge can’t reduce it merely because 9% is well above recent market interest rates or because the interest amount can exceed the base damages amount if litigation drags on for more than 10 years (as here).  As with punitive damages, GAAP requires Ambac to exclude all this prejudgment interest from its litigation-recoveries book value.

The prejudgment period is measured from the time of the contract breach, or the time damages occur, if later, through the time of the final judgment.  The RMBS contract breaches all occurred when the RMBS were created, because the representations and warranties were false at the time of creation. That was between 2004 and 2007, with an average date of 2005 or 2006.  The damages started occurring as mortgages started defaulting in 2007, the banks refused to take back the faulty mortgages starting in 2007, ratings agencies started downgrading Ambac’s credit in January 2008, and, at the latest, concluded when Ambac declared bankruptcy in mid-2010.  If this case were tried to a verdict and a court were required to decide the prejudgment interest amount, the parties would argue strenuously about the start date, but we’ll use 2008 as a rough and safe-looking mid-point for all of these events.

Because BAC has delayed settling with Ambac for years longer than with Assured Guaranty or MBIA, its prejudgment interest liability is 3-4x higher, and the amount Ambac should receive in settlement, relative to the litigation asset on its books, is much higher.  As a percentage of the base damages amount, if Assured Guaranty had gone to trial in 2011 rather than settling that year, it could have expected 9%/year x 3 years = 27% in prejudgment interest. In 2013, MBIA could have expected 9% x 5 = 45% in prejudgment interest.  By mid-2020, Ambac can expect 9% x 12 = 108% of its base damages in prejudgment interest.

 

SIZING THE UPSIDE

So how much might Ambac receive in settlement of its remaining RMBS claims?  Even Ambac’s litigators, who have far more information than us, can’t know with any precision.  But we know enough to be confident that it should be far more than what is on Ambac’s books. Here is a SWAG.

Current RMBS litigation proceeds asset on Ambac’s books: $1,793m

+ Settlement discount built into book value (20% of expected damages amount = 25% of book value): $448m

+ Probability-weighted value of fraud punitive damages (which could be >100% of base damages, but the odds of a large award are low): 10% = $179m

+ Increase for inherent lawyer conservatism in estimating book value: 10% = $179m

= Expected base damages at trial: $2,600m

+ Prejudgment interest of 9% x 12 years: $2,808m

+ Attorneys’ fees and costs (as required by the RMBS contracts): $50m 

= Expected total judgments if litigated all the way to trial verdicts: $5,458m

Many of these factors could be off by significant amounts in percentage terms, but it is not worth the trouble of ranging them given the wide range required for the final factor.  The largest unknown is the size of the discount from Ambac’s expected trial recoveries that Ambac will accept to achieve settlements. Here is the range of outcomes assuming a 20-60% settlement discount.

Although choosing a discount number is necessarily a highly imprecise judgment call, a 30-50% discount seems a reasonable range, for three reasons.  First, on its face and considering a host of other actual settlements across a range of different lawsuits with different litigation postures, that range seems undemanding in a case where the defendant has a facially weak position and has already settled nearly-identical cases for very high amounts of the claimed damages.  The discount should be lower in the primary BAC case than in previously-settled cases because Ambac has already eliminated many of the risks of losing or being denied large categories of damages; it has survived three rounds of appeals of various summary judgment motions and pretrial motions.

Second, the numbers implied by a 30-50% settlement discount compare well with Ambac’s JPMorgan settlement at 40% above the recorded book value.  The BAC settlement premium should be higher than the JPMorgan premium because (1) prejudgment interest will have accrued for 4.5 extra years (which is worth an extra 4.5 x 9% = 40% by itself), (2) Ambac’s BAC claims have already survived three appeals, and (3) Ambac’s overall business is in a stronger position than it was four years ago, thanks to the rehabilitation exit, the reduction in remaining liabilities, and the higher book value.  All else equal, from a base JPMorgan premium of 40%, these factors should push the BAC premium above 80% and possibly above 100%. In the table above, a 50% settlement discount implies a $2.7 billion recovery that is “only” 52% above recorded book value, and a 30% settlement discount implies a $3.8 billion recovery that is 113% above recorded book value.

Third, the discount we choose for the actual settlement is necessarily tied to the discount we assume Ambac’s lawyers have already built into the book value.  In the numbers above, we assume Ambac has built a settlement discount of “only” 20% into the book value number (excluding pretrial interest and punitive damages), but we assume that Ambac will actually accept a 30-50% discount from the real trial recovery number (including pretrial interest and punitive damages).  We assume a higher real-life discount percentage than what we assume Ambac’s lawyers used for book value because we believe Ambac will in fact act differently when working with a $5.5 billion potential trial recovery that includes speculative punitive damages and pretrial interest than Ambac has assumed it will act with a starting amount less than half that size.  If one chooses to increase the size of the actual settlement discount, one should also increase the size of the discount embedded in the book value, which increases the size of the base trial recovery amount and leaves the ultimate settlement recovery roughly unchanged.  

Circling back to our first-cut estimate in the introduction:  If one assumes the Ambac lawyers and accounting estimated everything precisely correctly to produce their $1.8 billion book value, the extra 108% prejudgment interest is worth another $1.9 billion.  Our more detailed calculations produce an expected excess value range of $0.9-2.0 billion. The sanity-check number of $1.9 billion suggests the actual outcome may be closer to the top of our range.

To highlight what these numbers imply for Ambac’s stock value:  The range produces settlement payouts that are $19-$41 per share more than is included in Ambac’s current book value, roughly 100-200% of the current stock price.  Because this excess will be realized via near-term cash payments to Ambac and not merely adjustments to estimates for years’ worth of future payouts and receipts, equity investors should value it at 1.0x the cash received, not the 0.6-0.7x of adjusted book value at which they have been valuing the overall business.

Although we cannot know how much of this upside is already embedded in Ambac’s current stock price, it appears that very little of it is.  Ambac has been consistently trading near 0.6-0.7x adjusted book value, including over the past year while it has been out of rehabilitation. MBIA and Assured Guaranty tend to trade at the same multiple range, without the prospect of a large RMBS settlement ahead of them.

 

RISKS

Nothing is certain in litigation.  Stuff happens, stuff that surprises even the litigators who are working on the case, let alone those trying to peer in from the outside.  The tail-risk downside here is that the parties fail to reach settlement and BAC wins at trial. The primary source of that risk seems clear: BAC’s legal argument that, if Countrywide is found liable, those liabilities do not extend from the subsidiary Countrywide to its parent BAC.  Ambac could theoretically win a multi-billion judgment against an insolvent Countrywide that never pays. The strongest argument that BAC is likely to be found liable in addition to Countrywide – and, more tangibly, that BAC's defense doesn't destroy the settlement value is that BAC had the same defense available against MBIA and Assured Guaranty. If BAC can escape Countrywide’s liability to Ambac, then it could have escaped liability to the other two as well. No one is better situated than BAC to assess this legal argument’s strength, and BAC necessarily decided it was weak enough to justify paying MBIA and Assured Guaranty $1.7 billion and $1.6 billion early in those cases’ lives.  BAC’s parent/subsidiary defense is also automatically baked into Ambac’s book-value estimate of its settlement proceeds and therefore into our estimate of those proceeds.

The more likely downside, with a smaller negative effect on the stock value, is that the BAC settlement amount is below our expectations.  For the stock to fall, the settlement would need to be below the average Ambac investor’s expectations, not just ours, and that outcome seems unlikely.

 

Given the expected short time frame of this thesis, other potential risks – e.g., Puerto Rico somehow blows up again – seem unlikely to manifest themselves materially  in the interim.

 

I hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise hold a material investment in the issuer's securities.

Catalyst

A settlement with Bank of America is likely by mid-2020 and could occur much earlier.

    sort by    
      Back to top