Sunday, November 25, 2018

SocGen Regulators - Be Not Proud!

Last week, the news media made much of the latest penalty imposed by US authorities on French banking giant Société Générale SA (SocGen) for processing sanctions-violating transactions.

On the back of the settlements, the US Attorney General for SDNY, Berman, has been full of celebratory praise for the "outstanding work" done by his team and his fellow investigative bodies. 

But we have examined the settlements, and they don't seem at all impressive.  Rather, they seem the result of defective investigative work.  Current SocGen shareholders, and ADR-holders, should be vexed.

Snapshot of Holders of SocGen's Sponsored ADR, including
Oregon Public Employees Retirement System
(incomplete list) via Bloomberg LP
Let's explain.

The $1.4 billion settlement will come out of SocGen's shareholders' pockets  not from the employees involved in the long-enduring misconduct nor the supervisors overseeing it.  

Thus, shareholders have long paid the wrongdoers' (no doubt handsome) salaries and now pay for their bad decision-making.  Or, said another way, current shareholders are paying for a concealed, misguided scheme which, as we will see, spanned an 8 or 9-year period ending 8 years ago.  

To be clear, this was no idle incident.  As admitted to by SocGen, this was broad, intentional misconduct, spanning many years, comprising thousands of illicit transactions, deliberately implemented (with procedures drawn up) and concealed.  The wrongdoers were aware, throughout, that they were breaking the law.

Excerpts from the Statement of Facts mutually agreed-to by the US Justice Department and SocGen, include:

  • In total, SG engaged in more than 2,500 sanctions-violating transactions through financial institutions located in the County of New York, valued at close to $13 billion, during this period.
  • For example, a senior member of SG’s Money Market department back office (“MMBO”) wrote to another MMBO employee in 2004 that “[t]he American authorities have now identified the procedure we were using (two MT 202s) to ‘circumvent’ the OFAC rules.”
  • In total, SG processed over 9,000 outgoing transactions that failed to disclose an ultimate sanctioned party sender or beneficiary (“non-transparent transactions”), with a total value of more than $13 billion. The overwhelming majority of these transactions involved an Iranian nexus and would have been eligible for the U-Turn License. There were, however, at least 887 non-U-turn transactions with a total value of $292.3 million that were both nontransparent and violated U.S. sanctions. 381 of these transactions with a total value of $63.6 million were related to the Cuban credit facility conduct described below, while the remaining 506 transactions with a total value of $228.7 million involved other SG business with a sanctioned nexus.
  • Between 2003 and 2010, in connection with the Cuban Credit Facilities, SG engaged in 3,100 unlawful U.S. dollar transactions that were processed through United States financial institutions located in the County of New York, worth approximately $15.1 billion
  • Since at least 2002, SG engaged in the Concealment Practice in order to minimize the risk that sanctions-violating transactions would be detected and/or blocked in the United States. SG employees used cover payments for this purpose, in which SG would send one SWIFT payment message to the relevant U.S. bank, located in the County of New York, omitting the “beneficiary” field that would otherwise disclose the ultimate beneficiary of the payment, and listing only the bank to which the funds should be sent. SG would then send a second SWIFT message to the non-U.S. recipient bank, providing the name of the sanctioned party beneficiary to whom the funds should be remitted. Using this procedure (the “Cover Procedure”), SG would ensure that the sanctioned party beneficiary information was not disclosed to the United States bank that was involved in the transaction.

So, it was rather easy to get around the sanctions controls, and SocGen's employees did so many times .... which is hardly comforting.  But now that that's been uncovered we would, of course, have several SocGen employees awaiting criminal prosecution. Oh, no  there's none of that.  

Let's understand why.  

SocGen failed to self-report its misconduct  oops!

From the looks of it, SocGen didn't disclose the individual misconduct until after the statutory clock had run for bringing criminal claims.  

In other words, even after the so-called "Investigating Agencies" were onto them, they let SocGen self-report any individual violations (which they didn't do!), failed to follow up on a timely basis, and then simply fined the shareholders and declared that a success.  The Investigating Agencies seem to have outsourced their decision-making to the party being investigated, and that party obliged by sending the investigators down the wrong path.  

Now SocGen's stakeholders  which include US pension funds, and likely you and us!  are compensating US authorities for the misconduct of bank individuals 8 or more years back.  

Here are some choice (or inconvenient) extracts regarding the scope of the operation and the statute of limitations running, with our emphasis added.

  • Despite the awareness of both Group Compliance and senior SG management that SG had engaged in both the Concealment Practice and the unlawful U.S. dollar payments under the Cuban Credit Facilities, SG did not disclose its conduct to OFAC or any other U.S. regulator or law enforcement agency prior to the commencement of the present investigation.
  • SG did not disclose the Concealment Practice or the Cuban Credit Facilities during these discussions, and its proposals for the scope of that lookback did not include the time period, business lines, or geographic regions that would have revealed that unlawful conduct. It was only after SG performed a detailed forensic analysis based on the broader scope of investigation required by the Investigating Agencies that it disclosed, in October 2014, the Concealment Practice and the Cuban Credit Facilities to the Investigating Agencies.
  • As a result of this untimely disclosure, the statute of limitations for [Trading with the Enemy Act] or [International Economic Emergency Powers Act] violations relating to the Concealment Practice, and to much of the individual conduct involving the Cuban Credit Facilities, had already run by the time the Investigating Agencies learned of them.
Given enforcement agencies knew in 2014 that there were governance issues and large-scale unsustainable business practices ongoing at SocGen concealed from shareholders, why has it taken until 2018 to share that crucial information?  Regulators often have a specific mission that would encourage the dissemination of precisely this type of information, to ensure efficient and orderly public markets, and maintain the public's trust and all.  (The SEC's mission, for example, reads: "The mission of the SEC is to protect investors; maintain fair, orderly, and efficient markets; and facilitate capital formation. The SEC strives to promote a market environment that is worthy of the public's trust.")

It's worth reading the entire document, as there seems also to be evidence that US regulators could earlier have shut down the misconduct at SocGen, had they earnestly tried. 

High 5s all around - but little achievement
Manhattan US Attorney Berman warns: 
"Other banks should take heed: Enforcement of U.S. sanctions laws is, and will continue to be, a top priority of this Office and our partner agencies." (emphasis added)
But was it a "top priority?"

It took years to identify pretty basic noncompliance. Much of the misconduct was reported to them, or else it was missed.  There doesn't seem to be much enforcement here, aside from the fining of shareholders.

Top priority enforcement would mean identifying anomalous transactions early and shutting down promptly any misconduct before it escalates – not fining an institution's shareholders 8 years after the last of a series of illicit transaction has taken place, which themselves in many cases endured for 8 or 9 years.  The Investigating Agencies let SocGen define the scope of the investigation, and then followed the bait.

It seems very much like the fox was running the hen-house.  And it results simply in more pain for the punter.

Wednesday, November 14, 2018

Can Technology Freshen Up Stale CMBS Ratings?

Sears' recent bankruptcy filing underscored the challenges confronting shopping malls in the late 2010s. Because mortgages on these facilities often account for the lion’s share of CMBS asset pools, shopping mall performance needs to be top of mind for those analyzing (or rating) CMBS tranches.

New technology – originally targeted at investors analyzing retail sector stocks – might also be applicable to CMBS analysts.

Foot Traffic

Consider, for example, Advan Research. The company processes billions of daily foot traffic measurements from cellphone applications, and computes foot traffic data pertaining to 1,800 companies including both retailers and Real Estate Investment Trusts. Since many REITs own shopping malls, the company collects foot traffic data for these retail centers.

I asked Advan for data on a mall discussed in a previous post. A mortgage on The Mall at Stonecrest in Lithonia, Georgia accounts for almost all of the remaining collateral supporting Banc of America Commercial Mortgage Series 2005-1. Fitch rates the most senior remaining tranche, Class B, at Single-B. S&P assigns the same tranche a low investment grade rating of BBB-

Who’s right? The data from Advan suggests a downward trend in foot traffic at Stonecrest, as shown in the accompanying chart. Average estimated visitors for the five Saturdays in July 2017 were 19,816; for the five Saturdays falling 52 weeks later, the average fell to just 12,659. On the other hand, a similar comparison between October 2017 and October 2018 shows only a slight drop, suggesting that perhaps the decline in visits has been arrested. 

To the extent that Advan’s data can be relied upon, it seems to give us a more recently refreshed gauge on the shopping mall’s health than other data sources. Certainly, the trustee report is not giving us up-to-date guidance. The November report includes the following special servicer comments: 
Modification closed and funded 8/5/2017. The loan is currently paying as agreed. The loan matures in 8/2018 and the Borrower advises that the proposed adjacent 100 acre sports project has been put on hold due to lack of funding. Although the collateral is 97% occupied, the dark Kohl's and Sears may trigger some co-tenancy issues. The Borrower advises it is in the market seeking refinancing, but due to the current situation with the sports project and 2 dark anchors, refinancing may not be sufficient to pay off the loan in full at maturity. The Borrower has engaged CREMAC to aid it in its workout negotiations with the Lender/Special Servicer. A new appraisal has been ordered and received. Valuation is under review. Maturity Date extend to 8/1/18; principal reduction in the amount of $1,233,073.95 for a balance of $92,066,680.26; no change in rate of 5.603%. 
These comments do not appear to have been revised since the most recent term extension for the Stonecrest mortgage which was through August 1, 2018.

Social Media and Other Sources

In addition to reviewing foot traffic, analysts can monitor the web and social media for news about relevant shopping malls. For example, a local newspaper, the Springfield News-Sun, reported that nearly 100 cars in the mall’s parking lot were broken into on October 5, 2018. A nail salon employee at Stonecrest argued that the mall does not provide video surveillance of the parking lot, making it harder to identify and apprehend any wrongdoers. A search for #stonecrestmall on Twitter reveals that a shooting occurred at the center – but it took place three years ago.

While it is possible to use free tools like Google Alerts to monitor individual shopping centers, that approach might not scale well to a large portfolio. Specialized search services like Bitvore (for which I used to consult) enable analysts to track news on large numbers of positions, even allowing news searches by CUSIP number.

Cell phone activity, web content and social media posts offer new ways for rating agencies and other analysts to track CMBS mall collateral real time. Finding or compiling the nuggets of useful data from these information streams is a challenge that new technology firms can help solve.


This piece was written by Marc Joffe, who consults for PF2.  Marc Joffe is a Senior Policy Analyst at the Reason Foundation and a researcher in the credit assessment field. 

Thursday, November 8, 2018

KPMG's Big Announcement

If you're ever in the UK or Australia, you'll notice that local newspapers run what seem to be daily articles portraying popular dissatisfaction with the quality of audits being performed.

The Financial Times, back in August, ran a terrific series of articles entitled "The Big Flaw: Auditing in Crisis."  The FT's series, and much of the debate generally, has centered on two broad themes:
  1. the potential for consulting arms of the Big Four to jeopardize the independence of their audit function (to the degree they consult for clients they audit too)
  2. the lack of competition in the audit sector
These issues are serious and thorny.  They are not, however, new.  

Back in 2002, in the near aftermath of Enron's failure (then big-5 firm Arthur Andersen was the auditor) at a congressional hearing concerning WorldCom's failure, Congressman Bernie Sanders castigated an Arthur Andersen representative:
Mr. Dick, it appears very clearly that Arthur Andersen failed in their audit of WorldCom. You failed in the audit of Enron. You failed in the audit of Sunbeam. You failed in the audit of Waste Management. You failed in the audit of McKesson. You failed in the audit of Baptist Foundation of Arizona. What was Arthur Andersen doing? I mean, how do you—it is incomprehensible to me that a major accounting firm can have such a dismal record in trying to determine what the financial health of a company is. It’s almost beyond comprehension.
Recent, topical examples include perceived deficiencies in the audits of Taylor, Bean & Whitaker (Deloitte); Steinhoff (Deloitte); Wells Fargo (KPMG); GE (KPMG); Carillion (KPMG); Abraaj
(KPMG); Colonial Bank (PWC); Vocation (PWC); and Sino Forest (Ernst & Young).

In the news again this week is the 1MDB saga, said to be among the largest of a new generation of frauds. 1Malaysia Development Berhad (or 1MDB) went through three of the Big Four between 2010 and 2016. Each of E&Y, KPMG and Deloitte was either fired or resigned from the role.  KMPG and Deloitte signed off on 5 annual reports between them, with 1MDB reportedly announcing that its 2013 and 2014 audited financials should not be relied on.  Oh well.

Today, KPMG made a significant announcement

Having been criticized by UK accounting regulator, the Financial Reporting Council, for their audits having deteriorated to an "unacceptable level," KPMG decided to limit or stop all consulting work for those large UK clients for which it also acts as an auditor. 

(Interestingly, our understanding is that it was never shown to be the case that this specific conflict undermined the quality of the audit provided; but the possibility remains that it can undermine auditor independence ... and perception can be as important as reality.)

This raises a number of questions:

  • If the conflict is real, why only implement this procedure for large clients?  
  • Why only in the UK?  
  • If this makes sense for accounting firms, would it also make sense for other providers of financial services, like price providers or credit rating agencies: should rating agencies that rate corporates limit their ability to consult for them too (e.g., in the sale of analytics).

Thursday, September 20, 2018

S&P Maintains Investment Grade Rating on CMBS Tranche Mainly Collateralized by a Defaulted Loan

The Class B notes of Banc of America Commercial Mortgage Series 2005-1 (BACM2005-1) are currently collateralized by two commercial mortgages.  

One of these mortgages, a $92 million loan on the Mall at Stonecrest in Lithonia, GA accounts for 96.9% of the collateral pool and is in “special servicing” – a fancy name for workout. Yet S&P maintains an investment grade rating of BBB- on this risky instrument.

The most recent remittance report on BACM 2005-1 (available at CTSLink) includes the following language with respect to the Mall at Stonecrest mortgage:

The loan matures in 8/2018 and the Borrower advises that the proposed adjacent 100 acre sports project has been put on hold due to lack of funding.  Although the collateral is 97% occupied, the dark Kohl's and Sears may trigger some co-tenancy issues.  The Borrower advises it is in the market seeking refinancing, but due to the current situation with the sports project and 2 dark anchors, refinancing may not be sufficient to pay off the loan in full at maturity.  The Borrower has engaged CREMAC to aid it in its workout negotiations with the Lender/Special Servicer.

The “sports project” mentioned in the report is Atlanta Sports City, a 200-acre sports and entertainment complex planned for a plot adjacent to the mall.  If and when Atlanta Sports City opens, it will presumably generate substantial foot traffic in the vicinity of Stonecrest.  But construction has been delayed and there is no clear timeline for completing the project, leaving a large vacant parcel next to the mall for the time being.

Since the servicing note quoted above is dated September 4 and the maturity date was August 1, the Stone Crest loan would appear to be in default. This default follows an August 2017 loan modification at which time the maturity date was extended and principal was reduced by over $1 million.

So how can a CMBS tranche backed almost entirely by a defaulted shopping mall loan be investment grade?  Well, the Class B notes do benefit from “overcollateralization”: two subordinated bonds would absorb losses on the loan before the BBB- class is impacted.

Fitch appears to have a less sanguine view of this overcollateralization benefit:  they rate the notes at single B – deep into junk territory. In its latest update, Fitch reported:

The overall mall and collateral occupancy have continued to decline. As of the September 2017 rent roll, overall mall occupancy declined to 76.1% (from 85.5% one year earlier) after Sears vacated its 145,000sf non-collateral store in January 2018.

S&P’s relatively high rating could be the result of insufficient monitoring, an overly sanguine view of shopping mall collateral or some combination of both.

S&P’s last report on BACM 2005-1 is dated March 2, 2018. The write-up does not refer to press reports about the delay of Atlanta Sports City, so it is unclear whether this news was considered. Further, the certificates have not been downgraded, placed on watch or assigned a negative outlook since the latest remittance report appeared. Since that report indicates that the Stonecrest mortgage was neither repaid nor refinanced by its August 1, 2018 maturity date, some rating action would appear to be warranted.

Overrated Shopping Mall CMBS

In 2015, I argued strongly against inflated credit ratings on Commercial Mortgage Backed Securities, especially those with a collateral pool consisting of a single shopping mall loan. Because they lack diversification, such deals expose investors to event risk inconsistent with the AAA ratings assigned to the senior tranches in these deals.

With six NRSROs competing for generous fees on rating CMBS transactions, the ability for deal underwriters to engage in rating shopping is high and the incentives for rating agencies to lower their credit standards is strong. Assigning inflated ratings in any one asset class violates Dodd Frank’s universal rating symbol mandate, according to which symbols must have the same risk implications across all asset classes. Moody’s was recently sanctioned by the SEC for its apparent failure to apply universal rating symbols when rating CLO Combo Notes.

Although none of the single mall deals I listed in 2015 has experienced credit events thus far, they have yet to be tested by a recession.  In the meantime, we have seen abundant evidence that shopping malls are vulnerable. Brick and mortar retail faces a stiff challenge from Amazon and other online retailers. Several national retail chains have filed for bankruptcy or announced large-scale store closures, creating mall vacancies.

Back to BACM

Although BACM 2005-1 launched with a diversified portfolio securing the issued notes, it had a heavy retail weighting – loans in this category comprised 35.8% of the initial collateral pool. The Class B certificates received initial ratings of AA from both S&P and Fitch, levels that proved too optimistic given the performance of the collateral pool.

Thus far the deal has realized $193 million in cumulative losses, representing 8.4% of initial collateral. The failure of Stonecrest Mall SPE to pay off its loan on the original maturity date of October 1, 2014 has left Class B investors in the deal for a much longer duration than originally expected. This bond’s estimated final distribution date was March 10, 2015 according to the original prospectus.

What remains now closely approximates a single asset CMBS, but one with distressed collateral. Class B will probably pay off in full at some point since junior notes are available to absorb some amount of additional write-downs. But ratings are supposed to reflect a greater level of precision than the word “probably” communicates.  According to S&P, obligations rated 'BB', 'B', 'CCC', 'CC', and 'C' are regarded as having significant speculative characteristics. That seems to be a fair description of the BACM 2005-1 Class B notes.

This piece was written by Marc Joffe, who consults for PF2. Marc is a Senior Policy Analyst at the Reason Foundation and a researcher in the credit assessment field. 

Monday, April 16, 2018

A VIXing Problem

2017 was a banner year for equity investors across the globe, including in the U.S., as the S&P 500 index gained over 19% (excluding dividends). Moreover, those gains were accompanied by extraordinarily low levels of volatility. 

We are nearly three months into 2018, and to say that things have not been as smooth would be an understatement, as volatility has returned with a vengeance to equity markets. The increase in volatility was especially acute on the day of February 5, 2018, when the CBOE Volatility Index (VIX®) moved more than it ever had in history, closing over 20 points higher from the previous day’s close (rising from 17.31 to 37.32), while equity markets plunged (the S&P 500 fell over 4%).

The 116% spike in the VIX reportedly triggered the implosion of a popular exchange-traded product called the VelocityShares Daily Inverse VIX Short Term ETN (ticker: XIV). XIV enabled investors to bet that low volatility would continue to rule the day. Through XIV, they would essentially be betting against short-term VIX futures. The VIX spike and XIV implosion has piqued the interest of investors, regulators, lawyers and journalists. Interested parties wanted to know the reasons for the volatility spike and whether there was any wrongdoing involved.

Before we consider the possibility of misconduct, we should take a step back to consider what exactly the VIX measures. The VIX is a market-based measure of expected, or implied, volatility. The Chicago Board of Exchange (CBOE) derives the VIX by backing out the volatility that is implied by market quotes for a portfolio of out-of-the-money put and call options on the S&P 500 index (SPX). With the aggregated option quotes we can gauge the volatility for the underlying SPX. 

Back in May 2017, well before all of the recent VIX excitement, researchers Griffin and Shams released a paper entitled, “Manipulation in the VIX?” The Griffin paper details how the VIX settlement process can be gamed by simply posting bids and offers (that may never lead to actual trade executions) to affect the VIX settlement. This possibility is reminiscent of spoofing conduct reported across various markets such as FX, Treasury futures, precious metals futures, and equity index futures, where some market participants are thought to have influenced market prices by posting bids and offers that they have no intention of actually executing.  Griffin concludes that price and volume data patterns are consistent with a trading strategy whose purpose is affecting the VIX settlement. The paper notes: 
“In sum, our findings show that the VIX settlement appears susceptible to manipulation, and that the aggregate evidence aligns itself with what one would expect to see in the case of market manipulation of certain settlements. However, we cannot fully rule out all potential explanations without more granular data.” 
One week after the wild VIX ride of February 5, 2018, a DC-based law firm, Zuckerman Law, released a letter to the SEC and CFTC on behalf of its anonymous whistleblower-client, alleging manipulation of the VIX. The next day, former CFTC Commissioner Bart Chilton noted during a CNBC television interview that, “The VIX has been suspect for at least seven years.” 

And on March 9 of this year, the law firm Cohen Milstein filed a purported class action complaint on behalf of Atlantic Trading USA, LLC against unknown John Does, alleging manipulation of the settlement price for VIX futures and options. The complaint alleges that defendants “caused the monthly final settlement price of expiring VIX contracts to be artificial. They did so by placing manipulative SPX options orders that were intended to cause, and at minimum recklessly caused, artificial VIX contract settlement prices in the expiring contracts.” 

With the VIX being prone to manipulation, and billions of dollars’ worth of derivatives and exchange traded products tied to it, we’d like to see if there might be a better way of calculating expected volatility – a method that is not as prone to the vagaries of potential wrong-doers. 

PF2 consultant Joe Pimbley and PF2's Gene Phillips have done just this in a paper called Fix the VIX, which can be accessed here, courtesy of the Global Association of Risk Professionals (GARP). 

Our investigation finds that three approximation techniques being implemented by the CBOE in “calibrating” the VIX may not be necessary, and may exacerbate errors or increase the VIX’s susceptibility to manipulation or error. Of course, volatility won’t go away. But the VIX will more accurately capture it and describe it, with a couple of … fixes.

Friday, April 6, 2018

Tesla Bonds – Revved Up by Moody’s?

There’s one thing about equity analysts talking up Tesla and getting behind the hype: equity investors enjoy the upside if their optimistic scenarios come true. 

But bonds have only downside, and rating agencies are supposed to analyze various scenarios – the good the bad and the ugly – in coming up with bond ratings.

It doesn’t look like Moody’s did that when rating Tesla B2 and Tesla’s $1.8 billion bond issue B3 in August 2017.   Rather, they assumed as true Tesla’s optimistic production targets (or hopes) for Tesla’s Model 3 and rated Tesla based on those coming true.  To exaggerate how bizarre this approach is, had Tesla said they hope and expect to produce a million cars a day, perhaps Moody’s would have rated them Aaa!  Click, whirr

Crucially, Moody’s provided Tesla, the company and its bonds, ratings based on a picture of its future financial that exceeded its true financial position, before Tesla had met the goals that would warrant the rating.

Moody’s rating rationale reads as follows (with our emphasis added): 

"The B2 CFR reflects Moody's expectation that the launch, production ramp up, and market acceptance of the Model 3 will be successful enough to achieve approximately 300,000 unit sales during 2018 (a full-year sales rate averaging about 5,500 per week) with a gross margin approximating 25%. This level of sales and profitability would enable Tesla to strengthen its performance from sizable losses to an operating position that supports the B2 CFR. The B2 rating is further supported by Moody's expectation than in the event of severe financial or operating stress, Tesla's brand name, production facilities, and product lineup would have considerable value to another automotive OEM or technology firm targeting the electric vehicle and mobility markets."
To use Moody's language, in short, the achievement of the goals "would enable" Tesla to achieve the rating being provided now!  Moody's has assumed a future, rosier picture of the company, and based its current rating on the achievement of this rosy future, rather than waiting for the financial position to warrant the rating provided.  

Yet in its own press release, Moody’s rating analyst Bruce Clark (Senior VP) notes that "The major challenge facing the company during the next twelve months will largely be the considerable execution risks associated with the rapid ramp up in production of a totally new vehicle." 

So, why not see if Tesla can execute before rating Tesla B2, if the B2 rating is contingent on execution at a level far beyond what Tesla has ever yet achieved?   The answer, unfortunately, is that had they looked at Tesla’s actual then-current balance sheet, they would never have rated them B2, but probably in the Caa range.   And Tesla might have gone elsewhere for its second rating (it landed up getting a B- rating from S&P) or scratched the idea of issuing this bond. In fact, Moody’s essentially acknowledges this pressure, which to us seems to be a potential conflict of interest: “Without the proceeds from the [proposed] note offering, Tesla's liquidity position would be stressed.”

Moody’s didn’t exactly mark Tesla to market did it? Moody’s marked them to an optimistic future. 

It does make one wonder where the Moody’s opinion lies if Moody's is simply going to take Tesla's management’s assumption as a given.  A good job, if you can get it, but hardly an insightful opinion. 

As it happened, towards the end of March Moody’s noticed that Tesla was still far away from achieving its optimistic goals, having suffered some production hurdles and delays not atypical for a young company producing a new vehicle.   

Moody’s downgraded the so-called “long-term” ratings in March 2018, a little more than half a year after the bonds were issued.  The long-term ratings were ultimately based on very short-term expectations.  Click, whirr. 
"[Moody’s downgrade of] Tesla's ratings reflect the significant shortfall in the production rate of the company's Model 3 electric vehicle. The company also faces liquidity pressures due to its large negative free cash flow and the pending maturities of convertible bonds ($230 million in November 2018 and $920 million in March 2019). Tesla produced only 2,425 Model 3s during the fourth quarter of 2017; it is currently targeting a weekly production rate of 2,500 by the end of March, and 5,000 per week by the end of June. This compares with the company's year-earlier production expectations of 5,000 per week by the end of 2017 and 10,000 by the end of 2018." 
Oddly, now Moody’s is no longer hinging its rating to Tesla's current expectations: “The rating could be raised if production rates of the Model 3 meet Tesla's current expectations and if the company maintains good liquidity.” 

Bondholders ought to be frustrated. They have bought into a B2 corporate family rating of a company which clearly wasn’t yet B2 at the time of the issuance. They may well have taken Caa-like risk, but only been compensated for the taking of single B risk. 

Tesla has already been downgraded, and it is arguable whether the new B3 rating is well-founded, too. The bonds have been downgraded from B3 to Caa1 and have lost roughly 3% in value on the day of the downgrade. Altogether, the bonds are down roughly 10% in price terms since issuance. 

PF2 would like to thank Joe Pimbley for his contribution to this article.

Monday, February 26, 2018

Florida Shootings Require Cultural & Mindset Changes

Our failure to prevent the Florida school shooting illustrates a pervasive problem in modern societies: we often have access to ample warning signs but all-too-frequently fail to leverage this information to avoid disaster. The issue not only impacts law enforcement agencies, but our financial institutions as well. To more effectively handle all the intelligence available to them, organizations will require major structural and cultural change. 

The FBI and local law enforcement reportedly had more than enough information to legally disarm and detain confessed school shooter Nikolas Cruz before he killed 17 people at Parkland High on February 14. This is not the first such intelligence failure, and won’t be the last. Consider these examples:
  • 9/11 could have been prevented had the CIA and FBI done a better job of sharing and handling intelligence.
  • Russian intelligence warned the FBI about Tamerlan Tsarnaev long before he carried out the Boston Marathon bombing.
  • In France, authorities failed to act on multiple clues that would have enabled them to prevent the Paris bombings that claimed 130 lives in November 2015. 

In the financial industry, rating agencies and bank risk management teams failed to act in their or their clients’ best interests when they continued to create and sell residential mortgage-backed securities, despite the deterioration in mortgage lending standards, and the increasing and disturbing amount of mortgage fraud being reported by the FBI in its annual mortgage fraud reports.

A well-operating risk-management function, with a voice, would most likely have limited the potential for the cultural failures seen at the Royal Bank of Scotland, as detailed in the recently published report commissioned by the Financial Conduct Authority. The extraordinary activities of the gung-ho Global Restructuring Group at RBS in London could immediately have been stymied, as they posed reputational and business risks far outweighing the group's short-term revenue-generating interests. As the report explains: 
“GRG enjoyed an unusual independence of action for a customer-facing unit of a major bank. It saw the delivery of its own narrow commercial objectives as paramount: objectives that focused on the income GRG could generate from the charges it levied on distressed customers. In pursuing these objectives, GRG failed to take adequate account of the interests of the customers it handled and, indeed, of its own stated objective to support the turnaround of potentially viable customers.” 
These assorted failures suggest that we have a systemic problem with risk monitoring, or a failure to incorporate it appropriately within institutions. And because the problem is systemic it won’t be solved by firing a few bad apples. Instead, we need to understand and address the root cause. 

One feasible argument is that jobs involving risk monitoring and mitigation generally come with a relatively low social status and thus do not necessarily attract the most motivated applicants.

This phenomenon is epitomized by our (often unfair) stereotype of security guards: that they are ineffective and prone to sleeping on the job. Because security jobs are low paying, they don’t often attract type “A” individuals. The job itself is quite boring: most of the time nothing happens. While a more proactive security guard could find and act upon many clues during the course of his or her day, almost all of the extra effort will be for naught. At least 99 times out of 100 that suspicious backpack won’t contain an explosive device. 

Although bank risk managers and FBI call handlers undoubtedly have higher social status than security guards, they are most likely to be subordinated within their organizations. At a bank, monitoring credit risk is much less glamorous and lucrative than acquiring or merging companies, underwriting deals or trading securities. And, as with the case of the seemingly suspicious backpack, most clues won’t lead anywhere anyway: for every legitimate call law enforcement departments receive there are many that lead nowhere; a missed charge card payment, similarly, often doesn’t presage a mortgage foreclosure. 

Ideally, we should elevate the status of risk monitoring jobs and make them more exciting. More attention from senior management may help. Although most money-center banks took massive losses during the financial crisis, Goldman Sachs came out relatively unscathed. A major reason is that the bank’s Chief Financial Officer reviewed daily risk management reports and held a meeting in his office to call for immediate action once its was detected that mortgage backed securities had begun to underperform in 2006. Goldman is also an exceptional case in that it rotated fast-track talent between moneymaking and risk management roles, and it empowered risk management staff to veto certain trading activities. 

Although more high-level attention might help those charged with receiving and sifting through raw intelligence, the job is still a tedious one – akin to looking for a needle in a haystack. 

Conviction Dilemma 

In addition to the possibility that risk monitoring personnel are as a group less motivated, risk personnel tend to be a more introverted type than their front-desk colleagues. This may manifest in their being apprehensive when expressing themselves to their comparatively more aggressive colleagues, and potentially come off as being indecisive or speculative.  Leaders often like a strong, definitive opinion: “hedge this risk!” and may shun or ignore a more complex opinion coming from a more cautious analyst. 

In short, the personalities hired into risk-management roles often suffer from what we will term the “conviction dilemma,” which emanates from the work of Philip Tetlock and others, who studied predictive expertise. Tetlock's research findings informed his commentary that those whose expertise was valued and sought out, for example pundits on TV shows like The McLaughlin Group, were those who had vocal, unequivocal opinions, that could be articulated with utter conviction – but were often wrong. 

Altogether, even strong and motivated risk experts may be introverted and may be indecisive when expressing themselves. Playing a function that is considered as subordinated in management’s eye, they might struggle to make convincing and resolute “do this!” arguments, and management might therefore be less likely to take them seriously, and act on them expeditiously. 

Applying Technology 

In the 21st century, we have learned to assign boring or laborious jobs to computers. We can identify potential attackers earlier by entering all the clues law enforcement receives into shared databases, and we have state-of-the-art data science tools built for analyzing this mass of information. This approach need not violate privacy: social media posts, calls from tipsters and prior arrests are all legitimately available to law enforcement today. 

Palantir is among the most prominent of companies offering software that enables intelligence agencies to find needles in the haystacks of raw data they receive. Unfortunately, Palantir is not an inexpensive solution, and may thus be beyond the budgets of smaller law enforcement agencies. 

Governments and NGOs may wish to invest in the development of free, open source data analysis. Aleph is an open source tool that can analyze large volumes of unstructured data. Although designed for investigative journalists, it could be customized for use by law enforcement of for counterparty tracking. Whether they use licensed or open-source solutions, law enforcement and intelligence agencies should establish and apply technical standards for data sharing. Because financial firms are overtly competitive, data sharing of financial intelligence may be less appropriate between competing firms, but may be more prevalent within institutions. 

Often the information needed to prevent mass killings is hiding in plain sight. By improving organizational structures and leveraging technology, financial firms and law enforcement agencies can harvest more actionable data from legally available information. Armed with this data, they can prevent certain future acts of carnage. While no single policy solution – VaR levels or gun control included – can ever guarantee endless success, we need to be thoughtful and dynamic in going about limiting the frequency or even the magnitude of these catastrophes, and we would do well to use our tools effectively in pursuing the goal not only of making money, winning clients and awards, but also of limiting the downside.


This piece was co-written by Marc Joffe, who consults for PF2, and members of PF2’s staff. For more on this topic, visit our 2016 piece on the detrimental impact of short-term thinking patterns on conduct with financial firms. Among other things, we recommend a re-thinking of the design of incentive structures: “The approach we put forward here is the studious linking of profit-sharing to successful and honest risk-taking and business practices.”

Marc Joffe is a Senior Policy Analyst at the Reason Foundation and a researcher in the credit assessment field. He previously worked as a Senior Director at Moody’s Analytics. 

Tuesday, February 20, 2018

Apple or the United States: Which is the Stronger Credit?

In blogs like this, we sometimes like to begin discussions. 

Today, we're going to present an interesting situation. Moody's holds that the US is a Aaa credit, while Apple Inc. is rated Aa1 (both its debt and its corporate family ratings are Aa1). 

Aside from some recent hiccups, Apple is a growing company (see the chart). Its debt level sits at 44% of its annual revenues, and the expense of its debt is somewhat moderate, at 1.2% of its annual revenues. The US is growing its revenues too, albeit at a modest clip. The expense of its debt, however, is approaching 8% of annual revenues. 

Recent developments make us wonder whether the US is truly the stronger credit. This blog takes you through some of the bigger picture here, including elements of the credit rating processes that might be supportive of the rating agency's stance, despite the numbers perhaps telling a different story. We don't take a position here, as much as pose the question.  And we look forward to your thoughts.

Comparing Apples & Oranges? 

Recent fiscal developments have us thinking again about the US's credit rating.

February 9th’s budget and spending caps agreement tipped next year’s budget deficit over $1 trillion, according to the Committee for a Responsible Federal Budget, despite the fact that we are almost nine years into the current economic expansion and with the official (U3) unemployment rate at just 4.1%.  That bipartisan debt binge, in the wake of last year’s $1.5 trillion tax cut (over ten years), has some prudent budget-watchers scratching their heads. 

Last month, Reuters reported that “China’s Dagong Global Credit Rating Co, one of the country’s most prominent ratings firms, on Tuesday cut the local and foreign currency sovereign ratings of the United States, citing an increasing reliance on debt in the world’s largest economy.” Dagong cut the US’s sovereign rating to BBB+ (from A-), with a negative outlook. 
“Deficiencies in the current U.S. political ecology make it difficult for the efficient administration of the federal government, so the national economic development derails from the right track,” Dagong said. “Massive tax cuts directly reduce the federal government’s sources of debt repayment, therefore further weakens the base of government’s debt repayment.”
Looking at the numbers as of fiscal year-end 2016 (the most recent year with complete figures), Apple’s debt as a percentage of revenue and its interest expense as a percentage of revenue are much lower than the US’s corresponding levels.

Strictly based on the debt/revenue and interest expense/revenue ratios, an entity with US’s metrics would likely garner a below-investment-grade rating: according to Moody’s' ratings methodology for diversified technology companies, a negative EBIT/Interest Expense equates to a rating of Ca (the lowest rating on the subfactor scale), all else equal.

US’s Primary Balance/Interest Expense, our proxy for US’s EBIT/Interest Expense, is negative. We don’t have a proxy for the US’s EBITDA (earnings before interest, taxes, depreciation, and amortization), but let’s take our proxy for the US's EBIT and magically add back $2 trillion(!) worth of imaginary “depreciation and amortization.” Doing so gets us to an “EBITDA” of $1.7 trillion. In this hypothetical, its Debt/EBITDA is still over 10x, which would correspond to a rating of Ca (all else equal) for a diversified technology company.

Even in the rosiest possible proxy for EBITDA, in which we suppose USA has zero expenditures(!), and thus its EBITDA would simply equal its revenue, the hypothetical Debt/EBITDA would be 3x, which maps to a rating of Ba, all else equal, for a diversified technology company.

Viewing the US’s long-term fiscal situation with a more critical lens, we might consider the future fiscal impact of Social Security and Medicare – which are underfunded by $46.7 trillion over the next 75 years (per the Fiscal Year 2016 Financial Report of the U.S. Government jointly produced by the US Treasury and the Office of Management and Budget of the Executive Office of the President). 

Yet, Moody’s and many of the other rating agencies don't seem overly concerned, rating the US Aaa (stable outlook).

Into the finer details, now, there are several quantitative factors that might be equally or more relevant to these entities’ respective ratings (e.g. EBITDA margin or FCF/Debt for Apple; Interest Expense/GDP or Inflation for the US), but we tried to highlight metrics in this chart which are most comparable between the two debt issuers. 

Keep in mind that there are also key qualitative characteristics that Moody’s considers when rating corporates and sovereigns. For example, Moody’s gives significant weight to considerations such as management financial policy (“management and board tolerance for financial risk”) for diversified technology corporations, and institutional strength (specifically, “government effectiveness”, “rule of law”, and “control of corruption”) for sovereigns.  

  • The United States government has the legal authority to compel its “customers” (i.e. taxpayers) to pay more, if necessary (although political willingness might be another matter), while a corporation has no such relationship with its customers.
  • Apple lacks a printing press, whereas the USA can print its own currency. 
  • The US has a central bank that has the ability to monetize federal debt, whereby the Federal Reserve can essentially finance fiscal deficits by purchasing as much government debt as the federal government issues, although technically the Fed maintains its independence from political interference. 
  • (Moody’s does consider inflation as an important ingredient in its sovereign ratings, so the rating itself might suffer in a scenario in which the country printed its way of a fiscal hole, but there technically would still be no default.)
Altogether, Apple’s revenue is growing faster than the US’s (99% vs. 31%, over 5 years), it has much less debt than the US (relative to revenue: 44% vs. 304%), and its interest expense is much lower (relative to revenue: 1.2% vs. 7.7%).

So the question is, are the US’s other strengths or advantages -- its ability to raise more revenue through taxes if truly needed, and its unique ability to service its debts by either printing more of the world’s reserve currency or having its central bank purchase its debt -- enough to justify a Aaa rating?  In deciding this, one must consider not only the US's current trillion dollar deficits at a point in the economic cycle that does not beg for drastic fiscal stimulus, but also the fact that we have a Social Security and Medicare funding gap of nearly $47 trillion, in present value terms, that is only growing worse. (At this juncture, it seems that our elected officials have chosen to do nothing about the funding gap.)

Remember, the question is not whether or not the US will default, or how it may grow itself out of the current situation -- but whether the likelihood of there being an issue here is so remote as to warrant a Aaa rating, a rating higher than that of Apple's debt, which is currently well-supported.

It is worth considering Stein’s Law as we end this blog: "If something cannot go on forever, it will stop" or "Trends that can't continue, won't."  Interestingly, Stein meant this in the sense that there is no need for action or a program to make it stop, much less to make it stop immediately; it will stop of its own accord.

Lisa Benson Editorial Cartoon used with the permission of Lisa Benson, the
Washington Post Writers Group and the Cartoonist Group.  All rights reserved.

Thursday, February 8, 2018

Market Trading Investigations - Layering, Spoofing, Front-Running, Stop-Loss Triggering

Adding to our prior compilations of valuation issues, questions of fairness in market executions, and fee disclosure concerns, we are adding a new set that looks at investigations into, and allegations of layering, spoofing, front-running, barrier-running and stop-loss triggering in the financial markets.
  1. Jan 2018: Spoofing - Precious Metals; S&P Futures. CFTC Files Eight Anti-Spoofing Enforcement Actions against Three Banks (Deutsche Bank, HSBC & UBS) & Six Individuals 
  2. Jan 2018: Front-running - FX. HSBC front-running victim (Prudential Plc) 
  3. Jan 2018: Front-running - FX. Barclays front-running of HPQ in $8.3bn FX options trade 
  4. Oct 2017: Front-running - FX. U.S. jury finds ex-HSBC executive guilty of fraud in $3.5 billion currency trade 
  5. Aug 2017: Spoofing - Treasury Futures; Eurodollar Futures. CFTC Finds that The Bank of Tokyo-Mitsubishi UFJ, Ltd. Engaged in Spoofing of Treasury Futures and Eurodollar Futures 
  6. Jul 2017: Spoofing - Commodity Futures (multiple). CFTC Orders New York Trader Simon Posen to Pay a $635,000 Civil Monetary Penalty and Permanently Bans Him from Trading in CFTC-Regulated Markets for Spoofing in the Gold, Silver, Copper, and Crude Oil Futures Markets 
  7. Jun 2017: Spoofing - Precious Metals. CFTC Finds Former Trader David Liew Engaged in Spoofing and Manipulation of the Gold and Silver Futures Markets and Permanently Bans Him from Trading and Other Activities in CFTC-Regulated Markets 
  8. Mar 2017: Spoofing - Equities (cash and contracts for difference). Ex-DBS Trader Convicted in Singapore's First Spoofing Case 
  9. Jan 2017: Spoofing - Treasury Futures. CFTC Orders Citigroup Global Markets Inc. to Pay $25 Million for Spoofing in U.S. Treasury Futures Markets and for Related Supervision Failures 
  10. Dec 2016: Stop-Loss Triggering - FX. Australian Securities and Investments Commission (ASIC) accepts Enforceable Undertaking from Commonwealth Bank of Australia (CBA) for triggering FX customer stop-loss orders 
  11. Dec 2016: Spoofing - Futures (equity index; crude oil; natural gas; cooper; VIX). Federal Court Orders Chicago Trader Igor B. Oystacher and 3Red Trading LLC to Pay $2.5 Million Penalty for Spoofing and Employment of a Manipulative and Deceptive Device, while Trading Futures Contracts on Multiple Futures Exchanges 
  12. Nov 2016: Spoofing - Equity Index Futures. Federal Court in Chicago Orders U.K. Resident Navinder Singh Sarao to Pay More than $38 Million in Monetary Sanctions for Price Manipulation and Spoofing 
  13. Jan 2015: Layering & Spoofing - Equities. Canadian Man Charged in First Federal Securities Fraud Prosecution Involving 'Layering' 
  14. Jul 2013: Spoofing - Commodity Future (multiple). CFTC Orders Panther Energy Trading LLC and its Principal Michael J. Coscia to Pay $2.8 Million and Bans Them from Trading for One Year, for Spoofing in Numerous Commodity Futures Contracts 
  15. Dec 2012: Spoofing - Wheat Futures. CFTC Files Complaint in Federal Court against Eric Moncada, BES Capital LLC, and Serdika LLC Alleging Attempted Manipulation of Wheat Futures Contract Prices, Fictitious Sales, and Non-Competitive Transactions 

Monday, January 8, 2018

Blog 1 of 2018 / Markets for Consumer Data & User-Beware

Hello readers, and thanks for joining us for our first blog of the new year.

This year, in addition to our typical musings on financial markets, we'll be writing a little more on consumer data.

Hacks have been all the rage for a while already (Equifax and Yahoo! more recently; Target in 2013, Sony Playstation in 2011 and TJ Maxx in 2007 are not too-distant memories).

But our interest lies somewhere else: in what is happening behind the scenes with our data.

hiQ v LinkedIn

hiQ is a San Fran-based startup which was doing something pretty interesting: it was scraping data from LinkedIn and then selling that data or analyses done using that data.

LinkedIn didn't much like this, with hiQ's automated robots ("bots") bypassing LinkedIn's security measures to scrape the data, which LinkedIn felt undermined LinkedIn’s privacy commitments to its members.

They battled it out in court, with the court finding in August, probably reasonably, that LinkedIn could not stop hiQ from scraping the publicly-viewable information from LinkedIn's website.  In fairness, LinkedIn doesn't own its users' data -- it's our data! -- and therefore couldn't limit hiQ's ability to access or study it.

The ruling is interesting for several reasons, including some of the First Amendment-type arguments made by hiQ to support its right to scrape.  
“To choke off speech and the precursor of speech, the gathering of facts and the analysis of information, is a dangerous path down which we should not go,” 
         -- Harvard law professor Laurence Tribe, representing hiQ, reportedly told                 the judge.    
“hiQ believes that public data must remain public, and innovation on the internet should not be stifled by legal bullying or the anti-competitive hoarding of public data by a small group of powerful companies, ... It is important to understand that hiQ doesn’t analyze private sections of LinkedIn – we only review public profile information. We don’t republish or sell the data we collect. We only use it as the basis for the valuable analysis we provide to employers. ”
            -- hiQ said in a statement 

Okay, meh. But how about what happens next?  Like Microsoft (which owns LinkedIn) firms like Facebook, Amazon, eBay and Google, control and study copious amounts of customer data (and they sometimes get hacked and lose control of it).  But importantly they also sell it (as does hiQ).  We might like to think that they only sell aggregate data, but how would we know? 

Personal, Personnel Information

What interests us is that hiQ sells information about LinkedIn users to those LinkedIn users' bosses, including information, generated from scraping LinkedIn, about the likelihood of an employee leaving. hiQ's clients reportedly include companies like CapitalOne and GoDaddy, and hiQ's products include their Keeper product, which identifies, for employers, when their employees are at risk of leaving for another job. (For example, when employees are "looking around," they tend to make connections on LinkedIn.)

So that's a sale not necessarily of the more obvious, vanilla, personal information (name, address, date-of-birth), but user/employee/personnel's tendencies and movements. But it's almost certainly not aggregated: if it were aggregated, it would be worthless.  Sure, they're not selling the individual's vanilla data itself but they've done a basic analysis of individual's behavior and are selling the analysis.  Aggregated, it is not. 

hiQ would have no greater ownership interest in our data than LinkedIn would have.  Through hiQ's bots, we just have a simple work-around (imagine, for example, that LinkedIn were simply to buy a stake in hiQ).  If we knew that LinkedIn could "tell on us" to our employers -- and make money doing so -- would we have signed up?

We have the Latin expressions caveat emptor and caveat venditor to connote the short-hand principles of buyer-beware and seller-beware, when entering into transactions.  In 2018, the awkward expression caveat utilitor -- user-beware -- might just become part of our lexicon.

The value of Johnny's house, or Sandy's choice of handbags, is information that would help advertisers target Johnny or Sandy more appropriately.  If Johnny's house price is on the low end, all else equal one wouldn't push Maserati ads at him.  If Sandy is buying Louis Vuitton bags, well, maybe she would like this newly-released Prada bag or another Louis Vuitton bag.  But whose data is that, and do companies have a right to sell and profit from that information?  And how do we separate data, the sale of which may be limited on an individual basis, from analysis of data, which seems to be fair game.  Are these two both analyses?

  • Johnny's house was purchased for $200K this year
  • Last year, Sandy bought two of Brand X's bags and three of Brand Y's bags.

All the best for 2018.  Keep watching the Watchmen.

~ PF2