Blogs

Posted: March 12, 2019

SOS to GSEs: Your Bonds are A Beautiful Mess

Annnd we’re back. Have you missed the scintillating stories of mischief and manipulation spun by your favourite lawyers-turned-bloggers? If the answer to that isn’t a resounding yes, I’ll just assume you’re new around here.

You know what else is new to these parts? Government-Sponsored Enterprises! Yes friends, this is the start of a brand-new miniseries here on the Manipulation Monitor. In this series, I’ll be tracking the ins and outs of several new litigations against the “usual customers” (read: the big banks), this time in their roles as horizontal competitors and dominant dealers of bonds issued by Government-Sponsored Enterprises, or GSEs. Complaints to date include the following:

  • City of Birmingham Retirement and Relief System, Electrical Workers Pension Fund Local 103, I.B.E.W., and Local 103 I.B.E.W. Health Benefits Plan, individually and on behalf all others similarly situated v. Bank of America N.A., Barclays Bank PLC, Barclays Capital Inc., BNP Paribas Securities Corp., Citigroup Global Markets Inc., Credit Suisse AG, Credit Suisse Securities (USA) LLC, Deutsche Bank AG, Deutsche Bank Securities, Inc., First Tennessee Bank, N.A., FTN Financial Securities Corp., Goldman Sachs & Co. LLC, J.P. Morgan Chase Bank, N.A., Merrill Lynch, Pierce, Fenner & Smith Inc., and UBS Securities LLC, 1:19-cv-01704;
  • Alaska Electrical Pension Fund, on behalf of itself and all others similarly situated v. Bank of America N.A., Merrill Lynch, Pierce, Fenner & Smith Inc., Barclays Capital Inc., BNP Paribas Securities Corp., Citigroup Global Markets Inc., Deutsche Bank Securities, Inc. HSBC Securities (USA) Inc., HSBC Bank PLC, J.P. Morgan Chase Bank, N.A., Nomura Securities International, Inc., TD Securities IUSA) LLC, and Wells Fargo Securities LLC, 1:19-cv-01796-UA; and
  • Lincolnshire Police Pension Fund, Individually and on Behalf of All Others Similarly Situated v. Bank of America N.A., Barclays Bank PLC, Barclays Capital Inc., BNP Paribas Securities Corp., Citigroup Global Markets Inc., Credit Suisse AG, Credit Suisse Securities (USA) LLC, Deutsche Bank AG, Deutsche Bank Securities, Inc., First Tennessee Bank, N.A., FTN Financial Securities Corp., Goldman Sachs & Co. LLC, J.P. Morgan Chase Bank, N.A., Merrill Lynch, Pierce, Fenner & Smith Inc., and UBS Securities LLC, 1:19-cv-02045.

The World as I See It: An Overview of the GSE Bond Market

These newly-filed class action complaints, which have been accepted as related by Judge Jed S. Rakoff of the Southern District, all allege violations of the Sherman Act arising from a conspiracy by the defendants to fix prices and restrain competition in the market for unsecured bonds issued by various GSEs, including the Federal National Mortgage Association (“Fannie Mae,” or “Fannie”), the Federal Home Loan Mortgage Corporation (“Freddie Mac,” or “Freddie”), the Federal Farm Credit Banks Funding Corporation (“FFCB”), and the Federal Home Loan Banks (“FHLB”). The City of Birmingham and Lincolnshire Police cases focus specifically on Fannie and Freddie bonds (“FFBs”), while Alaska Electrical takes a more general approach, attacking the markets for all unsecured bonds issued by Fannie Mae, Freddie Mac, FFCB, and FHLB, which I’ll refer to more generally as “GSE bonds.”

While GSE bonds are not backed by the “full faith and credit of the United States,” these bonds do benefit from their association with the government, and are generally regarded as secure investments by investors (including our various plaintiffs). These bonds are sold on a secondary market, with primary, pre-approved dealers like the Defendants serving as intermediaries. The dealers purchase bonds from the GSEs at issuance by way of auction or syndication and re-sell them on the secondary market to investors, making their profit on the mark-up.

The secondary market for GSE bonds is an “over-the-counter” (“OTC”) market, which means that investors are required to communicate directly with a salesperson or employee of the dealer, either by phone or digitally, in order to receive a price quote. The result is a highly opaque market: while investors may communicate with several different dealers seeking price quotes, they cannot see GSE bond prices in real time, and thus cannot evaluate prices quoted by multiple dealers without a substantial delay. The opacity is exacerbated by the fact than most dealers place short expiration times on the quotes they provide, making it impracticable to “shop around” to more than a small handful of dealers at a time.

On June 1, 2018, Bloomberg reported that four confidential sources revealed that the DOJ Antitrust Division is conducting a criminal investigation into collusion among dealers to fix FFB prices. Specifically, the investigation focuses on illegal activity of bank traders suspected of coordinating their actions to benefit the institutions they work for. All three complaints mark the reveal of this investigation as the first time plaintiffs learned of their anticompetitive conduct, arguing that, because of Defendants’ fraudulent concealment, plaintiffs and the classes they represent were not aware of the misconduct and could not have discovered it through an exercise of reasonable due diligence. Accordingly, all Plaintiffs assert that the applicable statute of limitations on their claims were tolled until that date.

Frank D. Fixer: Plaintiff’s Proof of Foul Finagling

The two FFB-specific complaints, City of Birmingham and Lincolnshire Police, rely on much the same evidence to back up their claims, while Alaska Electrical takes a slightly different approach in its analysis across various GSE markets. Let’s look at each in turn, shall we?

Fixing the FFBs – The Details in the Fabric

The City of Birmingham and Lincolnshire Police complaints identify three key ways in which they allege that prices and other economic data confirm the existence and impact of Defendants’ conspiracy. The economic facts that they lay out in their respective complaints are the result of an analysis conducted by the Plaintiffs, and identify anomalies in FFB pricing that are inconsistent with normal, competitive market conditions. Plaintiffs further claim that these economic facts are statistically significant with a confidence degree of 95% — that is to say, the results are at least 95% likely to have been caused by factors other than chance or co-incidence. Plaintiffs’ analysis covered the period between March 1, 2010 through December 31, 2017, and, as will be described below, identified sharp changes in the market following an April 27, 2014 report by the Financial Times that DOJ criminal prosecutors were travelling to London to question FX traders—thus confirming that criminal prosecutions would not be limited to LIBOR.

1. Fixing prices of newly-issued FFBs in the week following each new FFB issuance.

Freddie and Fannie issue new FFBs at set intervals; the most recently issued FFBs are known as “on-the-run” FFBs, while all other, older FFBs with similar characteristics are known as “off-the-run” FFBs. In a competitive market, it would be expected that the difference between what Defendants paid to Fannie and Freddie to purchase the FFBs and the price at which Defendants sell those FFBs to investors will particularly small immediately after new FFBs are issued, and particularly on the first day, as the impact of new information (such as changes in interest rates, Fannie/Freddie creditworthiness, or liquidity) is nearly non-existent. Indeed, Plaintiffs’ review of pricing data shows that, after the class period, the difference in price for FFBs sold on “offer day” (day of issue) was very small, averaging 0.4 cents. Before April 27, 2014, however, the difference was 3.2 cents – nearly eight times higher.

Similarly, Plaintiffs compared the price differential for the week after offer days; this review showed that the increase in price averaged only 0.73 basis points over the baseline after April 27, 2014; during the class period, however, it was nearly four times higher, with an average increase of 3.01 basis points over the baseline. Plaintiffs also compared the prices defendants charged for FFBs in the class period to those charged for U.S. Treasury securities with comparable maturities; these securities carry a similar amount of credit risk as FFBs, and are affected by the same market conditions and macroeconomic factors, but Defendants prices for FFBs on offer days was highly abnormal as compared to Treasury securities. Specifically, Plaintiffs found that the prices charged by Defendants for FFBs over the yield offered by U.S. Treasury securities was nearly double that to non-Defendant FFB dealers. Further, after April 2014, the difference between Defendants and non-Defendant dealers was negligible.

2. Fixing prices for on-the-run FFBs artificially higher in the period leading up to a new FFB issuance

As previously discussed, newly issued FFBs of a given type generally have similar characteristics of existing FFBs, except that they mature later. Accordingly, the prices of newly-issued FFBs are closely correlated with the prices of those previously-issued, and, in a competitive market, one would expect to see lower prices for FFBs that were about to go “off the run” than for newly-issued FFBs. This is because the market for on-the-run FFBs is generally more liquid than the market for off-the-run FFBs, with investors preferring to purchase more liquid FFBs so they have a higher likelihood of funding an investor at market price should they decide to sell. Accordingly, in the days leading up to a new issuance one would expect the prices for FFBS about to go off the run to drop as investors prefer to purchase from the new issuance. During the class period, however, Plaintiffs’ analysis showed that notes about to go off-the-run experienced a statistically significant price increase in the days leading up to the new issuance. By way of further comparison, and to isolate against any potential macroeconomic factors affecting pricing, Plaintiffs compared prices in transactions between Defendants and investors against transactions between Defendants for the same instruments. This comparison showed that the price inflation for notes about to go off-the-run only occurred in transactions between Defendants and investors. And, as with the pricing of newly-issued FFBs, price inflation for notes about to go off-the-run dissipated after April 27, 2014.

3. Quoting agreed-upon, artificially inflated bid-ask spreads to investors throughout the class period

The allegations in this section concern the profits that Defendants earned on all FFB transactions through the class period. In a competitive market – and I am aware I am over-using this phrase, but it’s a handy one – dealers would compete with each other by offering narrower bid-ask spreads to their customers. When a dealer charge wider spreads, either by lowering the bid price or by raising the ask price, that dealer should, in theory, lose customers to rivals offering tighter spreads. Offering tighter spreads would make economic sense for dealers, despite the lower profits, because it would give them a better chance to earn the customer’s business. Conspiracy to fix bid-ask spreads could be accomplished by agreeing to offer a particular quote or to agree to ask a minimum bid spread; in either case, the dealers are better off because they can guarantee a consistently higher profit margin without the fear of losing their customers to competition from rivals. Defendants were able to accomplish this, Plaintiffs posit, because they jointly underwrote more than 64% of all FFB underwriting during the class period.

In support of this theory of artificially widened spreads, Plaintiffs analyzed bid-ask spreads quoted by dealers in the FFB market during and after the class period, with the (again, statistically significant) result that bid-ask spreads were more than 1.85 basis points wider during the class period as compared to post-class. Moreover, this discrepancy held true despite market conditions. For example, a more liquid market would generally be expected to result in decreased spreads; in the FFB market prior to April 27, 2014, however, increased liquidity actually correlated with wider spreads.

To further confirm their theory, Plaintiffs also analyzed “riskless principal” transactions, where a dealer purchases an FFB after it has already agreed to sell the FFB to another customer (or vice-versa), and therefore never bears any liquidity risk or risk from other changes in market conditions. In these unique transactions, the average bid-ask spread prior to April 27, 2014 was 13.4 basis points; in the post-class period, this dropped to 11.9 basis points. These numbers further increased when Plaintiffs reviewed transaction data with known dealers, showing that the Defendant dealers had bid-ask spreads a full 43% wider than non-Defendant dealers for the relevant period.

Geek in the Pink on the Gulling of the GSE Bonds

In contrast to the FFB cases, the Alaska Electrical complaint posits a much larger class period, running from January 1, 2012 through June 1, 2018, and their analysis similarly covers a broader period.

One feature of the GSE bond market addressed in more detail in the Alaska Electrical complaint is the failure of Defendants to separate their syndication and trading markets, with the result that the same desk, and largely the same people, were responsible for both underwriting new GSE bond issuances and for trading in the secondary GSE market. That is to say, the same people who were in constant communication with other dealers for the purposes of syndication were the very same people responsible for trading the bonds later on – a situation ripe for collusion, and, I assume, a nightmare for any legal compliance practitioner. This overlap, Plaintiffs posit, resulted in the “fundamental agreement” that Defendants would not complete against each other in the market for GSE bonds, but instead cooperate to maximize their own profits at the expense of their customers, many of whom they shared in common. This was carried out in person, through electronic chatrooms such as Instant Bloomberg, instant messaging, and telephone.

To evaluate the results of the alleged conspiracy, Plaintiffs made use of “screens:” statistical tools based on economic models that make use of data such as prices, bids, quotes, spreads, market shares, and volumes to identify the existence, causes, and scope of conspiratorial behavior. Where the FFB plaintiffs, Plaintiffs largely relied on differences during and after the class period to identify evidence of conspiracy, the Alaska Electrical complaint largely compares the “expected” market, as generated by models, to the prices actually observed.

The first “screen” utilized by Plaintiffs was a regression analysis modelling dollar bid-offer spreads as a function of a variety of market fundamentals and bond-specific characteristics. The analysis uses these factors, which have been previously identified in literature as fundamental drivers of bid-offer spreads, to determine if basic economic factors could explain the observed movement in GSE bond spreads. This analysis very accurately predicted the level of GSB bond spreads both before 2012, and from June 2, 2018 to the present. For the period between January 1 2012 and June 1 2018, however, the exact same factors and bond-specific characteristics applied to the same universe of data by the model produced results that were markedly different from actual spreads. This confirms, per Plaintiffs, that during that period the market was “under assault” by forces not accounted for in the macroeconomic and bond-specific data – i.e., the Defendants’ conspiracy.

The second screen utilized by Plaintiffs was a different type of regression analysis, one which adds an additional variable into the analysis that identified the conspiracy period. With the “Collusion Indicator” set to “yes” for the class period, the analysis tests whether the indicator is statistically significantly positive or not. Given that we’re now on page five of this blog post, I doubt many will be surprised when I relate that this Collusion Indicator was found to be related to the actual spreads for the bonds to a statistically significant degree.

The third screen was performed to analyze how consistent the spreads were for GSE bonds across time. A conspiracy to maintain an advantageously high bid-offer spread on a consistent basis, irrespective of the larger economic landscape, would theoretically manifest itself in spreads that were relatively stable over time. What the analysis found was that the GSE bond spreads were more predictable from day-to-day during the class period (when they were allegedly being manipulated) than they were outside of the class period. This difference was, again, statistically significant, confirming that there was an artificial pressure being applied to the GSE bond market during the core conspiracy years.

An additional three screens were performed by Plaintiffs in the course of this analysis, showing, respectively, that (a) GSE bonds reacted differently to external stimuli during the conspiracy years – for example, the level of liquidity in the market had a statistically smaller impact on spreads for GSE bonds during the conspiracy than before or after it; (b) yields for the GSE bonds were less stable during the core conspiracy years than before or after, as a result of Defendants pushing yields artificially low when selling, and artificially high when buying, causing greater volatility; and (c) a basic comparison of bid-offer spreads on GSE bonds to comparable U.S. Treasury bonds of comparable maturity show that the difference is larger during the conspiracy period than before – despite the preceding period containing the U.S. financial crisis, which had significant effect in credit risk, and thus bid-offer spreads, for GSE bonds. Average GSE bond bid-offer spreads were wider in the conspiracy period, while average Treasury spreads barely changed.

Finally, the Alaska Electrical complaint also provides a review of relevant academic literature, concluding that the results of the econometric analyses described above are exactly what that literature predicts would result from a conspiracy not to compete. Chief among the evidence relied on here are studies demonstrating that increased competition among dealers reduces spreads and prices paid by investors in financial markets, as well as reduces transaction costs for bond market participants. These studies provide further evidence, according to Plaintiffs, that the market for GSE bonds were subject to a competition-reducing conspiracy for the class period.

The Remedy:

After their respective analyses, all complaints provide an overview of other current market-manipulation schemes before delving into the question of damages. The Alaska Electrical plaintiffs identified fairly broad categories of damages, but come down to losses suffered as a result of Plaintiffs’ payment of supra-competitive prices and bid-offer spreads for GSE bonds. In contrast, the City of Birmingham complaint – though not the Lincolnshire Police complaint – take their analysis a step further, identifying – in addition to general allegations of overcharge and underpayment – specific instances of purchases by Plaintiffs from the various defendants. If you’re a regular reader of this blog (and if you’ve made it so far into this post, you really ought to be), you may recall that many of the motions to dismiss briefs filed in these antitrust actions attack plaintiffs for failure to identify losses with adequate particularity; it will be interesting to see what the briefing looks like when Defendants are faced with this combination of up-front analysis and specific instances of affected transactions – though even here, Plaintiffs still don’t attach specific loss numbers to the transactions they’ve pinpointed.

With that, I’ll wrap up this blog post. Expect future posts on the motion to dismiss briefing, discovery disputes, class action certification, and any other interesting tidbits that may arise.
PS – bonus point to the clever reader who can guess what artist your favorite antitrust blogger was listening to whilst writing this missive . . . .

This post was written by Alexandra M.C. Douglas.

We welcome your feedback. If you have questions or comments about this post, please e-mail John M. Lundin, the Manipulation Monitor’s editor, at jlundin@schlamstone.com or Alexandra M.C. Douglas at adouglas@schlamstone.com or call John or Alexandra at (212) 344-5400.

Click here to subscribe to this or another of Schlam Stone & Dolan’s blogs.

View posts