Morgan Securities LLC and JPMorgan Securities, Inc. (collectively, "JPMorgan"); Citigroup,

Transcription

1

2 Morgan Securities LLC and JPMorgan Securities, Inc. (collectively, "JPMorgan"); Citigroup, Inc., including but not limited to its subsidiaries Citigroup Global Markets Inc. and Citibank NA (collectively, "Citigroup"); William Blair & Company, including but not limited to its subsidiaries William Blair & Company, LLC and WBC Holdings, L.P. (collectively, "William Blair"); Bank of America Corporation, including but not limited to its subsidiaries Bank of America NA, Bank of America Securities LLC, Merrill Lynch, Pierce, Fenner & Smith Incorporated, Bank of America Capital Corporation, BofA Merrill Lynch Asset Holdings, Inc. and Bank of America Merrill Lynch (collectively, "BofA"); Morgan Stanley, including but not limited to its subsidiaries Morgan Stanley Smith Barney LLC, Morgan Stanley & Co. LLC and Morgan Stanley Capital Group Inc. (collectively "Morgan Stanley"); BMO Financial Group, including but not limited to its subsidiaries BMO Financial Corp., BMO Capital Markets Corp. and BMO Capital Markets GKST Inc. (collectively, "BMO"); Barclays PLC, including but not limited to its subsidiaries Barclays Bank, Barclays Capital Inc. and Barclays Investment Bank (collectively, "Barclays"); and Fifth Third Bancorp, including but not limited to its subsidiaries Fifth Third Bank and Fifth Third Securities, Inc. (collectively "Fifth Third") -- under the Illinois False Claims Act, 740 ILCS 175/1 et seq., and alleges -- upon personal knowledge with respect to first-hand industry experience, analysis and expertise, and upon information and belief with respect to all other matters -- as set forth below. Relator filed this action pursuant to section of the Illinois Code of Civil Procedure (735 ILCS 5/13 217), which provides a plaintiff the absolute right to refile a complaint within one year after the plaintiff has voluntarily dismissed the action. Relator filed the original complaint in this action on August 28, 2014; voluntarily dismissed the original complaint on January 13, 2016; and refiled the complaint under 735 ILCS 5/ on January 10,

3 INTRODUCTION 1. This case is about defendants' widespread fraud and collusion in the fees they charge and the interest rates they set for variable rate, tax-exempt bonds -- known as Variable Rate Demand Obligations ("VRDOs") -- issued by the state of Illinois, including any agency of State government, the system of State colleges and universities, any school district, community college district, county, municipality, municipal corporation, unit of local government, and any combination of the above under an intergovernmental agreement (collectively, "Illinois"). 2. Illinois has hired defendants as so-called "remarketing agents" to actively and individually market and price these bonds at the lowest possible interest rates. Instead of doing that, however, defendants have engaged in a coordinated "Robo-Resetting" scheme where they mechanically set the rates en masse without any consideration of the individual characteristics of the bonds, the associated market conditions or investor demand. Defendants "Robo-Reset" these rates in order to keep the bonds in the hands of their holders, and thus alleviate the need for defendants to remarket the bonds. 3. Defendants have engaged in this scheme so they can collect millions of dollars per year in remarketing fees without providing any of the remarketing services for which Illinois pays them. But even more egregiously, defendants failed to set the rates at the lowest possible interest rates, as their agreements with Illinois required. Instead, defendants have employed the Robo-Resetting device to collectively impose artificially high interest rates on Illinois VRDOs, the exact opposite of what Illinois hires them to accomplish. 4. Defendants benefit from keeping the VRDO interest rates artificially high because it causes VRDO investors (typically tax-exempt money market funds, which defendants in many instances own or manage) to hold onto the bonds rather than redeem them at face value plus interest. This "put" option is one of the defining features of a VRDO security. So is the responsibility of the remarketing agent to find another investor when the "put" option has been exercised. If the remarketing agent is unable to find another investor, a liquidity provider -- often the remarketing agent itself -- must step in and purchase the VRDO from the redeeming 3

4 investor. 5. Defendants' Robo-Resetting scheme thus allows defendants to extract from Illinois substantial fees for remarketing services they do not provide. And it allows them to extract even greater fees as the liquidity provider, typically through guaranteeing the VRDO with a letter of credit, when the risk of needing to draw on that letter of credit is essentially nonexistent. 6. As a result of defendants' fraudulent rate-setting activity, Illinois since at least April 2009 has paid hundreds of millions of dollars in overcharges by paying (i) remarketing fees for remarketing services defendants do not actually provide, (ii) inflated and collectively set VRDO interest rates, and (iii) excessive letter of credit fees for letter of credit services that are rarely, if ever, called upon. 7. Relator brings this case on behalf of Illinois to recover the money Illinois has paid in excessive or unsupported fees and inflated interest rates due to defendants' fraudulent ratesetting scheme. Relator also brings this case to stop defendants from continuing with this fraudulent scheme and to ensure that Illinois receives the proper interest rates and remarketing services it pays for in the future. PARTIES 8. Relator Edelweiss Fund, LLC is a limited liability company located in Delaware. Through its principal, Relator has more than 20 years of experience advising municipalities and other clients on issuing securities, particularly VRDOs and other types of municipal bonds. Through its work in the industry, Relator became suspicious that defendants and other VRDO remarketing agents ("RMAs") were, working in a coordinated fashion, systematically resetting the VRDO interest rates on an algorithmic or some other kind of mechanical basis (so-called "Robo-Resetting"), maintaining artificially high rates for these securities and charging Illinois and other states improper or excessive fees for remarketing and letter of credit services they were not actually providing. Relator confirmed its suspicions after performing an extensive forensic analysis of the interest rates and other market data -- for the April 1, 2009 through November 14, 4

5 2013 period 1 -- for the Illinois VRDOs (and VRDOs issued by other states) for which defendants have served as the RMA. 9. Defendant JPMorgan is a financial services company that does business in the State of Illinois and nationally. It is headquartered in New York City and since April 1, 2009 has served as the RMA for roughly 202 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $6 billion. 10. Defendant Citigroup is a financial services company that does business in the State of Illinois and nationally. It is headquartered in New York City and since April 1, 2009 has served as the RMA for roughly 35 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $2.1 billion. 11. Defendant William Blair is a financial services company that does business in the State of Illinois and nationally. It is headquartered in Chicago and since April 1, 2009 has served as the RMA for roughly 63 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $1.4 billion. 12. Defendant BofA is a financial services company that does business in the State of Illinois and nationally. It is headquartered in Charlotte, North Carolina and since April 1, 2009 has served as the RMA for roughly 170 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $3.8 billion. 13. Defendant Morgan Stanley is a financial services company that does business in the State of Illinois and nationally. It is headquartered in New York City and since April 1, 2009 has served as the RMA for roughly 33 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $2.2 billion. 14. Defendant BMO is a financial services company that does business in the State of 1 Relator's allegations concerning the number of VRDOs for which defendants served as the RMAs, the value of those issuances, and Relator's forensic analysis are based in part on data Relator obtained, which spans the period April 1, 2009 through November 14, However, the misconduct Relator alleges, as well as the resulting harm to Illinois, has occurred since at least April 2009 and continues through the present. 5

6 Illinois and nationally. It is headquartered in Toronto and since April 1, 2009 has served as the RMA for roughly 61 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $630 million. 15. Defendant Barclays is a financial services company that does business in the State of Illinois and nationally. It is headquartered in London and since April 1, 2009 has served as the RMA for roughly 32 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $2.2 billion. 16. Defendant Fifth Third is a financial services company that does business in the State of Illinois and nationally. It is headquartered in Cincinnati and since April 1, 2009 has served as the RMA for roughly 38 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $330 million. 17. Taken together, since April 1, 2009 defendants have served as the RMA for roughly 634 VRDOs issued by Illinois, which collectively had a value at issuance of roughly $18.7 billion. This represents approximately 65 percent of the VRDOs Illinois has had outstanding during this period. JURISDICTION AND VENUE 18. This Court has jurisdiction over the subject matter of this action under 740 ILCS 175/4(b). 19. This Court has personal jurisdiction over defendants because they can be found, reside and/or transact business in Illinois and Cook County. 20. Pursuant to 735 ILCS 5/2-101, venue is proper because defendants can be found in and transact business within Cook County. Throughout the time period relevant to the allegations of this Complaint, defendants engaged in substantial business transactions within the state of Illinois and committed many of the violations proscribed by 740 ILCS 175/3 in the state and this county, and continue to do so today. 21. Pursuant to 740 ILCS 175/4(b)(2), Relator previously provided to the Illinois Attorney General a copy of the original filed complaint in this action and written disclosure of 6

7 substantially all material evidence and information the Relator possesses. BACKGROUND 22. VRDOs are tax-exempt, variable rate bonds with interest rates reset on a periodic basis, typically weekly. From an investor's perspective, they are considered short-term securities because they include a "put" feature that allows the investor at each periodic reset date to tender the security back to the RMA at face value ("par") plus any accrued interest. VRDOs are attractive to issuers because they allow them to borrow money for long periods of time while paying short-term interest rates. They are likewise attractive to investors because they are a lowrisk, high-liquidity and tax-free investment. This tax advantage is the key feature that distinguishes VRDOs from the most closely analogous short-term debt instrument, 7-day AA non-financial commercial paper, and why investors historically have been willing to invest in VRDOs at an interest rate significantly lower than that of commercial paper. 23. VRDOs are individually identified by their Committee on Uniform Securities Identification Procedures ("CUSIP") number, which is a 9-character alphanumeric code assigned for tracking and trading purposes to U.S. and Canadian registered stocks and U.S. government and municipal bonds. VRDOs are tracked by CUSIP number on the Securities Industry Financial Markets Association ("SIFMA") swap index. The SIFMA index tracks the average interest rate for highly-rated VRDOs reset on a weekly basis. 24. VRDOs are primarily issued by state and local public entities such as municipalities, agencies, public universities and hospitals to raise money to fund various longterm projects or infrastructure. They also are issued by public entities on behalf of various 501(c)(3) organizations. Tax-exempt money market funds, which defendants in many cases own or manage, are the largest holders of VRDOs. 25. As of November 30, 2013, there were approximately 9,000 VRDOs outstanding in the United States with a collective balance of roughly $223 billion. Illinois was the issuer of roughly 575 of these bonds. 26. VRDOs typically receive liquidity and are secured by letters of credit provided by 7

8 highly-rated commercial banks to protect investors in the event the RMA is unable to find new investors for tendered bonds. In that instance, the obligation to purchase the tendered bond falls on the letter of credit provider, which in many cases is the RMA itself. Issuers currently pay letter of credit providers an annual fee of generally between 50 and 150 basis points of the VRDO debt balance for this liquidity and credit enhancement feature (one basis point equals one hundredth of one percent). Several defendants serve as both the RMA and letter of credit provider for a sizeable number of their Illinois VRDOs. 27. VRDO issuers like Illinois contract with RMAs to manage the bonds on an individual basis. RMAs have two basic jobs, for which issuers typically pay an average annual fee of roughly 10 basis points of the VRDO debt balance. 28. First, RMAs are required to reset (typically on a weekly basis) the VRDO interest rate at the lowest possible rate. This must be based on an individual determination of what rate the specific VRDO will bear considering the unique characteristics of the bond, the relevant market conditions and the particular investor demand for the specific bond at issue. 29. Second, RMAs are required to actively "remarket" at the lowest possible rate the VRDOs to other investors when the existing investor "puts" the bond back to the RMA for a return of its investment (at face value plus interest). 30. At least three sources set forth these twin obligations and detail the rights and responsibilities surrounding the VRDO issuer/rma relationship. 31. First, the Municipal Securities Rulemaking Board ("MSRB") has issued rules that cover RMA duties to VRDO issuers. MSRB rules are enforceable by various regulatory bodies, including the United States Securities and Exchange Commission and the Financial Industry Regulatory Authority. 32. MSRB Rule G-17 requires RMAs to deal fairly and honestly with VRDO issuers. It provides, in relevant part, that RMAs "must, in the conduct of their municipal securities activities, deal fairly with all persons and must not engage in any deceptive, dishonest, or unfair practice." 8

9 33. MSRB Rule G-18 requires RMAs to secure for VRDO issuers a fair and reasonable interest rate. It provides, in relevant part, that RMAs must "make a reasonable effort to obtain a price for the customer that is fair and reasonable in relation to prevailing market conditions." In adopting the rule, the MSRB explained that this standard means that RMAs "will exercise the same level of care as the professional would if acting for its own account, including the exercise of diligence in ascertaining prevailing market conditions." 34. A second source that sets forth the RMAs' twin obligations is the SIFMA Model Disclosures Pursuant to MSRB Rule G-17 ("SIFMA Model Disclosures"). These are disclosures that SIFMA advises RMAs to make to VRDO issuers to comply with their MSRB Rule G-17 obligation to deal fairly and honestly with the issuers. Notably, defendants play a very active role in SIFMA with representatives from Bank of America, JPMorgan, Citigroup, Morgan Stanley and William Blair currently sitting on the SIFMA Board of Directors SIFMA Model Disclosures make clear that RMAs are obligated to secure for VRDO issuers the lowest possible interest rate. They read, in relevant part, that an RMA is "required to set the interest rate at the rate necessary, in its judgment, as the lowest rate that permits the sale of the VRDOs at 100% of their principal amount (par) on the interest reset date." (Emphasis added.) 36. Finally, this language from the SIFMA Model Disclosures setting forth the RMA requirement to set the lowest VRDO interest rate on each rate determination date is standard throughout the VRDO industry. It is specifically contained in defendants' various remarketing agreements with Illinois, including remarketing circulars and other official statements Illinois issuers prepare that set forth the rights and obligations that define the RMA/issuer relationship. 37. For example, a remarketing circular dated February 2, 2011 concerning $479 million in VRDOs issued by The Illinois State Toll Highway Authority and remarketed by Citigroup and Barclays states, "Pursuant to the Remarketing Agreements, each Remarketing 2 See 9

10 Agent is required to determine the applicable rate of interest that, in its judgment, is the lowest rate that would permit the sale of the applicable series of Reoffered Bonds at such interest rate at par plus accrued interest, if any, on and as of the applicable effective date. The interest rate will reflect, among other factors, the level of market demand for such Reoffered Bonds." (Emphasis added.) 38. A sampling of additional VRDO offering or remarketing circulars containing this identical or near identical language includes: (i) a remarketing circular dated May 15, 2014 concerning $600 million in VRDOs issued by the State of Illinois and remarketed by JPMorgan; (ii) the remarketing memorandum dated August 14, 2009 for over $140 million in VRDOs issued by the Illinois Finance Authority for the University of Chicago Medical Center and remarketed by Barclays; (iii) the offering circular dated March 13, 2002 for over $100 million in VRDOs issued by Cook County, Illinois and remarketed by William Blair; and (iv) the March 5, 2010 remarketing circular for $20 million in VRDOs issued by the Illinois Housing Development Authority and remarketed by Merrill Lynch. 39. Defendants' remarketing agreements with Illinois specified defendants' representations to Illinois. Defendants billed Illinois for their purported services under these remarketing agreements. Relator did not have access to defendants' accounting departments but nonetheless knows about defendants' illicit remarketing practices. Relator has certain of defendants' remarketing circulars, which describe the specific services that defendants represented they were providing to Illinois under their remarketing agreements. The remarketing circulars also detail the specific services for which defendants made claims to Illinois, i.e. "any request or demand, whether under a contract or otherwise, for money" made to the State or to someone acting on behalf of the State (740 ILCS 175/3(b)(2)). The remarketing circulars thus evidence the false or fraudulent claims for payment or approval that defendants made to Illinois, as well as the false records or statements material to false or fraudulent claims made to Illinois. 40. Defendants have not complied with their obligation to actively and individually reset and remarket Illinois' VRDOs at the lowest possible rate. Instead, they have collectively 10

11 engaged in their Robo-Resetting scheme to minimize their need to provide remarketing services, secure artificially high VRDO rates, extract from Illinois improper and excessive remarketing and letter of credit fees, and ultimately cause Illinois to pay hundreds of millions of dollars in VRDO-related overcharges. 41. Defendants' claims to Illinois did not merely request payment, but also made specific representations about the services provided. Defendants' failure to disclose noncompliance with material statutory, regulatory, or contractual requirements made those representations lies or, at best, misleading half-truths. DEFENDANTS HAVE ENGAGED IN ROBO-RESETTING TO COLLECT IMPROPER AND EXCESSIVE FEES AND CAUSE ARTIFICIALLY HIGH ILLINOIS VRDO INTEREST RATES 42. Since at least April 1, 2009, Illinois has paid defendants tens of millions of dollars to actively and individually reset and remarket its VRDOs at the lowest interest rate possible. However, defendants have deliberately not performed the services for which Illinois has paid them. Instead of resetting the VRDO rates and remarketing them on an individual basis and with the purpose of finding the lowest possible interest rate, defendants have engaged in a practice of setting their VRDO rates mechanically and collectively, without any consideration of the unique attributes of each particular bond. The effect of Robo-Resetting is to maximize the likelihood that existing holders of each bond will retain their holdings -- at greater cost to Illinois -- even if alternate investors exist who may be willing to hold the same paper at a lower interest rate. The RMAs' job is to ensure the bonds bear the lowest possible interest rate; it is not to guarantee existing bondholders are happy with the rate they are receiving. 43. Relator's forensic analysis of defendants' rate-setting practices for the four-and-ahalf-year period of April 1, 2009 through November 14, 2013 reveals that for the vast majority of the VRDOs they manage -- and likely for all of them -- defendants group into "buckets" collections of unrelated bonds and set their interest rates collectively. They do this by applying to each VRDO in a particular bucket an identical pricing spread which moves the interest rate of each bond in the bucket up or down in lock-step fashion. In no case are defendants making an 11

12 individual determination of what the appropriate rate should be for a particular bond. Nor are they making any effort to secure the lowest possible rate for the bond. Defendants' collective rate setting and the absence of individual pricing considerations result in an equal absence of remarketing efforts directed at finding investors willing to hold paper at the lowest interest cost. 44. For the purposes of this pricing analysis, a particular bond was considered to be part of a bucket if it had the identical week-over-week rate change as the other bonds in the bucket for at least twenty-six weeks (but typically much longer) and at least 80 percent of the time (but typically much higher). These two bucket threshold "qualifiers" were necessary to account for the periodic sale of the bonds or their departure from the market as a result of being paid off or converted to fixed rates. They were also necessary to most accurately capture the extent of the Robo-Resetting in which defendants have engaged. 45. Relator's pricing analysis for each defendant included all VRDOs they managed with weekly reset dates and for which there were at least twenty rate resets within one twenty-six week interval during the April 1, 2009 through November 14, 2013 period. 3 This represents the lion's share of each defendant's total VRDO portfolio for the period. To avoid double-counting, once a VRDO was identified as part of a particular bucket, it was removed from the data set even though many of these bonds fell into more than one bucket over the four-and-a-half-year study period. 46. As shown immediately below, Relator's detailed pricing analysis unequivocally demonstrates that since at least April 1, 2009, each defendant has engaged in Robo-Resetting for its entire Illinois VRDO portfolio (as well as for its VRDO portfolios of other states). This is evident from the pervasiveness of defendants' bucketing practices and the statistical impossibility that any bucketing would have occurred -- even for only two VRDOs -- had defendants actively 3 VRDOs with daily reset dates were excluded from the analysis because they represent a small percentage of VRDOs and would have greatly increased, with little incremental gain, the already significant burden of conducting the forensic bucketing analysis. Likewise, VRDOs in the market for less than twenty-six weeks were excluded because of the twenty resets in twenty-six weeks threshold qualifier discussed above. 12

13 and individually priced the VRDOs they managed as they were obligated to do. JPMorgan 47. Relator's JPMorgan analysis included 1,377 VRDOs, which had a collective value at issuance of $49.3 billion. Illinois has been the issuer of 149 of these bonds (with a collective value at issuance of $2.8 billion). The analysis shows JPMorgan used bucketing to set the VRDO interest rates for at least 95 percent of the JPMorgan portfolio studied, including at least 146 of the 149 Illinois VRDOs in the study. 48. JPMorgan's pricing for these VRDOs can be broken down into four buckets, with the vast majority of the bonds -- 1,083 of them -- residing in a single bucket. 137 of the bonds in this single bucket were issued by Illinois, representing roughly 92 percent of the Illinois VRDOs in the JPMorgan study. Attached hereto as Exhibit A is a listing of JPMorgan's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis While each of the VRDOs in these JPMorgan buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 50. For example, with respect to the 1,083 VRDOs in JPMorgan's largest bucket, 941 of them had the identical interest rate change (at least 80% of the time) for a full year; 744 of them had the identical interest rate change (at least 80% of the time) for two years; and 634 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 51. Similarly, 1,004 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 947 of them had the identical 4 The Illinois VRDO listing for JPMorgan and each of the other defendants (Exhibits A through H) breaks out the VRDOs by whether they are reset on a weekly or daily basis. As noted above, only the weekly reset bonds in the market for at least twenty-six weeks -- which account for the majority of defendants' Illinois VRDO portfolios -- were included in the pricing study. 13

14 interest rate change 95 percent of the time (for at least twenty-six weeks); and 785 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). Citigroup 52. Relator's Citigroup analysis included 1,114 VRDOs, which had a collective value at issuance of $44.2 billion. Illinois has been the issuer of 32 of these bonds (with a collective value at issuance of $2 billion). The analysis shows Citigroup used bucketing to set the VRDO interest rates for at least 94 percent of the Citigroup portfolio studied, including at least 29 of the 32 Illinois VRDOs in the study. 53. Citigroup's pricing for these VRDOs can be broken down into six buckets, with the vast majority of the bonds of them -- residing in two buckets. 26 of the bonds in these two buckets were issued by Illinois, representing roughly 81 percent of the Illinois VRDOs in the Citigroup study. Attached hereto as Exhibit B is a listing of Citigroup's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 54. While each of the VRDOs in these Citigroup buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 55. For example, with respect to the 487 VRDOs in Citigroup's largest bucket, 435 of them had the identical interest rate change (at least 80% of the time) for a full year; 390 of them had the identical interest rate change (at least 80% of the time) for two years; and 335 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 56. Similarly, 442 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 426 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 373 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). 14

15 William Blair 57. Relator's William Blair analysis included 90 VRDOs, which had a collective value at issuance of $1.5 billion. Illinois has been the issuer of 58 of these bonds (with a collective value at issuance of $1.2 billion). The analysis shows William Blair used bucketing to set the VRDO interest rates for at least 99 percent of the William Blair portfolio studied, including at least 57 of the 58 Illinois VRDOs in the study. 58. William Blair's pricing for these VRDOs can be broken down into three buckets, with the vast majority of the bonds of them -- residing in a single bucket. 57 of the bonds in this single bucket were issued by Illinois, representing roughly 98 percent of the Illinois VRDOs in the William Blair study. Attached hereto as Exhibit C is a listing of William Blair's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 59. While each of the VRDOs in these William Blair buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 60. For example, with respect to the 77 VRDOs in William Blair's largest bucket, 71 of them had the identical interest rate change (at least 80% of the time) for a full year; 64 of them had the identical interest rate change (at least 80% of the time) for two years; and 73 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the full four-and-a-half-year study period. 61. Similarly, 76 of the VRDOs in this bucket had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 75 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). BofA 62. Relator's BofA analysis included 1,959 VRDOs, which had a collective value at issuance of $65.6 billion. Illinois has been the issuer of 138 of these bonds (with a collective 15

16 value at issuance of $3.4 billion). The analysis shows BofA used bucketing to set the VRDO interest rates for at least 68 percent of the BofA portfolio studied, including at least 103 of the 138 Illinois VRDOs in the study. 63. BofA's pricing for these VRDOs can be broken down into four buckets, with a majority of the bonds -- 1,089 of them -- residing in a single bucket. 97 of the bonds in this single bucket were issued by Illinois, representing roughly 70 percent of the Illinois VRDOs in the BofA study. Attached hereto as Exhibit D is a listing of BofA's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 64. While each of the VRDOs in these BofA buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 65. For example, with respect to the 1,089 VRDOs in BofA's largest bucket, 765 of them had the identical interest rate change (at least 80% of the time) for a full year; 164 of them had the identical interest rate change (at least 80% of the time) for two years; and 38 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 66. Similarly, 765 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 553 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 234 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). Morgan Stanley 67. Relator's Morgan Stanley analysis included 548 VRDOs, which had a collective value at issuance of $31.4 billion. Illinois has been the issuer of 22 of these bonds (with a collective value at issuance of $1.5 billion). The analysis shows Morgan Stanley used bucketing to set the VRDO interest rates for at least 83 percent of the Morgan Stanley portfolio studied, 16

17 including all 22 of the Illinois VRDOs in the study. 68. Morgan Stanley's pricing for these VRDOs can be broken down into five buckets, with a majority of the bonds of them -- residing in a single bucket. 19 of the bonds in this single bucket were issued by Illinois, representing roughly 86 percent of the Illinois VRDOs in the Morgan Stanley study. Attached hereto as Exhibit E is a listing of Morgan Stanley's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 69. While each of the VRDOs in these Morgan Stanley buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 70. For example, with respect to the 316 VRDOs in Morgan Stanley's largest bucket, 246 of them had the identical interest rate change (at least 80% of the time) for a full year; 143 of them had the identical interest rate change (at least 80% of the time) for two years; and 58 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 71. Similarly, 240 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 187 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 43 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). BMO 72. Relator's BMO analysis included 100 VRDOs, which had a collective value at issuance of $1.6 billion. Illinois has been the issuer of 56 of these bonds (with a collective value at issuance of $520 million). The analysis shows BMO used bucketing to set the VRDO interest rates for 100 percent of the BMO portfolio studied, including all 56 of the Illinois VRDOs in the study. 73. BMO's pricing for these VRDOs can be broken down into three buckets, with the 17

18 vast majority of the bonds of them -- residing in a single bucket. 52 of the bonds in this single bucket were issued by Illinois, representing roughly 93 percent of the Illinois VRDOs in the BMO study. Attached hereto as Exhibit F is a listing of BMO's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 74. While each of the VRDOs in these BMO buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 75. For example, with respect to the 90 VRDOs in BMO's largest bucket, 87 of them had the identical interest rate change (at least 80% of the time) for a full year; 76 of them had the identical interest rate change (at least 80% of the time) for two years; and 74 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 76. Similarly, 87 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 83 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). Barclays 77. Relator's Barclays analysis included 549 VRDOs, which had a collective value at issuance of $19.6 billion. Illinois has been the issuer of 30 of these bonds (with a collective value at issuance of $2 billion). The analysis shows Barclays used bucketing to set the VRDO interest rates for at least 85 percent of the Barclays portfolio studied, including at least 24 of the 30 Illinois VRDOs in the study. 78. Barclays' pricing for these VRDOs can be broken down into four buckets, with a majority of the bonds of them -- residing in a single bucket. 18 of the bonds in this single bucket were issued by Illinois, representing 60 percent of the Illinois VRDOs in the Barclays study. Attached hereto as Exhibit G is a listing of Barclays' entire Illinois VRDO portfolio 18

19 which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 79. While each of the VRDOs in these Barclays buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 80. For example, with respect to the 316 VRDOs in Barclays' largest bucket, 267 of them had the identical interest rate change (at least 80% of the time) for a full year; 197 of them had the identical interest rate change (at least 80% of the time) for two years; and 175 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 81. Similarly, 285 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 255 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 213 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). Fifth Third 82. Relator's Fifth Third analysis included 415 VRDOs, which had a collective value at issuance of $3.7 billion. Illinois has been the issuer of 33 of these bonds (with a collective value at issuance of $300 million). The analysis shows Fifth Third used bucketing to set the VRDO interest rates for at least 95 percent of the Fifth Third portfolio studied, including at least 32 of the 33 Illinois VRDOs in the study. 83. Fifth Third's pricing for these VRDOs can be broken down into six buckets, with the vast majority of the bonds of them -- residing in a single bucket. 25 of the bonds in this single bucket were issued by Illinois, representing roughly 76 percent of the Illinois VRDOs in the Fifth Third study. Attached hereto as Exhibit H is a listing of Fifth Third's entire Illinois VRDO portfolio which identifies for each VRDO: the issuer, CUSIP number, letter of credit provider, and whether it fell into a bucket under Relator's pricing analysis. 19

20 84. While each of the VRDOs in these Fifth Third buckets had the identical interest rate change as the other bonds in the bucket for at least twenty-six weeks, 80 percent of the time, the majority of the bonds were in their respective buckets for significantly longer than twenty-six weeks and at a significantly higher percentage than 80 percent of the time. 85. For example, with respect to the 312 VRDOs in Fifth Third's largest bucket, 284 of them had the identical interest rate change (at least 80% of the time) for a full year; 222 of them had the identical interest rate change (at least 80% of the time) for two years; and 247 of them had the identical interest rate change (at least 80% of the time) for the entire time they were in the market during the four-and-a-half-year study period. 86. Similarly, 297 of the VRDOs in this bucket had the identical interest rate change 90 percent of the time (for at least twenty-six weeks); 295 of them had the identical interest rate change 95 percent of the time (for at least twenty-six weeks); and 269 of them had the identical interest rate change 100 percent of the time (for at least twenty-six weeks). * * * 87. There is no economic or business justification for the Robo-Resetting pricing practice in which defendants have engaged. Nor is there even a remote statistical possibility that defendants' VRDO pricing -- if done actively, individually and in consideration of the relevant market conditions -- would result in this kind of widespread bucketing. Indeed, the statistical likelihood of even two actively, individually priced and marketed VRDOs residing in the same bucket for any meaningful length of time is zero. 88. This comes not only from a basic understanding of how the VRDO market is supposed to work. It also comes from a simulation Relator conducted based on (i) real-world changes in the SIFMA weekly VRDO pricing index, and (ii) the real-world distribution of these weekly rate changes across the entire VRDO market. 89. In setting it up to mirror both the real-world VRDO weekly rate changes and realworld distribution of these rate changes across the entire VRDO market, the simulation was 20

21 designed to account for any actual changes in the market or economy that might have had a realworld impact on how VRDO rates were set during the four-and-a-half-year period of study. 90. The simulation was constructed as follows: A model RMA was created with a simulated portfolio of 1,959 VRDOs. This replicated the total number of bonds in the BofA study -- the largest of the defendant portfolios studied -- for the period April 1, 2009 through November 14, For each week in this four-and-a-half-year period, the interest rate for each of these 1,959 VRDOs was reset in such a way that the average weekly rate change among this collection of simulated bonds was identical to the real-world average weekly rate change across the entire VRDO market as measured by the SIFMA index. In other words, if for a particular week SIFMA recorded an average VRDO weekly rate increase of 10 basis points, for that same week, the 1,959 bonds in the simulation would also have an average rate increase of 10 basis points. To replicate what is supposed to happen in the VRDO market -- where each particular bond is actively and individually priced and marketed on a weekly basis -- the weekly rate change for each of the 1,959 bonds was randomly assigned. But it was not only done in such a way that the average weekly rate change for the portfolio matched that of SIFMA. It was also done so that the standard deviation of the distribution of the weekly rate changes among the 1,959 simulated bonds was matched to that of the real-world distribution of weekly rate changes that actually occurred for that week across the entire VRDO market. The simulation, which covered the April 1, November 14, 2013 period, was run one thousand times. 91. Not a single bucket of even two bonds resulted from the one thousand simulations Relator conducted. Not only were there no buckets of even two bonds for twenty-six weeks, 80 percent of the time (the threshold qualifiers used in the bucketing analysis of defendants' VRDO pricing). There were no such buckets for any meaningful length of time. 92. This sharply contrasts with the pervasive level of bucketing among defendants' VRDO portfolios. Based on Relator's pricing analysis, this bucketing involves all or virtually all the VRDOs managed by BMO (100%), William Blair (99%), JPMorgan (95%), Fifth Third (95%) and Citigroup (94%), and the vast majority of VRDOs managed by Barclays (85%), Morgan Stanley (83%) and BofA (68%). It involves single buckets containing dozens, and for 21

22 most defendants, hundreds of bonds, or even more than one thousand bonds -- BofA (1,089), JPMorgan (1,083), Citigroup (487), Morgan Stanley (316), Barclays (316), Fifth Third (312), BMO (90) and William Blair (77). And it involves buckets where a significant number of bonds, and for most defendants the majority of them, were parked for the entire time they were in the market during the four-and-a-half-year study period -- JPMorgan (634), Citigroup (335), Fifth Third (247), Barclays (175), BMO (74), William Blair (73), Morgan Stanley (58) and BofA (38). 93. These numbers, particularly when compared to the complete absence of bucketing in the real-world simulation, empirically demonstrate that defendants have intentionally engaged in Robo-Resetting rather than pricing their VRDOs actively and on an individual basis. 94. Furthermore, this pricing analysis understates the extent to which defendants have engaged in Robo-Resetting to set their VRDO interest rates. For each of the defendant banks, Relator's bucketing analysis involved reviewing weekly rate changes, for four-and-a-half years, for anywhere from 90 VRDOs (William Blair) to 1,959 VRDOs (BofA). This resulted in a study of from roughly fifteen thousand to several hundred thousand rate changes for each defendant. Dealing with these kinds of numbers makes it extremely unlikely that Relator's analysis was able to capture all the bucketing in which defendants have engaged, particularly for those buckets that did not contain a noticeably large number of VRDOs. Nor would it capture other practices beyond bucketing in which defendants have engaged to automatically and mechanically set their VRDO interest rates without attempting to secure the lowest possible rate. 95. From all of this, there can be little question that defendants have been engaging in Robo-Resetting for all the VRDOs in their Illinois portfolios (and in their VRDO portfolios of every other state). And they have been doing so in a coordinated and cooperative fashion. Together, defendants have avoided carrying out the duties they contract to fulfill: find investors to hold Illinois VRDOs at the lowest possible interest rate. 22

23 DEFENDANTS HAVE COLLUDED IN THEIR ROBO-RESETTING SCHEME 96. Defendants conspired together to carry out their Robo-Resetting scheme. 97. The collusion among defendants in their Robo-Resetting scheme is evident from the bucketing that has occurred across the defendant banks. The likelihood of cross-bank bucketing should be even lower than intra-bank bucketing if RMAs are acting in the absence of collusion and complying with their obligation to set interest rates competitively and at the lowest rates possible. Yet, Relator's analysis found the same kind of lock-step price movements even across banks. Here are several specific examples of where this coordinated VRDO pricing among defendants is particularly borne out. Example One -- Cross-Bank Bucket As just one example, based on the same criteria used to identify single bank buckets discussed above, Relator isolated a cross-bank bucket of 440 VRDOs from defendants JPMorgan, Citigroup, Morgan Stanley and BofA, as well as from Wells Fargo, another major RMA but which largely conducts its RMA business outside Illinois. The average maximum 26- week matching rate for the bonds in this bucket over the four-and-a-half-year study period was 95%, broken out as follows: JPMorgan (143 VRDOs in bucket, with 93% matching rate); Citigroup (102 VRDOs in bucket, with 98% matching rate); Morgan Stanley (23 VRDOs in bucket, with 95.4% matching rate); BofA (100 VRDOs in bucket, with 92% matching rate); and Wells Fargo (72 VRDOs in bucket, with 97% matching rate). Example Two -- Cross-Bank Bucket As a second example, Relator isolated a cross-bank bucket of 231 VRDOs from defendants Citigroup, Morgan Stanley and BofA, as well as from Wells Fargo, that had closely matched interest rate changes for the entire four-and-a-half-year study period. Over the last year covered by the data set, the interest rate changes among these 231 bonds were virtually identical, with an overall matching rate of 91.2%. The average maximum 26-week matching rate for the bonds in this bucket over this period was 97% broken out as follows: Citigroup (170 VRDOs in 23

24 bucket, with 99% matching rate); Morgan Stanley (47 VRDOs in bucket, with 92% matching rate); BofA (4 VRDOs in bucket with 87.5% matching rate); and Wells Fargo (10 VRDOs in bucket with 91% matching rate) The graph below shows the cumulative rate changes for a majority of the VRDOs in this bucket, which moved in near-perfect lock step over the final year of the study period This data clearly shows the extent to which the rates on these VRDOs move together despite the fact they are set by different RMAs. Another way to measure the irrational and improbable nature (absent collusion) of this price-setting activity is to look at the correlations between the interest rates on pairs of VRDOs within the bucket. Over the final year of the study period, where these VRDOs moved most closely together, the average correlation between the interest rates on pairs of VRDOs in the full group of 231 VRDOs (26,796 correlations) is The average correlation is just as high considering only the interest rates on pairs of VRDOs remarketed by different banks (11,068 correlations). 24

25 Example Three VRDO Random Selection 102. These cross-bank buckets are just two specific examples of the numerous crossbank buckets that exist among defendants (and other large RMAs such as Wells Fargo). A third example of defendants' coordinated pricing comes from Relator's random selection of ten VRDOs from the defendant banks JPMorgan, Citigroup, Morgan Stanley, BofA, and Wells Fargo, which covered a wide cross-section of the market in terms of the state of issuance, the type of issuer, the relevant industry, the type of credit support, the credit provider, and the size of the debt. For this collection of fifty bonds (set forth in Exhibit I), Relator reviewed the weekly interest rates defendants set for the 2012 period and as shown in the chart immediately below found almost identical price movement across the board. 5 5 For some of these VRDOs, their charted rate streams were "time-adjusted" (i.e., moved forward or backward) between one and seven days to account for the fact that different VRDOs reset on different days of the week. 25