The warnings were already inside the building
In November 2006, the head of S&P’s residential-mortgage surveillance group circulated a spreadsheet about trouble in subprime mortgages. By January, the group was discussing a deflating housing bubble and vulnerable mortgage securities. Yet the major public rating actions would not arrive until July 2007. The intervening decisions became central to the account S&P acknowledged when it settled federal and state litigation in February 2015. [1]
The defendants were Standard & Poor’s Financial Services LLC and its parent, McGraw Hill Financial, formerly The McGraw-Hill Companies. The rating business was known as Standard & Poor’s Ratings Services. They agreed to pay $1.375 billion to resolve the Justice Department’s civil lawsuit and suits brought by 19 states and the District of Columbia. This was a civil settlement with an acknowledged statement of facts, not a criminal conviction or a trial verdict. [1][2][3]
A mortgage payment becomes an investment rating
Residential mortgage-backed securities, or RMBS, packaged home loans into securities whose payments depended on the underlying mortgages. Different classes of the same deal absorbed losses in different orders. A collateralized debt obligation, or CDO, could then hold mortgage securities or exposure to them, adding another layer between the homeowner’s payment and the final investor. A rating was the agency’s judgment about credit risk, not a government guarantee of repayment. [1][2]
That layering made the assessment of the underlying mortgages important to both products. If mortgage bonds were about to suffer substantial downgrades, a CDO backed by those bonds could be riskier than its existing rating assumptions suggested. S&P’s account acknowledges that it continued issuing and confirming ratings on CDOs substantially backed by subprime RMBS without adjusting its CDO criteria for anticipated negative mortgage-rating actions. The problem was the relationship between information already inside the firm and the ratings reaching the market. [1]
Investment banks arranging securities were also rating-agency customers. S&P’s published policies said issuer fees and commercial relationships were not supposed to influence its analytical opinions. The lawsuit focused on the gap between those assurances and the actual decisions described in company documents. It did not require treating every inaccurate forecast as fraud. [1][2]
A model update meets customer resistance
The agreed facts begin before the mortgage losses peaked. In 2004 and 2005, S&P was updating its CDO Evaluator model to version 3.0, known as E3. The then-head of the global CDO group set goals that included limiting adverse effects on one segment of the ratings business and improving the firm’s competitive position in another. According to the acknowledged account, the risk of losing transaction revenue affected the update process. [1]
An initially proposed default matrix was not adopted. Work on E3 continued, but the firm encountered another commercial obstacle in July 2005. Feedback from an investment bank warned that the new model would surrender an advantage S&P had held on lower-rated asset pools without giving it an offsetting advantage in higher-quality business. An internal report then described the rollout as slowed and reduced while the firm considered the negative feedback. [1]
These were not merely allegations about an unidentified corporate motive. The settlement incorporated a chronology of model development, client feedback and internal reporting. That record showed where business concerns entered specific analytical decisions. It did not establish that every S&P model adjustment, or every rating during those years, was commercially dictated. [1]
The mortgage alarms grow louder
The November 2006 spreadsheet reported that, in more than half of the subprime RMBS transactions S&P had rated that year, severely loans represented at least a quarter of the protection supporting the lowest-rated class. Some deals had already realized losses. This was a comparison with a layer of credit protection, not a statement that a quarter of all mortgages had defaulted. [1]
On February 7, 2007, surveillance staff recommended placing subordinate classes from about 30 transactions on public CreditWatch Negative and classes from about 20 more transactions on an internal watch list. Five days later, a committee that included staff from the new-issue business chose a narrower public action: 18 classes from 11 transactions. The acknowledged facts also recorded colleagues’ accounts that the surveillance-group head had complained of being prevented from making the downgrades her group wanted because of business concerns. [1]
By June, internal updates described mounting delinquencies and losses. Senior managers circulated an analyst’s warning about the possible scale of losses on 2006 mortgages, and the surveillance head warned that even highly rated securities could default. On June 29, S&P accelerated its revision of surveillance criteria in anticipation of broad negative rating actions. On July 10 it placed 612 classes of subprime RMBS on negative watch; large-scale downgrades followed July 12. [1]
The parallel CDO business supplied the crucial connection. From February 7 through the July 10 public announcement, S&P continued issuing and confirming CDO ratings without adjusting the criteria for the negative mortgage actions it anticipated. The agreed account therefore ties the timing of warnings, public mortgage-rating changes and continuing CDO assessments together. [1]
Investigators pursue the promises behind the ratings
The federal investigation began in November 2009. The United States filed its civil case in the Central District of California on February 4, 2013, announcing it the following day. The Justice Department said federally insured institutions had lost more than $5 billion on failed CDOs rated between March and October 2007. That was the government’s identified loss allegation at filing, not the amount of the eventual settlement or a judicial damages award. [1][2]
The department used the Financial Institutions Reform, Recovery, and Enforcement Act, known as FIRREA, to seek civil penalties for alleged fraud affecting financial institutions. Its theory was that S&P had represented its ratings as independent while revenue and market-share concerns influenced its work. State cases pursued related claims under their own laws. The federal-state resolution arrived two years later, rather than through a completed trial. [1][2][3]
What the money did, and did not, represent
The February 2015 agreement split $1.375 billion evenly. The federal $687.5 million was a FIRREA civil monetary penalty. The other $687.5 million went to the states and District of Columbia under allocations and legal characterizations specified in the agreement. The entire settlement should not be labeled a federal fine or a single investor-refund pool. [1][3]
At the same time, the company announced a separate $125 million settlement with the California Public Employees’ Retirement System, CalPERS, over ratings on three structured investment vehicles. That amount was additional to the federal-state deal. Combining the two produces $1.5 billion in announced settlements, but they resolved different claims and had different recipients. [4]
S&P acknowledged the statement of facts and withdrew its claim that the government had brought the lawsuit in retaliation for S&P’s 2011 actions on the United States’ credit rating. The agreement said the discovery did not support that retaliation allegation. The company emphasized that the settlement contained no findings of legal violations; the Justice Department emphasized the acknowledged conduct. Those statements describe different aspects of a negotiated resolution, rather than changing it into a verdict. [1][3][4]
A settled case leaves a detailed record
The agreement provided for dismissal of the federal action with prejudice and releases of specified civil claims, subject to its conditions and exclusions. It also required compliance with specified state laws and five years of good-faith cooperation with certain state information requests. That five-year information provision was a time-limited settlement term; it was not an expiration date for all legal duties or a declaration that later ratings could not be challenged. [1]
This is a retrospective account of the 2015 resolution, reviewed October 5, 2026. It does not present the original civil lawsuit as pending or the five-year information provision as a new current requirement. The durable record is more specific: it shows how client feedback, model changes, surveillance warnings and the timing of public ratings interacted before the broad public downgrades in July 2007. [1][3][4]
Sources
- Executed federal-state settlement agreement and Annex 1 acknowledged facts, February 2015; SEC Exhibit 10.34Filing / reportBack to text: ↑1↑2↑3↑4↑5↑6↑7↑8↑9↑10↑11↑12↑13↑14↑15↑16↑17↑18
- DOJ civil complaint announcement, February 5, 2013; complaint filed February 4Official sourceBack to text: ↑1↑2↑3↑4↑5
- DOJ federal-state settlement announcement, February 3, 2015Official sourceBack to text: ↑1↑2↑3↑4↑5
- McGraw Hill Financial settlement announcement, including separate CalPERS resolution, February 3, 2015SourceBack to text: ↑1↑2↑3