Credit agencies awarded top grades to mortgage securities while issuers paid for their assessments. The crisis exposed the consequences of commercial pressure and reliance on ratings.
By Matthew A. McIntosh
Public Historian
Brewminate
Introduction: When Triple-A Became a Question
On July 10, 2007, Moody’s downgraded 399 separately rated portions, or tranches, of securities backed by subprime mortgages issued the previous year. Standard & Poor’s placed 612 subprime mortgage-security tranches on negative watch that day, then downgraded 498 of them on July 12. These initial actions concentrated on lower-rated claims, but their significance extended into structures that investors had regarded as exceptionally secure. Some mortgage-security tranches served as collateral for collateralized debt obligations, whose own senior portions carried the highest available ratings. Deterioration in one part of this arrangement raised questions about protections elsewhere. The revisions exposed a widening discrepancy between the assessments attached to mortgage investments and the performance of the loans supporting them.
For institutional investors, a high rating had practical consequences well beyond the reassurance conveyed by a familiar symbol. Investment mandates could restrict purchases to specified rating categories, while financial regulations incorporated recognized agencies’ judgments into specific requirements governing eligible assets or capital. Moody’s, Standard & Poor’s, and Fitch consequently occupied a position in which private assessments helped determine how financial institutions deployed their resources. Their authority depended partly on accumulated reputations, but public rules and contractual arrangements gave those reputations additional force. The judgments themselves concerned creditworthiness, with methodologies differing in their treatment of default probability and expected loss; market price and the ability to sell an instrument remained separate questions. Structured finance complicated the relationship between those distinctions and the apparent familiarity of the rating scale. By assigning losses first to junior claims, a transaction could provide senior investors with substantial protection even when the underlying borrowers were risky. The adequacy of that protection nevertheless depended on assumptions about losses occurring together, making the highest-rated portions sensitive to errors that a single rating symbol could scarcely explain.
The commercial relationship behind these assessments created another difficulty. Under the predominant issuer-pays model, the organization seeking a rating, or the bank arranging the transaction, paid the agency responsible for evaluating it. Major arrangers selected assessors and brought them recurring business, giving those clients influence that extended beyond any individual security. A stricter assessment could increase the protection required for a proposed transaction and reduce its commercial appeal, while the arranger retained the opportunity to approach another agency. Analytical independence had to operate within a business relationship that could reward accommodating judgments.
Explaining the failures of 2007–2008 requires examining how that pressure interacted with the substance of the assessments. The housing downturn affected a financial system in which highly leveraged institutions depended on continued access to funding, and many sophisticated participants underestimated their exposure. Ratings also conveyed useful information. Research on mortgage-backed securities found that assessments responded to differences in underlying credit risk, even as risk-adjusted standards deteriorated near the peak of issuance. That combination makes the historical problem more demanding than a general accusation that the agencies sold worthless opinions. Neither a mistaken forecast nor a conflicted payment arrangement establishes the motive behind every decision. The argument developed here is that issuer selection became consequential when transactions were constructed around rating criteria and their assessments acquired authority through investment practices and regulation. Commercial incentives could weaken scrutiny precisely where complex products required sustained attention to changing evidence. The resulting history concerns how an assessment paid for by an interested party became a basis for decisions whose consequences reached far beyond the transaction’s immediate participants.
From Selling Information to Charging Issuers
Commercial credit reporting was already an established business when bond ratings appeared. Lewis Tappan’s Mercantile Agency, founded in New York in 1841, gathered information about businesses for merchants considering sales on credit. Early subscribers could visit a reporting room and hear a clerk read an account of a prospective trading partner’s circumstances. They paid for access to knowledge collected beyond their own commercial networks, establishing a business in which investigating debtors generated revenue from the people contemplating whether to trust them.
Railroad financing brought a related informational problem into the securities market, where buyers needed to evaluate enterprises operating far beyond their communities. Henry Varnum Poor’s History of the Railroads and Canals of the United States of America, published in 1860, assembled financial and operational information that would have been expensive for individual investors to collect. Such compilations made comparison easier, although their readers still needed to interpret what the figures implied for certain investments. John Moody advanced this publishing tradition in 1909 with Moody’s Analyses of Railroad Investments, combining detailed evaluation with letter grades that summarized his conclusions. The significance of the innovation lay in making an assessment recognizable across numerous securities while retaining a larger body of explanatory material. A railroad’s commercial success alone could not establish the quality of every obligation it issued, because creditors possessed different claims against its income and property. Moody’s service offered analysis of investments as well as information about enterprises. Other publishers entered the grading business, including Poor’s in 1916 and Standard Statistics in 1922; those companies merged in 1941 to form Standard & Poor’s, while Fitch developed a competing service. Through manuals and subscriptions, these firms principally charged readers for assistance in evaluating securities, and the grades that later became their most conspicuous products initially belonged to a broader business of financial publishing.
Selling research to investors created a direct connection between the agencies’ income and the usefulness of their publications, but it did not eliminate errors in judgment. Studies of foreign government borrowing during the interwar period show that agencies frequently reacted after financial distress had become apparent, subsequently issuing extensive downgrades. The subscription era consequently supplies no reliable baseline of uniformly timely or prescient assessment. Its record establishes that forecasting difficulties existed before issuer payment became predominant, requiring historians to investigate both the financing of analysis and the separate question of how well analysts understood the obligations they examined.
The early 1970s brought a substantial alteration in the industry’s revenue arrangements. Moody’s introduced charges for corporate bond ratings in 1970, and Standard & Poor’s adopted issuer payment in 1974, leaving an interval during which the two firms financed comparable work differently. Explanations for the transition remain partly interpretive, because identifying circumstances favorable to a business decision does not establish precisely why each company made it. Improved photocopying offered one plausible influence: subscribers could circulate research to other users, allowing its benefits to spread without corresponding payments to its publisher. Charging readers alone could become an increasingly difficult way to recover the expense of producing information that many people wanted to use. Another proposed influence was Penn Central’s bankruptcy in 1970, which disrupted confidence in corporate borrowing and could make issuers more willing to purchase outside assessments of their financial standing. Borrowers also had an incentive to obtain recognizable grades when potential purchasers faced restrictions on the securities they could acquire. These developments made issuer financing commercially attractive through several routes, rather than demonstrating that one technological innovation or corporate failure caused the entire change. Agencies continued selling research and data to subscribers, but fees from the subjects of their assessments increasingly supported the production of ratings that circulated more broadly.
The new income could finance additional analysts and more extensive coverage, a benefit emphasized in Richard Sylla’s historical account. Agencies maintained that accepting payment for investigation remained compatible with preserving professional judgment, because unreliable work would damage their standing with investors. In 2003, the SEC recorded defenses invoking rating committees and established fee schedules as safeguards against inappropriate influence. Those arguments explain why participants could regard issuer financing as a workable arrangement, while leaving the effectiveness of the claimed protections to be established through evidence.
The different adoption dates permit a more direct examination of whether compensation affected grading. John (Xuefeng) Jiang and his coauthors studied corporate bonds issued between 1971 and 1978, comparing Moody’s and Standard & Poor’s assessments of the same obligations. Before Standard & Poor’s began collecting issuer fees, Moody’s generally assigned higher ratings; after the change, Standard & Poor’s grades rose sufficiently for that difference to disappear. The increases were concentrated among bonds associated with greater potential conflicts, including larger prospective fees or stronger incentives to secure favorable treatment. Using the same bonds as a basis for comparison helped distinguish the observed movement from changes in economic conditions that would affect both agencies. The findings support a connection between the payment arrangement and the resulting assessments without identifying the circumstances of every individual rating decision. Their importance for the longer history is that a measurable effect appeared in ordinary corporate borrowing during the 1970s, well before mortgage securitization reached the scale associated with the later crisis.
How Private Judgments Acquired Public Authority
Ratings had begun to influence American legal standards while the agencies were still principally selling publications to investors. Marc Flandreau and Joanna Kinga Sławatyniec have shown that courts used agency assessments as evidence of accepted financial standards of prudence before the banking regulations of the 1930s. Their research establishes an important chronology. Incorporation into public decision making grew out of an existing professional convention. During the Depression, federal banking authorities gave that convention a more explicit administrative role. The comptroller of the currency’s 1936 investment regulation prohibited national banks from purchasing securities whose characteristics were predominantly speculative, referred to recognized rating manuals, and required support from at least two manuals when eligibility was doubtful. Yet the comptroller’s contemporaneous explanation expressly denied that ratings conclusively established eligibility or relieved bank directors of responsibility. Sound unrated securities could qualify, and even a comparatively low-rated issue could be acceptable if its investment characteristics satisfied the examiner. The regulation acknowledged that smaller institutions often lacked the personnel and original information needed to investigate issuers themselves. Ratings consequently acquired official usefulness within a system that continued, in principle, to require judgment by bankers and supervisors.
The examination agreement reached by federal banking authorities in 1938 made this relationship more concrete. Its preferred category, Group I, included general market obligations in the four highest rating grades and unrated securities of equivalent value. Neither appreciation nor depreciation in that group entered the calculation of a bank’s net sound capital. Officials intended this treatment to prevent fluctuating market quotations from unnecessarily restricting credit to sound businesses. Here the attraction of ratings lay partly in their usefulness as comparatively stable classifications during a period when market prices could intensify financial distress.
A different institutional problem produced the designation that would eventually dominate American ratings regulation. In 1975, the Securities and Exchange Commission incorporated the term “nationally recognized statistical rating organization,” or NRSRO, into its revised net capital rule for broker-dealers. The immediate setting was the effort to impose more consistent capital protection after brokerage failures and the operational breakdowns of the late 1960s and early 1970s. Securities held by a dealer could lose value before they were sold, so the rule applied deductions, commonly called haircuts, when calculating the firm’s net capital. Certain highly rated debt received smaller deductions than other securities, allowing more of its market value to contribute to the regulatory calculation. A rating affected the financial resources a securities firm needed to maintain against its business. This approach drew on an established practice: the New York Stock Exchange had used ratings in members’ capital calculations since 1939. Andrew Smith and Robert E. Wright’s archival study shows that the SEC’s eventual choice emerged from a contested process in which officials considered alternatives and already possessed evidence of rating failures and payment conflicts. Commission staff had debated the usefulness of market values, and an earlier investigation of Penn Central had exposed weaknesses in assessments of its commercial paper. The adoption of NRSRO terminology reflected a particular administrative solution, adopted with knowledge of some of its limitations. Its subsequent importance exceeded the relatively specific capital calculation for which the category was first introduced.
Recognition depended largely on SEC staff correspondence rather than a comprehensive statutory registration program. Agencies sought no-action letters that provided assurance about whether their ratings could be treated as NRSRO assessments under the relevant rules. Moody’s, Standard & Poor’s, and Fitch became the initially recognized firms, while other agencies obtained recognition in subsequent years. The staff considered an applicant’s organization, resources, procedures, and independence, but acceptance among major market participants was a central criterion. That requirement placed newcomers in a difficult position. Recognition depended on an established following, while customers seeking ratings usable for regulatory purposes had a reason to prefer firms already recognized. An agency could publish assessments without securing this status, but its work would lack some of the uses available to recognized competitors. The resulting obstacle to entry could reinforce incumbent influence without requiring a law expressly prohibiting new rating businesses.
Once established, the designation spread into provisions governing activities well beyond securities dealers’ capital. The Secondary Mortgage Market Enhancement Act of 1984 created a statutory category of mortgage-related security whose qualifying conditions included a rating in one of the two highest categories from at least one NRSRO. That grade was one requirement among several concerning the underlying loans and the security’s structure, rather than a universal approval of mortgage investments. Money market regulation used ratings differently. Under the precrisis version of SEC Rule 2a-7, eligible rated instruments generally had to fall within the two highest short-term categories; qualifying unrated instruments could also be purchased if judged comparable. Fund advisers still had to make an independent determination that a security presented minimal credit risk. These distinctions matter because the familiar four-category definition of investment grade did not describe every regulatory threshold, and the applicable rules did not uniformly demand the highest possible rating. Private agreements added further conditions. Investment guidelines and contracts could specify acceptable ratings or name particular agencies, making their assessments useful even where no public rule required that exact provider. The Government Accountability Office later identified such contractual reliance as an additional difficulty for prospective competitors. These arrangements made a rating relevant to several separate decisions: whether an instrument met a statutory definition, qualified for a fund’s portfolio, or satisfied a purchaser’s contractual instructions. The commercial value of an assessment increasingly included its usefulness in meeting institutional requirements.
The growth of these uses preceded a comparable statutory system for supervising the agencies themselves. The SEC’s 2003 report described uncertainty about the extent of its authority over NRSROs, despite the established role of their ratings in its rules. Congress enacted the Credit Rating Agency Reform Act in 2006, and the resulting registration program began operating in 2007. Recognition had supplied regulatory standing for decades before this framework existed. The chronology reveals a consequential institutional imbalance. Public authorities could make extensive use of assessments while the arrangements for overseeing their production remained less clearly established.
International capital standards extended the range of applications during the years surrounding the mortgage boom. Basel II, first issued in 2004 and consolidated in 2006, permitted recognized external assessments to help determine risk weights under its standardized approach and in parts of its securitization framework. A more favorable classification could reduce the regulatory capital associated with a bank’s exposure, giving ratings a direct consequence for the treatment of its balance sheet. National supervisors determined which external assessment institutions qualified, using criteria that included objectivity and independence. Implementation proceeded at different speeds, and the American NRSRO designation was not a single worldwide authorization. In July 2007, Federal Reserve governor Randall Kroszner was still discussing the steps needed to implement Basel II in the United States; its complete framework cannot simply be projected backward onto American banks throughout the preceding expansion. Historical development was cumulative and uneven across jurisdictions. Each additional application gave financial institutions a practical reason to seek assessments that fitted an official classification, while leaving them responsible for decisions that the classification could not settle. A mistaken rating could consequently influence an asset’s regulatory treatment alongside the purchaser’s understanding of its credit quality.
From a Borrower’s Mortgage to an Investor’s Security
A residential mortgage began with a loan secured by a particular home, rather than with an instrument created for the securities market. The borrower undertook a repayment obligation whose demands depended on the interest rate and the schedule for returning principal. A loan could carry a fixed rate, or an adjustable rate that changed after an initial period; some products also postponed principal repayment. Evaluating that obligation required information about the household’s capacity to pay and the property’s value, including other debts secured against the same residence. A favorable credit score could coexist with little homeowner equity or weak documentation of income. Default also did not automatically mean losing the entire loan balance, since foreclosure proceeds might recover part of the debt after expenses. The resulting credit risk depended on the probability of nonpayment and the loss remaining after recovery, while early repayment presented a different problem for investors expecting future interest. During the expansion preceding the crisis, those characteristics were changing. Loan-level research subsequently found deterioration in subprime mortgage quality across successive years even after accounting for observable borrower and loan characteristics and economic conditions. Rising house prices had obscured some of that deterioration. The assets entering securitization required attention to the underwriting practices and verification behind their recorded characteristics.
Mortgage securitization was already an established practice when subprime lending expanded. Ginnie Mae began guaranteeing mortgage securities in 1970, and its guarantee carried the full faith and credit of the United States. Before the 2008 conservatorships, Fannie Mae and Freddie Mac supplied their own guarantees, legally distinct from Ginnie Mae’s federal backing. Private-label securities lacked those agency guarantees and relied more heavily on the collateral and contractual credit support. Their pools included prime jumbo loans and Alt-A mortgages as well as subprime credit, so “private-label” did not identify a single borrower category.
Preparing a private-label issue involved a sequence of purchases and transfers that could bring several firms into the financing of one household. An originator might obtain the borrower through a broker, make the loan, and sell it to an institution assembling mortgages for securitization. While loans accumulated, short-term warehouse financing could fund the inventory pending its sale or placement in a completed transaction. The sponsor assembled the pool, and an intermediate entity, commonly called the depositor, transferred the assets to the issuing trust. Sales could replenish the originator’s funds for further lending, but they also introduced obligations concerning the quality and eligibility of the mortgages delivered. Representations and warranties could require the seller to remedy a breach or repurchase an affected loan. Financial groups could perform several stages through affiliates and retain securities from the transaction. Compensation depended on the activity performed. An originator could earn fees and a sale margin, while a servicer received income for administering outstanding loans. The prospectus supplement for GSAMP Trust 2006-NC2, dated June 27, 2006, documents an actual arrangement of this kind. Its collateral comprised 3,949 mortgages drawn from New Century’s mortgage business, with approximately $881.5 million in scheduled principal. Goldman Sachs Mortgage Company served as sponsor, and its affiliate GS Mortgage Securities Corp. acted as depositor. The certificates offered against that collateral were divided into named classes, including senior securities carrying AAA ratings from Standard & Poor’s and Aaa ratings from Moody’s. These were assessments of specified claims created by the transaction. The individual households had neither received those grades nor become direct borrowers from the purchasers of the certificates.
The trust’s legal separation served a purpose distinct from assessing the mortgages’ economic strength. Transaction structures sought to keep the transferred assets beyond the reach of an originator’s or sponsor’s creditors if that intermediary became insolvent. This protection depended on the effectiveness of the legal arrangements and did not prevent borrowers from failing to pay. Collection also remained an ongoing task after issuance. A servicer administered the loans and dealt with delinquencies under contractual instructions. The GSAMP prospectus designated Ocwen as servicer, while other institutions performed supervisory and administrative functions. A borrower might continue sending payments to the same servicing firm even after ownership of the loan changed. Ownership and collection were separate questions, and continuity in the household’s experience could coexist with substantial changes in who ultimately received its payments. Investors depended on both the assets’ performance and the operation of the agreements that conveyed their proceeds.
Pooling and tranching performed different functions within this organization. A simple pass-through could give each investor a proportionate interest in the payments from a mortgage pool, spreading exposure across numerous loans. Tranching divided those payments and losses among classes with unequal rights. Consider a deliberately simplified pool of $100 million in loans financing $85 million of senior claims, $10 million of intermediate, or mezzanine, claims, and a $5 million first-loss position. Assume that this structure has no additional credit protection and that losses are allocated in reverse order of seniority. If the pool ultimately suffers $12 million in net principal losses, the first-loss position absorbs $5 million and the mezzanine class absorbs the remaining $7 million; the senior class loses no principal. At $20 million of losses, both subordinate positions are exhausted and the senior class bears $5 million. The borrowers’ obligations have not improved between these calculations, but the contractual allocation gives investors markedly different exposure to the same assets. Senior securities could accordingly merit stronger credit assessments than the average loan in their collateral. Actual transactions contained more elaborate provisions for distributing interest and principal, so the payment rules cannot all be reduced to a single instruction to pay senior investors first. The essential analytical task was to determine how a particular claim would perform under the transaction’s complete terms. A rating attached to that claim’s protection against loss, rather than to an improvement in the underlying households’ ability to repay.
Additional credit support complicated the calculation further. Overcollateralization meant that mortgage principal exceeded the principal of the securities it supported. Excess spread arose when income from the assets surpassed specified expenses and payments owed on the issued claims, leaving funds potentially available to absorb losses. Unlike an already funded reserve, that protection depended on income actually being earned and remaining available. Performance triggers could also redirect payments to preserve senior protection when delinquencies or losses crossed contractual thresholds. Assessors had to consider the timing of losses and repayments alongside their eventual amounts, because protection available at issuance could change during the transaction’s life.
Mezzanine mortgage securities could themselves become collateral for another issue, extending the distance between the homeowner and the eventual investor. A cash collateralized debt obligation, or CDO, acquired securities, which might include subordinate claims from numerous residential mortgage-backed transactions, and financed those purchases by issuing its own classes of debt and a residual interest. Mortgage payments first passed through the rules governing the original securities; proceeds reaching the CDO were then distributed under another set of contractual priorities. A portfolio of lower-rated mortgage claims could support highly rated senior CDO liabilities, provided that the assumed losses and the subordinate protection justified those grades. The new structure concentrated some risks as it redistributed them. Securities drawn from differently named mortgage pools could remain vulnerable to the same deterioration in housing conditions, limiting what their apparent variety accomplished. Purchases by CDOs also provided an outlet for mezzanine mortgage securities, helping arrangers complete deals whose subordinate portions needed buyers. Synthetic CDOs extended this activity without requiring the purchase of the referenced mortgage bonds. They used credit default swaps to create contractual exposure to those securities’ credit performance. Multiple contracts could reference an existing security, allowing gross exposure to expand without a corresponding increase in the number of mortgages financing homes. An investor could hold a claim whose losses depended on mortgage securities that the issuing vehicle did not own. Understanding the instrument required identifying both the reference obligations and the additional contractual arrangements through which their performance affected payment.
At the assessment stage, the reliability of the mortgage information became inseparable from interpretation of these contractual designs. Agencies received data describing individual loans and applied analytical procedures to estimate collateral performance and the protection available to particular securities. Those procedures depended on whether the reported characteristics accurately represented the assets. The SEC’s 2008 examinations found that the three agencies reviewed had publicly disclosed that they did not perform due diligence to verify the accuracy or quality of the loan data underlying the residential mortgage pools they rated during the review period. They relied on information supplied by sponsors and regarded verification as the responsibility of other participants. A detailed calculation could rest on an appraisal or income record that had not undergone independent checking by the agency performing it. The distinction matters for interpreting what an assessment established. Analysis of a transaction did not demonstrate that the underlying lending records were trustworthy. Securitization had produced transferable claims, but confidence in those claims still depended on facts generated before their elaborate contractual protections were designed.
Building Another Security from the First
The mortgage CDO boom grew from a change in collateral preferences within an established market. During the 1990s, CDO portfolios commonly contained corporate bonds or bank loans; later transactions assembled securities backed by several kinds of assets. These multisector portfolios offered the prospect of distributing exposure across businesses whose fortunes appeared sufficiently different. Disappointing performance in some sectors during the early 2000s encouraged arrangers and managers to favor residential mortgage securities, which combined substantial issuance with comparatively attractive yields and a reassuring recent record. An attempted response to earlier investment difficulties consequently helped concentrate later transactions around housing. The distinction between different CDO families is essential here. Collateralized loan obligations, or CLOs, held corporate loans, while structured-finance ABS CDOs purchased claims already produced through securitization. Even within mortgage-related CDOs, high-grade portfolios and mezzanine portfolios differed in the securities they acquired. Those descriptions referred to collateral composition, rather than guaranteeing that every security issued by the vehicle possessed the same standing. By the middle of the decade, an established financing technique was increasingly being applied to assets whose behavior depended on an expanding mortgage market.
Many mezzanine mortgage bonds occupied narrow intervals in their original pools’ allocation of losses. Once losses reached one of those intervals, a comparatively small further deterioration could consume much of the bond’s principal. The percentage lost by the underlying mortgages and the percentage lost by the mezzanine investor were different quantities. A CDO buying these bonds acquired claims with concentrated sensitivity to certain outcomes, even when the households behind them were numerous. Its analysis needed to account for that sensitivity instead of treating the collateral’s investment-grade designation as a complete description of its behavior.
Collateral selection also continued beyond the point at which many CDOs first received ratings. In a managed transaction, the collateral manager operated within an investment mandate established by the governing documents, which restricted purchases and trading. The rules could impose concentration limits and requirements concerning the quality of eligible assets. Reinvestment periods, often lasting as long as five years in the structured-finance ABS CDOs examined by Larry Cordell, Yilin Huang, and Meredith Williams, permitted the purchase of replacement assets as earlier holdings repaid principal. Mortgage prepayments made this an important feature. Without replacement purchases, the collateral supporting a deal could contract rapidly. Consequently, the year in which a CDO had been issued did not necessarily identify the origination years of all mortgages ultimately influencing its payments. A transaction established earlier in the expansion could subsequently acquire securities backed by loans made in 2006 or 2007. Management introduced another set of judgments concerning which assets to retain and how to satisfy the transaction’s conditions. Different investors could also prefer different choices. Holders of senior claims valued stronger protection, while residual investors had an interest in the income remaining after contractual payments. Coverage tests could redirect proceeds toward senior protection when specified conditions deteriorated, constraining the consequences of those competing interests. Contemporary supervisory analysis accordingly treated a manager’s experience and record of complying with covenants as relevant assessment inputs. Rating a managed vehicle involved evaluating the permitted range of future portfolios and the controls governing their selection. The assessment concerned an arrangement for making subsequent investment decisions as well as the securities initially assembled.
At Moody’s, assessment of the second transaction depended heavily on judgments already made elsewhere within the firm. The Financial Crisis Inquiry Commission found that its CDO analysts relied almost exclusively on existing mortgage-security ratings to establish probabilities of default for that collateral. Former managing director Gary Witt explained that the CDO group accepted grades supplied by the mortgage-securities group. The later calculation consequently inherited the adequacy, or inadequacy, of the earlier assessment. Witt proposed an investigation in early 2005 to examine whether assumptions used at the two levels were consistent, but disagreements over acquiring the necessary software prevented the approved project from proceeding. This episode identifies a specific institutional weakness beyond inaccurate information supplied by lenders. Repeated assessment within the same organization could carry forward an analytical error without providing an independent test of its foundation.
The treatment of diversification presented a related difficulty because mortgage securities already represented pooled exposures. An original residential mortgage transaction might contain loans drawn from several regions, reducing the importance of any single borrower or locality. Another nationally diversified mortgage pool could contain a broadly similar geographical mixture. Combining claims from both did not produce the same additional protection that combining genuinely distinct regional exposures might provide. Their remaining vulnerability could be dominated by conditions common to both portfolios, including the performance of mortgages made during the same lending expansion. Research by Joshua Coval, Jakub Jurek, and Erik Stafford explains how overlapping locations and origination periods made the assumptions used in resecuritization consequential. Errors concerning mortgage performance affected the first securities and then entered calculations governing the claims issued against them. The relevant uncertainty extended beyond the likely loss on an individual asset to the likelihood that several assets would deteriorate together. Contemporary analysis by Ingo Fender and John Kiff had already demonstrated that alternative correlation assumptions could produce substantially different assessments of a senior CDO claim. Their exercises were illustrations of model sensitivity, rather than measurements proving that every transaction was misrated. They nevertheless established that apparently modest technical choices could influence the amount of protection deemed necessary for a highly rated security. The second structure’s creditworthiness could depend heavily on assumptions whose accuracy was difficult to establish from the relatively short experience of the collateral.
Some vehicles went further by purchasing tranches issued by other CDOs. Transactions concentrated in such claims were commonly described as CDO-squared, although other ABS CDOs could also hold CDO securities. Cordell and his colleagues found that 628 of the 727 transactions in their reconstructed sample contained some CDO collateral. Recursive holdings were consequently more widespread than a separate category of conspicuously complex deals might suggest. Assessors needed to reconstruct the intervening allocations of payments and losses connecting the original assets to the final claim, since each additional contract could change how deterioration reached its holders.
Warnings about these distinctions were available before the mortgage crisis became a general financial emergency. In 2005, Fender and Janet Mitchell emphasized that a rating based on expected loss or default probability did not describe the entire distribution of possible outcomes. Two instruments with comparable assessments could expose investors to different magnitudes of loss when adverse events occurred. The broader economics of tranching also mattered. Protection against ordinary fluctuations could leave senior investors exposed chiefly to severe conditions affecting many assets together. Coval, Jurek, and Stafford’s analysis of economic catastrophe risk demonstrated how securities could retain favorable credit classifications while concentrating losses in circumstances when investors most valued dependable payments. Applied to mortgage CDOs, that insight directs attention to what would happen after several contractual protections came under pressure at once. These findings make the boundary of the rating’s claim important. A grade could summarize an agency’s estimate of credit loss while leaving substantial questions about the investor’s exposure to an economy-wide contraction unresolved. Assessing those questions required information about the collateral’s composition and its performance under severe conditions, reaching across each stage of securitization.
Choosing the Assessor: Payment and Ratings Shopping
Mortgage securitization concentrated the power to commission ratings in the hands of a relatively small group of banks. Although a trust formally issued the securities, the arranger usually managed the relationship with the agencies assessing them. A succession of legally distinct issuers could consequently produce recurring business from the same commercial source. The SEC’s 2008 examination illustrated this concentration. Twelve arrangers accounted for 80 percent of both the number and dollar volume of 642 sampled subprime mortgage securities transactions. The relevant customer relationship extended beyond any individual trust or bond. Losing an assignment could matter because it suggested losing access to later offerings from the same bank. Conversely, an agency’s willingness to accommodate an arranger could have commercial value across an entire series of transactions. The resulting dependence was strikingly consequential because investors who would ultimately bear the securities’ losses generally did not control this initial selection of the assessor.
Payment also depended on how far an engagement proceeded. In its June 2008 description of structured finance ratings, the SEC explained that an agency typically received its fee only when the rating was issued, although abandoned assignments sometimes generated a breakup fee. An agency could undertake analytical work that failed to produce the expected revenue. The incentive operated through completion. Retaining the assignment helped determine whether an assessment became a paid service. That arrangement made the choice among prospective raters commercially significant before any final opinion reached investors.
Before publication, the proposed security could undergo repeated revisions in response to an agency’s analysis. Arrangers commonly approached this process with intended grades already in mind, asking what collateral and protection would support them. The Committee on the Global Financial System’s 2005 study treated such exchanges as an integral feature of structured finance assessment, while recognizing the concerns they raised about independence. An analyst who identified insufficient protection could require changes that made a transaction safer for its senior creditors. The arranger might accept a lower grade, replace assets, or strengthen the protection available to a particular class. Explaining those requirements could serve an essential analytical purpose, and the existence of a conversation does not establish that an agency compromised its standards. Ratings shopping introduced a further choice within this process. As the SEC subsequently described it, an arranger could seek preliminary assessments from several agencies and then select one or two for an entire transaction or for particular tranches. A comparatively demanding preliminary judgment might disappear from the set of assessments accompanying the marketed securities. The arranger could respond to disagreement by improving the proposal but could also obtain the desired designation by choosing an agency whose requirements were easier to satisfy. This distinction matters because the published grade did not necessarily reveal the alternatives considered during preparation. Even a careful reader of the final rating could lack the information needed to distinguish broad agreement among potential assessors from the successful selection of a favorable opinion.
Vasiliki Skreta and Laura Veldkamp showed why this selection process could generate optimistic published ratings without requiring dishonest analysts. In their theoretical model, individual agencies produce estimates that are unbiased, yet issuers prefer to disclose favorable assessments. The ratings investors observe consequently differ from the full distribution of opinions that agencies might supply. Greater asset complexity can increase disagreement and make this opportunity more valuable to an issuer. Their argument identifies a problem in the organization of disclosure rather than establishing the motives behind any precrisis rating. It also explains why methodological diversity, potentially useful to investors, could become less informative when the party selling the security controlled which assessments entered public view.
Commercial pressure also appeared explicitly in the documentary record. The Senate Permanent Subcommittee on Investigations reproduced an exchange from May 3, 2006 concerning S&P’s proposed treatment of silent second mortgages, additional borrowing that could weaken the equity position suggested by the first mortgage alone. UBS banker Robert Morelli warned S&P manager Peter Kambeseles that a more punitive methodology might lead UBS to use Moody’s and Fitch for its CDOs instead. S&P’s Thomas Warrack acknowledged that the more conservative approach would raise credit support requirements, while expressing concern about losing substantial business. The threatened withdrawal joined an analytical dispute about mortgage risk to the agency’s prospects for future assignments. Its significance extends beyond the possibility of pressure on a single grade. Changes in general criteria could affect numerous transactions and the economics of assembling them. An agency considering stronger protection faced a customer who could compare its requirements with those of competitors and transfer business accordingly. This exchange alone does not establish the eventual model decision or prove that every objection from an arranger lacked analytical merit. It does establish that an alternative agency was available as a bargaining instrument when a proposed assessment became commercially inconvenient. The underlying question was whether S&P would defend its interpretation of the risk despite the prospect of exclusion from subsequent deals.
Internal safeguards did not consistently keep these commercial considerations outside analytical work. The SEC found that some senior analytical personnel participated in fee negotiations until policy changes during 2007, and that market share concerns reached employees responsible for rating criteria. Its report nevertheless stated that the reviewed communications did not establish that methodology or model decisions were made to attract or retain business. This qualification preserves the difference between exposure to a conflict and demonstrated distortion of a particular judgment. Organizational separation had practical weaknesses, but identifying those weaknesses requires greater precision than treating every commercially aware analyst as compromised.
Empirical research further distinguishes selecting a favorable opinion from inducing agencies to become more accommodating. John Griffin, Jordan Nickerson, and Dragon Yongjun Tang found that AAA CDO tranches rated by both Moody’s and S&P defaulted more frequently than tranches rated by only one of those agencies. That result challenged a simple shopping explanation in which selecting one favorable assessment should produce the weakest securities. Their study instead identified upward departures from model results when a competitor used more lenient assumptions, with those adjustments associated with later downgrades. Two published ratings could reflect a competitive process that reduced scrutiny, rather than two judgments insulated from one another. A separate investigation by Jie He, Jun Qian, and Philip Strahan examined the importance of issuer size in mortgage securities. During the 2004–2006 boom, securities from larger issuers carried higher initial yields than similarly rated securities from smaller issuers and subsequently experienced greater price declines. The authors interpreted these findings as evidence that investors priced the risk of more inflated ratings for important customers. Neither study provides a complete record of the intentions behind individual decisions, and observed differences require attention to transaction characteristics and competing explanations. But they show why the commercial relationship deserves investigation through rating adjustments and market outcomes as well as through threatening correspondence.
Competition among assessors could consequently reward qualities other than the rigor of their risk estimates. Patrick Bolton, Xavier Freixas, and Joel Shapiro’s theoretical analysis explains how issuer selection, agencies’ incentives to attract business, and investors’ willingness to trust published grades can combine to weaken the results of competition. In their model, favorable ratings become more likely during booms and when investors place greater trust in ratings. Applied cautiously to the mortgage market, this suggests that adding another available assessor could expand the opportunity to secure a desired opinion unless investors could also evaluate the assessments that were rejected. Independence depended partly on who controlled the comparison among agencies and what information survived that comparison. Analytical independence could be weakened even when a transaction followed an agency’s formal procedures and obtained several published opinions. Assessing it requires reconstructing the decisions that determined which judgments were published, whose requirements were revised, and what investors could learn about those choices.
The Assumptions Inside the Models
The analytical work began by translating information about existing mortgages into estimates of future performance. The SEC’s account of the agencies’ methods distinguished models of default probability, loss severity, and the cash available to meet payments on the securities. Estimating how many borrowers might fail was only part of the task; an assessment also required judgments about the losses following those failures and their effect over time. Statistical relationships derived from earlier loans supplied starting points, while forecasts of economic conditions determined how those relationships would be tested. Several approaches used simulations to generate possible outcomes, although the agencies’ techniques differed. Every such exercise depended on decisions about what the historical record could establish and what required an additional assumption. A large number of calculations could explore the consequences of a chosen framework extensively without demonstrating that the framework captured the relevant risks. The crucial analytical question concerned the relationship between the mortgages being rated and the conditions under which the model had learned to predict their performance.
Historical coverage was one limitation that an expanding database did not automatically resolve. In October 2008 testimony, former S&P mortgage ratings executive Frank Raiter recalled that the agency’s initial statistical LEVELS model, introduced in 1996, drew on approximately 500,000 loans with five years of performance data. A subsequent version expanded the sample to roughly 900,000 loans. He reported that later versions using much larger datasets were not released, attributing the failure to implement improvements to management’s emphasis on controlling expenses. His retrospective account identifies model maintenance as an organizational decision whose consequences could persist as new mortgage products entered the market.
More fundamentally, the relationship between reported characteristics and actual creditworthiness could change. Uday Rajan, Amit Seru, and Vikrant Vig investigated this problem using securitized subprime mortgages originated between 1997 and 2006. They found that loan pricing became increasingly dependent on information readily communicated to investors, while other information available to an originating lender became less influential. Borrowers with similar reported characteristics could consequently represent different risks as the lending process changed. A model estimated when securitization was less extensive systematically underestimated later defaults, particularly where less easily transmitted information mattered most. Their findings explain a source of prediction error arising upstream from the agency, as changes in lending altered the information on which an assessment relied. The problem also involved combinations of weaknesses. Federal Reserve official Roger Cole’s March 2007 testimony described increasingly common loans that combined high leverage with limited income documentation and additional mortgage borrowing. Treating each characteristic as an isolated adjustment could miss the implications of their conjunction. A household might have little equity available to absorb a setback while its reported income supplied an uncertain measure of repayment capacity. Incorporating those interactions required evidence about how the combinations performed, including combinations that had become widespread only recently.
House price projections exerted particularly extensive influence because they affected the resources available to borrowers and the value recoverable by creditors. The Financial Crisis Inquiry Commission described Moody’s M3 Prime model, introduced in 2003 for prime, jumbo, and Alt-A transactions, as simulating loan performance under 1,250 scenarios. Across those scenarios, prices increased on average by approximately 4 percent annually, with little weight assigned to a sharp nationwide decline. This finding concerns that particular model and its scenario distribution and should be understood alongside the differences among agencies’ methods. It nevertheless shows how a consequential judgment could reside in the relative importance assigned to possible outcomes rather than in their complete omission. A model might contain declining markets yet still underestimate danger if the declines were too mild, too brief, or too unlikely within its assumed distribution. Research by Kristopher Gerardi, Andreas Lehnert, Shane Sherlund, and Paul Willen helps clarify this distinction. Using information available before the collapse and examining contemporary analyst reports, they found that market participants generally understood that substantial price declines would cause severe mortgage difficulties but assigned such declines a low probability. Their analysis encompasses the broader mortgage market rather than rating agencies alone and cannot identify the assumptions of every rating decision. It does challenge an account in which mathematical sophistication simply concealed ignorance of the connection between prices and foreclosures. The historical failure included expectations about which economic conditions deserved serious attention. Knowing what a severe decline would do and judging its likelihood were separate tasks, with potentially different weaknesses.
Refinancing created another dependence on the economic construct. A borrower who repaid an existing mortgage through a new loan could appear to have completed the original obligation successfully even if continued borrowing remained essential to maintaining homeownership. Christopher Mayer, Karen Pence, and Shane Sherlund found that unconventional mortgage features, including teaser rates, were not the principal explanation for defaults through mid-2008; house price declines and weakened underwriting mattered more. Refinancing had helped some borrowers escape problematic terms, but tighter lending subsequently restricted that option. Their findings caution against attributing the crisis chiefly to scheduled payment resets. They also show why earlier loan outcomes needed interpretation alongside the availability of replacement credit.
Recovery forecasts also depended on how a delinquent loan passed through foreclosure and sale. Jay Siegel’s 2003 explanation of Moody’s Mortgage Metrics included carrying costs in its estimates of loss severity, emphasizing the interest accumulated while a defaulted loan awaited resolution. Servicers could advance money to preserve payments to certificate holders during delinquency, but reimbursement from eventual liquidation proceeds would reduce the funds remaining from the sale. The model accounted for the time required to foreclose and dispose of a property, including differences associated with state law. Notice requirements and redemption rights could affect how long the process lasted. A projection of the property’s selling price was insufficient by itself to determine the creditor’s recovery. The period between missed payments and liquidation could enlarge the shortfall even where two properties ultimately fetched comparable prices. This analysis shows how institutional details entered calculations that investors encountered principally through a final grade. It also creates a demanding standard for evaluating those calculations. Reasonable assumptions about ordinary resolution times might become unreliable when distressed properties accumulated and sales became harder to complete. Forecasting the consequences of a housing downturn required attention to the course of default resolution as well as to the decline in collateral values.
Agencies did revise their methods as mortgage lending changed. On September 6, 2006, Moody’s announced a subprime extension of Mortgage Metrics developed from approximately two million loans with ten years of performance history. Its announcement specifically emphasized the combined effects of high leverage and weaker documentation, including their interaction with newer loan products. Such statements demonstrate recognition of important problems, although an announcement establishes an analytical intention rather than successful prediction. Adam Ashcraft, Paul Goldsmith-Pinkham, and James Vickery subsequently tested ratings against mortgage characteristics and outcomes across transactions issued between 2001 and 2007. They found useful information in the ratings, alongside declining standards after adjustment for risk near the market’s peak. Deals containing more low-documentation loans also performed worse than their initial assessments implied, including earlier cohorts less affected by the crisis. These results indicate that the deficiencies cannot be explained entirely by an unforeseeable late shock. They also distinguish the presence of relevant variables from the adequacy of the protection assigned to the risks those variables represented.
Evaluating the models ultimately requires an account of how their results became final judgments. Professional discretion could correct weaknesses in a mechanical estimate, especially where unusual loan characteristics or unreliable information demanded further scrutiny. Its effectiveness depended on a reasoned explanation that another analyst could examine. The SEC’s 2008 review found that agencies did not consistently document the rationale for departures from model results, making some rating decisions difficult or impossible to reconstruct. That finding concerns the ability to evaluate the process, without establishing that every undocumented adjustment was mistaken. A rigorous assessment needed both a defensible predictive framework and an intelligible record of its application. Where that record was incomplete, subsequent observers could struggle to distinguish a considered correction from an assumption that had escaped adequate challenge.
What the Agencies Knew about the Loans
Evidence about mortgage quality was distributed among institutions that had different reasons for collecting it. An originator’s file could contain documents absent from the electronic record supplied for rating, while a purchaser’s review could identify exceptions that remained in a confidential report. Reconstructing the agencies’ knowledge requires attention to the movement of information, including the point at which it stopped. Published descriptions of analytical procedures did not always resolve that uncertainty. In its 2008 examination, the Securities and Exchange Commission found that one agency’s criteria report still described an extensive review of origination and servicing operations even though its mortgage ratings group had discontinued formal reviews of origination practices. The agency detected and corrected this discrepancy through an internal audit. Nevertheless, the episode reveals how an investor could misunderstand the investigation supporting a rating without any change in the letter grade itself. Establishing what an agency knew consequently involves examining its actual procedures alongside its public explanations.
Later scholarship demonstrates that misleading loan information had measurable consequences. Tomasz Piskorski, Amit Seru, and James Witkin found false disclosures about borrowers’ housing equity in approximately 7–14 percent of the loans they studied, with misrepresented loans defaulting at a rate about 70 percent higher than otherwise comparable mortgages. Their retrospective reconstruction does not establish that rating analysts recognized those inaccuracies when the securities were issued. It does show why erroneous descriptions of collateral cannot be treated as incidental paperwork. An assessment could be internally consistent while evaluating a substantially different exposure from the one investors actually acquired.
The records of Clayton Holdings document defects discovered before mortgage purchases were completed. Investment banks and other buyers employed the firm to examine loans offered by originators, using underwriting guidelines and permitted exceptions as the basis for review. A loan could satisfy those guidelines, depart from them with adequate compensating factors, or receive Grade 3 because its deficiencies lacked sufficient compensation. Across 911,039 loans reviewed from January 2006 through June 2007, approximately 28 percent fell into that last category. Buyers nevertheless waived in 39 percent of the Grade 3 loans, accepting them despite the recorded exceptions. These figures describe reviewed samples rather than a representative survey of American mortgages, and the standards applied differed among clients and sellers. Nor did a failed guideline test necessarily establish fraud or predict default; a purchaser might reasonably accept some exceptions in exchange for other protections. What deserves attention is the separate decision to accept an identified deficiency after the reviewer had documented it. A buyer’s willingness to purchase a loan did not itself supply an analytical reason to treat missing documentation, an unsupported valuation, or another exception as immaterial. For a rating analyst, access to the exception and the grounds for its acceptance would have allowed a more informed judgment than the purchaser’s decision alone. The Clayton evidence locates consequential knowledge within the purchasing institutions, before addressing the more difficult question of its transmission to agencies.
That transmission became a disputed issue in the Financial Crisis Inquiry Commission’s investigation. Clayton’s former chief executive, D. Keith Johnson, described discussions with rating agencies about the company’s findings and its exception-tracking capabilities. His successor, Paul Bossidy, subsequently challenged accounts implying that Clayton had delivered definitive warnings and detailed loan findings before the market deteriorated. The commission’s staff defended the witnesses’ testimony but agreed with an important limitation in Bossidy’s letter. Clayton had not supplied agencies with client-specific reports or data. The staff distinguished those confidential records from aggregate information combining results across clients. Timing mattered as well. The completed report covering reviews through June 2007 could only have been discussed afterward, and staff placed the relevant meetings in the third quarter of that year, when the subprime disruption was already underway. Earlier contacts nonetheless complicate a simple claim that agencies had no exposure to Clayton’s work until then. A contemporaneous Moody’s meeting record, cited in the staff memorandum, documented an explanation on January 11, 2007, of the due diligence process, exception tracking, and the underwriting of a loan file. Another record, dated July 27, referred to discussion of information Moody’s might not be receiving. These encounters establish opportunities to understand the review process and its informational gaps; they do not demonstrate possession of every exception affecting a particular security. The distinction materially changes the historical charge. The records support scrutiny of agencies’ willingness to obtain and use information about underwriting failures, while leaving the knowledge supporting individual ratings to be established from the relevant transaction files.
Fitch’s examination of mortgage files in late 2007 exposed another problem: a faithfully recorded number could describe an implausible borrower. Its November 28 report described a targeted review of forty-five subprime loans, emphasizing high leverage, stated documentation, and early payment difficulties. Reviewers identified apparent fraud or misrepresentation in almost every file, although this small, deliberately adverse selection could not establish prevalence across securitized mortgages. Some warning signs were detectable before funding; others emerged through servicing. Fitch distinguished transcription errors from information correctly copied from a file that misrepresented actual credit risk, including unreasonable stated income. Checking whether a database reproduces an application accurately does not establish that the application describes a credible capacity to repay. The report’s late date prevents using its discoveries as proof of Fitch’s earlier knowledge of those defects. Its broader significance lies in demonstrating how fuller records could alter the meaning of familiar indicators, and why accurate data entry provided insufficient assurance about the information behind a rating.
There is direct evidence that concerns about lending practices existed before the broad downgrades. The Senate’s investigative report reproduced an August 7, 2006, communication in which Richard Koch, an S&P employee working on servicer evaluation, described extensive appraisal and underwriting fraud. Other communications that September expressed anxiety about mortgage products and the prospect of banking distress. Public warnings also preceded the collapse. Moody’s chief executive later testified that the agency had warned about weakening practices and adjusted assumptions beginning in 2003. Such evidence makes a portrayal of unsuspecting analysts difficult to sustain, although individual employees’ concerns cannot establish company-wide knowledge of every defective loan. Recognizing unreliable underwriting created a reason to investigate how much confidence transaction assessments should place in information produced by that underwriting. Determining whether the investigation was adequate requires evidence about the procedures applied, rather than an inference that issuing a warning discharged the agency’s responsibility.
The qualifications attached to due diligence itself sharpen that inquiry. According to Clayton executive Vicki Beal, the firm generally charged for each loan reviewed, with payment independent of its findings, subsequent securitization, or loan performance. Its reviewers compared electronic information with loan files, while clients determined the scope of work and selected samples, generally comprising 10–20 percent of a pool. The firm did not authenticate the underlying documents, and its reports belonged exclusively to its clients. This arrangement could produce useful, candid findings without producing comprehensive assurance for everyone who later relied on the mortgages. Independence in grading an exception did not ensure that the sample captured the relevant defects, that supporting documents were genuine, or that a rating committee received the result. These limitations also caution against reducing the entire episode to purchased dishonesty. The surviving records show reviewers identifying problems that other institutions then had to interpret or accept. The agencies’ responsibility was to assess whether the information available could sustain their conclusions, including what remained uncertain because of restricted access or limited review. Where those conditions were weak, a defensible assessment required additional evidence, explicit allowances for uncertainty, or a narrower claim. The historical failure involved how institutions acquired, qualified, and communicated knowledge as much as whether an individual assessor recognized a troubling loan.
Inside the Agencies: When Analytical Standards Met Revenue Targets
A rating analyst worked within an organization that defined professional success as well as acceptable credit risk. Management could encourage meticulous investigation while also rewarding rapid completion and attentive service to important clients. Former Moody’s employees described a growing tension between those expectations after the company became independently traded in 2000. Eric Kolchinsky, a former managing director, recalled a shift toward commercial priorities, while other witnesses associated Brian Clarkson’s advancement with stronger attention to market share. Raymond McDaniel and Clarkson disputed this interpretation, arguing respectively that the culture had not fundamentally changed and that commercial considerations had always mattered. Their disagreement prevents treating the public listing as a demonstrated moment when professional integrity disappeared. It nevertheless directs attention to an institutional question that cannot be answered by identifying who paid the fee: what happened when an employee’s analytical demands interfered with the organization’s preferred pace of business? The surviving evidence offers several ways to examine that conflict without assuming a uniform response across agencies or departments.
The expansion of Moody’s CDO business illustrates the scale of the demands placed on staff. According to the Financial Crisis Inquiry Commission, the volume of CDOs it rated increased sevenfold between 2002 and 2006, while staffing rose only 24 percent. Annual revenue from that business climbed from $12 million in 2003 to $91 million in 2006. The imbalance was not identical throughout the company. Staffing for residential mortgage securities approximately doubled alongside growth in that business. These differences make departmental capacity more informative than a general claim that every agency analyst faced the same workload.
S&P communications show why completed transactions could misrepresent departmental capacity. In December 2004, Gail McDermott reported that her staffing calculation assumed sixty-hour working weeks and still identified insufficient resources. She also described departures and employee health problems, challenging the inference that getting work finished demonstrated sustainable capacity. An October 31, 2006, message from the head of the CDO ratings group made the conflict still clearer. Satisfactory revenue and client-service figures coexisted with concerns about understaffing and declining analytical quality. These complaints preceded the broad collapse in mortgage ratings, giving them evidentiary weight beyond later recollections of an unsuccessful system. They also distinguish the appearance of productivity from the conditions under which judgment was exercised. A department can continue delivering ratings while reducing the time available to investigate unusual provisions or reconsider assumptions. If managers assess performance principally through completed business, deterioration in those less visible activities may remain poorly measured until losses make it undeniable.
Richard Michalek’s testimony supplies a detailed account of how such pressure could alter the substance of review. A lawyer working on Moody’s structured transactions, Michalek described increasing reliance on comparisons with previously approved documents as caseloads expanded. Reviewing changes against a familiar contract could conserve time without sacrificing quality, provided the earlier document supplied a sound standard. In his account, investment banks and their counsel increasingly selected the precedent, sometimes choosing documents that another reviewer had examined less thoroughly. He illustrated the consequences through provisions governing removal of a collateral manager. A requirement to establish gross negligence could replace a stronger provision allowing removal for ordinary negligence. The disputed wording affected investors’ ability to respond to managerial failings, making its significance greater than a stylistic preference in drafting. Once the weaker wording appeared in an approved transaction, a demanding reviewer had to explain why an apparently similar new deal deserved different treatment. Michalek also recalled performance evaluations that praised his attention to detail while criticizing insufficient attention to the continuing relationship with bankers and managers. These are a participant’s recollections, and they cannot establish identical practices throughout Moody’s or measure how much resulting losses reflected documentation. Their value lies in describing a recognizable mechanism through which standards could deteriorate incrementally. An earlier concession acquired authority as precedent, and the employee challenging its repetition bore the burden of justifying additional work. Under those conditions, apparently routine decisions about documentation could accumulate into substantive changes without a single instruction to issue an unjustified rating.
Senior management’s own reflections suggest that influence also operated through ordinary persuasion. An October 2007 presentation prepared for Moody’s board, examined during the House’s 2008 hearing, acknowledged that competition could penalize ratings quality and that safeguards against lowering standards did not fully resolve the problem. Its discussion of “Rating Erosion by Persuasion” recognized that arguments from market participants could improve judgment in some circumstances and distort it in others. This admission described something more complicated than an analyst knowingly exchanging a favorable opinion for a fee. Repeated interaction could make a client’s interpretation seem reasonable, particularly when an analyst had limited time to develop an alternative and when accommodating the argument helped complete a transaction. The presentation came after the disruption had begun, so it should be read as an internal effort to explain an emerging failure rather than a precise warning delivered before the boom. Even with that chronological qualification, it shows executives considering how judgment could weaken despite formal protective arrangements. A rule separating fee negotiation from analysis could address one channel of influence while leaving the analyst exposed to persistent arguments about what counted as acceptable risk.
Compensation evidence places an important limit on this interpretation. The SEC’s 2008 examination found policies prohibiting analysts from being paid or evaluated according to revenue from the issuers or issues they rated, and investigators found no indications that compensation violated those policies. Salaries generally reflected experience and seniority, while bonuses combined individual performance with the firm’s overall success. The record does not justify assuming that an analyst received a direct financial reward for each inflated grade. Organizational incentives could still operate through workloads and expectations about cooperation, but those mechanisms require evidence of their own rather than substitution of the company’s financial interest for an employee’s demonstrated motive.
Donald MacKenzie’s research adds a further complication by examining how expertise was divided inside rating organizations. Mortgage securities and CDOs became the responsibility of separate specialist groups, whose established methods shaped how new instruments were assigned for evaluation. In one interview account, CDO specialists at an agency undertook an exploratory calculation that suggested substantially greater dependence among securities than conventional assumptions allowed. They did not formally develop or circulate the result, partly because the relevant products fell outside their responsibility and partly because they considered their calculation rudimentary. This account does not document a management order suppressing a proven finding. It identifies a different organizational weakness. Expertise capable of raising a consequential objection did not necessarily carry authority to require its consideration. Specialization ordinarily improves professional work by assigning tasks to people with relevant skills, but its benefits depend on arrangements for examining problems that cross departmental boundaries. Otherwise, an analyst can conscientiously complete an assigned task while concerns arising elsewhere remain outside the decision process. Read alongside the staffing and performance evidence, MacKenzie’s interpretation suggests that providing more employees would not by itself have corrected every failing. Effective review also required a recognized route through which an uncertain but serious challenge could become a shared analytical problem.
An employee’s objection could become a reason to revise a transaction or an obstacle to meeting a deadline, depending on the response it received. Managers controlled assignments and resources, and their willingness to support further investigation helped determine which interpretation prevailed. The central test of internal independence was whether a concern could delay approval and receive substantive consideration without being dismissed as poor client service. Where completed business counted more clearly than the quality of unresolved judgments, formal commitments to rigor left employees facing incompatible demands. The resulting responsibility extends beyond the individual who signed off on a rating to the organization that determined how much investigation was practical and which objections deserved institutional support.
Why Investors Wanted the Ratings
For an institution deciding which securities to buy, an agency supplied services that were difficult to reproduce transaction by transaction. Its published analysis could help a purchaser evaluate unfamiliar collateral and compare an offering with other investments. The Committee on the Global Financial System’s 2005 report recorded these advantages in interviews conducted before the crisis, when market participants were still explaining why they valued ratings. Investors cited the agencies’ experience across transactions and jurisdictions, their knowledge of legal structures, and their continuing surveillance of outstanding securities. Some valued an external assessment even when they maintained their own models. This was a substantive demand for expertise, rather than evidence that every purchaser lacked analytical ability. Hiring internal specialists to duplicate all these functions would involve costs, and obtaining a second opinion could remain useful after that expenditure. The report also recorded a tradeoff between the expense of detailed cash-flow analysis and the modest additional yield available on certain highly rated investments. Reliance could increase precisely where investors expected the financial reward from further investigation to be small.
Higher returns gave those calculations immediate commercial significance. The International Monetary Fund observed in 2008 that structured credit generally offered wider spreads than conventional securities carrying similar ratings, indicating that prices reflected risks beyond default, including liquidity and market risk. For a buyer willing to accept those exposures, the combination of an approved credit grade and additional income could appear attractive. The yield difference nevertheless cannot establish how much a purchaser understood about hidden mortgage weaknesses. Recognizing that an instrument was less liquid did not mean anticipating the losses that defective underwriting and falling property prices would eventually produce.
Investment mandates introduced a separate reason to want a rating, rooted in the relationship between asset owners and the professionals managing their money. A pension trustee or fund shareholder could not continuously inspect every transaction yet needed an intelligible way to limit the manager’s discretion. Research by Ramin P. Baghai, Bo Becker, and Stefan Pitschner explains how ratings served that purpose by making portfolio restrictions explicit and allowing owners to check whether managers followed them. Their study of investment documents, including U.S. records extending back to 1999, distinguishes mandates specifying certain agencies from those defining grade thresholds, allocation limits, or permitted exceptions. Such provisions could influence a purchase even when the manager possessed a more elaborate private assessment. The rating helped establish whether the investment fitted the agreed instructions, giving the asset owner a means to monitor delegated decisions. This governance function also created a potential discrepancy between satisfying the mandate and delivering the protection its authors intended. A security offering additional yield within the permitted category could satisfy the contract’s express limits while carrying risks that the category described poorly. Neither a knowledgeable manager nor a conscientious trustee had to regard the grade as an exhaustive account for that discrepancy to matter. Pressure to secure an allocation could further shorten the period available to resolve it. The IMF reported that, during intense demand in late 2005 and early 2006, issuers compressed the interval between releasing a prospectus and opening subscriptions, and investors purchased on the strength of ratings without fully investigating underlying risks. In that setting, the available classification could support a prompt decision while a more demanding examination remained unfinished. The institutional usefulness of the rating helped explain its influence beyond any particular purchaser’s confidence in the agency’s forecasts.
Prices provide an important check on the claim that investors surrendered all judgment to the agencies. Thomas Mählmann examined 3,254 floating-rate tranches from 617 asset-backed CDOs issued between 2000 and 2007, asking whether initial yield spreads contained information about subsequent performance beyond the ratings themselves. Across the sample, they did. Securities with different prices could later perform differently despite their placement within the same rating categories. This finding indicates that the market incorporated information not fully captured by the published grades. But the result was uneven. Predictive information in spreads came principally from tranches below AAA and, to a lesser extent, from the lowest-priority AAA tranches; spreads on the highest-priority AAA claims provided no comparable predictive information. The usefulness of spreads also weakened for later issues and transactions with more complex collateral. Mählmann interprets these patterns as evidence of diminished due diligence, especially in highly rated, complicated investments near the end of the boom. Aggregate pricing cannot reconstruct the beliefs or effort of every buyer, but it makes a uniform portrait of professional investors untenable. Some exercised judgment beyond the rating, while the apparent protection associated with the highest classifications coincided with weaker evidence of independent discrimination. Understanding demand requires attention to what investors bought and where they stood within the transaction, as well as whether their organization qualified as sophisticated.
Prominent purchasers also challenged the agencies’ accommodating treatment of issuers. A Moody’s email dated July 11, 2007, discussed in the subsequent House investigation, reported complaints from PIMCO and Vanguard about methodologies and deteriorating standards. The PIMCO representative described earlier meetings questioning the agency’s assumptions, while Vanguard’s representative said concerns had emerged approximately eighteen months before the email. These reported conversations do not establish how either institution positioned every portfolio or whether its warnings accurately anticipated subsequent losses. They do establish that investor demand included pressure for more demanding assessments, complicating the suggestion that buyers universally preferred favorable grades regardless of quality. Purchasing institutions could want higher income and still object to an agency’s failure to resist arrangers. The difficulty was converting that objection into dependable discipline over assessments that remained useful for conducting investment business. An agency’s commercial success could consequently coexist with criticism from some of the institutions its opinions were supposed to serve.
The Warnings before the Downgrades
By January 2007, the agencies were confronting evidence that recently securitized mortgages were failing unusually early. Moody’s January 18 report, Early Defaults Rise in Mortgage Securitizations, compared subprime pools at the same six-month stage of their development. The proportion of collateral in foreclosure, lender-owned property, or realized loss had risen from 0.84 percent for first-quarter 2005 securitizations to 2.59 percent for second-quarter 2006 transactions. That combined measure described advancing distress rather than losses already charged against investors’ principal, but the comparison made the deterioration difficult to attribute simply to the normal aging of loans. The report nevertheless entertained two explanations with different implications for ratings. Unusually weak borrowers might be defaulting first, leaving stronger borrowers behind, or the early failures might reveal a generally inferior pool whose difficulties would persist. It also reported that repurchase demands were straining mortgage originators, some of which had already entered bankruptcy. A published warning contained both adverse evidence and a possible explanation under which ultimate losses might remain manageable.
Selective responses preceded the extensive rating revisions of the summer. In his 2010 Senate testimony, Peter D’Erchia, formerly responsible for S&P’s structured finance surveillance, said that the agency had begun monthly reviews of recent mortgage-security vintages late in 2006. He described a February 2007 change permitting CreditWatch placements before significant realized losses appeared, while defending the earlier caution on the grounds that delinquent loans did not invariably produce losses. His retrospective account helps identify the disputed issue. S&P acknowledged extraordinary deterioration and reported taking action, but the adequacy of that response depended on how quickly its reviews extended across the affected securities.
Trading supplied another warning during the opening months of 2007. The ABX.HE indices tracked credit default swaps referencing baskets of subprime mortgage securities, allowing participants to trade protection without purchasing the bonds. Ingo Fender and Martin Scheicher documented an initial spread increase early that year, followed by much greater disruption beginning in June. Their evidence also complicates any account of a unanimous market verdict. The AAA portions of the first two index series remained near par on June 1, while lower-rated segments had weakened. Liquidity and investors’ willingness to bear risk affected quotations, especially as the crisis developed, preventing prices from serving as direct measurements of expected credit losses. The significance of the early repricing was the appearance of observable disagreement with existing assessments, rather than proof that every outstanding rating required an immediate reduction. That disagreement provided a reason to investigate deteriorating collateral even where an index quotation could not establish the appropriate grade for a particular security. Regulatory intervention added a more concrete warning on March 7, when the Federal Deposit Insurance Corporation ordered Fremont Investment & Loan to cease unsafe practices that included inadequate income verification and qualifying adjustable-rate borrowers using introductory payments without adequately assessing their ability to pay the fully indexed rate. Fremont consented without admitting or denying the allegations. The order addressed a lender’s conduct rather than assigning grades to its securities, but it made shortcomings in mortgage underwriting a matter of public supervisory action.
Moody’s publications during March and April reveal how acknowledgment of worsening performance coexisted with confidence in much of the rated debt. As the Financial Crisis Inquiry Commission’s preliminary staff report later documented, the agency’s March 7 assessment maintained that considerable further deterioration would be necessary before most mortgage bonds rated A or higher faced losses. A separate March 23 study examined the consequences for structured finance CDOs through hypothetical changes in their mortgage-security collateral. For structures with higher subprime concentrations, the exercises produced potential downgrades of ten or more rating notches, while effects at lower concentrations were generally less severe. The report expressly distinguished these illustrative scenarios from expectations about future performance, and its use of a generic structure limited what it could establish about an individual transaction. These qualifications matter because knowledge that a damaging outcome is possible differs from a judgment that it has become sufficiently probable to require a rating change. By April 20, Moody’s was generally projecting cumulative losses of 6 to 8 percent on loans backing 2006 subprime securitizations, with stronger and weaker pools potentially falling outside that range. It still anticipated few material downgrades among bonds rated A or higher unless losses substantially exceeded those expectations. These assessments locate the disagreement in the probability assigned to severe deterioration and in the confidence placed in the protection available to different securities. The agency had publicly demonstrated that it could analyze a destructive scenario, yet its central expectations continued to support comparatively limited revisions. The interpretive problem is how analysts decided when incoming evidence justified moving from a sensitivity exercise to a changed forecast, particularly when several successive reports already described worsening conditions. An adverse scenario could inform readers about vulnerability while leaving the ratings intact; only a reassessment of its likelihood and consequences would alter the agency’s operative judgment.
Reassurance also came from institutions outside the ratings business. Speaking in Chicago on May 17, 2007, Federal Reserve chairman Ben Bernanke recognized weakened underwriting and expected delinquencies and foreclosures to increase further during that year and the next. He nevertheless anticipated limited effects on the wider housing market and no significant transmission to the rest of the economy or financial system. His assessment establishes that expectations of a bounded subprime disturbance extended beyond firms paid to rate securities, complicating any explanation that attributes contemporary optimism entirely to issuer influence. But its relevance has limits. A forecast about aggregate economic effects could not establish the safety of a particular mortgage bond, and a ratings agency remained responsible for investigating the obligations on which it had issued an opinion.
The practical difficulty lay in converting accumulated warnings into systematic reviews of outstanding ratings. The SEC’s July 2008 examination report found weaknesses in surveillance and reproduced an internal communication from February 3, 2007, in which a senior manager described a request to take earlier action on poorly performing transactions while reporting insufficient resources for the existing workload. At one examined agency, a July 2005 internal explanation indicated that revised assumptions generally prompted a fresh review only after a transaction had been flagged for performance problems; staff and modeling limitations hindered broader reconsideration. These findings point to a specific obstacle beyond disagreement over the economic outlook. Newly recognized concerns did not automatically reach every security assessed under earlier assumptions. They also require care, since deficiencies documented at certain agencies cannot establish an identical sequence of delay throughout the industry. The contemporary reports and subsequent investigations support a more demanding account of the months before the mass downgrades. Warning signs had acquired numerical, market, and public institutional forms, but uncertainty about ultimate losses allowed cautious revisions and reassuring forecasts to continue alongside them. Assessing the agencies’ conduct requires asking when their reasons for retaining a rating ceased to fit the available evidence, and whether their monitoring arrangements could detect that change promptly enough to make the published judgment useful.
From Failed Assessments to a Financial Panic
The downgrades of 2007–2008 affected securities already incorporated into financial institutions’ investment portfolios and financing arrangements. Efraim Benmelech and Jennifer Dlugosz’s study of Moody’s structured finance ratings recorded 8,109 downgrade actions in 2007 and another 36,880 during 2008 through September 22. These totals count actions rather than distinct securities, since a single tranche could be downgraded repeatedly. The average reduction also grew more severe, from 2.5 rating notches in 2006 to 4.7 in 2007 and 5.6 in the available 2008 observations. Large revisions made it harder to treat the original classifications as dependable estimates that required only occasional adjustment. Institutions had to reconsider positions acquired under substantially different assessments of their creditworthiness, often while related holdings elsewhere were being questioned. The significance of this reassessment depended on the obligations built around those assets. An investor able to wait for payments faced a different predicament from one whose creditors could demand cash immediately.
A lower rating could change the conditions under which an institution held or financed a security. Investment mandates might require portfolio adjustments, while agreements incorporating credit thresholds could demand additional protection or restrict continued financing. The effect depended on the applicable rules and contracts, so a downgrade did not compel every owner to sell immediately. Nevertheless, simultaneous reassessment across a substantial class of investments could concentrate the need to reduce exposure precisely when fewer buyers were willing or able to acquire it.
The breakdown in valuation became publicly conspicuous in Europe during August 2007. On August 9, BNP Paribas Investment Partners announced that it had suspended valuation and subscriptions and redemptions in three funds, with the suspensions effective August 7. Its explanation was that liquidity had disappeared in parts of the American securitization market, making fair valuation of the funds’ underlying asset-backed securities impossible irrespective of their quality or credit rating. The announcement did not establish that every security in those portfolios would default. It demonstrated that a manager could no longer calculate a reliable fund value from assets whose ratings had previously helped make them acceptable investments. A grade still attached to a bond could not supply the transaction prices needed to settle withdrawals fairly among investors. The Financial Stability Forum subsequently observed that some buyers, having depended heavily on agency assessments, lacked independent means of evaluating structured products once those assessments became suspect. That institutional gap helped explain why concern about some mortgage exposures could inhibit trading in a wider range of securities. Establishing a defensible price now required information and analytical capacity that could not be assembled as quickly as demands for redemption arrived. The BNP episode consequently illustrates a change in the practical meaning of uncertainty. Doubts that might once have prompted a discount could instead interrupt the process through which investors expected to recover their money. Confidence in ratings had supported transactions, but its loss exposed how difficult those transactions became when participants could no longer agree on valuation.
Asset-backed commercial paper made the interruption more dangerous because financing had to be renewed frequently. Programs often financed longer-lived assets with short-term debt, making continued access to purchasers essential. Daniel Covitz, Nellie Liang, and Gustavo A. Suarez found that one-third of programs experienced runs within weeks of the turmoil’s onset in 2007. Their definition captured programs that stopped issuing despite substantial amounts of existing paper coming due. Exposure to subprime mortgages and weaker liquidity support helped explain which programs lost financing, but the early retreat also contained an indiscriminate component that could not be attributed solely to individual programs’ characteristics. Investors could protect themselves by declining to renew rather than establishing the exact losses a portfolio would ultimately suffer. As promised bank support was called upon, pressure returned to sponsoring institutions’ balance sheets. So declining confidence in securitized assets affected both the vehicles holding them and the banks expected to provide funds when private purchasers withdrew.
Secured borrowing introduced another route from asset deterioration to financial contraction. In a repurchase agreement, securities serve as collateral for short-term financing; the borrower must fund the difference between their value and the amount the lender advances. Markus K. Brunnermeier explained how falling prices and increased collateral discounts could force leveraged holders to reduce positions, with their sales depressing prices and creating further financing pressure. A rating reduction could intensify doubts about collateral, although neither a formal downgrade nor a universal contractual ratings trigger was necessary for this process to operate. The historical evidence also rules out treating the repo market as a single mechanism that behaved identically everywhere. Adam Copeland, Antoine Martin, and Michael Walker found that margins and funding in the tri-party segment remained surprisingly stable for most borrowers during the crisis, while Lehman Brothers suffered a sharp funding decline in September 2008. Variations of this kind matter when explaining how doubts about mortgage securities became funding pressure. The form of borrowing helped determine which creditors could withdraw and how much warning a borrower received. Dependence on continued financing created vulnerability, but its consequences differed among institutions and market segments.
AIG provides a particularly clear case in which a credit judgment acquired an immediate contractual consequence. Its Financial Products business had sold credit default protection on CDO obligations, and collateral provisions could respond to changes in reference securities’ values or ratings as well as to AIG’s own credit standing. Those distinctions are essential to the sequence of events. The Congressional Oversight Panel later emphasized that the initial collateral calls beginning in June 2007 principally reflected falling values of the reference obligations, rather than ratings triggers. By the following year, losses and cash demands associated with securities lending were adding to the pressure on the company. AIG entered September 2008 with serious liquidity difficulties already in progress. On September 15, S&P, Moody’s, and Fitch lowered its long-term debt ratings, activating further requirements under its financial agreements. According to the Government Accountability Office, AIG Financial Products estimated that additional collateral demands and transaction termination payments would require more than $20 billion within a short period. This figure described an estimated financing requirement, including payments arising from contract termination, rather than a tally of mortgage losses or collateral already delivered. The downgrade intensified a shortage of immediately available funds and helped bring the company to the emergency government assistance arranged the next day. An assessment intended to warn counterparties about weakening creditworthiness had also accelerated the demands through which that weakness became harder to survive. Individual creditors could reasonably seek greater protection while their combined requests placed an already strained institution under still greater pressure. AIG’s experience does not make the revised rating unjustified; it demonstrates the instability created when confidence in an institution and its ability to meet current obligations were connected through agreements that required more cash as its standing declined.
Lehman’s bankruptcy extended the panic through exposures that were not confined to mortgage securities. The Reserve Primary Fund held $785 million in Lehman-issued securities and became unable to meet redemption requests on September 15, 2008. On September 16, its net asset value fell below one dollar per share, an event commonly described as breaking the buck. The loss showed that an investment treated as a place to keep readily available cash could transmit the failure of a major financial institution directly to its shareholders. Here the immediate disturbance involved Lehman debt and demands for redemption, demonstrating why the September panic cannot be reconstructed as a succession of mortgage-bond downgrades alone. The collapse of confidence had reached institutions whose usefulness depended on investors expecting both capital preservation and prompt access to funds.
Later research clarifies why the withdrawal of confidence could become self-reinforcing even among investors holding claims on the same portfolio. Lawrence Schmidt, Allan Timmermann, and Russ Wermers found that money market fund flows during the week of Lehman’s failure were consistent with coordination under incomplete information, including investors’ reactions to the anticipated behavior of others. The implication is that a decision to withdraw could reflect concern about subsequent redemptions as well as a private estimate of the assets’ eventual losses. That dynamic distinguishes a financial panic from the orderly recognition of disappointing mortgage performance. Ratings failures contributed to the disturbance by undermining an established basis for judging securities, while certain rating changes intensified it where contracts attached immediate financing consequences to the new grades. Short-term funding arrangements and the ability of creditors to leave supplied additional mechanisms through which those shocks became dangerous. Correcting an inaccurate assessment was necessary, but correction could occur within institutions already organized around assumptions that no longer held. The resulting crisis exposed the accumulated consequences of earlier judgments in a setting where waiting for clearer evidence was itself difficult. Obligations matured, counterparties sought protection, and some investors expected to withdraw before others. Responsibility for that vulnerability consequently extends from the production of credit assessments to the financial arrangements that made their sudden reversal so disruptive.
Who Bore the Losses? The Unequal Recovery
Financial stabilization and household recovery proceeded on different schedules. An institution could regain access to funding while a family continued to struggle with obligations incurred before the collapse. Rising asset prices could restore the value of an investment portfolio without replacing a lost job or returning a foreclosed home to its former owner. These differences matter when evaluating the consequences of the failures examined here, because the disappearance of acute panic supplied only one measure of improvement. Households entered the downturn with unequal resources, and their remaining resources influenced whether they could participate in the subsequent expansion. A family able to retain its assets could benefit when prices recovered; another that had exhausted its savings faced the task of rebuilding its position. The distribution of damage depended on the losses sustained during the crash and on what people were left able to do afterward.
Large dollar losses among affluent households should not be confused with the greatest loss of economic security. Fabian T. Pfeffer, Sheldon Danziger, and Robert F. Schoeni used the Panel Study of Income Dynamics to follow families through the downturn and early recovery. Their study found that wealthier families suffered larger absolute losses, while losses measured as percentages of previous wealth were greater among less advantaged households. Between 2007 and 2011, approximately one-quarter of families lost at least three-quarters of their wealth, and more than half lost at least one-quarter. Following particular families made this evidence especially useful, since changes in the composition of income or wealth categories can otherwise complicate comparisons between survey years. The results establish that vulnerable households suffered severely during the initial destruction of wealth, rather than encountering hardship only after richer households had recovered. They also show why national totals cannot describe the practical burden of a financial loss. A substantial reduction in a large fortune could leave its owner with considerable reserves, while a smaller dollar decline could eliminate another family’s protection against an interruption in earnings. For households with very little wealth to begin with, even proportional measures have limits. The disappearance of modest savings may be more consequential for daily security than its contribution to aggregate losses suggests. The central distinction concerns the resources remaining available after a setback, as well as the monetary size of the setback itself.
Mortgage debt amplified the consequences of declining house prices. A hypothetical homeowner with a house worth $200,000 and a mortgage balance of $160,000 would initially possess $40,000 in home equity. Holding the debt constant for illustration, a 20 percent decline in the property’s value would erase that equity completely. The house would have lost one-fifth of its value, but the owner would have lost the entire stake in it. This arithmetic helps explain why a family relying heavily on a mortgaged home could experience a much larger proportional loss than the movement in property prices alone suggested.
Damage to household finances also affected people beyond the borrowers whose property values had fallen. Atif Mian, Kamalesh Rao, and Amir Sufi examined the geographically uneven housing collapse between 2006 and 2009 and its consequences for consumption. They found larger spending responses to housing losses in areas with poorer or more heavily indebted households, alongside greater constraints on access to credit. Their evidence concerns patterns across local areas, rather than an identical response by every household within them. Nevertheless, it demonstrates why the location and distribution of losses mattered for the wider downturn. When affected households reduced purchases, businesses serving those communities faced weaker demand, creating additional pressure on employment. Someone who had avoided an unsuitable mortgage could consequently suffer through the deterioration of the local economy. The harm spread through ordinary spending and work, extending the crisis beyond the ownership of mortgage securities or participation in speculative borrowing.
Employment could remain damaged even when familiar indicators suggested recovery. Danny Yagan’s analysis of longitudinal tax records connected individuals’ later employment to the severity of the downturn in the places where they had lived before it. He estimated that exposure to a local unemployment increase one percentage point larger during 2007–2009 reduced the probability of employment in 2015 by more than 0.3 percentage points. Effects were greater among previously lower-earning workers and older members of the sample. This finding does not establish that every displaced worker suffered permanent exclusion, but it identifies consequences lasting well beyond the financial emergency. The distinction between unemployment and employment is crucial. People who stop seeking work are no longer counted among the unemployed. A declining unemployment rate can coexist with continued losses in participation and earnings. For families depending primarily on wages, recovery required renewed access to remunerative work, not simply higher prices for assets they possessed in limited quantities.
Possession of assets shaped access to the rebound. In a 2019 assessment, Federal Reserve governor Lael Brainard noted that higher-income households’ wealth began recovering when financial markets stabilized in early 2009, whereas middle-income households’ recovery began in 2010 as unemployment declined. Research by Lisa Dettling, Joanne Hsu, and Elizabeth Llanes provides a longer comparison using working-age households grouped by their usual income. By 2016, average inflation-adjusted wealth for the highest-income tenth exceeded its 2007 level, while the averages for all three groups comprising the remaining 90 percent stayed below theirs. The categories describe distributions at successive dates; they do not mean that every family followed its group’s average path. Declining homeownership and reduced stock-market participation among lower- and middle-income families limited their ability to benefit from rising prices. The reduction in homeownership also reflected fewer families entering ownership, so the explanation extends beyond the experience of those who lost existing homes. Less access to employer-sponsored retirement plans restricted another route into financial assets. Rebounding markets could not restore wealth to families who no longer owned the relevant assets or had been unable to acquire them. Edward N. Wolff’s separate analysis reinforces the distinction between overall improvement and the position of a typical household. Using a measure of marketable net worth that excludes vehicles and defined-benefit pension entitlements, he found that real median wealth in 2016 remained 34.3 percent below its 2007 level, despite recovery in the mean. Restoration of an average consequently offered insufficient evidence that the distribution of opportunities to rebuild had recovered.
Class categories also conceal substantial racial differences. Fenaba R. Addo and William A. Darity Jr. examined wealth among occupationally defined working-class households during the recovery from 2010 to 2019. They found that fewer Black working-class households met their wealth-based threshold of middle-class security than white working-class households. Professional status likewise did not eliminate racial disparities in accumulated resources. Their comparisons show why occupational position or current earnings cannot substitute for an examination of wealth. Families with similar jobs could possess very different reserves, affecting their ability to withstand losses and benefit from an expansion.
These findings change the scale on which the consequences of inadequate ratings must be considered. The studies do not isolate the share of unequal household losses attributable to the agencies, and they cannot turn every foreclosure or employment setback into evidence of a particular rating error. They do establish that the financial collapse had consequences extending far beyond the investors who purchased the assessed securities. Favorable grades supported the financing arrangements examined earlier, while the subsequent contraction reached households with little influence over those arrangements and limited means of absorbing their failure. The ability to survive a loss also influenced who retained access to the assets whose recovery later increased wealth. This interaction gave the crisis a lasting distributive significance. Initial damage could weaken the capacity to benefit from improvement. Affluent households were exposed to substantial losses, but their experience cannot stand for that of families whose financial reserves or attachment to employment were more severely disrupted. An account ending with the stabilization of major institutions would consequently leave important consequences unresolved. Assessing responsibility requires attention to the prolonged insecurity that remained after financial markets had resumed functioning.
Investigating the Raters
The agencies became subjects of intensive federal examination while the financial crisis was still unfolding. On August 31, 2007, SEC staff initiated examinations of Fitch, Moody’s, and S&P, reviewing activities extending back to January 2004. More than fifty staff members participated, interviewing agency personnel and examining deal files, internal policies, and over two million emails and instant messages. Their July 2008 report opened aspects of the ratings process that outside observers could not reconstruct from published grades alone. Its presentation nevertheless imposed an important limit on identification: because examinations of individual firms were nonpublic, most findings did not name the agency concerned. The companies also received opportunities to explain their documents and sometimes disputed the staff’s interpretations. The resulting record supported scrutiny of organizational practices while requiring readers to distinguish documented deficiencies from assumptions about which company committed them.
The investigation encountered a gap between the period it reviewed and the period covered by the new NRSRO rules. The three firms became subject to those requirements upon registration in September 2007, although examiners studied conduct dating from the preceding boom. Congress had also prohibited the SEC from regulating the substance of ratings or the procedures and methodologies used to determine them, while permitting examination and oversight of specified practices. Consequently, retrospective criticism of an agency’s conduct could establish serious supervisory concern without automatically demonstrating a violation of registration requirements that had not yet applied.
Congressional hearings made the disagreements more visible by placing witnesses with different relationships to the ratings business before the same committee. At the House Committee on Oversight and Government Reform’s October 22, 2008, hearing, Fons and Raiter appeared alongside Sean Egan, whose competing agency operated on a subscriber-paid basis. A later panel brought together the leaders of the three dominant firms. Former employees could describe internal changes over time, while serving executives defended organizations whose reputations and future business were at stake. Egan supplied an alternative commercial perspective, but his position as a competitor also mattered when assessing his testimony. These circumstances made corroboration through dated records particularly valuable. Stephen W. Joynt acknowledged that Fitch’s ratings had failed to capture the full risk in many mortgage-backed securities and CDOs yet asked the committee to judge his company’s intentions from its own conduct rather than from striking emails associated with other firms. His request identified a legitimate evidentiary distinction even while leaving Fitch’s analytical failures open to criticism. The hearing’s importance extended beyond its most memorable exchanges. It assembled admissions and competing explanations in a setting where investigators could compare claims about professional standards with material generated inside the agencies before their public defenses were formulated.
Private litigation tested a different question: whether describing a rating as an opinion necessarily prevented a fraud claim. In Abu Dhabi Commercial Bank v. Morgan Stanley & Co., a federal district court in 2009 allowed claims involving ratings of the Cheyne structured investment vehicle to proceed. Judge Shira Scheindlin rejected the agencies’ First Amendment argument in the circumstances alleged, distinguishing ratings distributed to selected private-placement investors from communications to the public at large. She also concluded that the plaintiffs had sufficiently alleged that the agencies did not genuinely or reasonably believe their ratings had factual support. This was a decision about the adequacy of the pleadings, not a finding that fraud had been proved. Its significance was the conditional nature of legal protection. The description of an assessment as an opinion did not, in itself, dispose of questions about its basis or the circumstances in which it was communicated.
Investigations during 2010 and 2011 enlarged the documentary record and gave different institutions opportunities to reconstruct the decisions behind the grades. The FCIC’s June 2, 2010, hearing organized separate sessions around the ratings process, the agencies’ place in the crisis, and their business model. Its archive preserved interview material and committee documents alongside testimony, with follow-up submissions extending the inquiry beyond the public proceedings. The Senate Permanent Subcommittee on Investigations pursued detailed case histories of Moody’s and S&P, subpoenaing hundreds of thousands of documents from the agencies and financial institutions that had obtained their ratings. Nearly two dozen interviews and consultations with outside specialists supplemented those records, and the April 23, 2010, hearing released one hundred exhibits. In April 2011, the subcommittee’s majority and minority staff report reaffirmed joint findings that competitive pressures had affected ratings and that important analytical revisions had not been applied adequately to existing securities. Its conclusions carried the authority of a bipartisan legislative investigation, with legal liability remaining a question for proceedings governed by different standards. For historical interpretation, these inquiries’ principal contribution was the opportunity to reconstruct the sequence in which objections arose and decisions followed. Investigators could examine whether a commercial objection preceded a delayed revision, or whether a later explanation fitted the concerns recorded at the time. Such evidence narrowed the space for accounts built entirely around unforeseeable events without eliminating the need to establish each asserted connection between business pressure and analytical conduct.
The subsequent federal case against S&P concentrated on the relationship between representations of independence and the conduct underlying its ratings. The Justice Department filed its civil action on February 4, 2013, alleging fraud affecting federally insured financial institutions and seeking penalties under the Financial Institutions Reform, Recovery, and Enforcement Act. Its argument concerned allegedly misleading assurances that commercial considerations did not influence ratings, together with decisions that government lawyers attributed to the desire to preserve business. Those claims initially remained allegations requiring proof. On February 3, 2015, S&P and its parent agreed to a $1.375 billion settlement with federal and participating state authorities, accompanied by an acknowledged statement of facts. That statement described commercial considerations affecting proposed changes to the CDO Evaluator model and customer feedback influencing the timing of its updated version. It also acknowledged that S&P continued issuing and confirming ratings on CDOs substantially backed by subprime mortgage securities during a period in 2007 when adverse actions on the underlying RMBS were anticipated, without adjusting the existing CDO criteria to account for those expectations. These acknowledgments gave the historical record a firmer basis than the unresolved allegations alone. The settlement nevertheless did not constitute a trial verdict of fraud, and the parent company’s announcement emphasized that it contained no findings of violations of law. That distinction preserves the legal character of the resolution without erasing the evidentiary value of the facts accepted within it. The case demonstrated how an inquiry could move beyond the observation that securities performed badly and examine whether an agency’s description of its independence matched its actual decision-making. The payment settled specific claims; the accompanying record exposed practices relevant to evaluating the professional assurances on which purchasers had relied.
Accountability involved objectives that even a successful lawsuit could not fully combine. John C. Coffee Jr. argued that liability should principally encourage adequate investigation and timely revision of models, since rating agencies could not reimburse investors for the financial system’s aggregate structured-finance losses. He also warned that liability without meaningful limits could destroy useful intermediaries, making deterrence and compensation potentially conflicting goals. His analysis distinguishes establishing responsibility for specific misconduct from expecting litigation to repair an entire crisis. The investigations made it possible to evaluate claims of independence against records created before the collapse and explanations offered afterward. They also showed that evidence of a deficient process does not automatically establish an actionable misrepresentation, just as investment losses alone do not explain how a rating was produced. Effective scrutiny required access to the reasons for decisions and evidence of how commercial considerations influenced them. An assessor’s authority could no longer rest comfortably on a declaration of independence when the process supporting that declaration was difficult to examine. Investigation made those assurances testable, even where the available remedies could address only part of the damage.
Reform after the Crisis: Changing the Rules around Ratings
Congress gave the postcrisis response a statutory foundation on July 21, 2010, when the Dodd–Frank Wall Street Reform and Consumer Protection Act became law. Its provisions concerning rating agencies pursued two connected objectives: strengthening supervision of the firms and reducing the automatic consequences attached to their judgments. This combination reflected a difficult policy choice. Agencies would remain participants in financial markets, but federal regulation would demand greater accountability from them while requiring other institutions to exercise more responsibility for assessing creditworthiness. Implementation unfolded over several years, culminating in a substantial SEC rules package adopted on August 27, 2014. Several important provisions took effect in 2015, so adoption should not be mistaken for evidence that the entire framework was already operating successfully by the end of 2014.
Supervision acquired a permanent institutional home in the SEC’s Office of Credit Ratings, established in June 2012. Dodd–Frank required examinations of every NRSRO at least annually, accompanied by public summaries of essential findings and responses to deficiencies. The SEC had already issued its first report under this annual examination requirement in September 2011, before the new office began work. A fixed examination cycle gave the regulator repeated opportunities to check whether identified shortcomings had been addressed, while the public summaries made supervisory follow-up a continuing matter of record.
The 2014 rules addressed the points where commercial activity could enter the production of a rating. A prohibited conflict arose when personnel determining or monitoring grades, or developing and approving methodologies, also participated in sales or marketing or were influenced by those considerations. Smaller agencies could apply for an exemption where separation was inappropriate because of their size and an exemption served the public interest. The methodological provisions required board approval of rating procedures and consistent application of material revisions to the existing and future ratings to which those revisions applied. Changes affecting surveillance had to reach relevant outstanding ratings within a reasonable period, making continued monitoring part of the obligation rather than treating a methodological improvement as something useful only for new business. Annual internal controls reports also placed responsibility on management, requiring disclosure to the SEC of material weaknesses and a chief executive’s attestation. Management could not pronounce the control structure effective at year-end while such weaknesses remained. Yet these reports were nonpublic, and the SEC did not require an external audit of them. This arrangement strengthened direct supervisory access without giving purchasers an independently audited public assessment of each agency’s controls. Its potential value lay in making senior management answerable for how established procedures operated; its practical success would depend on examiners testing those assurances and requiring deficiencies to be corrected.
Disclosure extended beyond the agencies’ organizational arrangements to the evidence available when a security was offered. The reforms standardized performance statistics and required information about the assumptions and limitations accompanying rating actions. For covered rated asset-backed securities, issuers or underwriters had to disclose the findings and conclusions of third-party due diligence reports they obtained, including in offerings that were not publicly registered. Providers’ certifications described the work performed and its results, enabling users to see more clearly what had actually been examined. But the prescribed certification form did not dictate how a review must be conducted. These provisions addressed the circulation and visibility of investigative material; they did not establish a universal government-directed examination of every underlying loan. Their significance depended on whether the disclosed scope was adequate for the security being rated and whether the findings influenced its assessment.
A more fundamental alteration concerned the use of ratings in federal regulation. Section 939A required federal agencies to review relevant rules, remove references to or requirements of reliance on credit ratings, and substitute appropriate standards of creditworthiness. The Office of the Comptroller of the Currency supplied a concrete banking example in June 2012, adopting revised investment rules effective January 1, 2013. Under the new approach, an investment-grade security was one whose issuer had adequate capacity to meet its financial commitments over the projected life of the exposure. Banks could still consult agency ratings, but they were expected to supplement them with analysis proportionate to the investment’s complexity and their own risk profile. Management remained responsible for the investment decision even when it used third-party analytical services. National banks had already been responsible for prudent investment decisions; the specific change was that an external grade no longer supplied the regulatory definition of investment-grade status. When defending a purchase to its supervisor, a bank needed to substantiate the creditworthiness that made the investment permissible. This relocated the determination required for regulatory eligibility, although federal instruction could not instantly create analytical capacity or alter established working habits. Nor did Section 939A itself rewrite private investment mandates, whose use of ratings rested on contractual choices. Frank Partnoy’s later critique argued that mechanistic reliance persisted despite the statutory campaign to remove regulatory references. His interpretation cautions against measuring success solely by counting deleted provisions, since investment practice and supervisory expectations could preserve dependence under different language. The distinction between legal revision and operational change is particularly important here: determining whether reliance had diminished required examining how institutions reached decisions after the rules changed, rather than assuming that a new definition of creditworthiness demonstrated a new method of evaluating it.
Competition received a separate intervention through amendments to Rule 17g-5 adopted in November 2009. For covered structured-finance ratings, the hired agency had to obtain an arranger’s commitment to make rating information available to qualifying NRSRO competitors through password-protected websites. The intention was to enable an agency that had not been commissioned to produce its own initial rating and challenge the assessment selected by the arranger. Access belonged to competing raters under specified conditions, rather than constituting general public access to the transaction files. The SEC’s December 2014 annual report nevertheless stated that no unsolicited initial ratings had been produced using information supplied through these websites. Agencies were publishing unsolicited commentaries, but that activity did not amount to the competing initial assessments the mechanism sought to encourage. The result suggests that reformers needed to distinguish permission to examine a transaction from the incentives and resources required to publish an assessment of it.
Changing who selected the initial assessor remained a more ambitious and unsettled possibility. Senator Al Franken’s proposed assignment system would have used a board to allocate initial structured-finance ratings, reducing the arranger’s discretion to choose the agency. The enacted legislation instead required a study of assignment and alternative compensation arrangements, with subsequent rulemaking dependent on further determinations. The SEC staff’s December 2012 report examined a system in which issuers would still pay reasonable fees even though a board selected the initial rater. It separated the source of payment from control over the appointment, illustrating that those two aspects of the commercial relationship could be altered independently. Staff identified potential benefits from selection outside the issuer’s control but also concerns about measuring agency performance fairly and encouraging investors to interpret an assigned rating as government endorsement. The ability to purchase supplemental ratings could also preserve opportunities for shopping after the assigned assessment. The report recommended a roundtable as the next step, rather than immediate establishment of a board. Meanwhile, a January 2012 Government Accountability Office study had examined seven proposed compensation models and found that none had yet been implemented. These investigations expanded the policy alternatives without demonstrating that a replacement system had resolved its own operational difficulties. By the 2014 adoption of the principal SEC reforms, the central American response consequently rested on stronger controls and reduced mandated reliance, while issuer selection and payment continued. Reform had changed the obligations surrounding the transaction more decisively than the commercial arrangement through which the assessor obtained its work.
Could Another Payment System Produce Better Assessments?
Judging alternative payment systems requires a definition of success more demanding than obtaining lower grades. An agency that classifies almost everything as dangerous might avoid conspicuous optimism while providing little help in distinguishing stronger borrowers from weaker ones. Useful assessments must identify differences in creditworthiness and respond appropriately when the evidence changes, while their production requires sufficient resources for investigation. A payment arrangement consequently has to be evaluated through the quality of the work it supports, including the effort devoted to cases whose risks are difficult to establish. The question is how commercial rewards affect that work, rather than whether a different category of customer can simply be presumed to prefer better analysis.
Research on subscriber-financed ratings supplies evidence of promise, but its findings resist a universal ranking of payment models. In a 2013 study, Jess Cornaggia and Kimberly J. Cornaggia compared Moody’s corporate assessments with those of Rapid Ratings, a subscriber-paid firm. Rapid Ratings identified default risk sooner and produced a better ordering of firms by credit risk. Moody’s remained slower even after the researchers accounted for its preference for stable ratings. That comparison showed that a subscription service could provide useful warnings, although differences in analytical methods accompanied differences in financing. Han Xia’s 2014 examination of Egan-Jones added a different mechanism. After the investor-paid challenger began covering a corporation, S&P’s ratings became more responsive to credit risk and its changes conveyed more information. Xia interpreted this improvement as evidence that competition could strengthen an incumbent’s concern for its reputation. Later research complicated the proposition that investor payment consistently produced superior assessments. Nan Qin and Lei Zhou’s 2025 article found that S&P’s corporate bond ratings were more stringent and accurate than Egan-Jones’s, whereas Fitch’s were less stringent and had similar or lower accuracy. Egan-Jones revised ratings more frequently and made fewer large downgrades, but also reversed its decisions more often. These findings make the distinction between promptness and accuracy consequential. Frequent revision may supply earlier information while also producing more changes that subsequently require reversal. The studies concern corporate borrowers and do not establish how an alternative payment system would have performed across the mortgage securities assembled before 2007. Those studies support testing subscriber-backed analysis as a source of better information while cautioning against attributing every difference between agencies to their customers.
Subscription revenue nevertheless presents a financing problem precisely because useful information can circulate beyond the person who purchases it. Marco Pagano and Paolo Volpin favored investor payment, but recognized that purchasers could leak or resell ratings, allowing others to benefit without contributing to their production. Restricting access could help preserve revenue while limiting the public availability that made conventional grades widely usable. They also considered pressure from large investors seeking to postpone downgrades of securities they already held. A buyer’s interest before acquiring a security need not remain identical after a substantial position has been accumulated. That possibility must be distinguished from an established record of manipulation. The SEC’s December 2012 study reported that commenters had supplied no empirical evidence that subscriber-payment conflicts had influenced ratings, while noting the limited number of structured products assessed under that model. The reasonable implication is that customers’ interests require examination in each arrangement. Potential conflicts do not prove that two systems produce equally poor results.
Collective financing offered ways to address the difficulty of charging everyone who benefited from an assessment. One designation proposal examined by the GAO allowed interested agencies to rate an issue, with issuers depositing fees through a third-party administrator and investors directing their distribution according to the research they valued. Maintenance payments would continue over the security’s life, giving holders repeated opportunities to allocate compensation. Published grades would remain publicly available, while supporting research would primarily go to security holders. A separate proposal pooled creditors’ resources so that an agency or independent board could commission ratings through competitive bids. Selection could consider the proposed investigative work as well as price, and a rolling fund would help finance assessments before buyers paid fees on a new transaction. These arrangements attempted to make research a collectively financed service while giving its users influence over the allocation of work or money. Their institutional machinery was consequential because the administrator would have to turn a broad interest in reliable information into actual purchasing decisions. Investors directing payments after issuance might reward useful analysis, but an agency undertaking an assessment without assured compensation would also face a difficult budgeting decision. Financing uncertainty could discourage work on complicated issues whose analysis was costly even when its findings were valuable. A pooled commission could provide more predictable funding, provided that selection rewarded sufficient investigation rather than simply the cheapest bid. These were proposals for reorganizing demand, with expected benefits that depended on rules governing the intermediary and the decisions of participating investors.
Payment over time introduced another possible connection between an assessment and its consequences. Submissions recorded in the SEC’s 2012 study proposed deferring part of a fee or making compensation vest according to the performance of initial ratings and subsequent revisions. Anil K. Kashyap and Natalia Kovrijnykh’s theoretical analysis made otherwise unobservable analytical effort central to the design of compensation, with rewards depending on the rating issued and the project’s eventual outcome. Within their model, investor-commissioned ratings were more informative than issuer-commissioned ratings, although investors also purchased ratings more frequently than was socially optimal. Their analysis further identified a difficulty in rewarding initial accuracy. The resulting incentives could make an agency slow to acknowledge an earlier mistake. Performance rewards could also take years to calculate, requiring a method that distinguished a mistaken assessment from a loss consistent with its stated risk. Deferred compensation needed to reconcile rewards for careful initial investigation with incentives for necessary correction.
Individual careers could create incentives that a new funding source would leave intact. In a 2016 study, the Cornaggias and Xia found that analysts moving to firms they had rated awarded inflated and less informative assessments to those future employers before changing jobs. They found no corresponding inflation for the other firms rated by the same analysts, making the relationship with the prospective employer strikingly significant. Payment redesign would still require attention to employment negotiations and review of decisions affected by them, rather than assuming that the agency’s customer supplied every relevant incentive.
For choosing among these arrangements, the treatment of an unwelcome assessment offers a particularly useful practical test. A credible funding system would permit an agency to publish well-supported adverse findings even when they reduced the value of a subscriber’s holdings or made an issuer’s proposed transaction harder to sell. It would also finance continued examination after the initial assessment, when monitoring could become commercially unattractive to the parties paying the bills. This criterion directs attention to contractual protections and dependable research budgets, alongside the formal identity of the payer. It also requires scrutiny of the criteria used to distribute rewards. Payment for an absence of downgrades could encourage silence, whereas compensation designed to support justified investigation and revision would pursue a different objective. Comparative evaluation would need to examine the underlying evidence and the cost of producing it, so that apparent improvements were not purchased by abandoning difficult coverage or indiscriminately assigning pessimistic grades. The historical case supports experimenting with arrangements that make useful criticism financially sustainable. It supplies a weaker basis for declaring an untested replacement successful merely because it rearranges the movement of money. Better assessments would have to emerge from a demonstrable change in what agencies found it worthwhile to investigate and publish.
Would Different Paymasters Have Prevented the Crisis?
The following video is a 60 Minutes special on the 2008 financial crisis:
The most serious challenge to an explanation centered on issuer payment is that the same commercial arrangement produced very different results across markets. As legal scholar Claire A. Hill argued in 2010, companies also paid for ratings of ordinary corporate debt, yet those assessments did not generate the extraordinary proliferation of supposedly safe securities that characterized the mortgage boom. The identity of the customer cannot, by itself, explain the distinctive failure of structured-finance ratings. Hill offered an interpretation in which agencies accommodated issuers while also believing that financial engineering could justify the grades they awarded. In this account, commercial incentives and mistaken convictions reinforced one another; conscious disbelief in every favorable assessment was unnecessary. That distinction changes the historical problem. Explaining why an agency benefited from approving a transaction does not establish whether its analysts recognized the danger, underestimated it, or worked within an organization that discouraged them from investigating it adequately. The payment model remains relevant, but its effects require an account of how judgments were formed.
Private financial choices provide evidence that optimism extended beyond what industry participants told their customers. Ing-Haw Cheng, Sahil Raina, and Wei Xiong examined the personal housing transactions of professionals working in securitized finance during 2004–2006. On average, these individuals did not demonstrate superior foresight by timing the housing market or approaching their own purchases cautiously, and some groups increased their housing exposure particularly aggressively. Their behavior complicates a picture in which knowledgeable insiders uniformly understood the impending collapse while persuading others to assume its risks. The study nevertheless concerns securitization professionals, rather than a representative sample of rating analysts, and homeownership is an indirect measure of financial expectations. Personal circumstances can influence a housing purchase, so the findings cannot establish what every participant believed about a particular security. A complementary interpretation appears in the research of Christopher L. Foote, Kristopher Gerardi, and Paul S. Willen, who emphasized the substantial mortgage exposure retained by financial intermediaries and the losses those institutions subsequently suffered. They argued that optimistic expectations about house prices were central to understanding decisions that became disastrous after the market turned. Their evidence challenges the proposition that the principal institutions consistently transferred risks they privately understood to less informed outsiders. Retaining exposure does not establish that compensation encouraged prudent decisions, or that everyone within a firm shared the same understanding. Employees could benefit from transactions whose longer-term dangers were borne by their institution, while optimistic managers could accept risks that a more searching review would have questioned. These studies consequently support a serious possibility. An assessor financed by another constituency might still have shared the prevailing confidence in housing. Changing its customer would not necessarily have supplied the intellectual independence needed to contest that confidence.
Geography supplies a further complication. The three-commissioner FCIC dissent emphasized that housing booms and financial distress occurred across countries with different mortgage institutions, including some with much less reliance on American-style securitization. This comparison suggests that credit expansion and property speculation require explanations extending beyond the United States’ particular arrangements for purchasing ratings. It is not a controlled comparison, because international lenders and investors operated in overlapping markets and often consulted the same agencies. Nevertheless, redesigning American rating fees alone would leave substantial features of the international crisis unexplained.
Disagreement within the FCIC helps distinguish the importance of ratings from certainty about the reasons they failed. The majority treated the agencies as indispensable enablers of the mortgage-securities market that actually developed, alongside failures elsewhere in financial governance and supervision. Commissioners Keith Hennessey, Douglas Holtz-Eakin, and Bill Thomas also regarded erroneous ratings as essential to the crisis, but identified their underlying explanation as unresolved. They considered defective modeling, commercial incentives, and government-induced limits on competition as competing possibilities. Their dissent challenges confidence about the causal weight of the payment arrangement without requiring a defense of the grades themselves. This distinction matters for the question posed by a different paymaster. Establishing that favorable assessments helped sustain a dangerous market does not establish that a particular alternative contract would have produced sufficiently different assessments, sufficiently early, to prevent a systemic breakdown. Even a demonstrable improvement in ratings would leave further questions about how borrowers, financial institutions, and supervisors would have responded to that information. Some transactions might have become more expensive or failed to proceed, reducing the eventual losses; other participants might have sought different instruments or accepted greater risk without the same assurances. These are possible responses rather than established outcomes, and their uncertainty limits any categorical claim about prevention. The strongest counterpoint is a demand to explain each causal connection, rather than to move directly from a compromised commercial relationship to the conclusion that replacing it would have averted the entire crisis.
Accepting that challenge does not make the agencies’ organizational choices inconsequential. An assessment can be sincerely believed and still rest on inadequate investigation, especially when testing an attractive assumption threatens an important commercial relationship. The records examined in earlier sections include objections and unfavorable information available before the full collapse, making a defense based on universally unavoidable ignorance difficult to sustain. Those materials do not show that every loss was predictable; they do establish grounds for asking whether agencies gave contrary evidence the attention it deserved. Commercial pressure could affect the treatment of uncertainty even where no individual privately possessed an accurate forecast. The defensible interpretation is that issuer selection and payment contributed to conditions under which weak assessments gained authority and dangerous exposures accumulated, while shared beliefs and other institutional failures influenced the eventual consequences. Different paymasters might have reduced those pressures and altered the scale of the damage, but the historical record does not demonstrate that they would have prevented the crisis. Independence acquires substance when an organization can sustain scrutiny of its own assumptions as well as resistance to its customers’ demands. The criticism consequently rests on the quality of professional judgment and the institutional conditions required to support it.
Conclusion: Independence Had to Be Made Effective
The distinction between a security’s seller and its evaluator gave credit ratings much of their value. Purchasers could consult a specialist’s judgment without accepting the issuer’s description of its own obligations, while agencies accumulated authority through their experience and recognized standing. During the mortgage boom, that division of roles coexisted with relationships that gave arrangers considerable influence over the assessment process. The investigations showed how maintaining business could compete with the effort required to examine changing risks. Agencies’ reputations did not reliably resolve that competition, even though their future earnings depended in part on remaining credible. Coffee’s analysis helps explain why an established franchise could survive incentives to favor current revenue over the quality of particular judgments. The resulting failure concerned the arrangements supporting professional discretion. Formal separation from an issuer offered too little assurance when the issuer could influence which evaluator obtained the work and how demanding that work became.
Ratings also connected decisions made by institutions that were otherwise pursuing different objectives. Arrangers and portfolio managers could use the same grade for different purposes, selling a security or justifying its purchase. Supervisors also incorporated external assessments into judgments about the risks of regulated institutions. A common grading system made those decisions easier to coordinate, but it also allowed confidence in the assessment to spread beyond the circumstances of its production. A conclusion about a security’s creditworthiness could become a reason for limiting further inquiry, even where the investor’s exposure depended on matters the grade did not address. MacKenzie’s study of financial evaluation directs attention to the organizational practices through which such conclusions became accepted knowledge. That perspective makes it necessary to examine what users believed had been established when they encountered a familiar classification. The agencies were responsible for the evidentiary basis of their judgments and for explaining their limitations. Investors and supervisors retained responsibilities that an external opinion could not discharge for them, including evaluating whether an investment was appropriate to the institution that would hold it. Recognizing these separate duties does not distribute responsibility so widely that the raters’ contribution disappears. It identifies why their errors became consequential across an extensive financial system and why improving the assessments alone could provide only part of a remedy. Trustworthy delegation would have required users to understand what had been examined and to preserve their own capacity to question the result. The historical significance of the agencies lies both in the judgments they supplied and in the decisions other institutions authorized on the strength of them.
Postcrisis reform began to translate expectations of independence into obligations concerning conflicts, internal controls, and the handling of analytical methods. The SEC’s 2014 rules made aspects of the assessment process more accessible to supervision and required greater disclosure of ratings performance. Such measures created grounds for examining conduct that professional assurances alone could leave obscure. Their adoption nevertheless established requirements, rather than demonstrating that those requirements had achieved their intended effects. Effectiveness would have to be judged through the agencies’ subsequent practices and the quality of the work produced.
The question of who paid the assessor ultimately leads to the question of who could hold it accountable. An issuer could finance serious credit analysis, but that possibility did not establish that the prevailing arrangements consistently supported it. Accountability required a basis for determining whether a favorable judgment followed adequate examination, and whether commercial considerations had displaced reasons for caution. The crisis exposed the difficulty of relying on a respected intermediary when outsiders could see its conclusions more readily than the choices that produced them. It also showed that the consequences of those choices reached people who had neither commissioned a rating nor agreed to rely on one. Preventing a recurrence demanded attention to how institutions exercised authority over risks whose costs could extend beyond their customers. The counterpoint places an essential limit on this interpretation. Changing the payment relationship could not be assumed to remove shared misjudgments or the financial system’s other weaknesses. It does not remove the need to make professional independence usable at the moment an assessment influences a decision. Independence had to support a well-founded refusal to award a requested grade, with reasons that could withstand scrutiny, before buyers committed their funds.
Bibliography
Affiliate disclosure: Amazon links on this site are paid links. As an Amazon Associate I earn from qualifying purchases.
- Abu Dhabi Commercial Bank v. Morgan Stanley & Co. 651 F. Supp. 2d 155 (S.D.N.Y. 2009).
- Addo, Fenaba R., and William A. Darity Jr. “Disparate Recoveries: Wealth, Race, and the Working Class after the Great Recession.” The Annals of the American Academy of Political and Social Science 695:1 (2021), 173–192.
- Arner, Douglas W. “The Global Credit Crisis of 2008: Causes and Consequences.” The International Lawyer 43:1 (2009), 91-136.
- Ashcraft, Adam B., and Til Schuermann. “Understanding the Securitization of Subprime Mortgage Credit.” Foundations and Trends in Finance 2:3 (2008), 191–309.
- Ashcraft, Adam, Paul Goldsmith-Pinkham, and James Vickery. MBS Ratings and the Mortgage Credit Boom. Staff Report 449. New York: Federal Reserve Bank of New York, 2010.
- Baghai, Ramin P., Bo Becker, and Stefan Pitschner. “The Use of Credit Ratings in the Delegated Management of Fixed Income Assets.” Management Science 70:5 (2024), 3059–3079.
- Basel Committee on Banking Supervision. International Convergence of Capital Measurement and Capital Standards: A Revised Framework—Comprehensive Version. Basel: Bank for International Settlements, 2006.
- Beal, Vicki. “Testimony before the Financial Crisis Inquiry Commission.” September 23, 2010. Financial Crisis Inquiry Commission Records.
- Benmelech, Efraim, and Jennifer Dlugosz. “The Credit Rating Crisis.” NBER Macroeconomics Annual 24 (2009), 161–208.
- Bernanke, Ben S. “The Subprime Mortgage Market.” Speech at the Federal Reserve Bank of Chicago’s 43rd Annual Conference on Bank Structure and Competition, Chicago, May 17, 2007.
- BNP Paribas Investment Partners. “BNP Paribas Investment Partners Temporarily Suspends the Calculation of the Net Asset Value of the Following Funds: Parvest Dynamic ABS, BNP Paribas ABS Euribor and BNP Paribas ABS Eonia.” Press release, August 9, 2007.
- Board of Governors of the Federal Reserve System. “Discussion of Questions Arising under Regulations of the Comptroller of the Currency Governing the Purchase of Investment Securities.” Federal Reserve Bulletin 22:6 (1936), 421–423.
- “Investing in Securities without Reliance on Nationally Recognized Statistical Rating Organization Ratings.” Supervisory Letter SR 12-15, November 15, 2012.
- —-. Report to the Congress on Risk Retention. Washington, DC, October 2010.
- —-. “Revision in Bank Examination Procedure and in the Investment Securities Regulation of the Comptroller of the Currency.” Federal Reserve Bulletin 24:7 (1938), 563–566.
- Bolton, Patrick, Xavier Freixas, and Joel Shapiro. “The Credit Ratings Game.” Journal of Finance 67:1 (2012), 85–111.
- Bossidy, Paul T. Letter to Phil Angelides, chairman of the Financial Crisis Inquiry Commission. September 30, 2010. Financial Crisis Inquiry Commission Records.
- Brainard, Lael. “Is the Middle Class within Reach for Middle-Income Families?” Speech, May 10, 2019. Board of Governors of the Federal Reserve System.
- Brunnermeier, Markus K. “Deciphering the Liquidity and Credit Crunch 2007–2008.” Journal of Economic Perspectives 23:1 (2009), 77–100.
- Cantor, Richard, and Frank Packer. “The Credit Rating Industry.” Federal Reserve Bank of New York Quarterly Review 19:2 (1994), 1–26.
- Cheng, Ing-Haw, Sahil Raina, and Wei Xiong. “Wall Street and the Housing Bubble.” American Economic Review 104:9 (2014), 2797–2829.
- Coffee, John C., Jr. “Ratings Reform: The Good, the Bad, and the Ugly.” Harvard Business Law Review 1 (2011), 231–278.
- Cole, Roger T. “Subprime Mortgage Market.” Testimony before the U.S. Senate Committee on Banking, Housing, and Urban Affairs, March 22, 2007. Board of Governors of the Federal Reserve System.
- Comiskey, Michael and Pawan Madhogarhia. “Unraveling the Financial Crisis of 2008.” PS: Political Science and Politics 42:2 (2009), 271-275.
- Committee on the Global Financial System. The Role of Ratings in Structured Finance: Issues and Implications. CGFS Papers 23. Basel: Bank for International Settlements, January 2005.
- Congressional Oversight Panel. The AIG Rescue, Its Impact on Markets, and the Government’s Exit Strategy. June Oversight Report. Washington, DC: Government Printing Office, June 10, 2010.
- Copeland, Adam, Antoine Martin, and Michael Walker. “Repo Runs: Evidence from the Tri-Party Repo Market.” Journal of Finance 69:6 (2014), 2343–2380.
- Cordell, Larry, Yilin Huang, and Meredith Williams. “Collateral Damage: Sizing and Assessing the Subprime CDO Crisis.” Working Paper 11-30/R. Federal Reserve Bank of Philadelphia, May 2012.
- Cornaggia, Jess, and Kimberly J. Cornaggia. “Estimating the Costs of Issuer-Paid Credit Ratings.” Review of Financial Studies 26:9 (2013), 2229–2269.
- Cornaggia, Jess, Kimberly J. Cornaggia, and Han Xia. “Revolving Doors on Wall Street.” Journal of Financial Economics 120:2 (2016), 400–419.
- Coval, Joshua D., Jakub W. Jurek, and Erik Stafford. “Economic Catastrophe Bonds.” American Economic Review 99:3 (2009), 628–666.
- —-. “The Economics of Structured Finance.” Journal of Economic Perspectives 23:1 (2009), 3–25.
- Covitz, Daniel, Nellie Liang, and Gustavo A. Suarez. “The Evolution of a Financial Crisis: Collapse of the Asset-Backed Commercial Paper Market.” Journal of Finance 68:3 (2013), 815–848.
- Demyanyk, Yuliya, and Otto Van Hemert. “Understanding the Subprime Mortgage Crisis.” Review of Financial Studies 24:6 (2011), 1848–1880.
- Dettling, Lisa, Joanne Hsu, and Elizabeth Llanes. “A Wealthless Recovery? Asset Ownership and the Uneven Recovery from the Great Recession.” FEDS Notes. Board of Governors of the Federal Reserve System, September 13, 2018.
- Federal Deposit Insurance Corporation. In the Matter of Fremont Investment & Loan, Fremont General Credit Corporation, and Fremont General Corporation: Order to Cease and Desist. Docket No. FDIC-07-035b. March 7, 2007.
- Fender, Ingo, and Janet Mitchell. “Structured Finance: Complexity, Risk and the Use of Ratings.” BIS Quarterly Review (June 2005), 67–79.
- Fender, Ingo, and John Kiff. “CDO Rating Methodology: Some Thoughts on Model Risk and Its Implications.” BIS Working Papers 163. Basel: Bank for International Settlements, November 2004.
- Fender, Ingo, and Martin Scheicher. “The Pricing of Subprime Mortgage Risk in Good Times and Bad: Evidence from the ABX.HE Indices.” Applied Financial Economics 19:24 (2009), 1925–1945.
- Financial Crisis Inquiry Commission. The Financial Crisis Inquiry Report: Final Report of the National Commission on the Causes of the Financial and Economic Crisis in the United States. Washington, DC: U.S. Government Printing Office, 2011.
- Financial Stability Forum. Report of the Financial Stability Forum on Enhancing Market and Institutional Resilience. April 7, 2008.
- Flandreau, Marc, Norbert Gaillard, and Frank Packer. “To Err Is Human: US Rating Agencies and the Interwar Foreign Government Debt Crisis.” European Review of Economic History 15:3 (2011), 495–538.
- Flandreau, Marc, and Joanna Kinga Sławatyniec. “Understanding Rating Addiction: US Courts and the Origins of Rating Agencies’ Regulatory License (1900–1940).” Financial History Review 20:3 (2013), 237–257.
- Foote, Christopher L., Kristopher Gerardi, and Paul S. Willen. Why Did So Many People Make So Many Ex Post Bad Decisions? The Causes of the Foreclosure Crisis. Public Policy Discussion Paper 12-2. Boston: Federal Reserve Bank of Boston, 2012.
- Gerardi, Kristopher, Andreas Lehnert, Shane M. Sherlund, and Paul Willen. “Making Sense of the Subprime Crisis.” Brookings Papers on Economic Activity 2008, 2 (2008), 69–145.
- Griffin, John M., Jordan Nickerson, and Dragon Yongjun Tang. “Rating Shopping or Catering? An Examination of the Response to Competitive Pressure for CDO Credit Ratings.” Review of Financial Studies 26:9 (2013), 2270–2310.
- GS Mortgage Securities Corp. GSAMP Trust 2006-NC2: Mortgage Pass-Through Certificates, Series 2006-NC2. Prospectus supplement, June 27, 2006.
- He, Jie, Jun Qian, and Philip E. Strahan. “Are All Ratings Created Equal? The Impact of Issuer Size on the Pricing of Mortgage-Backed Securities.” Journal of Finance 67:6 (2012), 2097–2137.
- Hennessey, Keith, Douglas Holtz-Eakin, and Bill Thomas. “Dissenting Statement.” In The Financial Crisis Inquiry Report, 411–439. Washington, DC: Government Printing Office, 2011.
- Hill, Claire A. “Why Did Rating Agencies Do Such a Bad Job Rating Subprime Securities?” University of Pittsburgh Law Review 71:3 (2010), 585–608.
- International Monetary Fund. Global Financial Stability Report: Containing Systemic Risks and Restoring Financial Soundness. Washington, DC: International Monetary Fund, April 2008.
- Jiang, John (Xuefeng), Mary Harris Stanford, and Yuan Xie. “Does It Matter Who Pays for Bond Ratings? Historical Evidence.” Journal of Financial Economics 105:3 (2012), 607–621.
- Kashyap, Anil K., and Natalia Kovrijnykh. “Who Should Pay for Credit Ratings and How?” Review of Financial Studies 29:2 (2016), 420–456.
- Krebs, Tom. “Letters Regarding Testimony of D. Keith Johnson.” Memorandum to Wendy Edelberg. Financial Crisis Inquiry Commission, October 12, 2010.
- Kroszner, Randall S. “Basel II Implementation in the United States.” Speech to the New York Bankers Association Annual Washington Visit, Washington, DC, July 12, 2007.
- MacKenzie, Donald. “The Credit Crisis as a Problem in the Sociology of Knowledge.” American Journal of Sociology 116:6 (2011), 1778–1841.
- Mählmann, Thomas. “Did Investors Outsource Their Risk Analysis to Rating Agencies? Evidence from ABS-CDOs.” Journal of Banking & Finance 36:5 (2012), 1478–1491.
- Mayer, Christopher, Karen Pence, and Shane M. Sherlund. “The Rise in Mortgage Defaults.” Journal of Economic Perspectives 23:1 (2009), 27–50.
- McGraw Hill Financial. “McGraw Hill Financial and S&P Ratings Reach Settlements with DOJ, Attorneys General of 19 States and District of Columbia and with CalPERS.” Press release, February 3, 2015.
- Mian, Atif, Kamalesh Rao, and Amir Sufi. “Household Balance Sheets, Consumption, and the Economic Slump.” Quarterly Journal of Economics 128:4 (2013), 1687–1726.
- Michalek, Richard. “Credibility of Credit Ratings, the Investment Decisions Made Based on those Ratings, and the Financial Crisis.” June 2, 2010.
- Moody’s Investors Service. “Moody’s Expands Moody’s Mortgage Metrics to Include Subprime Residential Mortgages.” Global Credit Research, September 6, 2006.
- Moody, John. Moody’s Analyses of Railroad Investments. New York: Analyses Publishing Company, 1909.
- Pagano, Marco, and Paolo Volpin. “Credit Ratings Failures and Policy Options.” Economic Policy 25:62 (2010), 401–431.
- Park, John. The Impact of Subprime Residential Mortgage-Backed Securities on Moody’s-Rated Structured Finance CDOs: A Preliminary Review. Structured Finance Special Comment. New York: Moody’s Investors Service, March 23, 2007.
- Partnoy, Frank. “What’s (Still) Wrong with Credit Ratings?” Washington Law Review 92:3 (2017), 1407–1472.
- Pasley, Robert. Anatomy of a Banking Scandal: The Keystone Bank Failure-Harbinger of the 2008 Financial Crisis. London: Routledge, 2016.
- Pfeffer, Fabian T., Sheldon Danziger, and Robert F. Schoeni. “Wealth Disparities before and after the Great Recession.” The Annals of the American Academy of Political and Social Science 650:1 (2013), 98–123.
- Piskorski, Tomasz, Amit Seru, and James Witkin. “Asset Quality Misrepresentation by Financial Intermediaries: Evidence from the RMBS Market.” Journal of Finance 70:6 (2015), 2635–2678.
- Poor, Henry Varnum. History of the Railroads and Canals of the United States of America: Exhibiting Their Progress, Cost, Revenues, Expenditures, and Present Condition. New York: John H. Schultz, 1860.
- Qin, Nan, and Lei Zhou. “Are Investor-Paid Credit Ratings Superior?” Financial Management 54:1 (2025), 53–87.
- Rajan, Uday, Amit Seru, and Vikrant Vig. “The Failure of Models That Predict Failure: Distance, Incentives, and Defaults.” Journal of Financial Economics 115:2 (2015), 237–260.
- Rocco, Joseph A. Early Defaults Rise in Mortgage Securitizations. Structured Finance Special Report. New York: Moody’s Investors Service, January 18, 2007.
- Schmidt, Lawrence, Allan Timmermann, and Russ Wermers. “Runs on Money Market Mutual Funds.” American Economic Review 106:9 (2016), 2625–2657.
- Siegel, Jay. Moody’s Mortgage Metrics: A Model Analysis of Residential Mortgage Pools. Moody’s Special Report. New York: Moody’s Investors Service, April 1, 2003.
- Skreta, Vasiliki, and Laura Veldkamp. “Ratings Shopping and Asset Complexity: A Theory of Ratings Inflation.” Journal of Monetary Economics 56:5 (2009), 678–695.
- Smith, Andrew, and Robert E. Wright. “Sowing the Seeds of a Future Crisis: The SEC and the Emergence of the Nationally Recognized Statistical Rating Organization (NRSRO) Category, 1971–1975.” Business History Review 95:4 (2021), 739–764.
- Sorkin, Andrew Ross. Too Big to Fail: The Inside Story of How Wall Street and Washington Fought to Save the Financial System–and Themselves. New York: Viking Press, 2009.
- Sylla, Richard. “An Historical Primer on the Business of Credit Rating.” In Ratings, Rating Agencies and the Global Financial System, edited by Richard M. Levich, Giovanni Majnoni, and Carmen M. Reinhart, 19–40. Boston: Kluwer Academic Publishers, 2002.
- U.S. Congress. House of Representatives. Committee on Oversight and Government Reform. Credit Rating Agencies and the Financial Crisis. Hearing, October 22, 2008. 110th Cong., 2nd sess. Washington, DC: Government Printing Office, 2009.
- —-. Dodd–Frank Wall Street Reform and Consumer Protection Act. Public Law 111-203. July 21, 2010. 124 Stat. 1376. Especially sections 932, 939A, and 939F.
- U.S. Department of Justice. “Department of Justice Sues Standard & Poor’s for Fraud in Rating Mortgage-Backed Securities in the Years Leading Up to the Financial Crisis.” February 5, 2013.
- U.S. Department of Justice, participating states, McGraw Hill Financial, Inc., and Standard & Poor’s Financial Services LLC. Settlement Agreement. February 3, 2015. Annex 1, “Statement of Facts.”
- U.S. Government Accountability Office. Credit Rating Agencies: Alternative Compensation Models for Nationally Recognized Statistical Rating Organizations. GAO-12-240. Washington, DC: Government Accountability Office, January 2012.
- —-. Securities and Exchange Commission: Action Needed to Improve Rating Agency Registration Program and Performance-Related Disclosures. GAO-10-782. Washington, DC, 2010.
- —-. Troubled Asset Relief Program: Status of Government Assistance Provided to AIG. GAO-09-975. Washington, DC: Government Accountability Office, September 2009.
- U.S. Securities and Exchange Commission. Amendments to Rules for Nationally Recognized Statistical Rating Organizations. Release No. 34-61050. November 23, 2009.
- —-. “Definition of Nationally Recognized Statistical Rating Organization.” Proposed rule. Federal Register 70:78 (April 25, 2005), 21306–21324.
- —-. Division of Trading and Markets. Report to Congress on Assigned Credit Ratings, As Required by Section 939F of the Dodd-Frank Wall Street Reform and Consumer Protection Act. December 2012.
- —-. “Nationally Recognized Statistical Rating Organizations.” Federal Register 73:123 (June 25, 2008), 36212–36252.
- —-. Office of Credit Ratings. Annual Report on Nationally Recognized Statistical Rating Organizations, As Required by Section 6 of the Credit Rating Agency Reform Act of 2006. December 2014.
- —-. Report on the Role and Function of Credit Rating Agencies in the Operation of the Securities Markets. Washington, DC, 2003.
- —-. “SEC Adopts Credit Rating Agency Reform Rules.” Press release, August 27, 2014.
- —-. “SEC Charges Operators of Reserve Primary Fund with Fraud.” Press release, May 5, 2009.
- —-. “SEC Names Thomas Butler as Director of New Office of Credit Ratings.” Press release, June 15, 2012.
- —-. Staff Report: Enhancing Disclosure in the Mortgage-Backed Securities Markets. Washington, DC, February 3, 2003.
- —-. Summary Report of Issues Identified in the Commission Staff’s Examinations of Select Credit Rating Agencies. Washington, DC, 2008.
- U.S. Senate. Committee on Homeland Security and Governmental Affairs. Permanent Subcommittee on Investigations. Wall Street and the Financial Crisis: Anatomy of a Financial Collapse. Majority and Minority Staff Report. Washington, DC, April 13, 2011.
- White, Lawrence J. “Markets: The Credit Rating Agencies.” Journal of Economic Perspectives 24:2 (2010), 211–226.
- Wolff, Edward N. “The Declining Wealth of the Middle Class, 1983–2016.” Contemporary Economic Policy 39:3 (2021), 461–478.
- Xia, Han. “Can Investor-Paid Credit Rating Agencies Improve the Information Quality of Issuer-Paid Rating Agencies?” Journal of Financial Economics 111:2 (2014), 450–468.
- Yagan, Danny. “Employment Hysteresis from the Great Recession.” Journal of Political Economy 127:5 (2019), 2505–2558.
Originally published by Brewminate, 10.09.2026, under the terms of a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International license.
About the Author
