The GradeThey Helped Write:How S&P and Moody's Built the Conflict at the Heart of Global Finance

Picture a professorwho grades your exam, but you pay the professor directly. Not through your tuition. You pay them specifically for this grade. You can also hire them in advance to tell you how to structure your answers before you sit the exam. And if they give you an A on work that turns out to be completely wrong, they face no professional sanction, no financial penalty, and no legal liability. Now scale that arrangement to the market that determines whether trillions of dollars of debt is safe to buy. That is the credit rating industry. It has operated this way for decades, it helped cause the worst financial crisis since the Great Depression, and today it functions on almost exactly the same model it used then. The agencies are called S&P Global and Moody's. Between them they control about 80 percent of the global credit rating market. And the arrangement described above is not a scandal. It is the business model.

The theory is elegant. When a government, a company, or a bank wants to borrow money by issuing bonds, it needs buyers. Buyers need to know how risky the bond is. Not every buyer has the expertise or resources to analyse every bond independently, particularly in global markets where a single pension fund might hold thousands of different debt instruments across dozens of countries. Credit rating agencies exist to solve that problem. They are supposed to be independent analysts who assess the likelihood that a borrower will repay its debt, translate that assessment into a simple letter grade, and give investors a reliable signal.

An AAA rating, the highest available, is supposed to mean the probability of default is vanishingly small. Investment-grade ratings, from AAA down to BBB-, are supposed to mean the debt is suitable for conservative investors such as pension funds, insurance companies, and money market funds. Below that is speculative grade, colloquially called junk, where the risk of non-payment is meaningfully elevated. The whole system depends on one assumption: that the agency giving the grade has no financial interest in the grade it gives.

For most of the credit rating industry's early history, that independence assumption held because the model worked differently. The original model was investor-pays: the agencies sold their research to the investors who needed it, in the same way a financial newspaper sells subscriptions. The buyer of the information and the source of the revenue were the same party, which meant the agency had every incentive to be accurate. Investors would stop buying research that was systematically wrong.

That model began to change in the 1970s. The shift was partly practical: photocopiers made it easy for investors to copy and share research, eroding the subscription business. So the agencies switched to charging the issuers, the companies and banks whose debt they were rating. Within a generation, the entire economics of the industry had inverted.

The issuer-pays model means the bond issuer, the company or bank borrowing money, pays the rating agency for the rating it receives on that debt. The conflict this creates is not subtle. The bank wants an investment-grade rating because it makes the bond easier to sell and cheaper to issue. The rating agency wants to keep the bank as a client. The agency that consistently gives lower grades than its competitors loses clients to the competitors. The agency that finds ways to justify higher grades keeps and grows its business.

This is not a theoretical concern. It is a documented commercial pressure that shaped how the largest agencies operated for decades. Internal emails and testimony from the 2008 investigations made the pressure visible in its rawest form. Analysts who pushed back on aggressive ratings faced internal resistance. Rating committees were aware of the competitive landscape. And when a bank unhappy with a preliminary rating shopped the deal to a competitor who would grade it higher, the message to the first agency was clear.

Rating shopping is the practice where an issuer, before formally requesting a rating, informally gauges which agency will give the most favourable grade. The agencies provide preliminary assessments. The issuer then proceeds formally only with the agency whose grade is highest. The agencies that are bypassed get no revenue. Over time, the systematic effect of rating shopping is that issuers consistently pay the agency most willing to give optimistic grades, and those agencies grow faster than their more conservative competitors. The market, rather than correcting for generous ratings, rewards them.

The issuer-pays conflict is bad enough when the agency is simply rating a straightforward bond. It becomes something qualitatively different when the agency is actively involved in designing the product before rating it. That is what happened with structured finance, and it is the part of this story that deserves more attention than it typically receives.

Structured finance refers to complex debt instruments, collateralised debt obligations (CDOs), mortgage-backed securities (MBS), synthetic CDOs, and similar products, where the key design variable is how the cash flows and losses are split between different tranches of investors. The whole point of the structure is to take a pool of lower-quality assets and engineer a product where some portion of it qualifies for a high rating.

Here is where the agencies became something other than independent analysts. Banks and investment banks building these structures would approach the rating agencies before completing the design and essentially ask: if we structure it this way, what rating do we get? The agencies would model the structure and provide guidance. The bank would adjust the structure based on that guidance to maximise the proportion of the product that received an investment-grade or AAA designation. The agency would then formally rate the finished product.

This consulting service was a significant revenue stream. Moody's structured finance revenues grew from roughly $60 million in 1998 to around $900 million in 2006, representing the majority of its total revenue by the height of the mortgage boom. The structured finance division was not a peripheral business. It was the engine of growth, and its growth depended on maintaining relationships with the banks structuring deals whose rating quality it was simultaneously supposed to be assessing independently.

The scale of what the rating agencies enabled in structured finance is difficult to comprehend without specific numbers. Consider this: in 2006, approximately 64,000 structured finance products received AAA ratings from Moody's. In the same year, only 12 US companies carried Moody's highest rating. The agencies had produced more AAA-rated structured finance products in a single year than existed AAA-rated companies in the entire history of their ratings. This should have been a signal. It was not treated as one.

The models the agencies used to assess these products shared a common and critical assumption: that housing prices across the United States did not decline simultaneously. If defaults in Florida were uncorrelated with defaults in California, then a nationally diversified pool of mortgages provided genuine protection through diversification. The senior tranche really was insulated from individual loan failures. But the model did not account for a scenario where the entire housing market declined together, which is precisely what happened, because the housing market had been inflated by the same easy credit conditions that produced the mortgages in the pool.

There is an internal Moody's document from 2004 that has become something of a landmark in post-crisis investigations. An analyst wrote internally that the correlation assumptions in the CDO rating models were "extremely sensitive" to stress scenarios. He noted that in a nationwide housing downturn, the default correlations across the mortgage pool would increase dramatically, meaning the diversification that justified the AAA ratings would largely disappear. His analysis was technically correct. The products were being rated anyway.

When the US housing market began its decline in 2006 and accelerated into collapse in 2007 and 2008, the structured finance ratings that S&P and Moody's had issued proved to be comprehensively wrong. Securities rated AAA began defaulting or were downgraded to junk in the space of months. In some cases, CDO tranches that had been rated AAA as recently as early 2007 were rated D, meaning actual default, within 18 months. The instruments that pension funds, money market funds, and bank balance sheets were holding as safe, liquid, triple-A assets turned out to be worth fractions of their face value.

The scale of the rating failures was not marginal. Between 2007 and 2008, Moody's downgraded more than $2 trillion of debt securities it had previously rated investment-grade. S&P downgraded approximately $1.9 trillion. The downgrades were not gradual recalibrations of slightly optimistic views. They were cliff-edge collapses from the highest grade to near-junk across instruments that institutional investors had been required, by regulation, to treat as safe. The capital write-downs that followed those downgrades contributed directly to the insolvency of several major financial institutions.

In 2013, the US Department of Justice filed a civil lawsuit against S&P, alleging that it had knowingly inflated ratings on mortgage-backed securities to win business from Wall Street banks. The suit ran for two years. In 2015, S&P agreed to pay $1.375 billion across settlements with the DOJ and various state attorneys general. No individual executives faced criminal charges. No rating agency employee went to prison for their role in producing the ratings. The company's market capitalisation, even after the settlement, was far larger than the fine. S&P paid the cost of doing wrong out of the profits of having done it, and continued doing it.

You might reasonably ask why, after a failure of this magnitude, investors and issuers did not simply move to different rating agencies. The answer reveals something important about how the agencies acquired and retained their power, and why the market cannot self-correct around their conflicts even when those conflicts are fully visible.

S&P, Moody's, and Fitch are formally designated as Nationally Recognised Statistical Rating Organisations, a regulatory designation created by the SEC in 1975. The designation matters because regulations throughout the US financial system, and in many other jurisdictions, reference NRSRO ratings specifically. A pension fund required by law to hold investment-grade debt must hold debt rated investment-grade by an NRSRO. A money market fund with AAA requirements must use NRSRO ratings. Bank capital requirements under Basel accords reference NRSRO ratings. The agencies are not just market participants. They are embedded in the legal and regulatory architecture of global finance.

This regulatory embedding means that new entrants face a structural barrier that has nothing to do with the quality of their analysis. Even if a new rating agency produced consistently more accurate ratings than S&P or Moody's, it could not displace them because the regulations that require their ratings specifically require NRSRO ratings. A bond rated AAA by an upstart analytical firm remains, for regulatory purposes, unrated. The pension fund cannot buy it. The incumbent agencies hold a government-sanctioned monopoly position reinforced by decades of regulatory references that were never updated.

The structural conflict embedded in the issuer-pays model is compounded by a personnel dynamic that makes the agencies' commercial relationships with banks even stickier than the fee arrangement alone would explain. The revolving door, the movement of people between the agencies and the financial institutions whose products they rate, is well documented and runs in both directions.

Rating agency analysts who develop expertise in, say, structured credit products are valuable to the banks structuring those products. The analyst understands how the agency models work, what the rating committees are likely to require, and how to frame a deal to achieve a target rating. Banks pay significantly more than agencies. Senior structured finance analysts at Moody's or S&P regularly move to structuring desks at investment banks at salaries multiples higher than they earned at the agency.

The reverse also occurs. Former bank employees join the agencies, often bringing their relationships from the bank side. A senior structured finance professional moving from a major investment bank to a leadership role at Moody's carries with them an implicit understanding of how deals are structured, which banks are the important clients, and what the competitive landscape looks like. This is presented as enriching the agencies with market experience. It is also, straightforwardly, the integration of bank interests into the agencies' leadership.

The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 included an entire section, Section 931 through 939H, dedicated to credit rating agency reform. It was the most serious legislative attempt to address the structural conflicts in the rating industry since the NRSRO designation was created. It achieved some things. It did not fix the underlying problem.

Dodd-Frank required agencies to disclose more about their methodologies, established an SEC Office of Credit Ratings with examination authority over the agencies, created some legal liability for knowingly false ratings, and mandated that federal regulations begin removing references to NRSRO ratings where possible. These were all meaningful steps in the right direction, and none of them altered the issuer-pays model. The commercial structure that created the conflict of interest in the first place was left entirely intact. Congress reformed the disclosure regime around a fundamentally conflicted model without addressing the conflict.

The removal of NRSRO references from federal regulations, also mandated by Dodd-Frank, has proceeded far more slowly than the legislation intended. Many of the embedded regulatory references remain. The capital adequacy frameworks used by US banks still reference external credit ratings in ways that effectively require NRSRO designations. The dependence that made the agencies too important to discipline after 2008 was not structurally dismantled. It was modestly trimmed at the edges.

There is a deeper dimension to this problem that sits underneath the issuer-pays debate. Even if you solved the issuer-pays conflict entirely, moved to an investor-pays model, prohibited revolving door employment, and forced full disclosure of all consulting arrangements, you would still have a system where private companies issue grades that function as legal thresholds in the regulation of trillions of dollars of investment activity. That arrangement has structural problems that are independent of the commercial conflict.

When the law says that pension funds can only hold investment-grade debt, and that investment grade is defined by reference to private companies' opinions, you have embedded private commercial judgements into public regulatory requirements. The agencies did not ask for this power. It accreted around them over decades as regulators found it convenient to outsource the analytical work of defining safe from risky. But the consequence is that the agencies' opinions now have a force that ordinary market signals do not. A ratings downgrade from investment grade to junk does not just reflect an analytical view. It mechanically triggers forced selling by every institution whose mandate prohibits below-investment-grade holdings.

The 2011 US credit rating downgrade by S&P, the first in American history, illustrated both the power and the absurdity of this arrangement in the same moment. S&P downgraded US government debt from AAA to AA+ during the congressional debt ceiling standoff. Markets moved sharply. Treasury yields actually fell, because global investors flooded into US bonds as a safe haven despite the downgrade, demonstrating that the market's opinion of US debt was not aligned with S&P's opinion of it. The agency's grade was dramatically out of step with the market's behaviour, and yet the grade retained its regulatory force. That is not a system that describes reality. It is a system that creates it.