Ratings Agencies: The Forgotten Constituency of Financial Crisis Interventions

What all the people who were harping about what we were doing at the time missed is the centrality of the rating agencies as a constraint on how bailouts were structured, because a financial institution cannot operate without at least an investment-grade rating.

— Jim Millstein, US Treasury Chief Restructuring Officer, 2009–2011

Credit ratings downgrades are commonplace in financial crises. Downgrades are a regular consequence of bank runs and other financial panics and can themselves contribute to a crisis. Downgrades can trigger contractual provisions that force banks to produce substantially more liquidity; they can make a bank lose eligibility to borrow at standing central bank facilities or make the bank’s debt ineligible for purchase by certain classes of investors. Anticipation of credit rating downgrades can force crisis-fighters to rush their announcement or implementation of a crisis response.

Yet, all too often, those very policy responses fail to initially consider the ratings agency reaction function, which can cause deleterious instability or uncertainty—and often force crisis-fighters back to the drawing board to respond anew. Bank stakeholders and other market participants are looking to the ratings agencies (among others) as interpreters and appraisers of the government response—and are often anxious to see the resulting ratings impact.

Drawing on historical case studies, this article highlights some key considerations for crisis-fighters designing emergency interventions that fall broadly into two buckets: (1) Giving intervention design consideration to the high-level mechanics of ratings agency models, and (2) updating crisis communications to give the ratings agencies greater predictability of potential policy outcomes.

Regarding policy design considerations, policymakers need to understand credit ratings agency models at a high level and how their emergency assistance programs will fit into them. If taking government assistance will exacerbate ratings pressure, or allow for procyclical ratings downgrade cliffs to remain, crisis-fighters may have accomplished little.

Second, crisis communications need to keep the credit ratings agencies in mind—particularly the agencies’ desire for a clear policy response function. Policymakers may themselves not precisely know their preferred or likely but policy, but they should be as clear as they can on their policy response framework while retaining their desired level of policy optionality. Effective crisis-time communication will already include attempting to speak to bank counterparties, the rest of the complement of bank stakeholders, and to the market broadly, but simple tweaks can often go a long way toward improving the ratings agency reception.

Historical interventions that failed to sufficiently take the ratings agencies into account have required changes ranging from additional communications to substantial reconfiguring of program structure. Given the importance to crisis-fighting of avoiding ratings erosion at a specific financial institution, the financial system writ large, or the macroeconomy broadly, satisfying the ratings agencies in the first instance is of paramount importance. An analysis of case studies of financial crisis interventions around the world and throughout history from the Yale Program on Financial Stability’s New Bagehot Project reveals that credit ratings agencies are often an afterthought for crisis-fighters as they make intervention design decisions. However, those experiences suggest there are better policy outcomes to be had from earlier consideration of the ratings agencies as a key policy constituency.

Meeting the model

Lending means leverage

Central bankers, typically wanting to minimize their market footprint, usually have a clear preference for crisis interventions to take the form of loans rather than asset purchases.[1] They often also demand seniority in the capital structure and substantial degree of overcollateralization. These are great principles for preventing central bank losses and minimizing moral hazard—yet they are all “ratings negative.” Every dollar (or euro, yen, etc.) of seniority that the central bank takes pushes down existing creditors in the seniority waterfall. The same goes for every dollar encumbered of previously unencumbered collateral.[2]

The US government’s efforts to rescue AIG in 2008 are illustrative. AIG was a $1 trillion international insurance company that had provided substantial credit insurance across the financial system, particularly on subprime mortgage-backed collateralized debt obligations. By the fall of 2008, AIG thus found itself in a reinforcing spiral of needing to meet margin calls as the value of insured assets fell, getting its ratings downgraded as a result of the new financial demands it was facing, and then facing further margin calls because of those downgrades.

After the failure of Lehman Brothers in September 2008, the Federal Reserve demonstrated a willingness to “lend freely” to backstop AIG by establishing an $85 billion Revolving Credit Facility, effectively taking AIG’s underlying businesses as collateral. The Fed added the $38 billion Securities Borrowing Facility in October. Yet, AIG’s new borrowing meant that its leverage was increasing, which neither the market nor the ratings agencies liked. (Other initial terms, such as the onerous interest rate and short duration, also didn’t help.) AIG’s credit rating had already fallen from AAA to A, and its new borrowing meant its leverage ratios were no longer consistent with an A-rated company. Further downgrades would have meant further collateral calls and a collapse in new and recurring business. Financial companies depend on maintaining investment-grade ratings in practice and, often, by contract—and insurers are generally held to even higher standards.

Facing the threat of downgrading, the Fed had to get more creative and untether itself from its preference for lending operations over purchases. The Fed set up two facilities—Maiden Lane II and Maiden Lane III—to directly purchase the assets that were causing the liquidity spiral at AIG. AIG retained the junior-most loss tranche of each of these investment vehicles. So, the Fed was still protected against loss by AIG and the collateral, but the funding leverage had moved to these Fed vehicles rather than sitting on AIG’s balance sheet. This was economically similar to Fed lending, but helped restore AIG’s leverage ratios and ease ratings pressure.[3]

Capital must be capital

Similarly, crisis-fighters in market economies often want to avoid government ownership of financial firms, which can be particularly tricky when government capital injections are needed. Moreover, they’re often under political pressure to extract a “good deal.” These considerations can tempt crisis-fighters to inject capital in the form of more-senior capital, such as preferred shares—or even Tier 2 capital, such as subordinated debt. Yet, for all the reasons policymakers tend to like such securities, ratings agencies dislike them.

Even when regulators take care to make sure the injected capital counts as regulatory capital, ratings agencies are looking for capital that is the least debt-like, and thus more able to absorb potential losses and not prioritize dividends or interest payments. Using more “senior” forms of capital proved a drag on ratings agencies’ assessments of the government capital injected into Citigroup and AIG during the Global Financial Crisis (GFC), leading to the restructuring of the capital injections for both institutions.

Accounting for procyclicality

As crisis-fighters well know, crisis outcomes are endogenous to the crisis response. Given that credit ratings downgrades are procyclical, the design of any crisis response that uses credit ratings to determine eligibility may benefit from incorporating expectations of ratings procyclicality. For instance, it is widely accepted by central bankers looking to charge a premium interest rate on emergency lending that the premium should be set relative to “normal” times; setting the rate at premium to the elevated crisis rates would be self-defeating. It may be beneficial to apply the same logic to other terms of the intervention,[4] inclusive of ratings-based thresholds.

For example, during the COVID-19 pandemic, the Fed introduced the Corporate Credit Facilities to provide liquidity to the corporate bond market through primary and secondary market purchases. As initially announced on March 23, 2020, the facility focused exclusively on investment-grade companies and exchange-traded funds (ETFs) thereof. In the weeks that followed, Fed facilities notwithstanding, corporations continued to be subject to credit ratings downgrades, with many becoming “fallen angels”—investment-grade companies losing that ratings status by getting downgraded to high-yield, or “junk,” status. Moreover, while there was some spillover benefit to ineligible companies, there are many cases wherein crisis-fighters may not be able to rely on positive spillovers to avoid cliff effects in their responses. Indeed in this instance, as Wall Street Journal chief economics commentator Nick Timiraos would later write, there was “a growing chasm between companies inside the central bank’s lending perimeter and those outside, including junk bonds,” as that market remained effectively closed to its borrowers and its spreads continued to widen.

Recognizing the problem, the Fed on April 9, 2020, expanded eligibility for both its primary and secondary market facilities to include these fallen angels, and its secondary market facility to include high-yield bond ETFs.

The upshot

Acknowledging the ratings agency reaction function earlier in the intervention design process is likely to lead to more effective interventions. As shown in these examples—where the initial preferences were for lending over purchases, senior capital over more loss-absorbing forms, and an exclusion of non-investment-grade assets—such acknowledgement would benefit crisis-fighters by sooner untethering them from their preferred, but less effective, policy. In short: crisis responses may need ex ante adjustments to “meet the models” of the ratings agencies.

Predictability of outcomes

Financial crises typically require that policymakers retain policy optionality: the freedom and leeway to change their policy response as necessary to best respond to changing crisis conditions. This is in tension with the models that govern how ratings agencies make their decisions. Policymakers would be well served by integrating the importance of the predictability of their policy responses—to the extent possible—in assessing how ratings agencies will respond to a given policy. Ratings agencies may keep ratings unduly depressed for lack of clarity over the government's intent to intervene. Thus, while “constructive ambiguity” may maximize policy freedom, it risks leading to self-reinforcing destabilization.

US GFC case studies

Debt Guarantee Program

During the GFC, crisis-fighters in the US rallied behind their “systemic risk exception” authority to provide open-bank assistance to the entire banking system at once. They automatically enrolled all banks in the Temporary Liquidity Guarantee Program (TLGP), wherein the banks’ term debt was, for a fee, insured by the Federal Deposit Insurance Corporation (FDIC). This was intended to put investors at ease about providing longer-term funding to banks while the system’s solvency was still in question.

The FDIC is backed by the “full faith and credit” of the United States government, which was rated AAA. Yet, when the US government released its initial proposal for the debt guarantee, the ratings agency S&P wrote (emphasis added):

To qualify for rating substitution treatment according to our criteria, the terms of a guarantee must be unconditional, irrevocable, and timely. However, in the case of the TLGP, it is uncertain whether the FDIC's payment of interest and principal under its guarantee would have to be made on a timely basis […] Unless the current proposal is amended to provide a guarantee of timely payment of interest and principal, we would be unable to rate the debt of financial institutions qualifying for the FDIC guarantees at the 'AAA' rating of the U.S. government.

The FDIC met with S&P and its competitor Fitch about their ratings methodologies, ultimately changing the guarantee program’s approach in the final rule, in part to achieve the transfer of the government’s credit rating to the guaranteed debt. S&P followed the release of the final rule with another note, this time saying:

We had expressed reservations about whether payments would necessarily be made on a timely basis under the guarantee provisions included in the FDIC's Interim Rule […] However, the guarantee provisions were revised in the Final Rule the FDIC's board approved today, and we now believe that timely payment of interest and principal are addressed.

AIG: Whatever it takes

When AIG continued to wobble into 2009, the Fed and Treasury issued a joint press release to express a “whatever it takes” mentality and reassure the ratings agencies that official support would continue to come to AIG as necessary. While US policymakers continued to strongly prefer to avoid any kind of nationalization, they put out a statement that was remarkably suggestive of standing fully behind AIG, saying (emphasis added):

The steps announced today provide tangible evidence of the U.S. government's commitment to the orderly restructuring of AIG over time in the face of continuing market dislocations and economic deterioration. Orderly restructuring is essential to AIG's repayment of the support it has received from U.S. taxpayers and to preserving financial stability. The U.S. government is committed to continuing to work with AIG to maintain its ability to meet its obligations as they come due.

As the US Treasury’s chief restructuring officer would explain in 2018, “The government’s support wasn't an unlimited guarantee, but the release was worded to suggest that if AIG needed more, we would do more.” He added, “We issued that press release to satisfy PwC, to get AIG a clean [audit] opinion to avoid a ratings agency downgrade.”

Post-GFC Europe

Implementations of the principles of the European Union's Bank Recovery and Resolution Directive (BRRD), even before the legal introduction of the framework or outside of its formal purview, demonstrated opportunities to allow some level of predictability while also maintaining policy optionality.

In the case of the banking and insurance firm SNS Reaal in the Netherlands in 2013, before BRRD implementation, the decision not to use broad expropriation powers on senior debtholders was cited by ratings agencies and commentators as a significant factor in the containment of contagion risk across the sector. This was an explicit self-curtailment of write-down powers by the Dutch government alongside its publication of a comprehensive response package. S&P responded by revising the outlook on SNS Reaal’s senior unsecured debt to developing from negative. A news report citing a Fitch analyst had previously cited the dangers of the potential write-down, and ratings agency Moody’s had placed SNS Reaal’s senior unsecured debt under review for further downgrade. Moody’s cited the increased risk of senior debt being included in burden-sharing should multiple rounds of support be necessary. Accordingly, the Dutch finance minister’s clear articulation in a letter to the parliament of how it would use its intervention powers expressed not only the desire to protect senior debt—an effort which itself would be buttressed by avoiding a ratings downgrade—but also the fear of erosion of its ability to do so through continued deterioration of the financial situation.

In 2015, Moody’s issued a report criticizing the fragmented approach to adopting the BRRD across the European Union. In the report’s press release, a Moody’s senior credit officer said (emphasis added):

National deviations seen to date point to a more complicated and fragmented structure that will delay the process of resolving banks and make outcomes less easy to predict for market participants. No one model is clearly superior, but the lack of consistency has the potential to lead to greater deviation in outcomes.

The ratings agencies are concerned with consistency and predictability not solely in the style of intervention, but also in the range of potential outcomes. Policymakers, for their part, are concerned with maintaining their right to use whatever tool necessary to stabilize their financial institutions, without causing avoidable moral hazard. The tension between these two goals can be reconciled when policymakers clearly define a range of options available for consideration. A 2020 S&P report centered the importance of EU governments retaining their “options” to support troubled banks, which they feared was eroding with taxpayer bailouts rather than creditor losses. Notably, the report included both formal aspects of the BRRD and “creative, unintended routes,” but lamented inconsistencies in timing.

Outside BRRD jurisdiction, the Swiss authorities’ unorthodox write-down of Credit Suisse’s (CS) additional Tier 1 (AT1) securities—without completely wiping out shareholders—in the UBS acquisition of CS indicate the possibility of policy optionality within a defined range. The CS intervention, however, was an unexpected outcome for the market, as few predicted that AT1 creditors would suffer losses without shareholders experiencing losses as well. In the substantial volatility of the AT1 market following the CS intervention, S&P issued a report discussing the implications for the broader AT1 market. Of particular interest to us:

the Credit Suisse events are a reminder that the authorities have the capacity to put in place alternative solutions, such as facilitating a merger or acquisition, while still potentially triggering loss absorption for AT1 investors. The effect on AT1 investors will depend on the terms and nature of the instrument, and the legal and regulatory flexibility of the jurisdiction.

These comments show a ratings agency’s receptiveness to “alternative solutions” for instruments that by design have a range of possible outcomes. While this policy option looks likely to be phased out given the uncertainty it caused, it demonstrates that garnering the favor of the ratings agencies need not simply mean the government must provide a full “bailout.” Swiss financial regulator FINMA’s postmortem on the Credit Suisse resolution directly states that policymakers studied reports from the ratings agencies while designing restructuring measures.

In the examples discussed in this section, policymakers had access to a range of tools that were clearly documented and from which they ultimately chose. Evidence that the intervention strategy occurred within “bounds” of government power proved a positive for the ratings agencies, despite the at times significant policymaker discretion exercised in its approach. Consideration of the usefulness of communication and documentation of available tools by policymakers is thus warranted—and can be done outside of crisis-time—to help maximize crisis-time policy optionality. Predictability and policy optionality can exist in tandem given government credibility and communication of government latitude. Clear articulation of the “bounds” available to policymakers does not inhibit, and perhaps strengthens, the capacity for creative usage of policy tools during acute crises.

Conclusion

The ratings agencies are valuable stakeholders to consider as tools for bolstering government credibility and promoting the restoration of confidence. Policymakers have historically adjusted policy decisions and crisis communications retroactively to meet agency models and to improve reception by the ratings agencies. Proactively incorporating the response of these critical stakeholders earlier in the program design process saves valuable time during the acute phase of a crisis and can help limit deleterious economic effects and their perpetuation. It also leverages the role of ratings agencies as a source of information and guidance for the market participants that policymakers are ultimately looking to influence. Just as a strong rating instills market confidence in a financial instrument, a ratings agency’s satisfaction with an intervention contributes to its effectiveness. Ultimately, consideration of the credit ratings agencies in the design of emergency assistance programs improves policy outcomes and reinforces policy optionality by enhancing credibility, which is of value to any policymaker in a moment of crisis.

[1] For instance, before turning to open market purchases, the Federal Reserve rolled out repurchase agreement (repo) operations in response to the September 2019 repo market dysfunction and the 2020 COVID-19 pandemic. Similarly, after the Bank of England (BoE) was forced to step into the gilt market with asset purchases in 2022, then-executive director for markets at the BoE said, “If temporary central bank liquidity assistance might be required, much our preferred approach from a financial risk perspective was to lend against collateral, either via the banking system, or directly.”

[2] The Swiss parliament’s investigative commission on the Credit Suisse failure recently reported facing this very conundrum. While the ultimate outcome of the Credit Suisse rescue were perhaps perceived as unconventional, the parliamentary commission noted (p. 274) that the continuous provision of central bank emergency liquidity would have been insufficient as a lone policy measure because, given its seniority, it would have inter alia put undue pressure on the firm’s credit rating.

[3] To be sure, the newly passed Emergency Economic Stabilization Act of 2008 (EESA) that authorized the Troubled Assets Relief Program (TARP) also helped on this front, as the Treasury was then able to inject capital into AIG as well.

[4] For a further discussion of this idea, see: Kelly, Steven. “Central Bank Liquidity Assistance: Challenges of Franchise and Asset Values in Banking Crises.” Speech, May 20, 2024. https://www.atlantafed.org/-/media/documents/news/conferences/2024/05/19/financial-markets-conference/presentations/kelly-remarks.pdf