HDThe Debt
As of year-end 2019, nonfinancial business debt (BD) and household debt (HD) as a share of GDP were at similar levels of around 74 percent, and yet Federal Reserve Financial Stability Report suggested that BD posed greater risks to financial stability than HD2. Since the onset of the pandemic, the size of aggregate BD has increased considerably as a result of roughly $1.25 trillion of new issuance, while HD has grown by less than $100 billion. This note looks through the lens of fire-sale risks to show why nonfinancial BD is more concerning for financial stability than the HD. We examine BD and HD according to holder types, credit quality, and market liquidity features, and we highlight the areas of greatest concern for fire sales. Although extensive policy measures have helped ease the strains, fire-sale risks, especially with BD, remain salient at least in the near term.
HDThe Debt
Although the COVID-19 pandemic is not over and it is too early to draw definitive conclusions, our analysis takes into account the economic impact seen in recent months, such as the increase in the level and riskiness of business debt in the first half of 2020. We also address how several of the recent Federal Reserve policy measures have helped mitigate the fire-sale risks we have identified, and we discuss where such risks remain. In addition, in the wake of the COVID-19 crisis, business leverage may continue to increase relative to HD. Therefore, the BD fire-sale risks that we raise in our analysis may be an even greater concern in the post-COVID-19 period.
Fire-sale vulnerabilities refer to spillover effects and pecuniary externalities imposed on the rest of the financial system by financial institutions engaging in forced sales of assets at prices below fundamental values. As such, they capture the risk that shocks to individual institutions or assets prices get transmitted to other institutions and amplified through cascading sales in falling markets. Such risks are closely related to the fire-sale tendencies of institutions holding these assets and to the liquidity features of the underlying debt, which are generally not explicitly accounted for by simple measures of vulnerability, such as lending standards, credit quality, and leverage.
The rest of the note is organized as follows. Section 2 describes our framework for assessing fire-sale vulnerabilities in the household and business sectors. Sections 3, 4, and 5 present our analysis of the key determinants of fire-sale risks in these sectors, namely the debt holder profile, credit quality, and market liquidity. Section 6 discusses the recent market turmoil and its implications for fire-sale vulnerabilities going forward.
Existing literature modelling fire-sale risks assumes that in periods of stress banks (and asset managers) sell assets in equal proportion5. In contrast, our analysis takes into account the possibility that in periods of stress financial institutions might sell their most liquid holdings (liquid securities trading in centrally-cleared markets) earlier and/or in greater proportion relative to less liquid assets (whole loans, less liquid securities traded in OTC markets, etc.). There is empirical evidence that supports our approach. For example, Haddad, Moreira and Muir (2020) show that investors tried to sell safer, more liquid securities to raise cash during acute phase of the COVID-19 crisis. We focus on three main factors that, in our view, determine the likelihood and severity of fire-sale risks in debt markets: 1) holder profile, 2) credit quality, and 3) market liquidity.
Similar to debt at the triple-B cliff, information-sensitive debt or debt that is difficult to price, which is typically the case for low- or unrated issuers, may be more susceptible to fire sales than the debt of highly rated, publicly traded companies.
The third, and final, determinant of fire-sale risk we consider is the ease with which debt can be sold in times of stress. Two factors are at play here: the degree of standardization of the debt and the level of market liquidity. Regarding debt standardization, an active secondary market does not exist for some bespoke forms of debt held by banks on their balance sheets, such as credit cards or bank loans. While the lack of a market for such debt may increase the price impact of forced sales, it makes it harder for banks to sell the debt in times of stress, incentivizing banks to sell other forms of debt that are more easily brought to the market first. In addition, because non-standardized debt is not generally marked to market on a regular basis, forced sales of such debt are less likely to cause direct losses to other financial institutions holding similar debt and to trigger further rounds of fire sales.
Regarding the level of market liquidity, debt in the form of relatively illiquid securities, such as high-yield corporate bonds, may generate considerable fire-sale losses to those who hold it, especially in response to downgrades or large investor redemptions. This result is obtained because when an institution holding these securities is forced to sell, the lack of liquidity in the market and the forced selling puts greater downward pressure on the prices of the securities. In contrast, forced sales of more liquid debt, such as agency MBS or Treasury securities, may generate only small fire-sale externalities, because the price impact of trades in deep markets tends to be small or because the markets for these securities enjoy explicit government support in times of stress (e.g. MBS).
While the risk is judged to be low, we wanted to quantify the fire-sale risk. One way to quantify potential fire-sale losses at banks is to utilize the results of the annual stress test (DFAST). The Severely Adverse Scenario is designed to generate losses that implicitly account for spillover effects and other non-linearities that are expected to be at play in deep recessions but that are difficult to model explicitly. The difference between the DFAST losses under the Severely Adverse Scenario and those from the Adverse Scenario may thus serve as a proxy for fire-sale losses in times of severe market stress. That said, some types of debt may be hit particularly hard in the severely adverse scenario and yet be less prone to fire sales, as only a relatively small fraction of this debt is traded in a standardized form in secondary markets, e.g. credit card debt8.
Moving on to GSEs, they are the second largest holder of household debt (37%), being especially prominent in the residential mortgage segment of HD, where they hold or guarantee roughly two-thirds of all mortgage balances10. By contrast, because GSE are established to support the mortgage market, they only hold negligible amounts of business debt. Of the GSEs, Fannie Mae and Freddie Mac are currently under the conservatorship of the FHFA and enjoy the full backing of the federal government, similar to that of Ginnie Mae11. Funding pressures and fire-sale risks at GSEs are therefore low.
Table 3 summarizes the current breakdown of outstanding BD and HD by rating category. HD has generally higher credit quality, with prime borrowers accounting for 66% of outstanding debt, compared with 49% in the high investment grade category of BD. The amount of debt at the speculative-grade boundary is also higher in the BD sector (25%) than in the HD sector (19%). Combined with the fact the holders of BD are much more sensitive to rating downgrades than those of HD, the triple-B cliff poses greater fire-sale risks in the BD sector.
To estimate the fraction of this debt that is at risk of being downgraded to speculative grade, we rely on historical downgrade frequencies observed in episodes of significant stress emanating from the corporate sector. J.P. Morgan (2018) estimate that when the dot-com bubble burst in March 2000, around 10% of outstanding BBB bonds experienced a downgrade to speculative grade over the following year and around 25% experienced a downgrade over the next 3 years. Using these downgrade rates, around $140 ($345) billion worth of corporate bonds held by mutual funds and insurers are currently at risk of becoming fallen angels over the next year (three years) should the current recession becomes prolonged.
Although banks do not face the same fire-sale risks associated with fallen angels as insurers and mutual funds do, credit downgrades can have significant impact on banks as well. In addition to causing direct losses on outstanding loans due to a simultaneous increase in default rates, an economic shock causing a wave of corporate downgrades would likely also involve additional liquidity pressure on banks. The newly downgraded firms may face tighter funding conditions in primary debt markets and resort instead to drawing on bank credit lines and looking to banks to roll over their existing debt. Additionally, downgrades can trigger a breach of loan covenants, limiting the amount of liquidity newly downgraded firms can access under their existing credit agreements with banks13.
Table 5 summarizes the current outstanding debt by market liquidity. According to our current best estimates, 49% of outstanding BD is in the form of securities (either corporate bonds or otherwise securitized debt), compared with 74% in the HD sector. The remaining 51% of BD and 26% of HD are individual loans (not in securitized form) on balance sheets of banks and other financial institutions. Turning to market liquidity, we estimate that around 54% of HD is liquid, primarily in the form of GSE-backed MBS, while only around 35% of BD is liquid (IG corporate bonds). Thus, we judge that, should fire sales occur, the price impact and the associated spillover losses would be higher in the BD sector.
Conditions also deteriorated significantly in some household debt markets. Notably, in mortgage markets, primary mortgage rates increased sharply in early March amid deteriorating secondary market liquidity and elevated volatility. MBS spreads widened significantly and exceeded levels last seen during the European debt crisis (Figure 5). To restore orderly market functioning in the primary and secondary mortgage markets, on March 15 the FOMC announced that over the coming months the Committee will increase its holdings of agency mortgage-backed securities by at least $200 billion. After the Fed announced the MBS purchases, mortgage market functioning gradually improved and spreads returned to normal levels. 041b061a72