The COVID-19 pandemic has inflicted a severe and unprecedented shock on the economies of Europe and the world. Against this background, the General Board of the European Systemic Risk Board (ESRB) decided at its meeting on 2 April 2020 to focus its attention on five priority areas where coordination among authorities or more broadly across the EU is likely to be particularly important to safeguard financial stability. One of those priority areas was the procyclical impact that downgrades of corporate bonds might have on markets and entities across the financial system (ESRB 2020).
Risks for financial stability from Covid-19 and the rationale for a cross-sectoral fallen angel study
To contain the spread of the coronavirus, drastic lockdown measures were implemented by many nations across the globe between March and July 2020. While slowing the spread of the virus, the lockdowns have also hit economic activity, with consumption and production falling sharply in response. Sectors such as tourism, aviation, hospitality, and entertainment have been particularly severely hit by the restrictions. The lockdown-induced falls in aggregate consumption, investment, and world trade have generalised the recession into the ‘Great Cessation’. With bleak cash flow outlooks, numerous companies that had exploited the low financing costs available in recent years to build leverage are now facing difficulties in meeting their payments on issued bonds and may even default on such payments. Credit rating agencies have already begun to downgrade a range of companies as well as securities like collateralised loan obligations (CLOs). Bonds transitioning from BBB (or higher) to BB (or lower) cross the threshold from ‘investment grade’ to ‘high yield’ (speculative),1 thus becoming ‘fallen angels’. To date, 16 euro area non-financial companies have seen their bonds become fallen angels, corresponding to €31.5 billion BBB bonds downgraded (out of a total amount outstanding of BBB non-financial corporation (NFC) bonds in excess of €650 billion). The downgrades have occurred at a slower pace than witnessed during the global financial crisis, but at a higher pace than during the euro area sovereign crisis (see Figure 1).
When bonds transition to fallen angels, it has important implications for investors. For instance, passive funds tracking investment grade corporate indices are forced to remove the fallen angels from their portfolios. Other funds with fixed mandates (for example, but not only, passive funds) or self-imposed constraints from their investment strategies may face similar requirements to sell these bonds. While funds have been given more time by IG bond index providers to make the transitions and smoothen the rebalancing, the adjustments will eventually have to be made.
Beyond the investment fund sector, other financial institutions may also have incentives to sell fallen angels: the larger credit risk attracts higher capital charges for banks and insurance companies and it may become uneconomical to hold the bonds, on a risk-adjusted basis. Moreover, there may be self-fulfilling negative spirals of ‘fire sales’ (Cont and Wagalath 2016a, Khandani and Lo 2011), in which downgrades force investors to sell their bonds, which depresses prices, leading to increased selling pressure by other investors in turn. When too many investors are ‘running for the exit’ (Pederson 2009) at the same time, prices can drop significantly below fundamental values and require time before readjusting (Duffie 2011, Shleifer et Vishny 1992).
In times of market turmoil, it may be difficult to disentangle price drops that reflect a change in fundamentals from those due ‘merely’ to order imbalance and too many sellers competing for too few buyers. Amid enormous market volatility and uncertainty during the global financial crisis, the ABX.AAA dropped below 30 cents on the dollar, although it did recover most of the losses in the post-crisis period (see Figure 2). The extent of the market disruption can thus be very sizeable and depends, beyond the overall degree of uncertainty, primarily on the amount of sellers and the desired sales volume relative to the amount of willing buyers and average trading volumes. Considering the size of a position relative to the available market depth is of crucial importance in risk management, as the liquidation value (i.e. the value at which the position can be converted to cash) for large positions can be considerably smaller than its market value.2
Figure 2 Closing prices of the ABX.HE.AAA 2007-01 series
By its very design, an analysis that seeks to understand and quantify what could happen under theoretical behaviours of market participants in a hypothetically stressed market must be imbued with uncertainty. Our study sought to embrace this uncertainty by providing ‘ranges of losses’ under various behavioural and liquidity assumptions as opposed to a single point estimate. More specifically, the analysis first considered two scenarios of severe and very severe downgrades, ranging from approximatively 25% BBB downgraded to 45% BBB downgraded. Second, three sets of increasingly severe behavioural assumptions were applied:(i) only passive investment funds sell their fallen angels (mild behavioural assumption); (ii) in addition to (i), active funds, insurance companies and pension funds also sell a portion of their holdings (severe behavioural assumption); and (iii) passive, active and pension funds, as well as insurers, sell all of their holdings (extreme behavioural assumption). It is assumed that banks, hedge funds, distressed debt funds or sovereign wealth funds would step in as buyers. In the extreme scenario where insurers, passive funds, active funds and pension funds all seek to sell their fallen angels, it is indeed likely that markets would break down and freeze, meaning sellers are no longer able to find buyers. Finally, the third layer of uncertainty concerned the price impact parameters which specify by how much prices decline for a given amount of sales. A range from low market liquidity with high price impacts to high market liquidity with small price impacts captured the uncertainty in this parameter. All together, the analysis comprises twelve scenarios. For a more detailed description of the approach taken, and applicable caveats, we refer the reader to the technical note.
As shown in Table 1, our simulation study suggests that the system-wide initial losses from the yield shocks could lie between €146 and €213 billion, which would be accompanied by fallen angel volumes of €232 to €443 billion. The volume of sales that is triggered depends critically on the behavioural assumptions and ranges from €30 to €198 billion euro for the increasingly severe behavioural assumptions under scenario 1. Similarly, in scenario 2, the volume of sales ranges from €65to €373 billion euro. These sales could in turn generate losses that range between €2 billion (the mildest behavioural assumption in scenario 1) and €85 billion (the most severe behavioural assumption in scenario 2). Overall, the analysis suggested that the impacts from fire sales could generate an additional loss of the order of 20–30% on top of the initial loss (which reflects the change in fundamentals).
Table 1 Summary of findings from the system-wide large-scale corporate bond downgrade simulation
Initial losses from downgrades (in all rating categories), volume of fallen angels, volume or sales and lower and upper bounds for losses resulting from fire sales (€ billions)
Sources: ESAs, Bank of England and ESRB Secretariat calculations.
Note: Owing to data aggregation issues, it was not possible to provide a breakdown of the losses into those on bonds issued by non-financial corporations and those on bonds issued by banks.
The magnitude of these fire-sale losses depends on so-called “portfolio overlaps” (Cont and Schaanning 2019), which quantify the degree of common assets between two portfolios. Figure 3 shows the network of portfolio overlaps for European corporate bonds, which gives rise to this price-mediated indirect contagion. The network displays the European insurance sector (dark green), investment fund sector (bright green), banking sector (blue) and pension fund sector (yellow) with the largest ten overlaps highlighted in orange. The figure reveals two interesting features. First, the most significant overlaps are between the insurance and investment fund sector; indeed, the pension fund sector is relatively disconnected from the rest of the network. Certain sectors in a number of countries present very considerable overlaps, exposing them to fire sale risk. If, for instance, French insurers were to engage in a fire sale of assets, this would have the most important repercussions for German and Italian insurers, as well as German banks and Luxembourgish and German investment funds. This relationship is symmetric, i.e. French insurers would be equally affected from a fire sale by Luxembourgish investment funds.
Figure 3 Cross-sectoral portfolio overlaps in the European corporate bond network
Lessons and implications for the future
We can draw a number of lessons from this analysis for the future:
- While credit rating agencies focus on individual securities and issuers, policymakers also need to take a system-wide view, especially when they have macroprudential responsibilities. The analysis demonstrates that, within reasonable uncertainty bounds, this is achievable and yields important insights that are not available at the idiosyncratic level – for instance, the potential for and extent of contagion can be measured and quantified only when taking this genuinely system-wide perspective. While important in and of themselves, country-specific or sector-specific analyses are not capable of uncovering international and cross-sectoral links which can become important channels for contagion in systemic crises.
- The dissemination of information on portfolio overlaps can improve institutions’ own risk management. The German banking sector, for instance, may be unaware of the fact that it has very similar portfolio holdings in European corporate bonds to the French insurance sector and that both sectors are thus exposed to fire sales risk from the other.
- Our findings suggest that future financial stability analyses may benefit greatly from cross-border and cross-sectoral system-wide stress simulations. Only this type of analysis has the power to reveal interdependencies between sectors and countries. This is of particular importance when modelling dynamics and the reactions of critical institutions to stress. Current approaches, using aggregate-level data, are unable to differentiate between concentrated and large exposures on the one hand and distributed smaller exposures on the other hand. In this sense, our approach offers a first step to building a bridge between macroprudential oversight and microprudential supervision: the system-wide analysis allows uncovering critical nodes and links of the financial infrastructure, which should then attract higher supervisory scrutiny. However, this requires different datasets to be linked. For instance, as no European regulator currently has access on a standing basis to granular cross-sectoral data, it is not possible to tell whether the overlaps identified in Figure 1 arise from a concentration in the various countries’ largest institutions, or whether these overlaps are due to the entire sectors of the countries being similarly exposed.
- One may even go a step further and peek beyond the borders of the EU: while asset markets and the financial system are global, cross-jurisdictional supervision and oversight are still in their infancy. Important hurdles are, for instance, the legal inability, or general reservations, to share confidential data across jurisdictions. However, it is technologically possible to conduct the above discussed analysis on portfolio overlaps at the global level without revealing confidential information across jurisdictions and thus safeguarding the confidential nature of the information. International institutions and organisations such as the Bank for International Settlements, the Financial Stability Board or the IMF may be suitable candidates for coordinating such analyses in the future.
Authors’ note: We are grateful to the following persons who contributed to the analysis in the report: Antoine Bouveret, Casper Christophersen, Geoff Coppins, Massimo Ferrari, Christoph Fricke, Camille Graciani, Sandra Hack, Emilio Hellmers, Petr Jakubik, Michiel Kaijser, Maximilian Ludwig, Luca Mingarelli, Ángel Monzon, Benjamin Mosk, Filip Nikolic, Stefano Pasqualini, Daniel Pérez, Elena Rancoita, Matthias Sydow and Anna Vinci. Any views expressed are those of the authors and do not necessarily reflect the official stance of the ESRB, its member institutions, or any institution to which the authors may be affiliated.
Cont, R and L Wagalath (2016a), “Fire sales forensics: measuring endogenous risk”, Mathematical Finance 26(4): 835–866.
Cont, R and L Wagalath (2016b), “Risk management for Whales”, Risk.net Cutting edge.
Cont, R and E Schaanning (2019), “Monitoring Indirect Contagion”, Journal of Banking and Finance 104: 85–102.
Duffie, D (2010), “Presidential Address: Asset price dynamics with slow-moving capital”, Journal of Finance 65(4).
Ellul, A, C Jotikasthira and C Lundblad (2011), “Regulatory pressure and fire sales in the corporate bond market”, Journal of Financial Economics 101(3): 596–620.
ESRB (2020), “A system-wide scenario analysis of large-scale corporate bond downgrades”, ESRB technical note.
Khandani, A E and A W Lo (2011), “What happened to the quants in August 2007? Evidence from factors and transactions data”, Journal of Financial Markets 14(1): 1–46.
Pedersen, L H (2009), “When everyone runs for the exit”, International Journal of Central Banking, December.
Shleifer, A and R Vishny (1992), “Fire sales in finance and macroeconomics”, Journal of Economic Perspectives 25(1): 29-48.
1 For Fitch and S&P; these are equivalent to the transition from Baa to Ba using Moody’s scale.
2 See Cont and Wagalath (2016b), where the authors illustrate how the ‘London Whale’ trade at JP Morgan, which had a reported value-at-risk of around $500 million, ended up generating a liquidation loss of $6.2 billion.