Regulating the doom loop

Spyros Alogoskoufis, Sam Langfield 03 October 2019

a

A

Sovereigns are exposed to bank risk and banks are exposed to sovereign risk. During crises, this two-way risk exposure interacts in a ‘doom loop’ to create additional risk endogenously (Farhi and Tirole 2018). At a leaders’ summit in June 2012, euro area governments recognised the imperative of breaking the doom loop. But bank regulation still treats sovereign debt as risk-free and does not penalise concentrated portfolios.

This column, based on a model of portfolio reallocation by euro area banks following hypothetical regulatory reforms (Alogoskoufis and Langfield 2019), sheds light on two questions. First, would banks reduce portfolio concentration in response to reforms? Second, would they reduce exposures to sovereign credit risk? Simulations show that the answer is never an unambiguous and simultaneous ‘yes’ to both questions under reforms envisaged by the Basel Committee on Banking Supervision (2017). In fact, under plausible conditions, portfolios become substantially more concentrated or risky, highlighting potential unintended consequences of regulatory reforms that focus only on concentration or credit risk in isolation.

Reform design can be improved by incentivising banks to reduce both concentration and credit risk. Yet all designs face a common constraint: the incompleteness of euro area sovereign bond markets. With the current investible universe, it is impossible to assemble a portfolio with levels of concentration and credit risk that are sufficiently low to avoid a doom loop. The intuition is that a low-risk portfolio would be highly concentrated (in German bonds), whereas a low concentration portfolio is moderately risky. A new asset that embeds both properties – either through fiscal or contractual innovation – would therefore make financial markets more complete. We show that regulation can complement a common safe asset by enhancing incentives for banks to reinvest into it.

Portfolio concentration and credit risk

Most euro area banks’ sovereign portfolios are heavily home biased. Brunnermeier et al. (2017) show that national doom loops can be avoided if moderately well capitalized banks were to diversify their portfolios. For this reason, several policy experts – including Véron (2017) and Gros (2018) – have advocated a regulatory regime that encourages banks to diversify, for example with positive risk-weights for concentrated sovereign portfolios. A bonus feature of diversification in the euro area is that cross-border portfolios do not bear currency risk. 

However, diversification can give rise to contagion. Banks are vulnerable to the re-pricing of both domestic and foreign sovereign debt when such securities are widely used in repo markets (Bolton and Jeanne 2011) or when equity levels are low relative to sovereign portfolios (Brunnermeier et al. 2017). Under these conditions, an international doom loop can occur in which sovereign distress anywhere affects banks and sovereigns everywhere. This has significantly more devilish implications for financial stability than merely national doom loops, since in the latter case sovereign distress affects only domestic banks. 

Outstanding repos backed by euro sovereign bonds amount to approximately €4.5 trillion;1 the average euro area bank holds nearly 200% of its book equity value in sovereign bonds.2 Hence, the dark side of diversification – contagion – is an empirically relevant concern. Breaking national doom loops while avoiding an international one requires banks to hold sovereign portfolios that embed low credit risk in addition to low concentration. This is the central insight against which regulatory reform ideas should be evaluated.

Measuring sovereign portfolios

The European Banking Authority regularly discloses banks’ sovereign bond holdings. We use data on mid-2017 holdings for a sample of 95 euro area banks; the appendix uses data from other vintages. Guided by theory, we measure portfolios according to their concentration and credit risk. Concentration is measured as the sum of squared deviations from portfolio weights given by the ECB capital key (which is based on the relative economic and population size of member states). Credit risk is measured as the five-year expected loss rate using a stochastic model of sovereign default introduced to the literature by Brunnermeier et al (2017). (Our main paper also reports alternative measures of concentration and credit risk.)

Figure 1 plots these two portfolio measures for our main sample. While there is wide cross-sectional dispersion, no bank’s portfolio has both low concentration and low credit risk. In fact, such a portfolio cannot be created from existing securities. The minimum value of the expected loss rate (ELRate), 0.5%, can be achieved only with a portfolio comprised exclusively of German sovereign bonds – but this portfolio has a very high deviation from the ECB capital key (KeyDeviation). Conversely, a portfolio weighted by the ECB capital key has a material ELRate of 4.4%. This insight provides early intuition for the tension between concentration and credit risk.

Figure 1 Concentration and credit risk in sovereign portfolios

Note: Crosses refer to the sovereign portfolios of sample banks as of mid-2017, measured by concentration (i.e. KeyDeviation, the sum of squared deviations from the ECB capital key) and credit risk (i.e. ELRate, the five-year expected loss rate). Banks are grouped by the rating of their home country. As a benchmark, the black diamonds denote the two corner solutions: min(KeyDeviation)=0, at which ELRate=4.4%; and min(ELRate)=0.5%, at which KeyDeviation=18.5.

Modelling endogenous portfolio reallocation

How might portfolios change in response to regulatory reform? Despite the abundance of reform ideas, there is little analysis of the impact on sovereign portfolios. Schneider and Steffen (2017) provide insightful quantitative assessments of the impact of reforms on capital requirements. However, they assume that banks maintain their current sovereign portfolios – the elasticity of holdings with respect to capital requirements is taken to be zero. Hence, such analyses characterise only a special case of banks’ reaction functions. 

To provide a more general characterisation, we model endogenous portfolio reallocation in response to regulatory reform. In the model, banks deviate from their extant portfolio allocation insofar as reallocation achieves lower capital requirements. Corner solutions are characterised by banks choosing a sovereign portfolio allocation that globally minimises capital requirements. We also quantify the full range of intermediate elasticities whereby banks partially reallocate portfolios. 

While portfolio allocation is endogenous in the model, total holdings of sovereign bonds are inelastic with respect to their regulatory treatment. This assumption is motivated by the insight that banks use euro area sovereign bonds as liquid stores of value and as collateral. In addition, regulation requires banks to hold liquid assets, such as sovereign bonds, to comply with liquidity requirements.

These conditions generally allow for multiple solutions because banks can minimise capital requirements with many different portfolio allocations. For example, a regulatory penalty for holding securities rated below a critical threshold can be avoided by any arbitrary portfolio comprising securities rated above that threshold. To establish unique solutions, we focus on three illustrative cases, which apply insofar as portfolio reallocation can lower banks’ capital requirements:

  • Prudent case: banks reinvest into the lowest-risk sovereign bond that attracts the lowest capital charge. This provides a limiting case of the most conservative portfolio allocation under a given regulatory reform.
  • Base case: banks reinvest into the sovereign bond that most closely matches their initial portfolio characteristics and attracts the lowest capital charge.
  • Imprudent case: banks reinvest into the highest-risk sovereign bond that attracts the lowest capital charge. This provides a limiting case of the greatest credit risk exposure that banks would assume under a given regulatory reform.

Substantial portfolio reallocation is likely to affect relative prices. One would expect the price of sovereign bonds to increase (decrease) when banks are on aggregate net buyers (sellers). Our model abstracts from these general equilibrium effects on relative prices. In that sense, our results can be interpreted as a lower bound on the unintended consequences of regulatory reform.

Results for post-reform portfolio allocation

The model is applied to two regulatory reforms envisaged by the Basel Committee on Banking Supervision (2017). In the first (shown in Table 1, Panel A), risk-weights are an increasing function of single-name holdings. In the second (Panel B), risk-weights are set according to sovereign credit ratings. (In Alogoskoufis and Langfield (2019), we also apply the model to quantitative restrictions on sovereign portfolios.)

Table 1 Options to reform the regulatory treatment of sovereign exposures

Note: The table reports calibrations of two ideas to reform the regulatory treatment of sovereign exposures, as mooted by the Basel Committee on Banking Supervision (2017).

For each of these proposed reforms, the model determines a new set of sovereign portfolios which can be summarised by their concentration and credit risk. Figure 2 plots the results for concentration-based risk-weights, and Figure 3 for ratings-based risk-weights. In each figure, the fan represents the cross-sectional dispersion across bank portfolios. The horizontal axis captures the extent of reallocation: at 0%, banks just hold their mid-2017 portfolio; at 100%, banks hold the capital requirement minimising portfolio. This can be interpreted as the elasticity of portfolio allocation with respect to regulation.

Concentration-based risk-weights unambiguously reduce concentration. Nevertheless, portfolios remain concentrated relative to ECB capital key weights even after 100% reallocation (Figure 2, top panel). Moreover, in the base case and particularly in the imprudent case, concentration-based risk-weights are consistent with banks increasing their sovereign credit risk exposure (bottom panel). Such an outcome could strengthen the doom loop and lead to its international propagation. 

Figure 2 Concentration-based risk-weights and portfolio reallocation

Note: The fan represents the cross-sectional dispersion across 95 euro area bank portfolios of concentration (measured by KeyDeviation, i.e. the sum of squared deviations from the ECB capital key) in the top panel and credit risk (measured by ELRate, i.e. the five-year expected loss rate) in the bottom panel. The horizontal axis captures the extent of reallocation: at 0%, banks just hold their mid-2017 portfolio; at 100%, banks hold the capital requirement minimizing portfolio.

By contrast, ratings-based risk-weights can exacerbate portfolio concentration (Figure 3, top panel), although credit risk is generally declining in the prudent and base cases and stable in the imprudent case (bottom panel). High concentration – even in ostensibly low-risk sovereigns – can be problematic insofar as idiosyncratic sovereign credit risk is time-varying. Thus, concentrated holdings of foreign sovereign debt can generate cross-border contagion.

Figure 3 Ratings-based risk-weights and portfolio reallocation

Note: The fan represents the cross-sectional dispersion across 95 euro area bank portfolios of concentration (measured by KeyDeviation, i.e. the sum of squared deviations from the ECB capital key) in the top panel and credit risk (measured by ELRate, i.e. the five-year expected loss rate) in the bottom panel. The horizontal axis captures the extent of reallocation: at 0%, banks just hold their mid-2017 portfolio; at 100%, banks hold the capital requirement minimizing portfolio.

Common safe assets

A feature of reform designs mooted by the Basel Committee on Banking Supervision (2017) is that they target only concentration or credit risk in isolation. This drives our findings concerning the potential unintended consequences of regulatory reform. In principle, it is possible to improve on the Basel designs, for example by setting risk-weights as a function of both concentration and credit risk.

However, even a first-best design is constrained by a missing feature of euro area sovereign debt markets. A portfolio with both low concentration and low credit risk can only be assembled if the investible universe is expanded to include a security that entails both properties. We refer to such an area-wide and low-risk security as a ‘common safe asset’.

In the absence of fiscal innovation, a common safe asset must be financially engineered through contractual innovation. Contract design options include ‘pooling-then-tranching’ existing sovereign debt (as in the feasibility study of the High-Level Task Force on Safe Assets 2018) or ‘tranching-then-pooling’ it (as suggested by Buti et al. 2017 and analysed by Leandro and Zettelmeyer 2018). With appropriate calibration, either option can be designed to have negligible levels of both concentration and credit risk. 

Figure 4 Portfolio reallocation into a common safe asset

Note: The fan represents the cross-sectional dispersion across 95 Eurozone bank portfolios of concentration (measured by KeyDeviation, i.e. the sum of squared deviations from the ECB capital key) in the top panel and credit risk (measured by ELRate, i.e. the five-year expected loss rate) in the bottom panel. The horizontal axis captures the extent of reallocation: at 0%, banks just hold their mid-2017 portfolio; at 100%, banks hold the capital requirement minimizing portfolio. Note that in this application of the model there is no variation across the prudent, base and imprudent cases, because a portfolio comprised of a common safe asset always represents the unique solution to the constrained optimization problem.

Would banks reinvest in such an asset? In a final step, we apply our model to a world in which a common safe asset exists alongside national sovereign debt. Surprisingly, neither of the two reforms envisaged by the Basel Committee on Banking Supervision (2017) would provide strong regulatory incentives for banks to reinvest into a common safe asset, since they allow multiple solutions to the constrained optimisation problem. To strengthen regulatory incentives, one option would be to set a positive risk-weight floor on all single-name sovereign holdings (either on its own or in combination with other reforms). In this way, holding a common safe asset uniquely minimises capital requirements, regardless of the reallocation rule that banks adopt. Therefore, at 100% reallocation, sovereign portfolios are comprised entirely of the common safe asset, implying minimal levels of both concentration and credit risk, as shown in Figure 4.

Policy conclusion

Mainstream approaches to reform the regulatory treatment of sovereign exposures face a tension between concentration and credit risk, as summarised in Table 2. This tension implies that reform could backfire. If portfolio concentration is too high, banks will be vulnerable to time-variation in idiosyncratic sovereign credit risk. If credit risk is too high, banks will be vulnerable to both domestic and foreign sovereign debt re-pricing, potentially giving rise to an international doom loop.

Table 2 Summary of results

Note: The table summarises the main results in Alogoskoufis and Langfield (2019).

All regulatory designs are constrained by the incompleteness of euro area sovereign debt markets, which make it impossible to assemble a portfolio that has sufficiently low concentration and credit risk. Hence, breaking the doom loop requires the portfolio opportunity set to include a common safe asset. Well-designed regulatory reform complements such an asset by incentivizing banks to reinvest into it. This policy conclusion supports the approach of Bénassy Quéré et al. (2018), who advocate a common safe asset for the euro area alongside regulatory reform. The complementarity between safe assets and regulation was also acknowledged by the European Commission (2019) in its contribution to the recent EU27 leaders’ meeting in Sibiu. If policymakers act on this insight, Europe’s devilish doom loop may yet be broken.

References

Alogoskoufis, S and S Langfield (2019), “Regulating the doom loop”, International Journal of Central Banking, forthcoming.

Basel Committee on Banking Supervision (2017), “The regulatory treatment of sovereign exposures – discussion paper”.

Bénassy Quéré, A, M Brunnermeier, H Enderlein, E Farhi, M Fratzscher, C Fuest, P-O Gourinchas, P Martin, F Pisani, H Rey, N Véron, B Weder di Mauro, and J Zettelmeyer (2018), “Reconciling risk sharing with market discipline: A constructive approach to euro area reform”, CEPR Policy Insight 91 (summary available on Vox).

Bolton, P and O Jeanne (2011), “Sovereign default risk and bank fragility in financially integrated economies”, IMF Economic Review 59(2): 162-194.

Brunnermeier, M, S Langfield, M Pagano, R Reis, S Van Nieuwerburgh, and D Vayanos (2017), “ESBies: Safety in the tranches”, Economic Policy 32(90): 175-219. Summary available on Vox.

Buti, M, S Deroose, J Leandro and G Giudice (2017), “Completing EMU”, VoxEU.org, 13 July.

European Commission (2019), “Preparing for a more united, stronger and more democratic Union in an increasingly uncertain world”, contribution to the EU27 leaders’ meeting in Sibiu on 9 May 2019.

European Systemic Risk Board (2015), “Report on the regulatory treatment of sovereign exposures”.

Farhi, E and J Tirole (2018), “Deadly embrace: Sovereign and financial balance sheets doom loops”, Review of Economic Studies 85(3): 1781-1823.

Gros, D (2018), “Does the euro area need a safe or diversified asset?”, CEPS Policy Brief 2018/03. 

High-Level Task Force on Safe Assets, Sovereign bond-backed securities: a feasibility study, report of the European Systemic Risk Board (summary available on Vox).

Leandro, Á and J Zettelmeyer (2019), “The search for a euro area safe asset”, Peterson Institute for International Economics Working Paper 18-3 (summary available on Vox).

Schneider, Y and S Steffen (2017), “Feasibility check: Transition to a new regime for bank sovereign exposure?”, study provided at the request of the Economic and Monetary Affairs Committee, European Parliament. 

Véron, N (2017), “Sovereign concentration charges: A new regime for banks’ sovereign exposures”, study provided at the request of the Economic and Monetary Affairs Committee, European Parliament.

Endnotes

[1] Source: International Capital Market Association data.

[2] Source: European Banking Authority data.

a

A

Topics:  EU institutions Financial regulation and banking

Tags:  sovereign risk, euro area, doom loop, Vox debates

Research Analyst, ECB

Principal Economist, Financial Research Division, ECB

Events

CEPR Policy Research