Sovereign debtors and their national banking systems are closely linked through a range of direct and indirect channels. These include banks’ claims on sovereigns, semi-automatic links between sovereign and bank credit ratings, public backstops, collateral in banks’ operations, and the effects of fiscal distress on the overall economy – and thus the quality of bank loans (CGFS 2011, Bank of Italy 2013a). Because of these close interlinkages, problems arising in the banking sector can trigger downward spirals in which increased sovereign risk, banking system difficulties, and the deteriorating economic situation spill over and feed on each other. Recent examples include Spain and Ireland, where a banking crisis – triggered by the burst of a real estate bubble – caused sovereign difficulties. A similar spiral can be activated by sovereign problems (see e.g. the Greek case).
During the crisis, euro-area banks’ holdings of domestic government bonds recorded a significant increase, fuelling an intense debate. Several commentators have argued that this surge in ‘home country bias’ has reinforced the bank-sovereign loop, and that it is due to the preferential prudential treatment granted to domestic sovereign exposures (they are typically assigned low or zero risk weights and are exempted from the rules limiting large exposures), which should therefore be removed.
The increase in banks’ holdings of government bonds is mainly a consequence, rather than a cause, of the crisis
To understand this debate, it is useful to take stock of a few facts.
- First, in most euro-area countries, bank holdings of domestic government bonds consistently declined over the decade before the global financial crisis, up until the Lehman default in September 2008 (Figure 1).
Figure 1. Domestic government bonds held by banks in the Eurozone (as % of total assets)
Source: Angelini, Grande and Panetta (2014).
- Second, after September 2008 home country bias started to increase in most countries. This coincides with the aggravation of the crisis, well ahead of the outburst of the euro-area sovereign debt crisis.
- Third, data available for Italy show that these trends have been common to other sectors of the economy. The recent increase in home country bias has been even more pronounced for insurance companies and pension funds (Figure 2).
Figure 2. Domestic government bonds held by selected institutional sectors in Italy (as % of total assets)
Source: Angelini, Grande and Panetta (2014).
- Fourth, Italian banks have been steadily selling government bonds since last June (Figure 3).
Figure 3. Italian banks: cumulated net sales/purchases of domestic government bonds since January 2011 (€ billion)
Clearly, the preferential prudential treatment of sovereign exposures cannot explain these patterns; it may have contributed to the increase in the home country bias during the crisis, but other factors may have had more important effects.
- First, in this period banks and other financial institutions – fearing a break-up of the Eurozone – shifted their asset-liability management from hedging on a currency basis to hedging on a national basis (Battistini, Pagano and Simonelli 2013).
- Second, in the event of a domestic sovereign default, banks would be in serious difficulty, due not only to large direct exposures (even in Germany, large banks’ holdings of domestic sovereign bonds at the end of 2010 was around 150% of total regulatory capital), but also to the pervasive nature of sovereign risk: as we document in the next paragraph, when the sovereign runs into trouble, the entire domestic economy does. This suggests that banks may have felt that by investing in their sovereign debt they incurred little if any additional risk; a ‘simul stabunt, simul cadent’ type of story is likely to be part of the explanation of the resurgence in the home bias.
- Third, for Italy there is evidence that risk-adjusted returns on government bonds have exceeded those on loans since the end of 2011, due to the considerable deterioration of loan quality (Bank of Italy 2013b). This has likely been the case in other countries under stress.
- Finally, between 2011 and 2012, when wholesale funding markets had all but frozen, banks’ demand for government bonds was fuelled by precautionary hoarding of liquid assets.
Summing up, in normal (non crisis) times banks have no incentive to load up on government paper, regardless of the favourable regulatory treatment. The heavy purchases of domestic sovereign paper were concentrated in the crisis period; banks began to sell sovereign paper as soon as the economic situation began to normalize. Overall, this evidence is consistent with the view that the increase in banks’ holdings of government bonds has been mainly a consequence – rather than a cause – of the crisis, and must be interpreted in the light of the process of fragmentation of euro-area financial markets during the crisis, largely driven by redenomination risk.1
Sovereign credit risk affects both banks and non banks
The various links between a sovereign debtor and the domestic banking system briefly summarized in the introduction would lead to a hypothesis that the sovereign-bank relationship is “special”. If this were the case, we would expect the credit risk premia of sovereigns to be more closely related to those of banks than to those of non-financial companies. A simple correlation analysis of premia on CDSs suggests that this is not the case. Data for all the main euro area countries show that banks are about as correlated with sovereigns as are non-financial issuers (Figure 4; similar charts for other countries are in the companion paper to this column). Within each country the level of the sovereign-bank correlation and that of the sovereign-non-financial correlation are similar.2 Furthermore, there is no clear structural shift in their patterns following the onset of the sovereign debt crisis at the end of 2009, or since its aggravation in the summer of 2011.
Figure 4. Sovereign/bank correlation, sovereign/non-financial companies correlation, and banks’ holdings of government bonds in Italy (left panel) and The Netherlands (right panel) (60-day correlation of daily changes in CDS premia and % of total assets)
Source: Angelini, Grande and Panetta (2014).
Note: ρ denotes the correlation between the sovereign/bank correlation and banks’ holdings of government bonds.
This evidence does not allow us to draw any firm conclusions about causality. Nevertheless, it suggests that the risk of a government’s insolvency is a factor that permeates the entire national economy and is transmitted to all of the country’s private institutions.
Loosening the banks-sovereign link requires sound fiscal policies and good bank supervision
The above considerations do not imply that loosening the banks-sovereign nexus is undesirable. Indeed, there is no doubt that achieving this objective would improve financial stability. Recently, several commentators have suggested that the capital regulation of bank sovereign exposures should be revised with this objective in mind (Gros 2013, Uhlig 2013, Weidmann 2013). These proposals start from the observation that in most countries, banks’ exposures to the domestic sovereign benefit from a preferential prudential treatment (they are typically assigned low or zero risk weights and are exempted from the rules limiting large exposures). Therefore, a tightening of this treatment would reduce banks’ incentives to lend to their governments, contributing to loosen the sovereign-bank link.
In our view, the policy option of imposing tighter prudential rules on banks’ sovereign exposures is viable in a steady state, but complex in a crisis situation. In the present phase, it could be highly procyclical, as it would risk fuelling instability on government bond and bank funding markets; this problem is acknowledged even by the proposal’s advocates, who generally argue for a medium-long term implementation. But even abstracting from this consideration, a revision of the prudential treatment of sovereign exposures would raise a number of concerns. First of all, some conceptual issues remain unresolved. Where countries are responsible for fiscal as well as monetary policy, the sovereign always has the ultimate option of inflating its debt away, i.e. to trade-off default risk with inflation risk (Goodhart 2012). This does not necessarily apply to monetary unions. Should the preferential treatment of sovereign debt be maintained in countries with fiscal and monetary sovereignty, and abandoned in monetary unions? Or, should it be revised in both cases to account for default as well as inflation risk, which – at a sufficiently high level of abstraction – can be seen as the same risk? Second, a number of operational issues would need to be addressed. In particular, what measure of sovereign risk should be adopted in practice? Obvious solutions (e.g., relying upon external ratings) would conflict with the objective of gradually reducing the reliance of the regulation on rating agencies – arguably a widely sha