خطر سرریز اوراق قرضه دولتی / بانکی در بحران بدهی اروپا

This paper investigates contagion between bank and sovereign default risk in Europe over the period 2007–2012. We define contagion as excess correlation, i.e. correlation between banks and sovereigns over and above what is explained by common factors, using CDS spreads at the bank and at the sovereign level. Moreover, we investigate the determinants of contagion by analyzing bank-specific as well as country-specific variables and their interaction. Using the EBA’s disclosure of sovereign exposures of banks, we provide empirical evidence that three contagion channels are at work: a guarantee channel, an asset holdings channel and a collateral channel. We find that banks with a weak capital buffer, a weak funding structure and less traditional banking activities are particularly vulnerable to risk spillovers. At the country level, the debt ratio is the most important driver of contagion. Furthermore, the impact of government interventions on contagion depends on the type of intervention, with outright capital injections being the most effective measure in reducing spillover intensity.

Due to the absence of a common European policy framework for handling the banking crisis as well as missing bank resolution mechanisms, several European governments were forced to rescue troubled banks headquartered in their countries during the financial crisis. Various measures have been taken, ranging from equity injections in troubled banks to the setting-up of bad banks (Petrovic and Tutsch, 2009 and Stolz and Wedow, 2010). Invariably, these rescue operations have increased national debt burdens and caused a deterioration of public finances (IMF, 2009). One consequence of the risk transfer from the private sector to sovereign treasuries has been an increased interdependence of banks and countries, causing negative feedback loops between their financial conditions. With the rise of the sovereign debt crisis in Europe, the link between bank and country risk has intensified further, especially for the countries that were quickly identified as vulnerable, namely Greece, Ireland, Italy, Portugal, and Spain (the GIIPS countries). This increased interdependence is depicted in Fig. 1. The two panels of Fig. 1 respectively show the CDS spreads of the 15 countries in our sample and the average bank CDS spread for each country. The graphs illustrate that heterogeneity is present in both the level of and the comovement between sovereign and bank CDS spreads. The link between the risk profile of banks and countries varies over time and is partly influenced by shocks in the economy or the banking system. A major shock stemming from the banking system was the demise of Lehman Brothers in September 2008, which provoked a substantial increase of CDS spreads for banks and also for certain countries, typically smaller countries with large banks or countries where banks had to be rescued. The sovereign debt crisis further intensified the link between bank and country risk. The sovereign debt crisis is usually considered to have started at the end of 2009, when the newly elected Greek government announced that the country’s budget deficit was much larger than previously reported. In the case of Greece, two bailout packages were put together under the surveillance of the “troika” (European Commission, ECB and IMF), one of them including a substantial write-off of Greek debt in the books of private investors. Later, further rescue packages were implemented for Portugal and Ireland, both under the supervision of the troika. A series of credit rating downgrades of the affected countries followed, causing bond and CDS spreads to widen considerably (see IMF, 2009, IMF, 2010 and IMF, 2011).1 During the sovereign debt crisis, banks in Europe were and remain confronted with stress in their capital and liquidity positions. A substantial number of banks had to rebuild their capital buffers after the losses they incurred in their securities (mainly asset-backed) and lending portfolios, especially those with real estate exposures. A general lack of trust hampered the access of banks to money market funding, which was eventually alleviated, at least temporarily, by non-conventional longer-term refinancing operations set up by the ECB. Further, the European Banking Authority (EBA) decided to conduct a sovereign stress testing exercise and required that banks execute detailed capital rebuilding plans before mid-2012. The disclosure of detailed information on banks’ exposures to sovereign risk in the EBA (and former CEBS) stress testing exercises provided valuable information to market participants to gauge the risk profile of European banks. Overall, the consequence of the continued stress in the banking system and the vulnerability of certain European sovereigns is that the financial conditions of banks and sovereigns became increasingly intertwined. Considering this increased interaction between sovereign and bank credit risk, the objective of this paper is threefold. First, we analyze whether we find empirical evidence of contagion. We investigate the time-varying intensity of the risk spillovers using excess correlations as our preferred contagion metric.2 Second, we attempt to explain the contagion effect by investigating the relationship between excess bank/sovereign correlations and both bank and country characteristics. While there have been several papers investigating the determinants of either bank risk or sovereign risk in isolation, there is less evidence on the potential mutual contagion effects. Although these excess correlations themselves do not give an indication of the direction of the spillover, the particular structure of our database will allow us to relate bank- and country-specific characteristics to the excess correlations. By analyzing a number of relevant variables and the interplay between bank and country characteristics, we are able to identify critical interactions that are related to bank/country contagion. This allows us to tackle a series of relevant policy questions concerning the banking system as well as the financial condition of sovereigns. Third, our setup allows us to analyze the existence of channels through which default risk can spread between banks and countries. As a result, we contribute to the literature on the sovereign debt crisis by empirically confirming the importance of an asset holdings channel, collateral channel and guarantee channel. The main findings of this paper can be summarized as follows. We document significant empirical evidence of contagion between bank and sovereign credit risk during the European sovereign debt crisis. In 2009, when the sovereign debt crisis emerged, we find significant spillovers between banks and their home country for either 51%, 65% or 73% (depending on the year of comparison) of the banks in our sample. We find similar results when focussing on the relationship between non-resident banks and sovereigns or when analyzing the relationship between banks and GIIPS countries. Second, we are interested in business models that can allow banks to minimize contagion exposure. We find that the degree of contagion is significantly linked to bank capital adequacy, and this effect is economically very significant. Furthermore, the higher a bank’s reliance on short-term funding and the lower the involvement in traditional banking activities, the higher the intensity of spillovers between banks and sovereigns. Third, making use of the EBA stress test disclosures, which include bank-specific information on sovereign debt holdings, we find evidence in favor of an asset holdings channel and a collateral channel. Higher sovereign debt exposures lead to more contagion, and this effect is more pronounced for banks that excessively rely on short-term funding. Fourth, our results also confirm the presence of a guarantee channel. Average excess correlations are higher between a bank and their home country. Furthermore, the default risk of large banks is more strongly related to the default risk of the home country. Additionally, the default risk of banks in countries with higher debt levels is more strongly related to the default risk of the home country. Taken together, these observations indicate that there is a guarantee channel at work. The remainder of this paper is structured as follows. Section 2 reviews the literature on contagion and more specifically the European sovereign debt crisis. In Section 3 we describe the data and the methodology. Section 4 reports our empirical findings, including robustness checks. Section 5 summarizes the conclusions and policy implications.

This paper provides empirical evidence on risk spillovers between banks and sovereigns during the European financial and sovereign debt crisis. Whereas there is a substantial literature exploring the determinants of bank or sovereign credit risk separately, empirical evidence exploring contagion between the two is scarce. This paper contributes to this topic by examining the pattern of contagion in the sovereign-bank nexus in Europe, by investigating which bank-specific and country-specific determinants drive contagion and by empirically identifying different contagion channels. We define contagion as “excess correlation”, i.e. correlation over and above what is explained by fundamental factors. Our preferred measure of sovereign and bank credit risk is CDS spreads. After controlling for common factors (market risk, economy-wide credit risk, term spread changes, and volatility), we document significant evidence of bank/sovereign contagion. In the year 2009, when the sovereign debt crisis emerged, we find significant spillovers between banks and their home country for either 51%, 65% or 73% (depending on the year of comparison) of the banks in our sample. We find similar results when focussing on the relationship between non-resident banks and sovereigns. Furthermore, these numbers increase to 82%, 92% or 100% when only considering spillovers between the banks and the GIIPS countries. This is significant evidence of contagion between banks and countries in the period covering the bank crisis as well as the sovereign debt crisis in Europe. We exploit the cross-sectional differences between bank/sovereign excess correlations by relating them to bank- and country-specific variables. We include a broad set of measures intended to capture the strategic choices inherent in bank business models. The capital adequacy level of banks has a strong and economically significant effect; we find that an increase in the Tier 1 ratio reduces the excess bank-country correlation significantly. Furthermore, the lower the banks’ reliance on short-term funding sources (measured as the proportion of short-term funding in total debt) and the more it focuses on traditional banking activities, the lower the intensity of risk spillovers between banks and sovereigns. These findings support the new regulatory Basel III framework which imposes more stringent capital adequacy ratios and new liquidity measures. At the sovereign level, we find that higher debt-to-GDP ratios significantly increase the degree of bank/sovereign contagion. This finding motivates the recommendation that public finances need to be consolidated, especially in countries with high debt levels. A credible commitment to reduce debt levels over time will probably require efforts at the domestic level as well as enforceable coordination at the European level and, perhaps, some form of (partial) debt mutualisation or debt forgiveness. We also confirm the existence of an asset holdings channel, a collateral channel and a guarantee channel. Using EBA disclosures of banks’ sovereign exposures, we document (i) that bank default risk is more strongly related to country default risk when the bank has relatively more sovereign debt of that country on its balance sheet and (ii) that this effect is stronger when country default risk rises or when the bank is relatively more dependent on short-term funding. Furthermore, higher excess correlations of large banks towards their home country and a strong and positive impact of debt-to-GDP ratios on the excess correlation between countries and resident banks corroborate the existence of a guarantee channel. This is further confirmed when analyzing actual government interventions, which amplify the excess correlations between banks and their home country. We also show that not all government interventions have a similar impact on the excess correlations between non-resident banks and countries. More specifically, capital injections reduce the excess correlation between sovereigns and non-resident banks. This implies that capital injections are perceived as a more efficient intervention to reduce spillovers, since uncertainty about the actual size of the intervention remains an important issue when using government guarantees. In terms of policy implications, our results suggest several actions to alleviate the contagion between bank and sovereign risk. The ambition of policymakers and supervisors should be to (1) decrease the probability of contagion and (2) when contagion occurs, decrease the intensity of the risk spillovers. In order to achieve these objectives, action in three dimensions is necessary: make banks more robust, make public finances more resilient and weaken the bank–sovereign link. On the bank side, the degree of capital adequacy turns out to be crucial. Moreover, banks should be restricted in their reliance on money market funding. These elements are at the core of the internationally agreed Basel III rules that will be phased in gradually. Our results thus lend support to the use of more stringent capital and liquidity constraints and policymakers and supervisors should provide incentives to banks to adjust their business models accordingly. Furthermore, since we find evidence in favor of the asset holdings channel, there might be scope for concentration limits in sovereign bond portfolios in various dimensions. On the sovereign side, making public finances more sustainable and ensuring that resolution mechanisms are in place to deal with distressed banks are important policy objectives. Additionally, our results on the impact of government interventions reinforce recent calls for resolution mechanisms that operate at a supra-national (European) level, as country-level initiatives always intensify the relation between banks and their home country. This requires a so-called banking union at the European (or Eurozone) level, implying that not only bank supervision should be executed at the European level (e.g. by the ECB), but also that deposit insurance and bank resolution, and the associated burden sharing arrangements, have to implemented on a European scale.