Search form

Sovereign portfolios or banks’ location: What channels sovereign risk into banking systems?

Chiara Angeloni, Guntram Wolff 19 April 2012

The European sovereign debt crisis has strengthened debate over the link between sovereign and banking risk. Uncertainty around the creditworthiness of some governments in the Eurozone has translated into higher instability of the banking systems. The link between sovereign and banking risk is shown graphically in the positive correlation existing between banks and sovereign credit default swaps (CDSs) in Figure 1. For the Eurozone countries, higher bank CDS prices are associated with higher sovereign risk, with the two financial instruments moving together during 2011.

Figure 1. Correlation between sovereign and banking CDSs

Source: Bruegel calculation with data from Datastream and Macrobond.

Note: Weekly averages over the period January 2011- February 2012. Banking CDSs by country are calculated as weighted averages of CDSs of the individual banks considered for each country. The graph for Greece is not displayed since it is characterised by hyperbolic pattern. The same graphical result holds not only for periphery countries but also for stronger EU economies such as Germany and France.

The interdependence between banking and sovereign risk may be the result of a two-way causality. On the one hand, higher banking risk translates into higher sovereign risk because the possibility that a given government has to rescue the domestic banking system becomes more likely (Gerlach et al 2010, Acharya 2011). On the other hand, a deterioration of sovereign creditworthiness impacts negatively on banks’ assets (BIS 2011).

In Angeloni and Wolff (2012). we analyse the spillover effect from sovereign risk onto banking risk. In particular, we focus on banks’ balance sheet composition in order to detect whether the banks’ holding of sovereign debt securities impacts banking system performances across countries.

The patterns of bank stock prices over the last year have been quite diverse across the European countries. Looking at the performances of the banking system indexes, we can see that all of them have been characterised by a negative trend over 2011; in particular, the most negative values are associated with those banks located in the so-called periphery countries (Figure 2).

Figure 2. Average stock price performances over 2011, by country

Source: Bruegel calculation with data from the EBA, December 2011, Datastream and Macrobond.

Using data from the July stress test and the December Capital Exercise of European Banking Authority (EBA), we provide a more detailed analysis of the balance sheet of these banks. The tests give us information for two different points in time, namely December 2010 and September 2011, and allow us to compare the choices of risk management taken by these banking systems during the last year. Two main results are worth noting:

First, individual banks have reduced their holdings of Greek, Irish, Italian, Portuguese, and Spanish government debt securities (Table 1). Comparison of the aggregate exposure toward periphery countries’ sovereign debts, as reported in the two EBA exercises, shows that the banking systems in the analysed countries are less exposed to sovereign risk in September 2011 than at the beginning of 2011. In addition, at the individual bank level, the exposure towards periphery country sovereign debts in percentage of Core Tier 1 is characterised by a dramatic decline. On one hand, for banks located in strong EU countries, such as France and Germany, this reduction is quite even for the sovereigns of the different geographical areas; on the other hand, banks headquartered in the periphery countries have mainly reduced their holdings of domestic sovereign debt, in accordance with banks’ home-bias preference toward domestic government securities. To some extent, the reduction is also driven by an increase in Core Tier 1 capital.

However, since the start of the ECB’s longer-term refinancing operation (LTRO) in December, there is increasing evidence that this positive development of fewer government bonds in banks has been reversed.

Table 1. Difference in exposure of banks to sovereigns and in level of Core Tier 1 capital as reported in December’s Capital Exercise and in July’s stress test by the EBA (billions of euros), the same set of banks

Source: Bruegel calculation with data from the EBA, July and December 2011.

The holdings of government debt securities can be expected to have played a role in the performance of the analysed banks in financial markets. Indeed, a higher exposure toward periphery country sovereigns is expected to be associated with poorer stock price performances on financial markets during the last year. Figure 3 plots the weekly average growth rate of stock valuations in the period October–December 2011 against the total exposure towards periphery countries’ sovereign debts as well as against the individual exposure towards Greek, Irish, Italian, Portuguese, and Spanish government securities. As the scatter plots show, these series are not characterised by a strong correlation.

Figure 3. Change in stock market index to exposure to Greek, Irish, Italian, Portuguese and Spanish sovereigns in percent of Tier 1 capital

Source: Bruegel calculation with data from the EBA, December 2011, Datastream and Macrobond.

In our paper we perform a more detailed regression analysis, with the inclusion of some controls, such as the risk-weighted asset and the Core Tier 1 ratio of each bank. We find that the statistical significance of the exposures toward each individual periphery country varies according to the time window in which the stock price average change is measured. In the period October–December 2011, exposures toward Italian and Irish government debt, as reported in the EBA December Exercise, turn out to have a negative impact on banks’ stock prices. When the period of analysis is July to October, in accordance with the empirical results of Wolff (2011), the holdings of Greek sovereign debts are found to impact negatively upon banks’ stock price, while the coefficient of exposure towards Italian debt fails to be statistically different from zero. Again, the average stock price growth rate in the period April–July turns out to be negatively correlated with Greek exposure. These empirical results suggest that banks’ performances are only weakly related to the riskiness associated with their holdings of government debts. Rather, bank stock prices seem to incorporate the risk associated with the country in which the banks are located. Thus, banks headquartered in the same country, but characterized by different levels of sovereign exposures, have performed equally poorly in 2011.

In conclusion, the analysis of banks’ balance sheets has shown a common tendency of banks in the Eurozone to reduce their level of holdings of periphery country sovereign debt during much of 2011. In particular, due to the home-biased composition of government debt portfolios of banks, such reduction has been massive for banks located in Greece, Ireland, Italy, Portugal, and Spain. With the LTRO, however, these trends have been reversed. The holdings of sovereign debt poorly explain the performance of bank stock prices in 2011. Rather, the financial market seems to incorporate the country risk associated with the location of the individual banks.