Monday, August 16, 2010

Sovereign Debt Part 5D: European Banks, What if Things Go Really Badly?

by Bill McBride on 8/16/2010 03:28:00 PM

CR Note: This series is from reader "some investor guy".

Part 5D. What if Things Go Really Badly? More Than Half of Europe’s Banks Would Need Capital. A quarter would be insolvent. This part was moved a bit later to allow a more thorough analysis of the European Bank Stress Tests.

There are two questions I’ve been asked repeatedly since the European stress tests were announced.

Q1. Was there much sovereign stress in the European bank stress tests?

NO. The most glaring oversight, in the opinion of the author and many other analysts, is assuming there would be no sovereign defaults, and thus not showing any losses on the bank’s long term holdings (in the banking category vs the “trading book”). According to the Committee of European Banking Supervisors, the sovereign stress scenario results in “39 billion euro associated with valuation losses of sovereign exposures in the trading book “.

“The haircuts are applied to the trading book portfolios only, as no default assumption was considered, which would be required to apply haircuts to the held to maturity sovereign debt in the banking book.”

Yes, as you may have seen in other reports, the “sovereign stress scenario” contained no sovereign defaults. Why on earth would this be the stress case? “The setting up of the European Financial Stability Facility (EFSF) and the related commitment of all participating member States provides reassurance that the default of a member State will not occur, which implies that impairment losses on sovereign exposures in the available for sale and held-to-maturity in the banking book cannot be factored into the exercise.” (source).

For there to be no sovereign defaults: the EFSF and IMF plans would have to be big enough to subsidize a lot of sovereign problems, they would have to stay in place for a long time, member countries would have to continue to subsidize them for a long time, and (the worst vulnerability) countries like Hungary and Greece have to accept the conditions which come with the offers of support.

Plenty of countries have turned down potential IMF support over the years. Hungary turned down additional support from the IMF and the EU, the week before the stress test results were released, (source). Hungary continues to say “no”, and “Hungary's defiance could conceivably spread to other governments currently being squeezed by the IMF and European authorities.” (source).

At least four forms of defiance might lead to default. First, a country might say that they can’t or won’t do the requested steps like austerity. They would rather just default. Second, a country might claim or actually believe they don’t need help. Then if markets freeze up due to problems elsewhere, they might not be able to roll over their debt, and they could default. Third, a country with a lot over the counter derivatives might come up with a reason not to pay them (see Part 5C for some examples of how sovereigns might try to get out of commitments). That’s a big potential problem for some European banks that don’t have collateral from sovereigns on their interest rate and currency swaps. Fourth, someone already receiving considerable help from the EU or IMF might not repay it. For example, a new government might be elected (as happened in Hungary).

Here is a list of the 30 countries whose sovereign debt was included in the stress tests. Six of them already are receiving help from the IMF. Note that most of the IMF support agreements end before the end of the Stress Test period, 12/31/2011. Most of the amounts outstanding are due to be repaid to the IMF quarterly installments 3-5 years later.

European Stress Test Participants and Existing IMF Support Agreements
As of August 5 2010, In Millions, source
Czech Republic      
PolandCredit Line7/2/20107/1/201113,69013,6900
United Kingdom      
Count6 Totals65,02641,88923,136

Official Opinion. Jean-Claude Trichet, President of the European Central Bank, claims that the stress tests were “an important step forward in restoring market confidence.” (August 5, NY Times: Trichet Sees Reasons for Optimism in Europe). OK. Let’s see what the bond and CDS markets have done since July 23rd when the results were released.

What the market thinks. The CDS market believes that stress test member governments have about half a trillion dollars of expected default losses, about one third of the worldwide total. Big difference. And the CDS and bond markets didn’t move much since the announcement of the stress test results. Greece’s CDS prices for Greece show a 50% chance of default within 5 years. Before the stress test, it was 53%. In other words, the people with real money riding on these outcomes were not impressed. If everyone in the CDS market believed the assumptions used in the stress tests, that pricing should be near zero.

European Bond Spreads Click on graph for larger image in new window.

CR Note: This graph of European bond spreads is from the Atlanta Fed weekly Financial Highlights (line added for stress tests).

Bond spreads would also have gotten much smaller, with debt from Greece, Spain, Portugal, and Italy having yields much closer to German debt. As can be seen from this CR chart from August 12, that isn’t occurring.

However, there some useful results of the stress tests. There is plenty of published data which is reasonably comparable from one bank to another. The regulators also put in general economic stresses like unemployment and slow growth, which caused a lot of losses to banks: 473 billion euros of impairment losses (source). That’s where the great majority of the losses in the stress tests arose.

Q2. How many European banks would fail a real sovereign stress test, or would fail in the Really Bad Scenario?

There are four sets of tests I would like to discuss:

1. The official European bank stress tests, from the Committee of European Banking Supervisors, where only 7 of 91 banks “failed” the tests, and would need more capital. Much of that data is here. As mentioned above, the official test doesn’t even match the expected level of sovereign losses, much less a real stress test.

2. Citi did a reanalysis if all of the banks’ sovereign bonds were marked to market and concluded that 24 of 91 banks would fail the Tier 1 capital test from the European regulators, on exactly the same economic assumptions as the regulators. In other words, about a quarter of the banks would need more capital to be considered solvent, according to Citi’s analysis.

3. Using a rather different method, and current credit default swap pricing, Some Investor Guy came to an almost identical conclusion as Citi, with an almost identical list of banks needing capital. Citi’s test can be regarded as a reasonable baseline.

Why is Citi’s scenario using the regulators’ haircuts and interest rate assumptions so close to Some Investor Guy’s scenario based on credit default swap data? Part of the reason is that both sets of losses were calculated from the same sovereign bond holdings for each bank.

I speculate that something else might also be occurring. Faced with political and public relations difficulties of having sovereign defaults of European countries despite a one trillion euro commitment that is supposed to prevent them, analysts at the central banks may have calibrated their sovereign stress assumptions to provide similar overall market value losses to what the CDS market currently expects. If so, the exact details are obscured by the stress tests using changes in ratings and yields (and multiple languages and currencies). If this is really what happened, I’m proud of the analyst(s) at CEBS who figured out how to set the sovereign assumptions and data collection so that someone like Citi or Some Investor Guy could easily do a real, expected value stress test. They keep their jobs, other people willing to wade through thousands of numbers get to do their own scenarios. Everybody’s happy.

4. In the Really Bad Scenario, the results are much worse for banks. 57 of 91 major European banks would need capital, and 23 of them would have their Tier 1 capital go below zero. In other words, they would be technically insolvent.

Banking and Debt CrisisIn the Really Bad scenario, defaulted sovereigns include Belgium, Bulgaria, Cypress, Estonia, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Malta, Poland, Portugal, Romania, and Spain. About $4 trillion of European sovereign debt would be in default, and losses to the European banking system would be about $568 billion, doubling the losses in the official stress tests. When added to the rest of the losses in the stress tests, losses to banks go to 1.041 trillion euros.

In this scenario, there is a strong correlation between where a bank is based, and whether it needs capital. Banks tend to hold a very disproportionate amount of their sovereign bonds from their home country. Thus, banks from Greece, Hungary, Ireland, Italy, Poland, Portugal, Romania, and Spain encounter a disproportionate amount of the damage in this scenario. The portion of outstanding bonds held by banks varies a lot. More than half of Greek debt is held by banks, but only about 15% of Belgian and Italian debt is at banks.

Please note that for the Really Bad scenario, the exact same baseline numbers were used for everything except sovereign losses. Same unemployment, GDP growth, private sector losses, etc. With widespread sovereign default, in the short run there would probably be more unemployment and lower GDP growth.

In Banking Crises, an Equal Opportunity Menace, Reinhart and Rogoff show that the cost of banking crises are greater than has been commonly assumed. “Banking crises dramatically weaken fiscal positions…with government revenues invariably contracting, and fiscal expenditures often expanding sharply. Three years after a financial crisis central government debt increases, on average, by about 86 percent. Thus the fiscal burden of banking crisis extends far beyond the commonly cited cost of the bailouts.”

Banking and Debt CrisisGiven the typical timing of banking and sovereign crises, it would take about 1-2 years for the crisis to reach its peak, and about 3 more years for recovery.

Chart source: page 40.

Full Disclosure: Some Investor Guy lives in the US and owns no sovereign debt from outside the US. He owns bonds from one of the banks involved in the stress tests. To his relief, that bank is solvent even in the Really Bad scenario. Since he didn’t name specific banks, it would be pointless to have modified the Really Bad scenario in any way to help that bank or the value of its bonds. This article is not an offer to buy or sell anything, but after all of the work assembling the data and doing calculations, Some Investor Guy might be persuaded to share the underlying analysis.

CR Note: This series is from "Some investor guy"

• Part 1: How Large is the Outstanding Value of Sovereign Bonds?

• Part 2. How Often Have Sovereign Countries Defaulted in the Past?

• Part 2B: More on Historic Sovereign Default Research

• Part 3. What are the Market Estimates of the Probabilities of Default?

• Part 4. What are Total Estimated Losses on Sovereign Bonds Due to Default?

• Part 5A. What Happens If Things Go Really Badly? $15 Trillion of Sovereign Debt in Default

• Part 5B. Part 5B. What Happens If Things Go Really Badly? More Things Can Go Badly: Credit Default Swaps, Interest Swaps and Options, Foreign Exchange

• Part 5C. Some Policy Options, Good and Bad

• Part 5D. European Banks, What if Things Go Really Badly?

Last 10 Posts