A Mixed Review for EU Bank Stress Tests

This week, we finally received some additional clarity on the extent and methodology of the European Union's banking stress tests. The new information is encouraging, while still woefully inadequate.

On the positive side, 91 EU banks will be included in the tests covering 65 percent of the EU banking sector and at least 50 percent of bank assets in each EU member state. The list of the included banks can be found in the table below.¹ This coverage significantly expands the list of 25 banks initially announced by the European Union, and it guarantees that all systemically important EU banks are included in the stress tests. The scope is thus comparable to the US stress tests in early 2009, which included 19 banks accounting for roughly two-thirds of US banking assets.² It is also noteworthy that the six largest Greek private banks are included in the stress tests.

Market pressure in the form of a stigmatization applied to nontested banks has clearly been effective in Europe.

More important, the EU stress tests will now include all the most likely "EU zombie banks," including all of Germany's (already) state-owned Landesbanken, Spain's Cajas, and the large Belgian and French banks. All of Europe's systemically important potential "rotten apples" are therefore included in the stress tests.

In addition, the EU stress tests will rely on commonly agreed macroeconomic scenarios (baseline and adverse) conducted and published on a bank-by-bank basis and differentiated for each member state. This is good news. It will force individual bank results into the open and take account of the vastly different economic circumstances in individual EU members. Accordingly, earlier concerns over the legal basis for forcing publication of individual bank stress test results in Germany, for example, should be dismissed. The publication of stress test results was, from the beginning, an initiative directly from EU leaders at the EU Council on June 13 (implicitly thereby overruling recalcitrant and "captured" national EU banking regulators). As expressed by Chancellor Angela Merkel when she was asked about the legal basis for publication of individual stress tests in Germany at the subsequent press conference,  "EU leaders would find a way" to publish these results.

In many ways the EU stress tests are structured in a manner similar to the US Federal Reserve Board's Supervisory Capital Assessment Program (SCAP), but it is, of course, debatable whether similarities with this program are an advantage or not. The two EU stress test scenarios focus on a two-year period (2010–2011), whereas SCAP focused on the two-year period of 2009–2010. The key macroeconomic variables in the EU stress tests are real GDP, unemployment, and consumer price developments, while in SCAP they were US real GDP, civilian unemployment, and house prices.

The EU stress tests will, in their "adverse scenario," assume a 3 percentage point deviation in GDP from the European Commission's forecast over the two-year scenario period.³ This is roughly comparable the US SCAP tests, which in their adverse scenario assumed a 1.3 percent (2009) and 1.6 percent (2010) lower real GDP growth rate in the US economy.⁴ The assumption of an aggregate 3 percent lower EU GDP, relative to the latest EU Commission forecast of 1 percent growth in 2010 and 1.7 percent in 2011 (annual average growth rates),⁵ implies that the European Union remains in (or close to) recession through 2011.

The EU stress tests will, unlike the US SCAP, also include a "sovereign risk shock," which will simulate a "shock on interest rates to capture an increase in risk premia linked to a deterioration in the EU government bond markets."⁶ The presence of a sovereign risk element in the EU bank stress tests is encouraging news.

So far, so good. But there are many remaining uncertainties surrounding the EU stress test.

First, we still don't know anything about the assumptions made for two out of the three macroeconomic parameters—unemployment rates or consumer prices. While obviously closely linked to the evolution of overall economic growth, this nonetheless leaves considerable ambiguity about the severity of the "adverse scenario."

Second, and perhaps more importantly given the importance of "tail-end events" for bank stress tests, it remains unclear what the "sovereign risk shock" will entail in terms of assumed haircuts for sovereign bonds. Here it is clear that the stress test focus on "a shock on [government bond] interest rates" in its sovereign component is a clever way to avoid explicitly stating the politically explosive number for just how large a haircut on sovereign bonds banks should be expected in the EU "adverse scenario." On the other hand, as anyone can calculate the "implied bond haircut" from the assumed increase in the interest rate, the avoidance of any explicit reference to a default is clearly a political exercise.

The stress test reliance on "differentiated scenarios" for each member state suggests that a similar "differentiation in assumed haircuts" will be applied to member states' sovereign bonds. It also suggests that in all likelihood only a few EU members' sovereign bond yields will be assumed to become stressed. Indeed, not only is it hard to envision sovereign impairment in the European Union outside Greece, Ireland, Portugal, and Spain (the GIPS countries, in the latest parlance), one might assume that German bonds (which achieved safe-haven status during the latest crisis) would actually strengthen as other euro members' bonds suffer.

Several "press leaks" suggest that differentiated haircuts are being assumed in the EU stress tests, with some reports stating that Greek bonds will suffer a 17 percent haircut and Spanish bonds 3 percent. Other reports suggest that the implied Greek haircut could be 20 percent, 8 percent for Spanish bonds and 5 percent for Portugal.

These implied haircut levels, however, are significantly below presently implied levels of Greek "recovery swaps," which indicate that investors would get back about 40 percent in a Greek default or restructuring. As a result, the "Greek sovereign element" in the EU bank stress tests looks superficially to be far less adverse than currently believed by financial markets. On the other hand, considering that Spanish banks (excluding the Cajas) are generally believed to be in good shape, the inclusion of any haircut on Spanish sovereign bonds in the stress tests seems excessive, even if the country no longer enjoys an AAA-credit, but only AA. Spain is not Greece and the stress tests will show that.

With respect to the size of an appropriately assumed Greek haircut, several things must be considered. First, as any Greek default/restructuring would inevitably have to result from a negotiated settlement between the Greek government and its EU/International Monetary Fund partners (unlike the unilateral Argentinean default in 2001), it is highly dependent on what the assumed "sustainable long-term Greek debt burden" would be after a default. Is it 60 percent or maybe 80 percent of GDP?⁷ This again depends on what one believes about the structural reform elements of the three-year IMF/EU program that Greece is implementing.

Then there's the issue about whether one should include in the average haircut estimate the ability of individual EU banks to offload Greek exposure to official sources in the form of European Central Bank (ECB) secondary bond market purchases or collateral. "Recovery swap rates" typically refer to the recovery rates for individual bonds, which are likely to be significantly lower than overall ultimate average recovery rates for entire bank portfolios of Greek sovereign debt held in May 2010. Assuming an average haircut on all Greek sovereign debt held by any given bank of, say, 17 percent, even with individual bonds returning as little as 40 percent, may be reasonable. Slavishly assuming that thinly traded derivative markets are correct is erroneous. More information about the precise operationalization of the European Union's "sovereign risk shock" is required to settle this issue.

It should be clear, though, that even a 17 percent haircut on Greek sovereign bonds will almost certainly force the six Greek banks included in the stress tests to gain access to the IMF/EU sponsored Financial Stability Fund (FSF), which is part of the country's three-year program aimed at ensuring the stability of Greece's private banking system.

We continue to lack any information about the precise type of capital requirements used to assess EU banks. The statement of the Committee of European Banking Supervisors (CEBS) merely stresses how "the objective of the extended stress test exercise is to assess the overall resilience of the EU banking sector," but is regretfully silent on whether the focus is on Tier 1 risk-based capital ratios or—as was the case in the US SCAP test—the portion of Tier 1 capital that is common equity. Without this information, it is not possible to discern what type of remedial actions EU banks that fail the stress tests must take. This is of critical importance to existing common equity holders, because they will see their holdings diluted if additional common equity must be raised.

In the US SCAP tests, more than half (10 out of 19) of stress tested banks "failed" in the sense that they were required to "enhance their capital structure to put greater emphasis on common equity"—in other words, to raise more common equity capital. The relatively large differentiation among member states and the bank-to-bank disclosures suggests that a considerable number of EU banking institutions, while not insolvent, will similarly be required to improve their capital cushions as a result of these stress tests. Such losses for existing shareholders in EU banks were further implied by ECB President Jean-Claude Trichet at his press conference on July 8 when he stated that EU banks "should retain earnings, turn to the market to strengthen further their capital bases, or take full advantage of government support measures for recapitalization."

Given the obvious current political advantages for EU leaders to be seen as tough on banks, it would be a grave error to underestimate the willingness of EU governments to force their banks to raise more capital in private financial markets or access available public funds. Moreover, the substantial public capital in both Spain's existing FROB bailout fund (up to €99 billion earmarked for the Cajas) and Germany's SOFFIN (estimated at around €50 billion), both of which have already been tapped during this crisis, suggests few operational obstacles to a speedy infusion of public money into private banks in need.

Finally, there is the issue of what will be published on a bank-to-bank basis on July 23. Will it include details about individual banks' exposure to particular individual asset classes (say Greek sovereign bonds)? Or will we simply get a single "black-box number" of how much capital each "failed bank" must raise? Certainly the success of the stress tests depends urgently on transparency.

One discomforting element here is the short time period for the EU stress tests.⁸ The US SCAP began February 25, 2009, and results were published on May 7 2009—72 days later. By contrast, the EU stress tests were announced June 13 and results will be published on July 23—40 days later. This compressed time table may indicate that relatively few details will be published, as national EU banking regulators may not have sufficient time to collect and analyze the data.

In summary, it remains too early to either hail or dismiss the European Union's bank stress tests.

>

Notes

1. Note that not all 27 member states are represented in the table. This is so if more than 50 percent of domestic banks, as in for instance the Baltic countries, are owned by other EU banks. Such banking groups have been tested on a consolidated level, i.e., so that subsidiaries and branches of an EU cross-border banking group have been included in the exercise as a part of the test of the group as a whole. Due to the accelerated consolidation among Spanish Cajas, more than 100 European banks are in reality part of the stress tests.

2. See the overview of the Supervisory Capital Assistance Program (SCAP).

3. The Wall Street Journal assumes a 3 percentage point deviation from the baseline in both year 2010 and 2011. This however, is in all probability an erroneous reading of the Committee of European Banking Supervisors (CEBS) statement, which reads: "The adverse scenario assumes a 3 percentage point deviation of GDP for the [European Union] compared to the European Commission's forecasts over the two-year time horizon."

4. The CEBS press release does not state whether the deviation from GDP is real or nominal.

5. See the spring 2010 forecast.

6. See the press release.

7. See Mussa (2010) for an elaboration of this issue.

8. I am indebted to my colleague Ted Truman for this point.