BCBS reviews sovereign risk weights for banks

If the noises coming from the Basel Committee on Banking Supervision (BCBS) and the European Central Bank (ECB) are to be believed,regulators are about to call time on one of the biggest and most lucrative arbitrage opportunities in modern finance.

The application of a zero-risk weight to sovereign debt in successive Basel reform packages has allowed banks to engage in a souped-up carry trade by holding such assets and reaping their returns, while setting aside no capital against the possibility of their default. 

Sovereigns, particularly those on the eurozone’s periphery, also benefited in the form of tighter borrowing costs. Without the need to assign any capital to the bonds, banks were able tolerate lower yields. The entire arrangement was propelled by the sheer rarity of sovereign defaults in the developed world, and the assumption that, in the eurozone at least, other countries would step in to prevent weaker sovereigns going bust. 

Cosy set-up

The eurozone sovereign debt crisis drastically undermined this cosy set-up. Regulators, focused on safeguarding the fiscal integrity of sovereigns from banking crises, now had to deal with risk running in the other direction. The Greek, Spanish, Italian, Irish and Portuguese government bonds lodged in bank balance sheets across Europe became a massive source of concern.

The question was, and remains, could regulators attach an appropriate risk-weighting system to sovereign bonds without pushing up the borrowing costs of heavily indebted countries and revealing further deficiencies in the banking system’s capital adequacy?

“Whenever the topic of sovereign risk weights was discussed among Basel members, it revealed a split in the committee,” says Marco Lichtfous, former deputy head of financial stability and prudential supervision at the Central Bank of Luxembourg, and an ex-representative of Luxembourg on the BCBS. “The European supervisors mostly wanted to leave the issue to one side to keep markets stable during the crisis, whereas supervisors from other continents wanted to enforce some market discipline and recognise the real risk exposure banks had toward sovereigns.”

No easy answer

This split now seems to have been resolved. In January 2015, the BCBS announced that putting together a sovereign risk-weight system would be on its agenda for 2015-16. Danièle Nouy, president of the supervisory council at the ECB’s new single supervisory mechanism, has also voiced her enthusiasm for a cap on banks’ exposure to their home country sovereign debt, deployed independently or in tandem with a risk-weights system. Regulators hope that the drawn-out nature of this process will dampen any ill effect on the market and banks themselves.

That hope may be misguided. “The Greek situation hasn’t really improved in the past five years. It is likely that peripheral eurozone countries will still be in serious fiscal difficulties in 2017, 2018 or whenever these changes are introduced. A lengthy implementation timetable will not reduce the pressure on banks who find themselves under-capitalised,” says Mr Lichtfous.

Just how significant will the impact be? According to Roberto Henriques, head of financials credit research at JPMorgan, banks may find the change quite brutal. “The risk weights applied to the vast majority of sovereign debt are not going to be excessively punitive,” he says. “We’re not going to suddenly see weightings of 50% or 100% for every government bond. It’ll be more like 10% to 15% for most. However, given that the capital allocation for these instruments at most banks is currently nil, the rate of change in the central requirements will represent a sizeable hurdle for banks to overcome.”

Shelving the subsidy

Thanks to the ECB stress tests conducted in 2014, European banks already have some idea of how sizeable that hurdle will be. For the purposes of the tests, the ECB asked banks to calculate appropriate risk weights for sovereign debt using a ratings-based approach.

Two European academics, Josef Korte at Goethe University in Frankfurt and Sascha Steffen at the European School of Management and Technology in Berlin, have attempted to replicate this approach. Using the probability of default, the loss-given default and the maturity assumptions set out in the basic internal modelling process allowed by Basel, the academics calculated that the risk-weight subsidy on offer for the European sovereign debt held by banks tested by the ECB totalled about Ä750bn as of June 2013, up from just under Ä450bn in December 2009. In hard capital terms, using the 12% adequacy ratio that most analysts expect banks to abide by once the Basel III reforms are introduced, that equates to a shortfall of Ä90bn. For some banks, particularly those deemed to be systematically important to the global market, the capital adequacy ratio will be even higher, leading to an even larger shortfall.

Put another way, using the 14.4% risk-weight calculated for an AAA rated sovereign such as Germany and applying a 12% capital adequacy ratio, banks holding Ä100m of bunds would have to put up Ä1.72m of capital towards this position. A similar amount of Italian debt, using the current highest rating of A- from Fitch, would attract Ä6.06m of capital. Whereas Ä100m of Greek debt, assigned a 191% risk weight using the B rating from Fitch and Standard & Poor’s (downgraded to B- in early February 2015), would require an eye-watering Ä22.9m of capital.

This assumes that banks will be able to model sovereign risk internally, at least to some extent. However, internal modelling is facing strong headwinds. The BCBS has launched investigations into risk modelling in trading books and banking books, asking banks to produce risk weights for identical portfolios. The exercise produced massive variations from bank to bank, both within the same jurisdiction and across different jurisdictions, and gave many on the BCBS a nasty surprise.

Installing floors

In response, Stefan Ingves, the BCBS chairman, announced in late 2014 that regulators will be installing capital floors, based on the simpler and generally more punitive standardised model of capital calculation. Banks will remain free to model their own risk internally, but if their risk weights are substantially lower than those generated by the standardised approach, the latter must be used instead. This arrangement is already in place for Swedish and Norwegian banks on some assets.

Mr Korte believes that regulators may ask banks to use just the standardised model when calculating sovereign risk weights, as the usual modelling factors, such as probability of default and loss-given default, have little meaning when applied to countries. “How can banks accurately assign a probability of default to a sovereign bond when sovereigns default so infrequently and when, particularly in the eurozone, there are so many political factors to consider?” he asks.

“How can a bank accurately assign a loss-given default value to the default of its own sovereign when such an event would in all probability mean the total collapse of that sovereign’s banking system? The variation from such modelling would be enormous.”

For Mr Korte and Mr Steffen, a better option may be to derive sovereign risk weights from credit default swap (CDS) spreads. “This would be a bit more nuanced than the standardised approach and would be more responsive to new market information. However, the link between CDS spreads and risk weights would need to be designed carefully, as CDSs can be prone to non-rational market sentiment. We should be careful not to replace one grossly erroneous distortion with another grossly erroneous distortion,” says Mr Korte.

What’s the cost?

There is a dispute over the impact that a risk-weight system would have on sovereign borrowing costs. Banks represent a meaningful portion of sovereign debt investor base, with European banks alone holding about Ä2000bn of European sovereign debt. As soon as capital allocation for this debt becomes a reality, banks will have to reconsider their balance sheet allocation.

“Either sovereigns offer a better return to compensate for the extra capital or they will have to find other buyers who might not take up the debt in such large quantities,” says Mr Henriques. “The blow will fall hardest on countries that already have high borrowing costs and a lot of debt to service.”

As a warning against attempts to tinker with the sovereign bond status quo too aggressively, Mr Henriques points to the dust-up over the effect of leverage ratio rules on the repo market. Initial guidance from the BCBS in 2013 suggested that banks would not be able to net off repo market exposures on their balance sheet when calculating their leverage ratio requirements. Analysts predicted that such a move would force banks to raise an extra $180bn of capital and reduce repo market activity, much of which is supported by government bond issuance. The BCBS eventually yielded and relaxed the rules.

Mr Korte, on the other hand, feels that sovereigns may actually benefit from any change. “Banks present a systemic risk to sovereigns. If investors believe that a German bank has adequately covered its Greek and Portuguese debt with a risk-weightings system so that the German government doesn’t have to protect it if these two countries fail, that might also reduce a little bit the borrowing costs of Germany.”

Risk limits

Another, less complicated option is to implement Ms Nouy’s idea of an exposure limit on own sovereign debt held by individual banks. This would serve to reduce concentration risk and would perhaps force banks to do more lending to participants in the real economy.

No specific limits have yet been suggested by the ECB, but analysis from Fitch shows that a cap of 25% of total capital resources would require a Ä1100bn rebalancing of sovereign bond holdings at European banks. A cap of 50% would produce a rebalancing of Ä800bn.

At first glance, the rebalancing process seems simple. Italian banks could take in the excess of German bunds held by German banks and German banks could take in the excess of Italian sovereign debt held by Italian banks. That would be an ideal outcome for the sovereigns’ point of view, as the demand for their debt would remain the same.

However, given the yield and risk differential between peripheral and core sovereign debt, such a mass transfer might not be equitable. Banks in more secure countries would be left with higher returning, higher risk debt, while those in lower rated jurisdictions would have safer but far less profitable balance sheets. Fitch estimates that a 50-basis-point decrease in the average yield of sovereign bond portfolios held by banks in Italy, Portugal, Greece and Spain would produce a 4.2% reduction in the banking sector profit across these four countries.

Moving away

As a result, banks may diversify out of sovereign bonds to a significant degree. Given a 25% cap, the required amount of rebalancing of Spanish debt within the eurozone banking system would equate to 20% of the country’s entire debt. This change would likely reduce demand for the asset type and force a painful repricing of borrowing costs for many sovereigns.

“The magnitude of the impact on sovereign borrowing costs depends on how any limit is calibrated and how quickly it is introduced,” says Bridget Gandy, co-head of the Europe, Middle East and Africa financial institutions group at Fitch in London. “Such a scheme has the potential to cause severe dislocation in the European sovereign debt market if implemented quickly, which is why we expect any implementation to be gradual.”

Risk limits could also potentially be damaging to smaller banks that operate in just one country by forcing them to diversify to riskier assets or take on sovereign debt from other countries that is less correlated to their activities than domestic sovereign debt.

In the end, risk limits could become surplus to requirements. Mr Korte says: “The moment you eliminate the favourable sovereign risk treatment, banks will cut down exposures anyway. It would be a de facto risk limit. It might make sense to introduce a cap now and plan for risk weights later. A cap would be pointless once risk weights are introduced.”