Standardised approaches pile up capital and data woes

The Basel Committee on Banking Supervision’s revamp of its standardised capital approaches will affect all banks, even those using their own models – part of a drive by regulators to enhance comparability. Dealers fear the cost will be higher capital requirements

Change afoot: banks are watching the development of the new standardised approaches

A 500% jump in market risk capital requirements; a doubling of capital requirements for corporate lending; and a "significant" rise for operational risk – these are the scare stories flying around as the Basel Committee on Banking Supervision overhauls its standardised capital models.

In the past, banks with approval to model their own regulatory capital requirements might have shivered in sympathy, but would not have lost any sleep. Times have changed. The standardised approaches are being overhauled so they can be used as anchors for the whole industry – modelling banks will have to calculate and disclose their standardised numbers, enabling apples-to-apples comparisons between institutions; modelled outputs will be floored at an as-yet-unspecified percentage of the appropriate standardised approach; and trading desks that lose modelling approval will have to apply the less risk-sensitive requirement instead.

It means banks of all sizes are now watching the development of the new standardised approaches, which cover counterparty credit risk, credit risk, market risk and operational risk (see boxes below and here). They don't like what they see, and some banks are not hanging around to find out what happens next – one European bank tells Risk it has brought forward plans to cut its interest rate swap portfolio by 20%.

Others warn the same thing will happen in other business lines: "The standardised approach is overestimating the true economic risk of portfolios. There is significant capital impact not only on the high-end contracts of banks but also on the bread-and-butter business on which the economy is relying for liquidity," says the head of market risk at a large European bank.

The biggest concern for bank trading businesses is the standardised model for market risk, part of the Basel Committee's Fundamental review of the trading book (FRTB), an ambitious project that kicked off in 2012. Three consultation papers have been published, and three quantitative impact studies (QIS) have tried to gauge the capital effect.

Nobody is looking at these as set numbers because a lot of businesses won't be viable anymore

For the standardised approach specifically, the effect appears to be a big jump. According to unpublished results from the most recent QIS, four bank participants found an average capital increase of between three and five times, when compared with the existing standardised model. Rates trading desks fared worst, with the banks seeing a five- to seven-fold jump.

Relative to internally modelled numbers, the banks also found total capital numbers would be up to five times higher, while capital consumption by rates desks was as much as 13 times higher, with equity and foreign exchange desks eating up to nine times more capital.

These numbers – standardised versus modelled – will have to be disclosed side by side in future, and it is not clear how investors and analysts, or even senior management, will respond to the vast gulf between them. Banks fear the higher number will be treated as the more truthful.

The other big unknown is what impact the planned framework of floors will have. If internal model banks have to meet a minimum capital requirement based on a 75% floor to the five-times-higher standardised figures, which has been mooted by some industry figures, it would result in an almost-fourfold increase in capital.

It's not just a problem for firms with large trading operations, either: "Although our trading portfolio is small, we did see a very significant increase if you compare the standardised approach to the current approach; it was around one-and-a-half to two times. On the trading side, the biggest impact was for interest rates – just the plain vanilla products where you don't expect a big hit," says a senior risk manager responsible for risk management and analysis at a mid-sized European bank.

One of the big questions is where the increase comes from. Many risk managers blame the new approach's treatment of correlation benefits. When calculating market risk capital under the existing methodology, a bank gets to subtract a certain amount from the exposure to reflect the benefits of imperfect but correlated hedges.

Under the current proposals, however, banks have to use correlations calibrated from stressed periods with conservative adjustments to reflect uncertainty. For instance, if a dealer had a US dollar-denominated interest rate swap hedging a euro swap, it would have to use a coefficient that gave limited benefit to the hedge. "This has unintended consequences, because you have a situation where increasing hedging increases risk. We have a lot of interesting examples [where] you have perverse incentives," said Eduardo Epperlein, global head of risk methodology at Nomura, at a conference conducted jointly by Cass Business School and consulting firm Capco on June 10 in London.

The correlation framework is based on maturity buckets and is also asymmetrical. If the two exposures were in the same direction, the dealer would have to assume a higher correlation, increasing the exposure; if they offset, a lower correlation is used. So for instance, under the current proposals, one- and three-year cashflows in the same direction are assigned a correlation of 0.85, whereas one- and two-year cashflows that offset each other have a lower correlation of 0.8. This leads to an odd situation in which maturities that are further apart are more correlated than those closer together.

"In practice, you would expect correlation between one-year and two-year to be higher. This leads to some odd effects, and if they were to be embedded into the approach, it could result in some odd behaviour in the business," says a director in the risk management team at a second large European bank.

Calls for change

As a result, banks are calling for a major revision. "There is right now a concern in the industry that the asymmetric correlations lead to too big a number. We are still hoping regulators could give up on that and give us something better," says a senior risk manager at a global bank.

Protests such as this have persuaded the Basel Committee to reinstate a scrapped fourth QIS, but it will be tricky to squeeze the exercise in before an end-2015 deadline for the rules to be finalised. Banks fear any changes will be relatively minor.

"The 2015 deadline is an ambitious target for finalisation of a fundamental revision and a robust framework. A number of important proposals in key areas are still being investigated, so I hope everyone takes the time to consider those even if it takes the process into 2016," says a senior risk manager at a third large European bank.

Another area that worried banks is the default risk capital charge – an add-on applied to all trading book instruments on top of any mark-to-market losses from changes in credit spreads and migration. Some industry calculations from the latest QIS show the charge is responsible for 20% of the five-fold jump in capital requirements under the revised standardised model.

The senior risk manager at the third European bank argues not all trading book instruments need to be subject to a default risk add-on. "Most of the default risk is already captured in the price movements in equities before issuer default. When the price deteriorates, the bank already starts to hedge that position. It doesn't sit on this exposure for a while until the issuer defaults. This is not how a bank operates," he says.

Some banks have already taken steps to minimise the damage the new standardised approach, combined with the capital floors proposal, could cause. When a Brussels-based bank found the new standardised methodology would hike its market risk capital by three times – with its interest rates desk hardest hit – it decided to accelerate an existing project to reduce the notional value of its multi-billion euro interest rate portfolio by 20% by transferring trades into LCH.Clearnet, where they can be compressed (Risk December 2014).

"Over-the-counter derivatives are very penalising in terms of capital charge. When we have the right to choose the currency of collateral we post, there is a benefit to us. In other cases, we will load them to LCH.Clearnet to get a gain in the capital charge," says the head of risk at the bank.

Amir Kia, a senior manager in the risk and regulation department at Deloitte in London, says others are waiting for the rules to be finalised before taking any major business decisions. "Everybody is looking at it as a work in progress. Nobody is looking at these as set numbers because a lot of businesses won't be viable anymore," says Kia.

The senior risk manager at the third large European bank makes the same point: "What we are concerned about are the areas that affect the markets as well as the economy, so for example, bond markets and credit default swap markets are likely to be impacted because of much higher capital demand from the new rules."

Calculation challenges

It is not just the model outputs that are causing headaches, though – banks see the process of actually calculating the new standardised capital numbers as a genuine challenge, despite the regulators' intent to keep the approach fairly simple.

In its October 2013 consultation – the second of the three – the Basel Committee proposed banks use a cashflow model to calculate regulatory capital for market risk. This would require banks to break down financial instruments into their constituent cashflows and then discount each cashflow using the risk-free curve for each currency plus the credit spread of each instrument.

This was rejected as too complex, so in the December 2014 paper, regulators instead proposed a sensitivity-based approach (SBA) that allows banks to value assets in their trading book using "price and rate sensitivities that are more likely to be available in their systems as inputs into the different asset class treatments" – for example, the per-basis point sensitivity of a position to a move in interest rates, known as DV01.

But despite the move, banks still see the SBA as too complex for many banks to use. "It is onerous. I don't see any reason why the standardised approach must be that complicated," says the head of market risk methodology at the first European bank.

Many banks say it will be tricky to set up their internal systems to calculate regulatory capital under the SBA. Most of this difficulty is due to the differences between the sensitivities required by the approach and those the bank already captures.

"In the SBA, billions of sensitivities are funnelled through regulatory prescribed buckets and risk weights and aggregated back up into a capital charge – they are not necessarily the same sensitivities we report every day," says the director in the risk team at the second European bank.

"You might have a spread-based sensitivity used to measure and monitor interest rate risk today, while the FRTB requires an outright sensitivity or sensitivities today in regulatory stress testing, where the methodology for market risk stresses is less dependent on the same set of buckets and risk weights or shocks as prescribed in the FRTB. All these different sensitivities will need to be clearly labelled within large data solutions at the banks," he adds.

The SBA also requires banks to calculate non-linear risks or second-order sensitivities such as gammas. This will be testing for smaller dealers, which often only capture first-order sensitivities when measuring options.

"The original spirit is to use risks banks already have in the data household. The way the regulators want us to calculate gamma risk is quite sophisticated – that is to use a full valuation approach. This is clearly something we don't have, which makes it overly complicated to build and forces banks into the avenues of internal models," says the head of market risk methodology at the first European bank.

The measurement of indexes is another area banks are grappling with. The proposed rules require banks to split indexes into their individual components, which are then measured for risk sensitivity individually by assigning them to tenor-based buckets. The constituents of the index then have to be adjusted for basis risk due to imperfect matching of the components.

"This is completely different to what banks do for liquid indexes at the moment," says Deloitte's Kia.

The current approach treats an index as a standalone instrument. The basis is so small for equity products that it is currently ignored, while for credit indexes, basis is treated as a risk factor in its own right. Kia says the new rules will require significant IT input.

"The approach they are using right now is not only difficult but it also requires
them to go to the IT infrastructure providers and make sure they have data to do this calculation. For instance, the data should be able to know the difference between indexes and non-indexes, break the position and net it and make sure the calculation is correct. That is a big challenge," he adds.

Based on a recent report by research services firm Celent, technology vendor Numerix estimates regulation-driven global investment in capital markets technology including projects such as the FRTB will total $27 billion over the next year.

 

Credit risk reboot

The Basel Committee on Banking Supervision published its latest draft of the standardised approach to measuring credit risk in the banking book in December 2014. It applies primarily to loan and bond portfolios. Like the other standardised approach reboots, the aim is to improve comparability, but regulators are also seeking to reduce reliance on external credit ratings.

Under the proposals, capital requirements for corporate exposures would be based on a look-up table in which risk weights range from 60% to 300% on the basis of two drivers – revenue and leverage. Exposures to residential mortgages attract risk-weights of between 25% and 100% – previously they attracted a flat 35% risk-weight.

A recent study by London-based consulting firm Risk Control shows that under the proposed risk weights, interest rates for corporate loans may have to increase to cover the extra capital consumption. The firm calculated the standardised numbers using data from a major bank with a significant presence in the Swiss loan market and found that – by itself – the risk-weight increase doubled the capital requirement. The increased capital costs would require the bank to raise its interest rates on these loans by up to a percentage point if it wanted to preserve its existing return on equity.

In some classes of specialised lending, such as income-producing real-estate and commodity trade financing, required capital rose by almost two-and-a-half times. Capital for residential mortgages, however, fell under the proposed model – to the surprise of some. "The proposals involved quite significant shifts in capital towards corporate lending while affecting exposures to mortgages and banks much less. This is hard to square with the nature of the recent crisis," says William Perraudin, a director with Risk Control and chair of finance at London's Imperial College.

The current calibration of the look-up tables is to blame, says Perraudin, and in the Swiss example it may have an outsized effect on smaller domestic banks.

"In Switzerland, there is a reasonable amount of activity in Cantonal banks in corporate lending. These are also standardised approach banks, so they will be directly impacted by it," he says.

In the Dutch market, smaller dealers also complain the capital requirements under the proposed standardised approach are punitive. "The standardised approach can have a big impact on us and a number of our peers if some elements of the method and parameters, as proposed by the committee, are not changed. We would be specifically hit hard in a couple of portfolios where the method proposed right now is relatively punitive," says a senior risk manager responsible for risk and capital integration at a mid-sized European bank.

For example, the bank has a large exposure to Dutch mortgages. Loan-to-value ratios (LTV) and debt-service-coverage ratios are the determining factors of the risk weight to be charged for both Dutch mortgages and residential real-estate loans. However, historically, LTV ratios for mortgages in the Netherlands are very high, which under the standardised approach would force the bank to apply a higher risk-weight than competitors using internal models.

The Dutch market has a number of social aspects that can lower default risk but are not captured by the standardised model, banks claim. For instance, the country's social security system can cover the mortgage repayments for a person who is newly unemployed. Foreclosure is also faster, which can limit losses to the lender.

"The one-size-fits-all approach of using the same percentage for each country is, in our view, an issue. An X% loan-to-value mortgage in country A is a different story when compared with X% in country B," says the senior risk manager at a mid-sized European bank.

 

Op risk overhaul

The Basel Committee on Banking Supervision's proposed changes to the standardised approaches to operational risk were published in October 2014, and represent a significant overhaul.

Presently, there are two standardised approaches to calculating operational risk capital, both of which use a bank's gross income as the base input. In the wake of the 2008 financial crisis, regulators felt this was not the most accurate approach, as losses – even those caused by operational errors – actually cut gross income, counterintuitively leading to lower operational risk capital requirements.

The revised approach replaces gross income with a so-called business indicator – a proxy for the scale and complexity of a bank's operations – which consists of items from the profit and loss (P&L) statement such as interest income, interest expense and fee income.

Some dealers argue the business indicator is also not an appropriate reflection of risk: "The business indicator is a compound indicator consisting of items from the P&L statement. There is no straightforward link with any business environment developments or with any internal control factors. The link between risk management and risk measurement is weak, so we may miss the opportunity to attract the right attention to the relevant operational risks," says a senior risk manager responsible for operational risk modelling at a mid-sized European bank.

Others complain the maximum risk-weighting applied to operational risk exposures is set to jump from 18% under the current rules to 30% in the new approach.

Separate risk-weighting of different business lines, such as retail and commercial banking, have also been scrapped – a move that is criticised by a number of risk managers.

Many banks have yet to calculate the full capital impact of the rules, but the operational risk modelling source says a preliminary assessment suggests it will be significant.