Ask someone on the street to define a bank, and they will probably tell you it is a business that accepts deposits and makes loans. So it might seem strange these trusty, common-or-garden bits of banking are giving regulators as big a headache as they have had in the post-crisis years.
While trading book instruments are often seen as riskier, contractual maturities and market-driven prices make them easier to model. Measuring the risk of banking book instruments, in contrast, involves a host of seemingly impossible questions, such as when customers will withdraw their money, buy a house or repay their loans.
The Basel Committee's interest rate risk in the banking book (IRRBB) working group, formed in 2012, is examining ways to standardise the capital treatment of these products. So far, the group is two years into its work, has not yet produced a consultation paper and members are said to be divided on whether it can be standardised at all.
"It is very difficult to have a completely standardised approach because there are so many variances across jurisdictions and banks, so I think there is sort of a general sense that a completely standardised approach is fairly difficult for interest rate risk," says a regulator close to the working group.
Industry sources say some members of the group favour an explicit capital charge – a so-called Pillar I approach – while others are increasingly favouring the more flexible Pillar II, which allows national supervisors to apply charges where they see fit. The cause for this shift is that regulators are struggling to find common ground on the basic, but big questions of how interest rate risk should be measured and how to treat non-maturing deposits (NMDs).
IRRBB is a major risk for many banks, particularly in a low-interest environment
As a result, the timetable for the IRRBB project has started to slip. Originally, the working group was meant to put out a consultation paper by the end of last year and a full quantitative impact study (QIS) at the beginning of this year. Now, the consultation document is expected, at the earliest, in March. Meanwhile, the QIS has been pushed back until later this year due to IT constraints banks are facing, according to the Basel Committee.
In a joint letter published in December, several industry organisations expressed concern that the new timeline may not allow QIS data to be taken into consideration in the consultation.
Still, discord among regulators would be a good omen for banks and industry lobbyists. Generally, they acknowledge the dangers of allowing interest rate risk to be treated differently for traded and non-traded assets – an arbitrage opportunity the Basel Committee is tackling as part of the banking book work and within its ongoing review of trading book rules. But they also argue banking book products do not pose enough of a threat to require a standardised treatment.
"You don't need a bazooka to shoot for a small target," says Gerhard Hofmann, vice-president of the European Association of Co-operative Banks in Berlin. "By putting IRRBB under a Pillar I charge, the Basel Committee would hit the bread-and-butter business of a large number of banks that are not internationally active, nor of systemic relevance. It really raises the question of whether regulators are shooting at the right target."
They are, says a spokesman for Germany's Bundesanstalt für Finanzdienstleistungsaufsicht (Bafin): "IRRBB is a major risk for many banks, particularly in a low-interest environment. Losses on IRRBB will not necessarily be compensated by gains on other exposures, so they should not be covered by capital allocated to existing Pillar I risks. This is already part of Bafin's approach for capitalising IRRBB and is one of the criteria against which we will assess the options on which the Basel Committee will consult."
Like the products from which it arises, IRRBB is as old as the hills, and regulators have been trying to find the best way to capture and capitalise on this risk since the early 1990s. The Basel Committee cites four main drivers of IRRBB risk, the first and most important being repricing risk. When interest rates rise, a short-term deposit funding a longer-term fixed-rate loan will reprice sooner, so the bank will be paying a higher rate on the deposit while continuing to get the same income from the loan.
The other sources of risk are: yield-curve risk, where repricing mismatches expose a bank to changes in the slope and shape of the yield curve; basis risk, where divergences occur between the various rates used as the basis for funding and funded instruments; and optionality, where the cashflow of an instrument can be altered by the buyer or seller.
The capital regime for capturing these risks has never been standardised internationally due to differences in local customer behaviour and the products offered. While it may be standard to repay a mortgage ahead of maturity in one country, for example, it may not even be possible in others.
Instead, national supervisors must set their own capital requirements via the Pillar II framework. Europe requires the banking book to be shocked by a 200 basis point move in interest rates, and if the economic value of a bank's equity falls by more than 20%, the institution has to report this to its national supervisor.
Towards the end of last year, the Basel Committee began updating the Pillar II treatment of IRRBB. While regulatory sources say this was always part of the work plan, others see it as a sign the Pillar II approach might be retained. German and UK supervisors are said to be the strongest supporters of a Pillar I approach, with those from Japan and the US supporting Pillar II. So far, Australia is the only country that applies a Pillar I treatment to IRRBB (see box, The Australian way).
"At the beginning of last year, the majority of Basel members were very much interested in a Pillar I charge. Now we are hearing there are changing opinions among members, possibly leading to different majorities," says Michaela Zattler, division manager for banking supervision at Germany's Bundesverband deutscher Banken (BdB) in Berlin.
It isn't hard to see why. For a start, there are two quite different ways of expressing interest rate exposure: the earnings-based method, whereby a bank measures the impact a rates move can have on its net interest income; and the economic approach, where the impact of rate movements on the present value of future cashflow is measured.
While both approaches give broadly the same results if everything were to reprice in line with changing market rates, in reality this is rare, meaning the two approaches can yield significantly different – sometimes opposite – results in practice, and cannot both be managed at the same time.
To use a crude illustration, imagine a bank has a £100 deposit paying 2% that reprices the first year, funding a £100 loan receiving 3% that reprices in year five. In the event of a 200bp rise in rates, the earnings approach would consider the effect of that shock on net income – the bank would be paying 4% on the deposit for four years, while receiving 3% on the loan until the fifth year, meaning a loss of £3 overall.
A bare-bones economic value example using a static gap analysis would put assets and liabilities into time buckets. For each bucket, the notional difference between the assets and liabilities repricing in that time period is calculated. The gaps are discounted using the current rate of interest, and then again using the shocked rate, generating a market value sensitivity number. It means little in isolation, as it doesn't take into account interest or other income flows; rather, the key is how this figure changes in response to shocks.
The large dealer banks each present their interest rate sensitivities differently. Bank of America Merrill Lynch, for instance, uses the earnings-based approach, while UBS uses economic value (see box, Interest rate sensitivities, selected banks).
Australian banks, which are subject to a Pillar I charge, all present their sensitivities in economic value terms in their reports, which shows they don't react in the same way to interest rate shocks – while most experience a loss when rates go up and gains when they rise, Westpac's results show the opposite.
Regulators tend to favour the economic value approach because the results are more comparable across banks and rely less on assumptions about future cashflow. There are numerous ways to construct an economic value approach, however, so industry groups are keen to ensure the chosen method chimes with their view of the risks.
"One of the issues we've got is there has never been an accepted understanding of what an economic value approach looks like, and that's something we've said we'd love to work with them on," says Simon Hills, executive director of prudential regulation and risk at the British Bankers' Association in London.
Other bankers say the mark-to-market style of the economic approach does not take into account the variable aspects of the banking book and leads to an unrealistic snapshot of IRRBB.
"There is a push by a lot of the regulators to implement effectively a mark-to-market approach to the banking book, which is what they have for the trading book. The banking book, by its nature, is not a mark-to-market beast as it is based on accrual accounting. It's really all about looking at your income over a cycle and not taking an instantaneous liquidation view of the world, as is appropriate for the trading book," says Shailesh Shah, chief risk officer for group treasury at UBS in Zurich.
While the regulators on the working group decline to speak publicly, their shifting thoughts on the issues can be inferred – with all the caveats that implies – by looking at the changing focus of their draft QIS. Despite the absence of a consultative document, a mini-QIS was filled out by a group of banks at the start of 2014, followed by queries to the industry on how to complement the economic value approach and whether a hybrid approach makes sense.
According to the BdD's Zattler, the hybrid methodology would start with the economic value approach and include an earnings overlay. Zattler criticised this approach at a Risk conference in London last October, saying she couldn't see how the two methods would work together.
Regulators admit a hybrid approach will be tricky: "It's technically very difficult to combine these two very different approaches, so it is very difficult to say what the outcome will be," says the regulator close to the IRRBB working group.
NMDs are another big source of debate. These include interest-bearing liabilities such as savings accounts and non-interest bearing liabilities, such as traditional checking accounts for which there is no way of knowing the maturity until customers choose to withdraw their money. There is also no set repricing pattern because banks do not automatically pass changes in interest rates to their clients.
Currently, banks tend to use internal models, based on historic data, to forecast expected maturities for their deposit base. This allows obvious scope for gaming, and regulators are said to be considering a maturity cap on NMDs as a result. Draft IRRBB guidelines from the European Banking Authority suggest capping maturity at five years, which some industry sources see as a clue to where the Basel Committee may end up.
"If a bank is trying to demonstrate to its regulator that its risk is low, there is a certain incentive to pick a modelling approach for NMDs that gives the smallest rate risk. Unsurprisingly, regulators don't like some of the results from those models, so they put up fences around the duration of deposits," says Leonard Matz, a Pennsylvania-based liquidity risk consultant.
Increased hedging costs
Banks and industry lobbyists have reacted furiously to the idea, warning it would force banks to limit the duration of their assets to match their liabilities, restricting their ability to lend, and potentially increasing hedging costs.
"While we support the regulators' objective to ensure sound management of NMDs and of the banking book in general, the assumption of a uniform and very short duration for NMDs is a cause for concern. While this might make sense from a liquidity risk standpoint, from an IRRBB management perspective it may distort how banks actually manage their risk. Banks might be forced to enter into unnecessary hedges because NMD balances are no longer recognised as offsetting assets, leading to more volatile earnings," says Jermy Prenio, deputy director at the Institute of International Finance in Washington, DC.
In addition, bankers complain it creates another gap between the economic and earnings-based approaches. If the economic value model has NMDs with short duration caps, a rise in interest rates will harm the bank as the liabilities may be supporting longer-term assets. However, if the same rate shock is modelled using an earnings approach, it would suggest the bank benefits.
It is another issue on which regulators are struggling to agree. "No decision has been made on NMDs, but a lot of ideas have been floated. The committee hasn't made any decision as this is one of the most difficult issues," says the regulator close to the Basel working group.
Some insight can once again be found in the group's QIS documents. In the template drafts given to banks in November, ahead of the full-blown QIS exercise this year, banks were asked to submit information based on their own estimates of NMD maturity, alongside a prescribed approach, possibly indicating a softening of the line on maturity capping.
"My feeling is they are recognising and acknowledging industry feedback, and this is partly reflected in the additional components in the draft QIS, compared to what they asked for before in the mini-QIS. The fact the timetable has been postponed means there are obviously internal discussions going on, based on feedback from the industry," says Bruce McLean-Forrest, managing director of treasury risk control at UBS in Zurich.
BOX: The Australian way
In 2008, the Australian Prudential Regulation Authority (Apra) implemented a standardised Pillar I capital charge for interest rate risk in the banking book (IRRBB), which includes some of the components now being considered by the Basel Committee on Banking Supervision.
Banks, for instance, have to calculate the maximum potential change in the economic value of the banking book for repricing and yield-curve risk, based on a 99% confidence level and a one-year holding period. This means the duration of investment capital, or equity, is capped at one year. Banks are allowed to calculate and assign maturities to non-maturing deposits – a freedom the Basel Committee's working group may restrict or remove – although banks must submit documentation with analysis to support their decisions.
"If anything, the regulation has just made banks analyse the balance sheet much more closely, understand the risk they have in there, and change the way they do business to mitigate some of these risks," says Craig Davis, head of KPMG's financial risk management practice for South-east Asia in Singapore.
"Rather than not lending, they have changed some of the products' features. For example, to mitigate basis risk, the product rate now changes in line with the underlying wholesale rate, without a window or a lag. Or, embedded optionality is better understood with product features changing to allow an effective market hedge," he adds.
Although there are no set standards for basis and optionality risk calculations, banks with permission to use internal models must have these approved by Apra. Internal models for basis risk must explain historical variation of margin between product interest rates and implied costs of funds, while optionality models must capture the non-linear price characteristics of positions with implicit options and include assumptions about variables such as the effect of interest rates on prepayment levels.
BOX: Interest rate sensitivities, selected banks
All figures are taken from recent bank earnings releases, annual reports or investor presentations. Some describe bank-wide impacts while others focus on the banking book.
Bank of America Merrill Lynch: The impact of a 100bp increase in interest rates on net interest income, excluding trading activities, would be a gain of $3.2 billion as of September 2014. A 50bp decrease would result in a loss of $2 billion. Net interest income, excluding trading activities, was $28.3 billion for the first nine months of 2014.
Barclays: A 200bp move upwards would increase pre-tax net interest income for non-trading financial assets and liabilities by £302 million as of June 2014. A 200bp decrease would result in a loss of £481 million.
BNP Paribas: An upward shock of 100bp would decrease banking book revenue by €126 million, or 0.3%, as of the end of December 2013.
Citi: A 100bp upward parallel interest rate shock would increase net interest revenue in its non-trading portfolio by $1.9 billion, while a 100bp decrease in long-term rates would cause a loss of around $150 million, according to data from September 2014.
JP Morgan: A parallel interest rate increase of 200bp as of the end of September 2014 would increase core net interest income by around $4.4 billion. Core net interest income, which excludes the impact of market-based activities, is reported at $30 billion for the first nine months of 2014.
Societe Generale: A 200bp increase in interest rates would increase net interest margin by €488 million, and decrease it by €391 million if there were a 200bp fall in interest rates as of December 2013.
UBS: A 100bp upward shift would lead to an increase of Sfr1.6 million ($1.78 million) in the present value of future cashflow in the banking book as of September 2014. A 200bp decrease in rates would increase the present value of future cashflow by Sfr176 million.