Dealing with funding on uncollateralised swaps

Many banks are now using their own cost of funding as a discount rate when pricing non-collateralised swaps trades. How are banks dealing with the difference in funding rates when quoting derivatives prices, and could this influence a client’s choice of dealer? By Christopher Whittall

Derivatives pricing has never been simple, but there were a few constants people used to be able to rely on. One of the most fundamental was the use of Libor as a discount rate to price derivatives trades. The financial crisis has caused this assumption to be thrown out of the window. The majority of banks now recognise that the overnight indexed swap (OIS) rate should be used to discount future cashflows on collateralised swap transactions (Risk March 2009, pages 19–22). Meanwhile, non-collateralised trades should theoretically be discounted at the bank’s own cost of funding. But this is far from easy – not least because individual banks fund at different levels, making standardisation and comparison all but impossible.

“If each bank discounts the cashflows using its own funding, the same swap will not have the same value from one bank to the other. It becomes a completely different game because, in effect, a bank has to move away from marking a swap to market towards marking to model,” explains Christophe Coutte, deputy global head of flow, fixed income and foreign exchange at Société Générale Corporate and Investment Banking in London.

The differences can be significant. If a bank assumes it can raise funds at Libor plus 100 basis points, a client receiving a fixed rate as part of an uncollateralised swap would collect a lower amount than if the dealer assumed a funding rate of Libor flat. Barclays Capital estimates the fixed rate on a 10-year US dollar interest rate swap would be around 4bp lower if the dealer funds at Libor plus 100bp rather than Libor flat. In theory, this means higher-quality banks with a lower cost of funding will be able to offer a better rate to clients that receive fixed through a swap. It’s not all one way, though: a client paying fixed rate would be likely to benefit by trading with a dealer with a higher cost of funding, as it is effectively lending the bank money.

A quick glance at Libor quotes roughly illustrates differences in borrowing costs between banks – the quotes for six-month euro Libor on June 2 ranged from 0.83% for UBS to 1.01% for WestLB. While this is not necessarily an accurate reflection of the levels at which banks can raise funds via the commercial paper or medium-term note markets, it does highlight variations between firms in the current environment.

Throw in the credit charges now built into swaps trades (Risk May 2010, pages 26–28) and the prices quoted by banks for an uncollateralised swap can differ markedly, making any comparison of prices difficult for customers. “At the moment, everyone knows where the mid-price is on a collateralised deal for a five-year interest rate swap. But by the time a bank has applied the credit value adjustment (CVA) and its own bespoke way of charging for funding, prices from banks for even plain vanilla deals could look very different on an uncollateralised basis,” says Clive Banks, head of derivatives marketing for European corporates at BNP Paribas in London.

This issue has been particularly pressing since the financial crisis. In a hypothetical trade where a bank pays fixed and receives floating – and the yield curve is upwardly sloping – the fixed rate will be higher than the floating rate at the inception of the transaction. At some point, the floating payments should become greater than the fixed rate paid by the bank, assuming the market follows the path predicted by the forward curve. 

In other words, the bank is effectively lending the client money over the first half of the swap and receiving it back in later years. In an uncollateralised swap, the dealer has to fund these payments itself – and since the crisis, the level at which banks can raise funding has risen significantly, with greater variation between firms. To illustrate, the difference between US three-month Libor and the US federal funds rate ballooned from around 8bp prior to the crisis to 366bp in October 2008. US three-month Libor was 32bp above the federal funds rate on June 16.

“It comes down to banks realising they are in the business of lending money, and just as a loan officer might make a decision to lend billions of pounds to some big utility, we’re going to have to make those same decisions when we’re doing a big swap with them,” says Richard Armes, co-head of European interest rates at Morgan Stanley in London.

Dealers say certain trades have always incorporated a funding element – typically those that include an upfront payment or are traded at an off-market rate. Theoretically at least, all trades should now be discounted at the rate at which banks can borrow, dealers say. 

“Even before the crisis, if a dealer was either receiving or paying a big upfront amount, the trade would not be discounted using Libor but a particular funding rate. That is obviously more prevalent now. This discrepancy in funding costs between overnight money and three-month Libor has increased the visibility of the funding cost for the derivatives book,” says Andrew Brown, head of corporate risk solutions for Europe, the Middle East and Africa at Royal Bank of Scotland (RBS) in London.

Whether dealers pass this cost on to clients that trade plain vanilla, on-market swaps is a different question, however. Some argue any bank that incorporates its cost of funding into a plain vanilla swap will be uncompetitive. As a result, some choose to swallow the impact internally.

“It is less obvious whether to pass on the funding charge to clients on simple products such as interest rate swaps where the pricing impact may be minimal,” says John Langley, co-head of the risk solutions group at Barclays Capital in London.

But even if the charge is not ultimately passed on to the client, banks need to be aware of the impact of funding in every single trade. Dealers are cagey about how they formulate the funding curves they use to discount uncollateralised swaps – although some suggest a blend of short-term, medium-term and long-term funding is used (see box, Building a funding curve). Many of the major dealers say they have set up specific desks to manage the funding risk of their derivatives portfolios, in much the same way a CVA desk manages the credit exposures.

“We externalise the funding implication from our trading desks, so the desk will either pay or receive funding benefit from our fixed-income treasury. A centralised desk aggregation is crucial to ensure you can offer the tightest bid-offer spreads, and it also makes it fair for clients: whenever they are paying or receiving funding, they are getting the fair level for the funding aspect of the deal. Also, the rates desk will know if we’ll pay more for 10-year money and therefore can give benefit to trades where people are paying us fixed,” says Banks of BNP Paribas.

So long as banks are consistent in their approach, the funding impact could benefit clients if they are paying fixed rate. In reality, corporate clients often prefer to receive fixed, creating potential problems for those dealers that have predominantly one-way business with corporates and hedge in the interdealer market.

Trades between banks are subject to credit support annex (CSA) agreements and so are backed by collateral. It is now generally accepted the future cashflows on these trades should be discounted using an OIS curve – creating a mismatch in dealer books between collateralised and uncollateralised transactions.

“The OIS curve may be around 30bp below Libor, while the uncollateralised world might be 100bp above Libor – but those numbers move. If you calculate the hedge ratio today between uncollateralised and collateralised hedges, it might be 95%. But if the OIS-Libor spread goes to zero, your hedge is going to increase a bit, which might tell you to increase your ratio to 97%. That spread is a risk and therefore you should think about whether you should hedge against that, which isn’t straightforward,” says Edward de Waal, head of structuring in the risk solutions group at Barclays Capital in London.

In fact, some suggest this problem could be exacerbated by the new pricing methodology. Clients might prefer to use high-quality banks when receiving fixed and poorer credit banks with higher costs of funding when paying fixed. “It may be to the advantage of clients to be able to say ‘for that kind of swap, I am going to use Bank X or Bank Y’. For dealers, it is not so good because there is a discrepancy in pricing methodology,” says Banks of BNP Paribas.

One potential answer is to use Libor-OIS basis swaps to hedge this mismatch. However, that market is still in the early stages of development. “The Libor-OIS basis does present a hedging issue as there is not much liquidity, but the market has grown over the past two years and I expect that to continue. At present, most liquidity is in the cross-currency basis swap market, so dealers can use a combination of cross-currency basis as a hedging proxy,” says Kevin Liddy, co-head of counterparty exposure management at RBS in London.

Ultimately, the mismatch between non-collateralised and collateralised swaps could spur dealers to ensure they have a more balanced client business – and this could work in favour of corporate end-users, with banks offering attractive terms to encourage clients to take the other side of the swap. “The big issue around funding is the mismatch between uncollateralised and collateralised counterparties. For example, if a bank receives fixed from one uncollateralised client, then in theory it has an axe to pay fixed from another uncollateralised counterparty. To the extent that you can create a balanced portfolio of market risk with your uncollateralised counterparties, you can be more aggressive from a pricing perspective,” says Langley at Barclays Capital.

This is all well and good, but there are some significant hurdles for banks – not least whether they should re-mark legacy swaps trades using their new funding curves. One dealer suggests the impact could be huge, particularly for those dealers with one-way exposures where client trades are hedged through the interbank market. Nonetheless, some stress the hit needs to be taken.

“You can’t ignore your existing stock business – you need to apply the same methodology to it. That may result in a gain or loss, but the size of that gain or loss is purely determined by what level of funding you put in to the curve, which is arbitrary. We approached this by incentivising a number of our clients across banks and corporates to amend their trades, and we have provided significant funding benefit as a result,” says Brown at RBS.

Other complicating factors exist. Specifically, many CSAs signed with corporate customers include thresholds, which mean the firm only needs to post collateral once a predetermined level has been breached. In other words, the bank needs to use its own cost of funding as a discount rate up until the moment the market value of the swap crosses the barrier, at which point the OIS rate would be used (see box, Taking CSA optionality into account). Termination options also need to be considered – will the trade be terminated before maturity, and what impact will that have on funding requirements?

“A bank has to decide if it is reasonable to assume these trades are going to stay on the book and it’s actually going to incur those funding charges for the life of the trade. Trades are often restructured, recouponed or wound down before they reach maturity. In some situations, it may make sense to assume a shorter effective maturity when charging for funding,” says Armes at Morgan Stanley.

Many dealers claim some of the less sophisticated banks are not taking all these variables into account. Others may deliberately not take funding into account to build up market share. Ultimately, however, those firms that aren’t considering the funding implications of derivatives trades may end up regretting it, say dealers.

“Many of the banks that didn’t use CVA models in the past three to four years had to take massive provisions when they eventually did adopt systems, as they always under-reserved. It could well be a similar story with funding,” says Gary Cottle, head of corporate risk solutions at RBS in London.

BOX

Building a funding curve

A bank would typically have an internal cost of funds that dictates where it lends to clients, as well as an external cost of funds that reflects the rate at which it can borrow in the market. The internal cost of funds – likely to be used to price uncollateralised derivatives trades – will generally be constructed from a combination of short-term, medium-term and long-term borrowing rates. “A bank’s internal cost of funds would have various inputs: where it can borrow benchmark size liquidity (three, five and 10 years) and short-term borrowings (three, six and nine months). Out of most banks’ public market borrowing, I would guess half of it is short-term paper,” says one head of long-term funding at a European bank.

The reliance on short-term money may ring alarm bells given the problems experienced by countless institutions when funding markets seized up during the credit crisis. Regulators are battling to prevent a rerun: the Basel Committee on Banking Supervision is drawing up liquidity risk management standards, which will force banks to eliminate structural mismatches between assets and liabilities and reduce the reliance on short-term, unstable funding.

However, the funding requirements of derivatives desks are unlikely to be top of the list of problems to tackle, argue some. “While not small in terms of absolute numbers, derivatives funding is a relatively small piece of a bank’s balance sheet. So when markets seize up, the fact you’re supposed to fund a forward-forward position in three years’ time is not going to affect your true solvency,” says Andrew Brown, head of corporate risk solutions for Europe, the Middle East and Africa at Royal Bank of Scotland.
Whatever the regulatory outlook, the ability to predict funding requirements for derivatives transactions in the future is difficult, says the head of long-term funding: “While banks should know their current cost of funding and balance-sheet liquidity, it is very hard these days for them to project forward what their cost of funding is going to be. I’m also not sure all banks incorporate funding costs accurately because it can be detrimental to the business – it is very expensive.”

 

BOX

Taking CSA optionality into account

Thresholds embedded within credit support annex (CSA) agreements can create headaches for banks. In theory, the future cashflows of the uncollateralised swap would need to be discounted at the bank’s cost of funding until the market value of the trade breaches a pre-specified threshold. At that point, the corporate would need to post collateral, and the trade would be discounted using the overnight indexed swap (OIS) rate. The impact on the price of the swap can be noticeable.

According to Barclays Capital, a client obtaining fixed rate on a semiannual 10-year sterling uncollateralised swap would receive 3.454% if the dealer assumes it can fund at Libor flat (for a trade executed on June 9). The fixed rate would drop to 3.42% for an uncollateralised trade assuming Libor plus 100-basis-point funding. However, this would rise to 3.437% if the client signs a CSA with a threshold of £50 million and 3.452% for a CSA with a £5 million threshold. In other words, the smaller the funding requirement for the dealer, the better the fixed rate the client receives.

“We use a Monte Carlo engine to calculate the funding of the trade in a more sophisticated manner where the credit terms are complicated. We can model the paths that go over the threshold limit –  which are priced off the OIS rate – while the paths that are below the threshold or the minimum transfer amount are priced at the uncollateralised curve. You can calculate the effect fairly precisely,” says Edward de Waal, head of structuring in the Barclays Capital risk solutions group.

Other optionality can be built in to CSAs – for instance, mutual breaks. Here, the approach between dealers can differ. “If you start pricing for breaks, you may well have to break the trade for economic reasons, even if commercially you do not want to,” says John Langley, co-head of the risk solutions group at Barclays Capital.

Andrew Brown, head of corporate risk solutions for Europe, the Middle East and Africa at Royal Bank of Scotland (RBS), takes a different view. “We price for mutual breaks, rolling breaks and expected call dates. When pricing, a number of factors feed into the decision – how highly rated the client is, which way around the portfolio is and how often we’ve dealt with them. These are all important factors to bear in mind,” he explains.

The lack of uniformity in CSA documents can lead to serious pricing complexities: thresholds can vary from client to client, different currencies can be posted, while breaks occur in some CSAs and not others. Richard Armes, co-head of European interest rates at Morgan Stanley in London, says CSA agreements are likely to become more standardised as a result.

“Where there is optionality in these CSAs for no good reason, I think there will be a widespread renegotiation and the vast majority of swaps will become standardised in terms of CSAs,” he says.

There is one breed of collateral agreement that is particularly troublesome for dealers: the one-way CSA. These are typically signed with sovereigns and supra-nationals, which have a higher credit rating than the bank counterparty. In these transactions, the dealer has to post collateral if the net present value (NPV) is negative, but does not receive collateral in return if the NPV of the trade moves in its favour. Some sovereigns have large derivatives portfolios, which mean the funding implications for banks can be significant.

The growing use of central counterparties might present one solution, suggest dealers, as this would require sovereign counterparties to post margin. “Sovereigns present two kinds of risks: credit risk and funding risk. The credit risk is unavoidable, but the funding risk can be mitigated, particularly as most of the largest funding exposures for dealers will come from sovereigns. If governments are sponsoring central clearing, it would seem perverse if they didn’t participate in it themselves,” says one London-based rates trader.

Dealers suggest sovereigns could even post their own government bonds as collateral – an easy option for most to comply with. This would not mitigate credit risk for the bank counterparty, but could help from another perspective as it would help the dealer to fund the trade.

“I doubt sovereigns will race to go on to central clearing – they tend to have large derivatives portfolios in one direction, and the initial margin number would be substantial. But if sovereigns want to persist with one-way CSAs, they will have to mitigate the funding costs, otherwise they will be charged. The idea of posting your own sovereign paper could be a good solution,” says Kevin Liddy, co-head of counterparty exposure management at RBS in London.