SYDNEY: 19 November 2014 – Today, I’d like to talk about benchmark reform in both foreign exchange (FX) and interest rates. I chaired the FSB work on FX with Paul Fisher who’s now at the PRA, but formerly head of markets at the Bank of England. Both of us were also involved in the interest rate benchmark work of the FSB, which was chaired by Jeremy Stein (formerly of the Fed) and Martin Wheatley of the FCA. I will also talk a little about what we’re doing here in Australia on the interest rate front.

The FSB published the recommendations of the Official Sector Steering Group on interest rate benchmarks which focussed on LIBOR in the middle of the year. There were two main aspects of the report. The first was around the calculation of a credit-based reference rate. The second was on developing risk-free alternative benchmarks which may be more appropriate than credit-based benchmarks for a number of financial products.

 In Australia, the main credit interest rate benchmark is BBSW, the Bank Bill Swap benchmark rate. The method by which BBSW was calculated here in Australia prior to a year ago was quite different from LIBOR. Just over 12 months ago, the methodology for BBSW was changed from the way it had been done previously. While both LIBOR and BBSW were submission-based, there was a significant difference in what a bank actually submitted. In LIBOR’s case, a bank submits the interest rate it thinks it can borrow at. In BBSW’s case, a panel of banks submitted what they think a prime bank, which in Australia’s case mostly (and certainly now) means what a major bank can borrow at. An important difference is that BBSW is about a generic bank, not any bank in particular.

So the incentives involved were somewhat different. Nevertheless in September last year, and in line with the recommendations in the IOSCO report on benchmarks, BBSW was shifted to an executable price methodology. It’s based on the best bid and offer across three platforms on which bank bills are traded. That is, it is now a traded price (or at least an executable one), rather than a submitted price.

In other parts of the world, consideration is being given to working out how to shift from a submission-based benchmark to a benchmark based on traded prices. One of the fundamental problems being the market isn’t traded and that’s a fundamental issue that overseas regulators are still grappling with.

The other main part of the FSB interest rate work focussed on the issue that credit based interest rate benchmarks are used more than is necessary, often because of historical accident. A bank interest rate is a combination of a risk free rate and a credit premium, because it’s a bank borrowing not a sovereign borrowing. When LIBOR started being referenced in contracts, it was basically fit for purpose, but it is now referenced in any number of contracts which it just really doesn’t need to be referenced in. So one of the other main focuses of the FSB work is to ask the question: is a bank-credit benchmark being used appropriately or would a risk free benchmark be more appropriate?

Then the question becomes if a risk free rate is more appropriate, what should it be? Around the world, there is a fair amount of work occurring on that issue at the moment. The Reserve Bank is working with the Market Governance Committee at AFMA to come up with some possibilities. One obvious possibility in Australia is OIS (overnight index swap) rates. Nevertheless there can also be the same problem with OIS if it’s not traded much either; there needs to be a robust way of calculating that each day. You don’t want to shift from one thinly traded market to another; it’s not going to solve the problem. So careful thought needs to be given to addressing that issue.

Let me turn to the FX market which has been my night job for most of this year. The group that Paul and I chaired put out a report at the end of September with recommendations about how to address the problems around the foreign exchange market, and particularly the benchmarks (or fixes) in foreign exchange. ‘Fixing the fix’ is roughly the subtitle of our work.

The primary foreign exchange benchmark is the 4pm London fix and that’s true for every single currency. So while there are fixes in our time zone here in Australia, the vast majority of contracts and products in Australia reference the London 4pm fix. There is relatively little use of FX benchmarks published in the Australian time zone.

The London 4pm fix is calculated by WM Reuters. It is calculated over a one minute window centred around 4pm. WM take the median transaction price in that window from either a Reuters or an EBS feed. In passing, it is worth noting that the problem with LIBOR was it was based on too few transactions; the problem with the FX benchmark is that is arguably based on too many transactions!

Paul and I were tasked with reducing the incentives for manipulation of the fix. We did this in complete ignorance of any of the investigations which have just come to light. The recommendations are based on our understanding of the way the market worked, informed by conversations with a large number of market participants.

The recommendations of the report focus on a few main areas. The first is the way the fix is actually calculated by WM. We recommended widening the window over which the fix is calculated. Two days ago WM announced that they’re going to widen it from one minute to five minutes. We also recommended they use more price feeds. As I just noted, currently they calculate only off either Reuters or EBS which only accounts for a relatively small share of traded volume in any currency pair. WM has agreed to follow through on our recommendations and have been co-operative throughout the whole process. In addition, IOSCO did a review of WM which suggested some areas of improvement, principally around governance, which WM has also agreed to follow through on.

The second set of recommendations focus on the market infrastructure around the fix. One of the things which became clear during our work is that the fixing business is often unprofitable for the banks involved. If I’m a customer with some business I want to do at the fix, I ring you - the bank - up around 15 minutes before the fix and say I want you to deal me at the fix price, no spread, no fee. Even though the bank doesn’t yet know the price, it agrees to do the transaction. At a minimum, this creates optics of dealers ‘trading ahead’ of the fix even where the activity is essentially under instruction from clients. One of the main issues around FX is that hedging and front running can be observationally equivalent. It is very hard to tell the difference between the two, including by observing the price action.

Hence, one of the report’s recommendations is that it would be reasonable to charge for this service. By charging for this loss-making business to more truly reflect the cost and risk, this should reduce incentives for manipulation. We acknowledge that competitive market forces have driven that price down to zero, but effectively one can view the fix business as an agency business, in which case a fee for service model might be appropriate. That is, going forward, if I’m going to execute for you, I’d be acting as your agent in the market, whereas in the past most of this business has been done as a principal where I take that risk on to my own balance sheet and I have to manage that risk.

Another recommendation from the report was to separate the fix business from the regular principal business. One important outcome of this would be to segregate the information flow. Another recommendation was to encourage the further development of  a number of market-based solutions which are effectively netting solutions, which accumulate all the fix orders and net them down as much as possible and then just execute the net in the market, rather than the gross flow.

So in terms of the market infrastructure, there are improvements that can be made to reduce the incentives for manipulation, but in the end it’s still a large business, a large volume going through a narrow pipe, and that still may give rise to some problems.

This leads to the conclusion that the behavioural side of things is going to be the key. In an OTC market, such as FX, defining what is, and is not, appropriate behaviour can get quite tricky. For example, if I want to complete a trade in the OTC market I actually have to share some information with my counterparty. However, I don’t have to tell them who my customer is to complete the trade. Hence the report contains some recommendations around appropriate information flow. It is relatively straightforward to articulate high level principles about information sharing. The problem is, the more precise and prescriptive, the more difficult it is. These difficulties arise for at least two reasons. Firstly, there’s a large grey area which is very hard to be precise in. Secondly as soon as you get prescriptive, then a trader can read this and say that’s what I’m not allowed to do but if it’s not written down here then I must be allowed to do it. So that becomes a problem as well.

The various foreign exchange committees around the world, including the Australian Foreign Exchange Committee of which I am the chair, are working on a common set of principles along these lines which we’re hoping to put out in the new year.

This then raises the issue of enforceability. It’s all very nice to have these codes of conduct but are they worth much more than the pieces of paper they’re written on? In Australia we endorse the ACI code (ACI is a global industry body). The ACI supports its code with a large amount of training. So one option that could be contemplated is for individual traders to (a) do that training, (b) attest that they’ve actually done the training, read the code of conduct and understood the principles.

To reiterate, behaviour is key, but it very much comes from the culture which should be coming top down rather than bottom up from the traders. So I think the culture issue is very much an institutional issue.

The other side of enforceability comes to the fact that the banks are acting as agents for their customers and there needs to be some enforceability coming from the customer side. One issue which came out of our work is that, particularly around the fix business, a number of the customers didn’t have a great understanding as to how that market operated. A large driver of flow through the fix comes from the passive asset management industry. Any time there is an asset portfolio which has more than one currency in it, generally the 4 pm fix is used to weight those different portfolios. The asset manager is being benchmarked against their performance on that portfolio, one component of which is the exchange rate. In some cases, their primary concern is that the tracking error was zero, not whether the exchange rate at which they were being dealt was the ‘appropriate’ price. From a fiduciary point of view in terms of the end customer, the person who has actually invested in that pension fund, there should be more focus on the actual exchange rate, rather than the tracking error, and on the FX execution. A number of asset managers do that, but there are some which don’t. A greater focus from users of the fix should also contribute to improved behaviour and practices on the part of those executing the business.

To conclude, the FSB report makes recommendations in a number of areas in order to reduce the incentives for manipulation, including on the calculation of the rate and on the market infrastructure. But in terms of FX benchmarks, as well as other benchmarks, conduct, behaviour and culture are critical. It is easy to say that, relatively easy to articulate some high level principles around it, but it is hard to (a) be prescriptive about what is, and is not, appropriate behaviour and (b) even harder to work out how to enforce that. 

This address was given at Allen & Overy on 19 November 2014. This work was funded, in part, by the Centre for International Finance and Regulation and in part by the Australian Research Council.