Nikki Marmery, editor, Credit magazine (NM): A good place to start is probably the area that's attracted most controversy lately: the subject of rule changes to indices. The addition of Fitch to Lehman's index coming at the same time as the autos downgrades attracted a lot of press as it caused some confusion over the autos' inclusion in the investment-grade index. But how are these rule changes determined and implemented, and to what degree is the buy side involved?
Ravin Onakan, executive director in index/portfolio strategies at Lehman Brothers (RO): Some of the members present here today have participated in the Lehman Brothers index advisory council. We consult major users at a session that takes place once a year and at the end, members vote on the issue. However the council members' decision is not binding: Lehman has the option to decide on what changes it implements.
Caroline Harrison, index analyst at Merrill Lynch (CH): A similar thing happens at Merrill Lynch. Throughout the year we look at suggestions that investors and users make. Around July we publish the changes we are considering and then ask investors. There's a public commentary period during which time anybody is free to participate. There's a possibility that we might bin a change or alter it slightly to make it more appropriate.
NM: So both of you brought Fitch on board at varying times, in response to these discussions you had with users and investors. What was the feedback during your discussion period?
RO: We had deliberations on both sides. Some felt that it would be good to add another rating agency into the equation; some felt that it was unnecessary. They voted and the majority went for the inclusion of Fitch, using the middle rating of three rating agencies as the new methodology.
CH: That's similar to the process we went through. We threw it out to investors a year ago and they reacted really positively. What they didn't agree with however was that we initially had suggested that we should use the lowest rating of the three agencies as the ratings methodology. Investors told us they preferred the idea of using the average, which is what we ended up doing.
NM: As investors, how did you take the inclusion of Fitch into the indices?
Peter Bentley, credit fund manager at Pimco (PB): I definitely viewed it as a positive. The frustration with some of the indices is that if you only have two rating agencies and you take the lower then you get a lot of bonds moving in and out of the index. It's much more consistent and preferable from our point of view to have as many sensible rating agencies used as possible and take the average. It's a good move for you guys compared with other index providers who don't do that, and take the lower of the three.
Ian Robinson, head of credit strategy at F&C Asset Management (IR): The biggest issue for us is when an index is created in such a way that we can't invest in it and clients don't allow us to follow that index. The addition of Fitch also raises other issues: it helps us go to clients and say, 'this is what's happening, this is how the market's moving, and we therefore suggest you change the guidelines to match'. One of the interesting things that Peter mentioned was the question mark over an outlying rating from one of the agencies. That's particularly important when some of the ratings have not been sought by the company but have been given by the rating agency. Then, there is more potential for difference and you don't want that to be the swing factor in whether or not a bond is included in an index or not.
NM: Richard, you're representing Fitch here. How do you respond?
Richard Hunter, head of Fitch Ratings' credit policy team for Europe, the Middle East and Asia (RH): It's good to hear all these nice words from people about Fitch's inclusion. We approached both Merrill and Lehman because our strategy is to make our ratings more valuable to people. It's good news for the profile of Fitch and good news for the investors in terms of their flexibility. The difference between the 'average of' and the 'lower of' methodologies is an interesting point; even the difference between the averaging algorithms used is interesting. How do they work? I guess that the algorithm doesn't just take each notch as being equivalent; it must reflect the different default likelihoods between each notch. I guess it must be a non-linear scale if you're going to average them, because if you were going to average the ratings on the basis of default you would want to distinguish as you go down the scale.
RO: We don't take into consideration the default probabilities, we use a linear application. If you were to take the example of, let's say, GM: Fitch gives it a rating of BB+, Standard & Poor's has BB, and Moody's has Baa2. We take a middle rating of that and that gives us a rating of Ba3, which pushes us into the high-yield index.
IR: Another interesting point is what happened to Railtrack and British Energy. There, there was a big disparity and it made an enormous difference to returns over the course of a year. In a 'lower of' methodology, if somebody goes off the chart, you can have ratings that are as much as 10 notches apart. There, an averaging system is going to pull the rating even further down, whereas a middle rating would keep it further up. There's a lot of detail that goes behind this, which I think can have a significant impact. The issue of average or middle does, at the margin, make a difference. Plenty of investors got stung on British Energy: it was an index rule that was logical in itself, but it had different repercussions because there were only two agencies involved in the index at the time.
PB: These are relevant issues that you've raised, Richard. I said before it doesn't generally matter if you get the guidelines right, but there are some extreme cases where it does impact the return on the benchmark. In those cases, nothing's ideal. From our point of view, the best method is one that's straightforward and easy to understand. Then you can position yourself and do your trades versus that benchmark with a degree of knowledge and clarity rather than trying to second-guess a complicated model of inclusion or non-inclusion. Given that, the linear model of ratings methodology is preferable because you can at least work out whether something's going to go in or out. Secondly, I think at the margin an average rather than a median is probably preferable on the basis that it incorporates the wide range of views that the ratings industries would have, instead of just defaulting to the middle of the three.
NM: What is the target from an investor's point of view? Do you want the ratings methodology to be as accurate as possible or is it more important to reduce ratings volatility as much as possible?
IR: What's most important to us is to make sure it replicates what we can do. You can't overemphasise that. A client is unlikely to let us invest in something that is triple-B from two rating agencies and single-C from the other because clearly there's some element of doubt with what's going on. We also manage a lot of insurance money and regulations don't tend to include Fitch or, if they do, they include it sort of on the side - 'if Moody's and S&P are not involved by all means have a quick look at Fitch'. It's a bit unfortunate from your perspective Richard, but from our clients' perspective that's going to drive the guidelines rather than the index.
RH: Whenever we address one issue - index recognition for example - something else comes up, like consultant guidelines. So we say, 'okay, let's chase down consultants next', then another issue comes up. It almost feels like the banging-the-mole-under-the-rug game. But we're making progress. The important thing is that we are seen by all the parties who do recognise us to be doing a decent job and on that basis it makes sense to have a broader array of informed opinion rather than a narrow range of opinion or a potentially unlimited array of less important opinion. A question for you Ian: an earlier point you made was about unsolicited ratings. Implicitly what you seemed to be saying is you would place less faith in unsolicited ratings or figures.
IR: The issue with unsolicited ratings is that it's difficult to know what the incentive is for the rating agency to ensure their rating is correct. And there might not be such timely changes, which may affect indices and when you can and can't invest. It may impede a decision that you would otherwise make and that can be problematic.
RH: There's a difference between solicitation and participation, whether the chaps get involved and have a regular discussion or not. They could participate like mad today, but something goes horribly wrong next month and they don't answer the phone; they'll just clam up because they've lost $500 million, have had to sack half the treasury team, and the other half are scanning the situations vacant column. Fairly low down the list is phoning back the rating agencies to say, 'by the way, just to keep you in touch ... '.
NM: Back on the subject of the rule changes: do you feel that you have sufficient input into how these rule changes are implemented?
PB: We do now. There's the index council at Lehman and I've seen flurries of emails about various index changes from Merrill. The only issue from our point of view is the timing: is it something that's just being talked about or is it imminent? As Ravin was saying, you'll go to an index council and something will be agreed in principle and then it'll disappear from our view and may come out at the other end as an index change.
NM: Did the timing of Fitch's inclusion, given what happened subsequently to GM and Ford's ratings, cause any problems for you?
PB: That's slightly different in that it was more than just the timing of the rule change. The change was announced and after that, as we all know, the rating changed on the auto companies, so they were set to go off the investment-grade index and then back in. Well, in the end that didn't happen. But it was more an unfortunate timing on the part of the rating agencies changing their view and the indices changing a few months later. There's not a lot you can do about that.
NM: Moving on to customised indices, how are they developed, and why are they growing in profile now?
CH: Customised indices are normally developed after we discuss with an investor or portfolio manager what they are trying to accomplish. Often a standard index can be used to meet those objectives. There are other occasions when investors want certain parts of the market to be excluded: it might be a specific rating, or clients might have a view on a sector they might want to leave out.
RO: We do very much the same thing. The global aggregate index, for example, is a rapidly growing index for us. But it is a multi-currency, multi-asset class index so it is likely that a European-based investor for instance will not have the expertise in, say, the US mortgage market, which is significant. So the client may come to us and ask for the global aggregate index excluding US mortgage-backed securities or other permutations. Maybe excluding certain currencies, certain asset classes or certain ratings.
IR: The problem with customisation is that it covers so many different aspects. When the credit market began in the UK and Europe it was quite small. By definition you needed an index that contains everything. Gradually, over time it has got big enough to divide it up in certain ways that suit you better, such as maturity.
NM: Do you use custom-built indices, Peter?
PB: Yes, to an extent. The client and the consultant will have an idea of what liabilities they're trying to beat and they want something that allows them to judge our ability to add value against that benchmark. There are a number of areas like those Ravin has highlighted where it is pretty difficult without a bespoke index to gauge how much value the investor is adding, so those kinds of developments are definitely welcome.
NM: Are there plans to increase your suite of custom-built indices?
CH: It's on a case-by-case basis. We build them according to individuals' requests. One thing that we've recently expanded a lot are custom-built indices for insurance companies as their parameters are slightly different. They may want to 'buy and hold' more than 'move in and out' as much as traditional asset managers. It's an area that we're exploring at the moment and we are just starting to bring it into Europe.
RO: We may talk to a certain investor base and feel a trend coming up and use that as a platform to talk to other investors. The need to match assets and liabilities has generated a lot of interest in the use of swap indices. We have clients who want us to combine an index that includes not just swaps and the credit element but cash as well, just to better capture the liability profile.
NM: How quickly can you turn them around?
RO: If we don't have a lot of requests for customised indices we can turn it around in a day. If we have a lot of clients making requests we do it in turn. Sometimes we don't have the technology in place to make a particular benchmark available in a standard platform.
CH: The requests from the client side are becoming increasingly sophisticated, probably for the reason that Ian mentioned, which is that the market's evolved to an extent that you can quite comfortably leave out big chunks and not have a huge impact in terms of your investment universe.
NM: Ravin, tell us what Lehman does in terms of constrained indices.
RO: There is renewed interest in the use of constrained indices now, as a result of the autos downgrades. We make them available for the high-yield market with indices in the range of 2% to 5% issuer cap but on the back of that we customise further for clients. Clients have different requirements: we have one client now who wants a 2.5% issuer constrained index, for instance - the only thing we didn't make available in the standard platform!
NM: So the 2% cap means that a certain issuer can't take up more than 2% of the index?
RO: Exactly. Take the example of GM for instance. With its entry, the European high-yield index would be close to something like 20% GM. Obviously, there's a lot of idiosyncratic risk in this and clients have come to us wanting to cap that impact.
PB: It's definitely a good thing in high yield. First of all, if you're trying to get a well-diversified universe, and you've got one or more issuers massively dominating the return to that universe, then that's not great. It's also a good thing from the perspective of just trying to manage the money, because if you have something that's a huge chunk of the universe, trying to buy that in huge size is not an easy exercise. GM is relatively liquid, but you can get cases where everyone's trying to chase a small amount of bonds because they're a large chunk of the index, and you get a situation where prices are artificially bid up just because they're a large chunk of the index. Again, this is difficult for the investor, but more importantly you end up doing what you shouldn't be doing for the underlying clients because of an index composition.
Thank you to all our participants.