Asset managers need to update their fixed income benchmarks each day to be able to judge how their portfolios are behaving in the market. They can lose serious money if the benchmark changes its exposure and their benchmark holdings are not up to date.
Benchmark data comes from files supplied by the banks. It is made up of name codes for securities, face value holdings, coupon payments and price. It can also include risk figures - duration, modified duration and convexity - as well as coupon, yield and margin to swap.
The problem is that each bank, each index provider, supplies this data to asset managers in a different format.
Andrew Colin, fixed income research director for StatPro Group, says this is creating a serious business risk. “In its current form, the way benchmark data is supplied is becoming a substantial business risk to the majority of fixed income institutions, not only for post hoc performance and attribution but also for live risk management.
“Without accurate, timely benchmark and, most importantly, detailed data, you’re effectively flying blind, with only the vaguest idea of where your portfolios have been and where they are going, relative to the rest of the market place.”
Colin points out that most modern applications rely on standardised software and data formats, and that it should be possible for fixed income managers to simply download the data from the internet.
“Curiously, for benchmark data, standardisation seems to have passed the banking world by,” he says.
The problem of non-standardised benchmark data is exacerbated by the emergence of more sophisticated risk management and attribution systems which specifically require security-level benchmark data – in particular returns.
“Benchmark data is usually adequate for a snapshot of market risk. The supplied exposures, prices and risks come straight out of the index provider’s computer and can be taken as definitive,” says Colin.
“The tricky stuff starts when you try using the data to get returns. Because this is where the traditional approach of supplying benchmark holdings files falls down. They don’t include security level returns. Nor is there any easy way to get them.”
StatPro carried out a project to ‘reverse engineer’ the fixed income benchmarks its customers use, using exactly the same data that the index providers supply. “We wanted to be sure we could replicate the published returns of each benchmark from the supplied security level holdings files,” says Colin.
StatPro achieved this, but revealed some problems in the process, he says. “The first problem was that the published description of how a benchmark works was often quite different to the way that the numbers are actually combined together to produce the published returns.”
Colin says it became clear that the index providers did not fully understand how their benchmarks worked: “That’s not an encouraging situation for anyone who wants to get from benchmark holdings files to returns data.”
Index providers are supplying firms with data that will be inadequate over the coming years, he says, and this could cost them money. For example, an asset manager mighty discover that a benchmark has changed its exposures and has shortened by a quarter of a year, while the yield curve has moved substantially.
“What is the cost if that manager hasn’t structured his or her flagship portfolio in time, simply because the benchmark holdings weren’t up to date?”
Colin argues that benchmark data has not kept pace with technology. “Technology has moved on but the benchmarks haven’t. We’re now running detailed attribution analyses and zeroing in on individual stocks rather than market sectors. We’re using multiple benchmarks over numerous currencies and asset types, and holdings are updated daily, rather than weekly or monthly.”
Much more detailed information is needed about the benchmark for the level of results demanded by the marketplace. In many cases this information cannot be supplied in its present form, Colin says.
“The old approach is about to be rendered horribly obsolete by the advent of powerful fixed-income attribution systems, for which it is vital to be able to complete the circle by moving from holdings to returns in a transparent manner. As we found, this just isn’t possible without a huge amount of unnecessary effort.”
The effort is unnecessary because index providers already have this information, he says. “Security-level verified weights and returns data are inside the providers’ software.
Yet the information is inaccessible, chiefly because making it available to customers would involve lengthy and expensive modifications to the application.
The main reason why it is inaccessible, however, is that nobody asked for it before. Ten years ago, few fixed income managers needed daily security returns.
The result, says Colin, is that the customer has to reproduce and validate the same numbers in an unnecessary expense of money, time and effort. “How much easier life would be if only the numbers about weights and return were available in the same way as prices and coupons, in a standard machine-readable format.”
So what should the asset management industry be demanding? Colin has three suggestions for index suppliers:
q Include security-level returns with raw benchmark data.
q Fully document the algorithms that produce the benchmark return. In other words, show how to go from security-level returns to the published benchmark return.
q Provide data in a standard, machine-readable format.
“If you are an index provider who puts those three ideas into practice, your customers will be delighted,” Colin concludes. “Index providers that cannot demonstrate a transparent route from holdings to returns are going to experience growing reluctance form customers to use their products.”