In the world of equities, much of the complexity of risk assessment and management has been simplified through the development of capitalisation-weighted indices and the spectacular growth of indexed funds on the back of this. At first sight, it appears a straightforward extension to apply the same approach to the management of fixed interest assets. Indeed, there has been a proliferation of bond indices produced by newspapers, investment banks and exchanges, competing for attention. Yet a straightforward transfer of the approaches taken in constructing equity indices to the bond markets can be deeply flawed. There are many differences in both the philosophical underpinnings of bond indices as well as their practical implementation that make them very different animals to deal with. 

An ideal bond index series would have the following characteristics: i) Use prices obtained from as wide a marketplace as possible; ii) Have a broad coverage to be useful for performance attribution; iii)    Probably have a component that is narrower focussing on more liquid bonds for use in derivative products; iv) Have the ability to be combined with analytical capabilities to be able to construct zero, par and forward yield curves; and tailor-made benchmarks v) Be able to combine sub-indices for government bonds, corporate bonds, emerging market debt and asset backed securities and other structured products in a consistent manner. Constructing bond indices that satisfy all of the people, all of the time, is clearly impossible, but satisfying most of the people has also proved to be more difficult than for the equity markets. There are good reasons why this is so. Key issues include; the nature of the passive investment decision inherent in the construction of an index; the sourcing of accurate pricing; the trade-off between coverage and liquidity in the index; the stability of the index; and the extent to which the data available can be used to tailor individual benchmarks for performance and attribution.

Indexing and portfolio construction

Any bond investor is essentially trying to avoid overvalued bonds where the risk of default far outweighs the extra yield. A well structured bond portfolio should be diversified across a range of credits with no undue concentration in sectors or issuers. A fundamental difference between equities and debt indices is that the amount of an issuer’s debt represented in a bond index is not very sensitive to the market’s pricing of its debt. The credit spread forms only a small part of a bond’s yield except for high yield or distressed debt portfolios. As a result, we have the perverse result that the weaker an entity becomes financially through the issuance of more debt, the more a capitalisation-weighted index will weight that entity, even though its attractiveness is reduced. In contrast, unsuccessful companies with decreasing equity prices will see their weightings automatically decrease in a capitalisation weighted equity index. Indeed, managers who stuck closely to the JP Morgan emerging market debt index weightings during the Argentinian crisis would have found themselves increasing weightings to a train wreck in slow motion. JP Morgan subsequently issued indices that have 10%  caps to the weighting any single issuer can have within the index. Such limits to specific exposures are a pragmatic solution, but suffer from the lack of any theoretical framework behind them.

Sourcing prices

Accurate pricing is a major issue in the bond markets and the quality and sourcing of prices is a major differentiator between index providers. Unlike equities, there is no single exchange price that can be utilised as the current or closing price for a bond. This is exacerbated by the sheer number of different issues with the total number of fixed interest securities a large multiple of the total number of equities. The global credit crash led to a complete freezing of activity on the secondary markets for a while, and led to a dichotomy in pricing between a thriving primary market and an illiquid secondary bond market.

Given that price sourcing is one of the most critical issues when it comes to indices, a clear distinction can be drawn between those indices produced by investment banks using their own prices as the only source, and indices that obtain prices from a number of independent sources. In recent years, there has been a growing acceptance that independent sources of prices are preferable, with providers such as Markit, iBoxx and EuroMTS, producing indices based on best prices obtained from a number of market-makers across a variety of different fixed income asset classes.

Trade-off between coverage and investability

A large percentage of the total market is rarely traded and index providers are faced with the choice of coverage versus liquidity. Very liquid indices with a narrow coverage are attractive from an index-providers viewpoint since they can form the basis of highly lucrative derivative products. However, a bond fund manager needs to access the whole market and his performance needs to be measured against that objective and risks controlled relative to the total market. This is a difficult conundrum to solve. The Lehman bond indices (now the Barclays Capital Indices) which are very popular in the US have gone for wide coverage but as a result, are not easily investable so to what extent are they a measure of performance? The original Lehman Brothers Euro-Aggregate Index initially had nearly 7000 bonds. Even with reductions to the number of bonds, indices with thousands of securities are not going to be replicated fully, although they can be sampled, or used purely as a benchmark comparison.

Indexing the newer markets

Whilst government bonds and corporate bonds are at least straightforward conceptually, more difficulties arise when indices have been constructed for the two other major bond sectors, namely, emerging market debt (EMD) and securitised and structured debt products such as Asset Backed Securities (ABS).

Emerging market debt indices suffer from all the problems that developed market government bond indices have as well as the fact that EMD has been such a rapidly developing marketplace that index providers have struggled to keep up with the pace of developments. The original EMD indices were based on the dollar denominated debt that had been issued. But the last few years have seen the increasing dominance of local currency debt. Initially, EMD fund managers reacted to the increased prominence of local currency bonds either by including them as an out-of-the-benchmark allocation in traditional USD-denominated portfolios or by switching to blended benchmarks. But investors are generally better served by making explicit recognition of the local currency as a separate asset class and awarding mandates to managers on that basis as a result of their own analysis of their strategic asset allocation. The creation of local currency indices gave investors the explicit tools to be able to do this. Whilst both dollar denominated debt and local currency debt will act as diversifiers within a portfolio, their behaviour will be different and a random mixture would not be easily modelled within an ALM study.

Pragmatic solutions to the problems of constructing EMD indices are clearly useful, but they can often lead to very different weightings of individual countries that arise from the assumptions that are made in the index construction. One question for example, is how easy is it in practise to invest in a particular country, and should that be reflected in the construction of an index? Traditionally, this has been an in or out issue, with weightings to countries such as China and India with capital controls set at zero. More recently, the World Bank has sponsored a local currency index known as the Markit iBoxx GEMX. This index is part of their attempt to develop local bond markets in emerging markets. The indices attempt to embed the Markit strategy of objectivity, transparency and independence, through having a provider not tied to an investment bank, using prices from multiple providers, and with a transparent and structured approach to index construction. The index looks very different from its competitors, with weightings of over 6% for China and India. These weightings are produced by multiplying a countries base weight based on capitalisation (with a cap at 10%) by an “investability” score, which denotes how easy it is to invest. This score is produced independently by CRISIL, an Indian rating agency owned by Standard & Poors.

Where next?

Going forward, it is clear that bond indices are still in their infancy compared to the role that equity indices play in the marketplace. But it is also clear that they have a number of very different roles to play for investors and as a result, there will always be room for many different providers that offer very different approaches to the philosophy, the construction and the pricing behind their indices. Whilst satisfying all the people all the time will certainly never be reached, satisfying most of the people most of the time is still an objective worth pursuing for the index providers as a group.