A collection of Futuregrowth thought leadership pieces, media articles and interviews.

Why Bond Indices are flawed

30 Mar 2015



Bond indices, globally and perpetually, are nearly always deeply and fundamentally flawed. Both equity and bond indices tend to be market-capitalisation weighted: the larger the issuer’s size the more they are included in the index.

Equity indices have the pretence of being representative of entire national economies, since companies eventually reach a size where they become public companies.  However, bond indices are inherently narrow as they only include issuers that borrow money, which is a small subset of all potential borrowers in an economy. The big problem is that innately a bond index will have a higher weighting to borrowers with more debt. This over-weighting of over-geared entities serves to make any credit-indices inherently dangerous: If you track the index then you end up with dominant holdings of less credit worthy companies.  

This problem is compounded because bond indices will narrowly represent only bonds issued in the public capital market, and will ignore bank and other forms of borrowings.

Further, there is an element of anti-selection in listed bonds because the instruments sold to investors tend to have weaker terms (covenants, security, etc.) than bank debt and, in the worst case, may actually be instruments banks themselves wouldn’t want to hold.  This is borne out as, according to a range of sources, bank loans tend to have a higher recovery rate in default than listed bonds sold to institutional investors.

Can funds use bond/credit indices? What are the challenges?
Pension funds, the dominant holders of bonds, tend to hold bonds as a match for long-term liabilities since bonds have less volatility than other asset classes such as equities, and also because bonds have volatility that more closely matches their pension liabilities. Thus, the decision to hold bonds is principally an interest rate risk decision (e.g. hold fixed-rate bonds to match projected fixed-rate liabilities).

The interest rate risk in bond indices (weighted modified duration), however, may not (and often don’t) bear any relation to pension funds’ actual interest rate risk target because, again, the indices are constructed by a limited, stilted universe of borrowers who happen to have issued listed bonds. Thus, pension funds have a “benchmark selection” challenge to deal with when choosing amongst funds.

In addition, the pension funds’ risk decision has become corrupted, as over the past 20 years the world-of-bonds has stupidly conflated two unrelated risks into the general category “bonds”: First, is the interest rate risk (term risk, price-volatility-for-changes-to-rates, yield curve shifts), and; Second, is the credit risk (the chance of not being repaid).  These risks are actually two separate asset classes:  Pension funds sought interest rate risk, but inadvertently ended up with embedded credit risk.  More recently investors have woken up to the differing risks and skill-sets required to manage interest rate risk versus credit risk.  Some investors have now wilfully chosen to distinctly seek returns from credit risk, apart from interest rate risk. But it generally remains true that risk measures, limits and budgets need to be developed that separate and manage the two issues separately. 

Futuregrowth is actively opposed to the formulation of official "credit indices" for the reasons above and because such indices will only serve to lead investors into an abyss of benchmark-cognizant credit investing – holding the instruments which are in the index simply because they are in the index. This is a fool's errand: Passive credit investing is roughly the same as a “passive” stroll in the Kruger Park – you will get eaten eventually.

"You might be interested in listening to an interview with Andrew Canter and Alec Hogg on BizNews: Warning to passive investors – bond indices by nature seriously flawed

How can you know if your credit manager is doing a good job?
Appreciating that investors and consultants want to know if asset managers who invest in credit are out-performing a “fair target” (something which earns more than risk-free bonds for the credit-risk taken), and lacking standardised risk-adjusted-return tools, we think there may be better answers than credit indices. For instance, one could devise agreed measures of likely portfolio volatility (arising from interest rate, credit, and other positions) to arrive at an expected outperformance over a relevant risk-free rate.  Alternatively a credit-spread-index could be constructed which is not an investable index per se, but is more of an indication of the credit spreads available in the market. These ideas needs more work to be practical, but could be overlain on fixed rate, floating rate, and inflation linked fund targets so that investors can choose their relevant interest rate risk target.

What about the JSE All Bond Index?
Given all that vitriole, you will be surprised to hear that we actually like the All Bond Index (ALBI), with its selection of 20 of the largest capitalisation and most liquid bonds in South Africa, for a number of reasons.

First, the ALBI comprises key elements of successful indices: It is independent, observable, replicable, liquid, transparently priced, and hard to sustainably manipulate. These are elements that all credit indices lack.

Second, the ALBI is presently credit-risk-free, since it is composed of RSA issued and/or guaranteed debt -- taking the assumption that the RSA is a "risk free" borrower in SA Rands. (The ALBI is 100% government guaranteed as at September 2015, and 92.38% of ALBI constitutes RSA debt [14 of 20] instruments. The remaining 7.62% of the index is comprised of explicitly government guaranteed Eskom and SANRAL debt: HWAY20, ES18, ES23, ES26, ES33, ES42). Thus, it is principally an interest-rate-risk measurement tool, and does not suffer from the fatal flaws of credit indices.

If it were up to Futuregrowth, the ALBI would include only RSA issued debt so as to be a pure interest rate/yield curve benchmark, and separate any “credit spread” from it entirely.  (There is a sub-set of ALBI called the GOVI which does just that but includes only 10 RSA bonds, and our view is that the ALBI should be replaced by a GOVI style index with more RSA guaranteed constituents).

Third, the ALBI is a relatively long-duration benchmark, which is suitable for retirement funds to use as a proxy for long-dated liabilities. Also, the ALBI’s interest rate risk is relatively stable because the government’s funding program continually issues longer dated bonds.  Thus, in general it is a useful and relevant proxy for funds' interest rate risk:return appetites.

"You might be interested in listening to an interview with Andrew Canter and Michael Avery on Classic FM: Could prescribed assets save us?