The federal bank regulator, the Office of the Superintendent of Financial Institutions (OSFI), wants bank-contingent capital in the bond index, and has a plan to achieve this objective.
Taking this step would be abusive to bond investors of all kinds, particularly retail investors who have — for the most part quite rightly — become enamoured of indexing their bond portfolios. But to understand the abuse, we must first understand the difference between a good and a bad index, and why including bank-contingent capital bonds, or “CoCos” in the bond index will turn it into a bad index.
Fixed Income Isn’t Working Like It Used To
To generate yield and manage risk for clients in today’s complex fixed-income
markets, take an active approach with active fixed income ETFs from
Franklin Templeton Investments.
The performance standard is paramount. Investment managers work in an industry in which their ability to do their jobs can to a large extent be measured objectively — the performance of a stock or bond portfolio can be measured precisely and compared with benchmarks.
The third element is, however, achieving increased emphasis, particularly among retail investors and their advisors — professional and otherwise. Something of a cult has formed around passive investment, which holds that all costs associated with investment are almost certainly wasted.
This new paradigm has had a significant effect: the Investment Company Institute reports that the market share of equity index funds (as a proportion of all equity funds) has increased from 4.0% in 1995 to 13.7% in 2009. In Canada, discussions of the “Bond Indexing Boom” go back as far as the late 1990s. It is difficult, however, to estimate the amount of bond assets currently indexed.
One complication, as President of Horizon ETFs Howard J. Atkinson has explained, is that indexation via over-the-counter derivatives, such as total return swaps, is much cheaper than straightforward indexing for large institutional funds.
While I am not aware of any Canadian bond ETF using this strategy, the HBP S&P/TSX 60 Index ETF (HXT) is able to offer — due to its use of swaps rather than direct holdings — a 0.07% management fee, and bill itself as “the lowest cost ETF in Canada [at] less than half the cost of its nearest competitor.”
Most Canadian bond index products are based on the DEX Universe Index or on clearly identifiable sectors thereof (such as “All Corporates” or “Mid-Term’). While explicit indexing is reason enough to consider the DEX Universe important, it is also quite clear that most active management strategies are benchmarked against this index (or its segments).
Decisions regarding the constituents of this index are therefore important not just to investors seeking a benchmark or passive strategy; it can also be argued that the index composition is important to the economy of Canada.
Good indices and correlation
The three purposes of an index can be achieved only through careful construction. Elements of an index should display a high degree of correlation within their segment and a lesser degree of correlation with other segments; the limiting factor in the degree of correlation is the desired broadness of the index.
The importance of correlation is emphasized by the second purpose of indices — providing a proxy for asset allocation purposes. A good index will have a predictable response to certain economic events. For example, long bonds will do poorly as inflation expectations rise (this old truism is one reason, perhaps, why Real Return Bonds are excluded from the DEX Universe Bond Index). Moreover, estimates of future sensitivity can be made by comparison to historical data, but only if the index chosen as a proxy has a reasonable level of homogeneity.
Thus, catastrophe bonds, for instance, are not included in the major indices. These bonds pay high interest, but are subject to capital losses after specified events, such as earthquakes and hurricanes. Catastophe bonds are better regarded as insurance contracts than actual bonds. And while they’ve enjoyed very good returns recently, this has less to do with interest rates and credit spreads than with a relatively benign hurricane season in the Gulf of Mexico, which is not usually a consideration when constructing a bond portfolio.
Similarly, bond indices in common use as broad market measures typically exclude junk bonds, since the correlation between this asset class and investment-grade bonds is not very high. Reilly, Wright and Gentry explain in a study published in the Journal of Applied Corporate Finance that the junk/investment-grade bond correlation is actually less than the correlation between investment-grade corporates and treasury bonds. Indeed, Reilly et al. show the correlation between junk and investment-grade bonds is so weak that “there appears to be no statistically detectable pattern.”
Modern portfolio theory holds that once a definable segment of the financial marketplace has a sufficiently low correlation with other segments, it can profitably be considered for inclusion in portfolios as a completely separate allocation. But a portfolio is not an index. Combining two poorly correlated financial instruments or segments — which may have wildly different responses to external events — may be good portfolio management, but it is poor index design.
A bad index
The DEX HYBrid Bond Index serves as an excellent example of an index that disregards the three principles Russell has outlined. This index combines the investment-grade portion of the DEX Corporate BBB Index and the non-investment-grade-rated DEX High Yield Index.
The first problem with this index is it combines two sectors with a low correlation, while not including sectors with higher correlation. Another problem is investment-grade bonds are included in the index according to a varying percentage of their market capitalization (30% as of August 2010). This has two effects. First, given similar market capitalization of issues, the DEX HYBrid Bond Index will have greater exposure to individual names of lower credit quality — a reversal of usual practice. For example, one popular fund caps single-issuer exposure at 10% for investment-grade bonds and at 5% for junk. Second, the relative weighting of the two groups will vary within the index over the long term in a manner that will not necessarily reflect their relative weight in the marketplace. This will also have the effect of changing the index’s response to economic stimuli.
Finally, the index’s name is a misnomer. The term “hybrid” is used in the bond market to denote a particular kind of investment, generally issued by banks and other regulated financial institutions, that has characteristics reflecting both regular debt and equity. The unwary might assume from the name of this index that such hybrid bonds are the constituents of this index, which is not the case.
Try as I might, I am unable to discern a purpose for this index based on benchmarking or asset allocation. It appears to have been initiated solely to serve as a platform for an ostensibly passive investment vehicle. In a piece that appeared in Financial Analysts Journal, John Bogle charged, “The [mutual fund] industry is a vast and highly successful marketing business, an industry focused primarily on salesmanship.” Sadly, DEX HYBrid Bond Index is an example of how this salesmanship is extending into index creation.
The TMX becomes defensive when its indices are criticized. My earlier criticism of DEX HYBrid Bond Index resulted in my receipt of a series of vituperative emails from a TMX official threatening a complaint to my professional association, and claiming my views were motivated by a grudge regarding an unsuccessful sales presentation in 2006.
OSFI and hybrid bonds
As I’ve discussed in previous articles, the panic of 2007 has motivated governments and regulators to seek ways of broadening burden sharing. Governmental purchase of equity in various hard-hit banks allowed the banks to avoid bankruptcy, thereby protecting holders of bank non-equity capital securities (such as preferred shares), Innovative Tier 1 Capital (“hybrid bonds”) and subordinated debt.
Thus, the OSFI has proposed that all future issuance of these instruments must contain a “Non Viability Contingent Capital” clause, which will require the full and permanent conversion of these instruments into common shares of the issuer should the Superintendent decide that the issuer has ceased, or will soon cease, to be viable. There is no necessity for the Superintendent to justify her views in any way, nor is there any path of appeal.
It is breathtaking to consider the amount of power granted thereby to a single bureaucrat. It shows the lengths to which governments are prepared to go to gain for themselves the powers ordinarily reserved for bankruptcy courts operating in the clear light of day. But that is more of a political question.
There is much to criticize in the plan. By making the trigger event non-viability, OSFI is preparing to deal with a crisis after the fact. If the conversion trigger occurs earlier in the decline of a bank’s fortunes, there is a greater chance that such a crisis would be averted. This is the path taken by the Swiss, who have good reason to avert crises rather than punish creditors afterwards.
Additionally, the draft proposal requires issuers to “provide a trust arrangement or other mechanism to hold shares issued upon the conversion for non-common capital providers that are not permitted to own common shares of the DTI due to legal prohibitions.” Some might consider this a clever way to circumvent the intent of legislators and non-legislated investment mandates. Others might choose a different adjective.
But the more immediate concern to fixed income investors is the question of the inclusion of these instruments in bond indices, which is all the more important in Canada since there is only one universal index in common use.
We were fortunate during the crisis not to have any of the major banks get into serious trouble, but this will not necessarily be the case during the next crisis. When this happens, speculation about OSFI’s intentions will add yet another layer of uncertainty to the analysis of hybrid bonds and result in their market values becoming uncorrelated with their peers. This will be in addition to fears of having creditor rights arbitrarily changed to suit bureaucratic convenience, as happened in the UK Bradford & Bingley nationalization.
In Canada, we’ve already seen the effect of arbitrary regulatory change on hybrid bond returns: this class of issue dropped by 7.5% in the six months commencing August 2010. Drops for issues with long periods until their first par call were even more dramatic: a TD Capital Trust issue dropped 15.6% in the complete absence of changes in the overall economy, bond market, or credit quality that could justify such a move. Virtually all the change was due solely to fears that OSFI would change the rules of the game by refusing to allow these issues to be included in Tier 1 Capital and thereby enable use of the “Regulatory Event” clause in the issue terms, allowing an immediate call at par.
This un-bond-like behaviour of bank regulatory capital should be of great concern to investors of all stripes, as these issues are currently included in the DEX Universe Bond Index and hence in most performance benchmarks and many ETFs. Indeed, these issues are often overweighted in popular ETFs, presumably because of their higher quoted yields.
With its refusal to grandfather extant regulatory issues when changing requirements for new issues, OSFI has cemented its reputation for operating with little regard for the capital markets. It should be clear that while bank hybrids and subordinated debt may well be good investments, they cannot and must not be regarded as bonds in the same category as senior debt. Their uncorrelated behaviour is likely to increase during a crisis, just when the safety of bonds — actual bonds, which at worst will default and give rise to a restructuring in a clearly defined process — is most desirable.
Investors and portfolio managers should therefore urge the TMX to remove these instruments from the main index — or, at the very least, incorporate them only as a distinct and easily removable group, as is currently the case with Maple bonds. Should this change not occur, the development and use of bondindices from other bond-indexing services ought to be encouraged. Investors in the UK have been successful in keeping some contingent capital out of major indices, despite similar pressure from UK regulators. This success should be replicated and reinforced in Canada.
James Hymas, CFA, BSc is president of Hymas Investment Management Inc.