It is nearly four decades since Bill Sharpe’s Capital Asset Pricing Model (CAPM). We think it is time to take his work seriously and to change many of our cherished beliefs on portfolio structures. There are two striking conclusions. First, we should exclude all references to asset categories such as emerging/ developed markets, domestic/ international, country, sector, industry, and capitalisation. Second, indexing strategies have moved away from the concepts behind CAPM. We need to get back to basics and stop passively managing to the active stock selection decisions of index committees.
These recommendations rest on the beliefs behind CAPM that investors are rational, risk-averse, and do not make persistent mistakes in maximising utility. This belief in investor rationality leads us directly to the well-known and compelling CAPM conclusion that holding the market portfolio will maximise your expected (risk-adjusted) return.
Implicit is the belief that prices appropriately reflect rational investor expectations about the return and risk of assets. Benchmark approaches that employ asset categories such as, domestic/ international, are essentially proposing an asset pricing model that suggests market prices are not in equilibrium.
Further, it is time to stress that in implementation investors should recognise the difference between passive investing and indexing. 1 passive investors aim to capture the returns associated with a broad market portfolio rather than blindly follow an index provider’s particular approximation of the market portfolio. From 1990–99 there were 195 additions to the S&P500 index, the benchmark with more money tracking it than any other and where tracking errors from indexed funds are measured by just a couple of basis points. To achieve that kind of tracking, managers must trade close to the time when stocks enter or leave the index. From pre-announcement to effective date, the average rise in price relative to the index was +7.5% and the “permanent” price effect on joining the index is about +3.5%. The cost to funds in terms of geometric mean return to minimise tracking error is considerable. This has nothing whatsoever to do with building wealth or with CAPM.
Those who do challenge the CAPM conclusion that the market portfolio is the unique efficient portfolio often assert that the underlying assumptions are unrealistic. Our view is that while the CAPM is an abstraction of reality, its assumptions are reasonable, and its conclusions are profound and difficult to challenge.
If we consider more pragmatic issues regarding the ideal strategic portfolio, happily we find that the CAPM-inspired market portfolio approach also has distinct cost advantages. A CAPM-based strategy meets the requirements of:
q Performance: it is a strategy designed to earn ‘top’ long-term returns; and
q Stability: it is a strategy that can be sustained well into the future.
Although we argue that the industry needs to move away from slavishly following every index change, we do acknowledge the need for some reference benchmark.
We have so far made no reference to commonly used asset categories. This omission is intentional and represents the implicit CAPM belief that over time rational, profit-motivated investors’ price assets, and asset categories, to appropriately reflect the return and risk expectations. Asset categorisation almost invariably involves arbitrary distinctions. This contrasts sharply with a capitalisation-weighted approach, which maintains the desired weight of each asset category by automatically incorporating investor expectations, through price. Alternatives to our capitalisation-weighted approach are often based on the belief that assets are mispriced by the markets. This belief typically manifests itself as a concern that prices, and therefore capitalisation, for a particular asset category are too high.
Benchmark responses to such pricing concerns are implemented in a number of ways: from ad hoc weight limits aimed at reducing exposure to specific assets, to more complex GDP-weighted approaches to country allocation. However, we find that most of these efforts are, in effect, asset pricing models that find fault with the prices investors have set in markets.
We can understand the desire to avoid establishing too large a position in expensive assets. Investors, however, should be aware of the substantial pitfalls associated with abandoning an approach based on a well-reasoned general model of equilibrium asset pricing, in favour of an asset pricing model that presumes the market is in disequilibrium but will eventually agree with the posited model.
Asset ‘mispricing’ is easily identified after the event. With the benefit of hindsight it is straightforward to suggest an asset pricing model that avoids or mitigates an unpleasant return experienced by an asset category. However, it is anything but straightforward to recognise:
q when, on a forward-looking basis, the collective forecast of market participants is incorrect;
q when prices will reflect the forecast.
Also, costs associated with non-capitalisation-weighted approaches can be significant. These costs represent a real hurdle to the active approach that is implied in a non-capitalisation-weighted strategy.
Of course, adopting a global cap-weighted benchmark with no domestic bias would lead, for most countries, to a very low weight to domestic equities and a very high exposure to foreign currencies. Our view on currency is that changes in spot rates represent uncompensated portfolio volatility. Without the benefit of a positive expected return, unhedged currency exposure increases the risk that liabilities will not be met. These beliefs lead us to the conclusion that currency exposure is best hedged.
We are sensitive to the fact that our conclusions rest on both a simplifying assumption regarding the currency exposure of the consumption basket, and the implicit assumption that the risk-reduction associated with hedging dominates any potential diversification benefit of maintaining unhedged currency exposures.
Capturing the return of the market portfolio of investable wealth-generating assets is best viewed as a continuous process, where income and inflows provide the means to respond patiently to the subtle changes in capitalisation that arise primarily from firms’ financing decisions. One of the under-appreciated benefits of this approach to investing is the extent to which it is self-rebalancing, and therefore minimises transaction costs. Cross-border mergers and reclassification of countries as emerging or developed are good examples of activity that can generate rebalancing unrelated to the thesis underlying the benchmark.
The Vodafone/Mannesmann acquisition highlighted one of the absurdities that can result. As Vodafone acquired Mannesmann, Mannesmann’s capitalisation ‘migrated’ from Germany to the UK. As UK institutional funds have a large domestic bias, managers ‘had’ to increase their weight to Vodafone. On the other hand, if Mannesmann had acquired Vodafone, the reverse would have been true, and yet the business of Vodafone/Mannesmann and all of its cash flows would have been exactly the same irrespective of whether it was listed in Frankfurt or London.
1 passive investors aim to capture the returns associated with a broad market portfolio rather than blindly following an index provider’s views. To that end, we suggest that investors avoid making continuous transactions in an attempt to track as closely as possible a specific benchmark. Instead, we suggest that investors resist selling securities, and simply add to under-weighted positions as cash flows and liquidity provide the opportunity.
Investors should take all reasonable steps to acquire and maintain desired exposures at minimum cost (eg, using crossing networks). In essence, we are saying that an obsessive interest in minimising tracking error against what is inevitably an arbitrary view of the market portfolio, will incur costs that harm the much more important long-run geometric return. Today’s indexing approach does lead itself to one, very simple, performance metric, tracking error. What we propose will involve more complicated metrics, and an increase in the oversight or governance ‘budget’. As well as tracking to a reference benchmark, managers and their clients will need to review turnover, transaction costs and measures to demonstrate the breadth of holdings the manager has acquired in the market portfolio.
In one sense, we are saying nothing new and are merely reiterating CAPM’s basic message. However we believe there are many issues in evidence today that make it important to reconsider the CAPM message in the current context. These issues include:
q the tendency to home bias;
q the disruptive and costly effects of index rebalancing;
q arbitrary definitions (multinational, developed, developed-emerging, emerging and pre-emerging); and
q the decreasing importance of countries and regions in explaining stock returns.
We would do better to follow the CAPM approach through ‘passive’ management of a global portfolio, rather than adopt an ‘index’ approach.
Alan J Brown is group CIO and chairman of State Street Global Advisors UK. A longer version of this appeared in the Journal of Asset Management Vol II