Brent Sheather: Asset allocation confounds the experts

With short-term bank interest rates around 3 per cent, more and more investors are having to consider a wider range of asset classes to achieve their investment objectives. But what mix is prudent?

Good question - and one which has confounded even the experts in recent times.

A few months back this column looked at one of the defining features of this financial crisis - that many assets which were previously thought to have a low correlation with one another all went down together when the crash came.

Diversification into commodities, hedge funds or venture capital - none of it was a big help in finding a port in the storm. In fact, it almost seems that the more effort you put into determining asset allocation, the more money you lost.

Asset allocation naturally dominates the strategic thinking of the gurus at the NZ Super Fund. But despite liberal helpings of private equity, commodities, property, timber and infrastructure, the portfolio fell by 25.1 per cent in the April 2009 year. The April data is unaudited.

The NZ Super Fund portfolio was just 17 per cent invested in bonds and cash at March 31, 2009.

In contrast the typical pension fund has for the past 30 years or so averaged a 40 per cent weighting in bonds and cash and, because of that higher bond weighting, the average pension fund's portfolio is estimated to have performed much better than that of the Super Fund - down by around 12.1 per cent in the April year, before tax and fees.

A comparison of longer-term performance does not flatter the fund either. Since its inception in September 2003, to March 2009, the fund has returned 2.09 per cent after fees. In the same period a balanced portfolio would have returned 3.38 per cent a year, admittedly before fees but also with less risk.

In fact in the April 2008 year the Super Fund would have been better off just concentrating its assets in an unhedged global share index fund, which would have returned negative 16.3 per cent, as against the fund's negative 25.1 per cent.

As an aside, the NZ Super Fund's concentration on asset allocation seems to have led to a few rather strange positions in terms of actual stocks in the portfolio.

With $11.5 billion in assets, you would think that the Super Fund's biggest shareholding would be a stock from among the world's 100 largest like, maybe, Exxon, General Electric, BP, etc.

In fact the biggest holding in the portfolio, at $230 million, is a small Australian motorway company, Transurban Group, with a market capitalisation of $5.5 billion, representing around half of 1 per cent of the Australian sharemarket, which itself represents just 2 per cent or 3 per cent of the world stockmarket.

By conventional standards the Transurban holding is very much overweight and, if this is typical of the management of the portfolio as a whole, it is likely to make the NZ Super Fund's performance vary widely from that of the market.

Modern portfolio theory assumes that holding the market portfolio - that is, everything in the index according to its relative size - is the least risky option. On this basis, because the weighting in Transurban is so high, this portfolio is likely to be more risky than the market portfolio.

This is also known as portfolio tracking error. In response, a NZ Super Fund spokesman noted the fund's very specific purpose, long-term horizon and high tolerance for illiquidity, allowing it to construct a portfolio substantially different from the norm so as to capitalise on illiquidity premiums.

This is based on the idea that assets that are difficult to buy, sell and value generate higher returns than a more liquid security like General Electric shares.

The spokesman also highlighted the rebound in the fund's fortunes of 8.8 per cent from April 1 to May 4. In the same period a balanced portfolio would have returned 6.2 per cent.

Back to asset allocation theory: the failure of correlation to mitigate losses in the past 12 months has been a disaster for people all over the world, not just the NZ Super Fund.

Consequently the investment consulting industry is now in something of a crisis, particularly the part of it which sold the "alternatives" theory as originally pioneered by the likes of the Harvard Endowment.

That theory was one of the classic "have your cake and eat it too" variety and it went something like this: "Alternative investments like private equity, venture capital, property, hedge funds and commodities have a low historic correlation with shares so they offer the potential to diversify a portfolio, as bonds might, but will give a higher return than bonds".

The strategy thus promised an investment nirvana - high returns with low risk. In hindsight it seems that low interest rates caused trouble everywhere, pushing retail investors into finance company debentures and institutional investors into alternatives.

In the May edition of the US Financial Analysts Journal, editor Richard Ennis develops this theme further, looking at current problems in asset allocation and recommends that high-powered investment managers and consultants "reduce power" and "keep it simple".

A return to a more basic overall asset allocation strategy is intuitively appealing to mums and dads with less computational resources than the NZ Super Fund and financial advisers with better things to do. The article "Parsimonious Asset Allocation" is the basis of this week's story.

Ennis first highlights two basic shortcomings of contemporary asset allocation: there are far too many categories to invest in and the difference between categories is becoming fuzzier.

It is thus difficult to understand, manage and control risk. Many investors, and perhaps some advisers too, thought finance company debentures could legitimately represent the low-risk bond component of a portfolio when, in fact, these debentures had risk profiles more like highly leveraged companies.

At the first sign of trouble their prices plummeted. Before the rise of the new categories, popularly known as "alternatives", a trustee or DIY investor had just three or four asset classes to think about and there were clear differences between, for example, bonds, property, local shares and overseas shares.

According to Ennis, some US pension scheme trustees today have to cope with up to 10 categories and, while computers can calculate the impact on riskiness of untold combinations, there is a limit to which an individual can evaluate the results.

Furthermore, the market is slanted towards complexity, as the investment consulting business is a lucrative one and the more complex one makes the asset allocation decision, the more indispensable the consultant. The logic was the more complicated the portfolio the better the job you were doing and, if no one else could understand it, so much the better.

Next up Ennis attacks the faulty assumptions on correlation which have got so many pension funds into trouble in the past 18 months. He notes that fixed income allocations have steadily declined over the years and investors have relied more and more on correlation forecasts to control risk.

The party stopped in 2008: "The assumptions of largely uncorrelated returns among stocks, hedge funds, commodities, private equity and real estate broke down utterly."

Ennis then explains the title of the article - by "parsimonious" he means an approach that relies on self-evident truths; hard empirical evidence rather than assumptions and a preference for the least complex solution to the problem.

He adds that as contemporary asset allocation is too complicated and the data changes frequently that a model based on historic data simply doesn't work. In two weeks' time we will look at Ennis' parsimonious solution to the asset allocation problem, particularly in the context of New Zealand retail investors.

* Brent Sheather is an Auckland stockbroker/financial adviser and his adviser/disclosure statement is available on request and free of charge.