Many planners and research houses have adopted the view that fund manager styles can be meaningfully classified along a simple value/growth spectrum (including growth at reasonable prices around the middle).
This mono-dimensional view is highly simplistic and ignores many factors which go into determining an investment style. Further, there is no reliable method for determining an optimal mix of value and growth at any point.
Some may simplistically represent them in a portfolio on a 50/50, 60/40, or other basis at all points in time. Surely the whole point of identifying two different sectors of the market is to be able to make use of the distinction by adjusting exposure to reflect the relative prospects for each.
From time to time, one hears that, say, “The prospects for value stocks are particularly strong”. What does research tell us about the possibility of forecasting sector outperformance?
Due to space limitations I will concentrate on only value stocks, and their performance relative to the market.
Let’s take the Ken French (of Fama and French) US Value Index, as it provides data since 1926, and contrast this with the S&P 500. (Note that French uses share price to book value — broadly net tangible asset value — to measure value.)
Value has produced a superior return, on average by 3.2 per cent per annum. Should value therefore routinely be over-represented in portfolios? Unfortunately, there are long periods during which underperformance occurs. For example over the 15 years to 2000 it produced a return lower by 1.6 per cent per annum.
If such long periods of superior and inferior returns cannot be predicted, it renders somewhat useless the knowledge that value tends to outperform.
We’ll test four hypotheses about forecasting value outperformance.
Hypothesis 1: GDP will be relevant.
It has been argued that investors require a higher premium to invest in value companies during a weakening economy. The graph shows smoothed (five-year trailing) GDP is compared with value’s subsequent excess return (that is, greater than the market).
There is a weak correlation of 0.35. (A weak correlation appears as a somewhat random scattering of dots. A strong one tends to see dots clustering in a band vaguely around a 45-degree axis.)
Hypothesis 2: Interest rates are relevant.
Making the simplified assumption that growth stocks are expected to exhibit higher future earnings, they may be susceptible to upward movement in bond rates (flowing into a higher discount rate for future earnings).
Comparing movements in yields (last year’s rate compared to the previous five year average) with value’s subsequent excess return reveals a slightly better, but still not strong, correlation of 0.39.
Hypothesis 3: Periods of especially attractive value will be relevant.
By measuring the average price-to-book of the 50 per cent of companies with the best value (low priced book) and the 50 per cent with the worst value, a ratio of these can be calculated. A high ratio indicates that value stocks are especially cheaply priced compared to growth. Does this flow into subsequent five-year outperformance?
The effect is minor, with a correlation of only 0.29.
Hypothesis 4: Mean reversion will be relevant.
When value has outperformed over a five-year period, does this suggest that it might underperform over the next five years?
No, it is utterly irrelevant, with a tiny negative correlation of -0.05.
We have been unable to find a strong method for predicting whether value will deliver a particularly superior or inferior return. This is not an immaterial problem for any portfolio construction model that uses the value/growth distinction as its essential element.
Further, as noted above, there are many other factors which are significant in understanding fund manager style. The Dimensional Australian Value Trust, MMC Wholesale Australian Share Fund and Tyndall Australian Share Wholesale Portfolio could all be reasonably described as value investors. Yet it would be absurd to think they followed the same, or even similar styles. Last year, we compared their stockholdings. Dimensional had 1.6 per cent portfolio overlap with MMC and 23.7 per cent with Tyndall. Tyndall had 0 per cent with MMC! It is hard to see how two managers can be described as following the same style when they have not come to a common decision on investing in a single share.
Another simple proof of their diversity is to analyse past performance, which has diverged markedly on an annual basis.
Incidentally, Tyndall had 44 per cent portfolio overlap with Colonial First State Wholesale Imputation Fund, which is viewed as a growth investor.
The many other factors relevant for meaningful analysis of investment style include:
n Large cap vs small cap bias;
n Stock universe mandates for example, industrial share funds;
n Quality filters (some managers only invest in quality companies while others will buy a poor company if the market price has discounted it more than is justified);
n Top down vs bottom up;
n Fully invested vs willing to cash up; and
n Willingness to short stock/market/currency.
These seem to be widely ignored, even by some research houses. This leads to absurdities such as Platinum being ranked less highly because it doesn’t confine itself to doing what someone, I suspect with less ability than Kerr Neilson, has deemed value investors are supposed to do. What they do works, but it doesn’t fit neatly into someone’s mental box.
There are other sophisticated and successful fund managers who publicly typecast themselves as definitive value managers, yet who privately acknowledge that it is an infantile categorisation — but necessary for a good research house rating. As always, the research houses are the funds flow police.
Our industry has adopted a simplistic tool to solve the complex problem of understanding manager styles. And there is no reliable method of applying the tool to achieve a superior outcome for clients.
Robert Keavney is chief executive officer of Centrestone Wealth Advisory .
