Past performances do not determine the future, so how do you predict trends?

Fund manager selection is a very awkward and often complicated task. It is quite common for successful managers to be poached by rival organisations to try to uplift fund performance. If a individual or fund of fund investor was to attempt to follow a fund manager it can mean significant transaction costs are incurred in the attempt to chase the stars.

The ability to be able to determine fund manager luck from skill is a particularly difficult task. Just because a managed fund beats a benchmark doesn’t necessarily mean the manager has exercised any degree of skill. While any number of funds can beat a benchmark over a period of time, few can demonstrate that they have done this with any degree of skill. Many studies judge skill by both the level of excess returns over an appropriate benchmark and the consistency with which this has been achieved month in month out. Watson Wyatt did some calculations based on the idea that successful managers have an information ratio of 0.5 which is to say that they can produce 1 percentage point of excess return for every 2 percentage points of tracking error. On this basis, it would take 16 years to be sure that skill, not luck, was the cause. The problem is that the tenure of any manager is most likely to be significantly less than that.

To highlight the complexity of this issue and try to demonstrate the problem, Watson Wyatt conducted an experiment in which they utilised the Caps survey of pooled pension fund managers from 1990. It created an equally-weighted single portfolio which invested only in the upper quartile of manager from the previous 3 years and backtested it to 1990. To contrast this it created a second portfolio based on the bottom quartile of managers for the same time period. The results were stark. The portfolio of best-performing managers lagged the sector average and the portfolio based on poorly-performing managers beat the average.

Managers and advisers want to be able to establish an approach in which they would be able to beat the index consistently, without the too much concern given to manager turnover. Some argue that indices are to blame, with the market capitalised indices, having an innate tendency to make investors have their largest weightings in the most overvalued stocks and their smallest weightings in undervalued shares.

Investment styles also are a significant source of fluctuation in the market. Some managers prefer a value orientated approach. They invest with the belief that the company is undervalued against its peers and provides an opportunity for investment. Momentum, growth and GARP (Growth At a Reasonable Price) are also significant styles. Reviewing the past 15 years one could see that Growth managers were in favour in the late 1990s with the telecom and technology stocks helping the boom, but when the bubble burst and the markets became more bearish than bullish the growth managers underperformed sharply in the first few decades.

A manager’s appetite to risk, also needs to be carefully considered. The Caps survey returns are not risk-adjusted, so it is perfectly plausible for managers to have exceeded their peers by taking significant risk bets in stocks or by taking a great deal of sector or country concentration. If one was to look at current tracking errors, many managers have become particularly cautious about taking significant stock bets in what has been a very volatile market. In fact the greater levels of risk in an asset class, the greater opportunity for excess returns within that asset class.

To be able to quantify luck versus skill it is important to measure the performance correctly. For example, if you have a UK growth manager, it is then appropriate to measure them against a passive UK growth benchmark. Otherwise, your results will be confounded by how growth stocks are doing and not by whether or not the manager is demonstrating superior skill. The information ratio is the most meaningful measure. To decipher superior skill you would need to look at both the numerator of the ratio (how much did you beat the benchmark on average by) and the denominator for the variation of the difference between the manager and the benchmark. But the vast majority of managers who have come up superior on that measure, that majority of superiority even if it historic periods, it will be luck. A figure of around 10% might be skill, if you are lucky.

Some such as Rob Arnott, Chairman of Research Affiliates and creator of the Research Affiliates Fundamental Index (RAFI) argue that market cap weighted, and equal weighted indices are significantly difficult for investors to replicate. Equal weighted indices can demonstrate outperformance but they can require significant weightings in illiquid stocks, whereas market cap weighted can make investors more biased towards the most overvalued stocks. Rob strongly believes in Fundamental indices which selects, ranks and weights companies, not by market capitalization, but by financial fundamental measures, such as sales, cash flow, book price and dividends.

Some research from Rob helps to demonstrate this. Over the period 1926-2004 the largest stock in the US index outperformed the benchmark over the subsequent year just 38 per cent of the time; over 10 years, the proportion dropped to 26 per cent. Over 10 years, the cumulative underperformance hit 40.1 per cent. The largest 10 stocks in the index also underperformed over one and 10 year periods.

Rob Arnott, Chairman of Research Affiliates, LLC, said, “Cap weighted indexes overweight over-valued companies and underweight under-valued companies - this creates a performance drag that isn’t replicated in fundamental indexes. We have done extensive research and have found that fundamental weighted indexes, on average, have outperformed the well-known cap weighted indexes in the US by some 200 basis points a year. In international applications, the benefit surpasses 300 basis points, on average”.

Research by Nomura shows similar effects in other countries and at the global level. Some of this is undoubtedly a value effect, but then other studies have shown value does work over the long term.