Twenty years ago, a portfolio of 25 to 30 stocks was considered fully diversified but is now thought highly aggressive
The average number of stocks in a mainstream active portfolio has been trending up over the course of the past decade, driven both by increases in stock volatility and the rise of tracking error as a risk assessment tool.
Twenty years ago, academics concluded that as few as 25 to 30 stocks produced a fully diversified portfolio and indeed it was maintained that after the first 15 stocks, each new addition to a portfolio added a decreasing level of diversity.
Academic research relating to any aspect of investment funds must be taken with a measure of scepticism. All such work is based on a set of assumptions, which allows a model to be formulated. It also reflects a historical moment in time referring to a specific market and set of circumstances.
Previous research also tended to refer to absolute risk rather than referring, as tracking error does, to a benchmark.
Things have changed since the 1960s and 1970s, when the pioneering academic work on portfolio construction was done and, according to Threadneedle, these changes have been behind the shift in the perception that now views a 30-stock portfolio as highly aggressive.
Most importantly, however, according to Threadneedle's communications director Richard Eats, mathematical modelling cannot be overlaid seamlessly onto a market which is rapidly changing and a system based both on fundamental, quantifiable factors and human behaviour.
To some extent, the power of academic work set the base number of stocks that a unit trust, and now an Oeic, can hold. Eats said that the 20 stock minimum written into the unit trust and now Oeic rules grew out of this work.
Despite this legislative base level, there has been a sea-change in the way investment houses think about the number of stocks necessary to achieve diversity and consequently in the level of stock diversification they offer.
Eats said analysis by Threadneedle shows that increases in volatility on both the stock and fund level have been the major driver behind the shift towards greater portfolio stock diversification.
However, Eats said, the simple fact there are more stocks does not mean more holdings in a portfolio are necessary to achieve controlled correlation to the index.
He added: 'I think the change has been driven by a number of factors. There has been a move towards sector analysis. Look at the big stocks in the UK: Vodafone does more business outside the UK than within, so does BP, Glaxo and HSBC. That is 40% of the index which behaves more like the global banking, pharmaceutical or oil sectors than what's happening in the UK economy.
'The second thing is that over the year there has been a tendency to move towards focused businesses. Conglomerates like Hanson or British American Tobacco, when it used to own financial services, now just focus on their core businesses. Companies have come to the market, such as Arm and Colt, which are almost one product companies.
'Conglomerates had diversity of earnings. Now earnings are much more focused. You therefore tend to get more volatility in a single stock than in a stock with three or four diverse businesses.'
Threadneedle quotes the recently published study by Harvard professor John Campbell that shows it is stock and not market volatility that has increased.
In his work, Campbell concludes that overall market volatility is no higher than it was during the past seven decades and that, compared to the 1970s, the 1990s were roughly 20% less volatile, something that may surprise many advisers and their clients.
Compared to the 1980s, it is 30% less volatile, Campbell found.
Analysis on a stock level showed quite the opposite ' prices were twice as volatile over the past decade as they were 30 years ago. Many stocks climbed sharply and many fell by an equal measure.
Campbell agreed with Eats that it is the focusing down of companies on core expertise that has led to the rise of sector investing, due to the greater correlation of stock performance within sectors, and that companies are more exposed to a downturn in their narrow business areas.
This, fund commentators suggest, has led the old 25 to 30 stock diversification argument to be altered to 50 or above.
One further factor has increased stock volatility, according to Eats, which again could be behind the tendency of mainstream portfolios to hold a greater number of stocks than in previous decades.
He said: 'The third factor is that there may be additional stock volatility from hedge funds. People get hit by shorts and then they close the short and it bounces. The short-term volatility of the stock tends to be exaggerated.'
The last point is something that is being increasingly blamed for stock volatility, although few figures have been brought out to prove or disprove the point.
Estimates from some of the largest investment banks have suggested that on a typical day hedge fund managers are responsible for around 20% of their trading, something that has seen a big rise in the commissions the institutions earn.
At a stock level, their influence on price can be far more extreme. It is estimated that around £35bn of hedge fund assets are managed from London. Although that is tiny in comparison with the vast assets of pension funds, hedge funds are far more aggressively traded and have the ability to gear up their stock bets.
Robert Burdett, the fund of funds manager at Credit Suisse Asset Management who recently joined the group from Rothschild Asset Management with Gary Potter and Kelly Prior, said the rise of tracking error has also been an important factor in the general increase in the number of stocks held in mainstream retail portfolios.
Burdett said: 'As a generalisation, funds are holding a greater number of stocks. If you look at the unit trust rules, you are limited to 20 stocks as a minimum. Finding a fund today that is even close to that is almost impossible. It would be considered the most aggressive of the aggressive funds launched.'
Potter said: 'We have seen the rise of tracking error in recent years. Five years ago you would have barely heard it mentioned.'
To achieve tight or controlled tracking errors it is generally felt a manager needs a greater number of stocks. That is particularly true if a manager wants to hunt outside of the largest capitalised stocks.
Potter said this creeping benchmarking, or semi-active, quasi- index tracking approach, often referred to as index-plus, has produced, in many cases, unexciting results, leading to the creation of focused funds free of tracking error constraints.
It is a measure that spread into the retail arena from the institutional world, according to Jeremy Lang of Liontrust, where the need to match liabilities to assets requires careful mapping. This mapping is based on assumptions for individual markets. Tracking error allows institutions to dictate more closely to fund managers the returns they expect.
It has also risen in importance as a tool for intermediaries, being one of a number allowing them to more closely match funds to clients risk tolerances.
Threadneedle uses tracking error as one of the key differentiators between its ranges of growth and select growth funds in the UK, Europe and the US. These funds correspond to core and more aggressive holdings.
Malcolm Kemp, head of quantitative analysis at Threadneedle, said groups tend to use systems like Barra to help construct portfolios with specific tracking errors, using the preferences of managers. For Threadneedle, the perimeters of fund mandates, manager preferences for stocks and quantitative measurements, come together in a feedback loop to construct funds with the risk profile sought.
This allows tracking errors to be reined in at times of uncertainty. Currently, the American Select and American Growth both have around 90 holdings as the risk has been taken down on both funds.
Kemp said: 'What we have seen here is that the whole curve has shifted upwards. For any given number of stocks, you have a higher tracking error. What people then do is hold more stocks in order to achieve the same desired tracking error.'
This can be clearly seen in the number of stocks in select and growth options Threadneedle offers in the US, UK and Europe. The European Select Growth fund, which is designed as the more punchy of the two, had 62 stocks at the end of June, compared to 83 in European Growth.
This results in a tracking error of just under 7% per annum for the Select Growth fund over the past three years and 5% per annum for the European Growth fund. UK Growth has 97 stocks and UK Select Growth has 70.
Kemp said: 'To come up with a precise number is pretty dubious but you can come up with an idea of where that number is trending in order to allow you to adopt a consistent computation of the number over time. Twenty five used to be the number in the textbook. If we now used the same sort of approach to calculate, that number would have gone up significantly. The general trend is discernable.'
There is a strong underlying set of principles in the numbers of stocks that produce diversity in different markets. JP Morgan Fleming fund manager Chris Complin, who works with Andrew Spencer on the group's style funds, warns that historic tracking error consistently comes out as higher than predicted tracking error, demanding caution in using the figures put out by systems such as Barra.
According to testing by JP Morgan Fleming centred on the US index, the S&P 500, tracking errors of certain levels can be calculated precisely for combinations of stocks, showing the trend upwards in tracking error as the number of stocks reduces. Complin said the top 50 names in the S&P 500 make up 55% of the index. A portfolio of those stocks cap weighted would produce a tracking error of 3.4%, 3.2% if equally weighted, and an absolute potential annual volatility of 15%. For the top 40 stocks covering 51% of the market, tracking error for a cap weighted portfolio is 3.89% and 3.68% for an equally weighted portfolio. The absolute risk is similar.
The pattern is clear. Randomly generated portfolios of 30 equally weighted stocks from the S&P 500 produces tracking errors of around 8%.
This naturally leads managers, according to research, towards core holdings in large cap stocks, often denying managers opportunities to seek returns in mid and small-cap areas. Often, according to a recent paper in the Journal of Portfolio Management by Goldman Sachs, this leaves managers struggling to outperform benchmarks as large-cap stocks are an arena in which active management is arguably least effective.
Carefully managed, however, this can result in a core around which risk can be added.
The Lang approach is an example of this. In his UK Growth fund, Lang controls tracking error with a spine of large-cap holdings, achieving a tracking error of between 4% and 8%, leaving him freer to follow special opportunities style investments than other managers.
For the adviser, however, according to Mark Dampier of Hargreaves Lansdown, the most important thing to look at is not the number of stocks alone. What is important, he said, is the number of stocks a manager can effectively run and that the process used is scaleable. It is this fit between number of stocks, fund size, market and investment process that matters in selecting a fund, he added.
For Albert Morrillo, a 35 stock portfolio produces the diversification and returns that satisfy Dampier. On the other end of the scale, maintaining around 200 holdings does not stop Anthony Bolton from continuing his strong track record on the Fidelity Special Opportunities fund, Dampier said.
The number of stocks held in a portfolio can also be driven by a fund's size.
Ten years ago, the average UK All Companies unit trust was much smaller than today. Larger funds tend to hold more stocks as they have liquidity constraints. A large fund not able to sell out of a holding completely in five days may be hard to turn around when a bet backfires.
Clarke replacing Balkham
'Deep-dive analysis of client behaviour'
Ways to mitigate April’s increases
The best equity income funds examined