Bearbull Portfolio: Why adding weak links helps your portfolio

That diversification is important is investing’s biggest truism. That said, not all diversification is created equal; some types are better than others. At one extreme, there is so-called random or naïve diversification, where no thought is given to comparing the components of a portfolio. By and large, this is how an insurance company behaves when it builds a portfolio of motor policies. It insures lots of similar risks, but it can still make money because it relies on the law of large numbers to guide its pricing. This is a powerful force in statistics, which tells us that the larger a data sample, the more likely that its average value will come close to its expected value.

So if industry-wide data tells an insurer that the claims rate on a type of motor cover will be one in 100 policies per year, the more policies of that type the insurer underwrites, the more likely that its claims experience will tally with prior probabilities.

Working along those lines can go some way in securities investing. Yet in a crucial way, a portfolio of stocks is different from a pile of motor insurance policies – the value of the stocks can rise and fall, the value of the motor policies can’t. That factor – the ability of stock prices to rise and fall – prompted one of the great insights in the history of finance: that it is not simply diversification that matters, but how an investor diversifies.

Spotting that insight – and proving it up – was done way back in 1952 by a 25-year-old PhD student from the University of Chicago, Harry Markowitz. We needn’t struggle with the formidable maths behind Markowitz’s work. Suffice to say he saw – fairly obviously – that the returns from a portfolio would be the sum of each component’s weighted return. Then came the clever bit. He reasoned – and proved – that the portfolio’s risk would be less than the sum of each component’s risk and the underlying reason why not is that the prices of the portfolio’s components will tend to move in varying directions.

For many years after his dissertation, Markowitz’s work was something of a curiosity – quite possibly brilliant, but of limited practical use because of its complexity when dealing with big portfolios. This helps explain why it was more than half a lifetime after its publication before Markowitz was finally awarded the Nobel prize (1990) for his efforts. Eventually, however, computer power came to the rescue. Today, give an Excel spreadsheet the data and it will handle the mean-variance analysis of a 100-stock portfolio instantaneously.

Nowadays therefore – and to varying extents – many investors behave in the way that Markowitz’s work recommends. Certainly, among the first things I want to know about a security in which I’m interested is how its likely behaviour will affect the returns of the portfolio in which it will be placed. In particular, I am asking, will its price behaviour smooth or amplify the portfolio’s returns? More formally, will it add to the portfolio’s volatility or reduce it?

Which brings us to where we left off two weeks ago. I had reasoned that putting more overseas holdings into the Bearbull Income Portfolio was long overdue. The fund’s miserable performance for several years was chiefly a function of being over-reliant on UK-listed value stocks. As an investment category, these have been deeply out of favour since the mid-2010s. There is no obvious reason why that sentiment should change soon, even though it will most likely switch at some point.

Hence Table 1, which shows the correlation coefficient in price changes between each pair of the seven variables shown (the Bearbull portfolio, the FTSE All-Share index plus five listed funds of overseas securities that might suit my purposes). The thing to know is that, where values are positive, price changes on average move in the same direction and the closer the coefficient is to 1.0, the closer the price changes (thus the actual value of 1.0 in the table where a fund’s price changes comprise both variables in the calculation). As values approach zero, the correlation gets weaker and when the coefficient is negative, prices tend to move in opposite directions. In portfolio building, negative correlation between pairs of assets is highly desirable, though in a portfolio consisting of equities it is rarely attained; even distantly-related equities tend to rise and fall in approximate unison.

Thus the delight in seeing the ultra-low correlation between returns from the Bearbull fund and contender investment trusts for Asian and European equities, JPMorgan Asia Growth & Income (JAGI) and JPMorgan European Growth & Income (JEGI). These correlations, based on monthly price changes from January 2010 to the present, are about as loose as equity price changes get – 0.02 for the Asian fund to the Bearbull returns and 0.07 for the European fund. Indeed, they are so low that I triple-checked the data. Given that shares in these two trusts also produced total returns superior to the Bearbull fund over the same period (see Table 2), then their inclusion in the Bearbull portfolio would have both improved its returns and – for what it’s worth – have reduced its volatility.

Useful though they are, correlations can be an unreliable guide to future price movements within a portfolio. As the warning would say on the can: ‘correlations don’t necessarily indicate a causal relationship’. Like all variables, they vary and a loose correlation over one period can give way to a tight connection in another. The point is made in the chart, which takes a longer stretch of price changes involving the Bearbull fund and the JPMorgan Asia trust going back to January 2000. However, rather than just calculating one figure for the whole period, the data is packaged into a five-year rolling correlation.

On that basis, the link between the portfolio and the investment trust stretches and contracts. Comparatively, it started out pretty close in the early 2000s, reflecting a period when equity markets were falling a lot. When the effect of the post-millennial dotcom crash started dropping out of the five-year data (from 2007 onwards) the correlation weakened considerably. Then it spent the next 10 years moving within a fairly narrow band until – perhaps coincidentally – Covid-related equity losses started seeping into the figures. However, within the context of equity-style correlations, the link has remained fairly loose. This prompts the best guess that it will that way in the future. Such a comment can also be made about the relationship between the Bearbull fund’s returns and those of the JPMorgan European trust, based on their own data set.

Given that these two also come with an income-fund-style dividend yield – 4.7 per cent for the Asia fund and 4.5 per cent for its European version – there seems adequate reason to put them into the Bearbull portfolio. Hopefully, they will bring some of that sought-after overseas performance from regions about which I confess to knowing little and which – perhaps more importantly – present retail investors with crushing administrative headaches when it comes to managing dividends.

As to how this is being done, that brings us to this week’s other big innovation. In effect, some 25 years after it was launched, the Bearbull income fund has had its second major capital injection. It started out in September 1998 with £100,000 subscribed. In round terms, £390,000 of value has been added, £248,000 of which has been in dividends received and distributed onwards; the remaining £143,000 in capital gains.

The new capital calls for a change to the way the fund accounts for its performance so as to make its future comparable with its past. Very conveniently, the original £100,000 subscription can be divided into 100,000 units with – obviously – a starting value of 100p per unit. Earlier this week, the fund’s market value gave a unitised amount of 242.88p. So, to raise the additional £100,000 – and maintain a continuous price history – 41,173 new units were issued. Thus, as we speak, the income fund now has £200,000 of paid-up capital divided into 141,173 units and its market value is £342,878 or, as just noted, almost 243p per unit.

Of the new money, two thirds has gone into the JPMorgan Asia and European trusts just mentioned. I am still toying with the idea of putting much of the remainder into iShares Core MSCI World (SWDA), an exchange-traded fund that tracks a US-dominated global index for developed countries. Its disadvantage is that it comes with a low yield but, as Table 1 shows, its returns have a loose-ish correlation with the Bearbull fund and it has a great performance record and low volatility. Sure, those characteristics may change, especially if and when the US tech giants morph into trundlers, as they will some day. Meanwhile I also want to look closely at some of the so-called dividend champions among US equities – the likes of AbbVie (US:ABBV), Procter & Gamble (US:PG) and 3M (US:MMM) – whose reliable income stream might also permit a shake-up in other parts of the Bearbull portfolio. More on this next week.

bearbull@ft.com