Before we move on to the fourth in this short series on investments in the main asset classes – property – let us draw breath and review the correlation between the classes we have looked at and discuss a little further the measurement and the concept of correlation.
Before building up the cross-correlations between the assets, it might be worthwhile looking at how the investment adviser can identify and, at least to some extent, quantify the level of these correlations.
If I were to tell you (noting there is a degree of subjectivity in the following numbers) the correlation between the assets we have discussed has been, over recent years, broadly in accordance with the grid shown below on the right, what could or would you surmise as regards the effect of diversification on portfolio volatility?
In fact, these levels sometimes change in nature with market conditions – a perfect example being the correlation between long-dated fixed-interest gilts and equity returns. Historically, these have been quite highly correlated. As we will discuss in a couple of weeks time, this is due to a large extent on the maintenance of a generally accepted long-term yield gap between the two asset classes.
In recent years, these two assets have become viewed more as competitors for investors' money and so have tended frequently to be very lowly correlated or even negatively correlated as a shift of money away from one asset (causing downward pressure on values) causes an influx of money into the other (causing upward pressure on values).
But what do we mean by, and how can we measure, high correlation, low correlation and negative correlation?
These measurements are readily and regularly made available to us by the main investment statistical and analytical providers (for example, Standard & Poor's Micropal) and by a number of product providers.
You will usually see correlation between assets (or between sectors, and indeed individual funds, if desired) denoted by a number between -1 and +1. If two assets have a correlation of +1, this means they have been perfectly correlated over the time period under review, meaning that the performance of one of the assets has been perfectly matched by the performance of the other.
This perfect correlation model never happens, of course, but if it did we would know that diversification between the two assets in question would produce absolutely no reduction in the overall volatility of the portfolio.
Assets which are said to be highly correlated (meaning diversification produces little reduction in overall portfolio volatility) will be indicated by a correlation factor between around +0.6 and +1 (but almost invariably between+0.6 and +0.8).
Correlation factors between around +0.4 and +0.6 indicate moderate correlation, factors between +0.1 and +0.4 indicate low correlation (and a high reduction in portfolio volatility as a result of diversification) and a factor on or close to 0 indicates no or very little correlation – meaning the performance of one of the assets has rarely, if ever, been necessarily reflected in the performance of the other(s).
If you have never come across this notation in the past, it is probably because you have never looked for them or have skipped over this part of the information provider's statistics in trepidation at the fancy descriptors frequently used.
I think you would find it useful to start looking again for these figures or better still proactively interrogating your information or product providers – the best ones will be delighted to help.
Negative correlation is quite unusual between investment options. Where it exists, it indicates that the performance of one of the assets under review will be mirrored by a reverse performance in the other asset(s).
Thus, for example – as I briefly mentioned above – equities and long-dated fixedinterest gilts have in recent years frequently been negatively correlated as a strong performance in equities has been the result of, or has resulted in, money tending to move away from the fixed-interest market, thereby depressing prices in that “alternative”.
When it exists at all in the investment world, negative correlation will rarely fall below -0.3, meaning that the assets have only been relatively mildly negatively correlated (meaning, in turn, that, mostly, the performance of one of the assets has not depended at all on the performance of the other).
Negative volatility (if it ever existed) between -0.3 and -0.6 would indicate moderate negativity with high negativity indicated by a factor between -0.6 and -1.0.
Perfect negative correlation is indicated by the factor -1.0, which would mean that the performance of one asset over a given period would result in a mirror image performance in the other asset. The table here summarises these generalisations.
Now, before we return to our asset class discussions, I feel it would be useful to ensure that we have not become too embroiled in any ideas that this summary represents mathematical theory with little use in practical portfolio planning.
Take a few seconds (if you have not previously been prompted to think about these concepts previously) to understand the usefulness of the implications of the following brief example.
If two investment asset classes have historically both produced very high returns over the longer term but have both been very highly volatile, then investment in either one of these options would have given investors a rocky ride in the short or medium term, only reliably rewarding those with patience or a strong acceptance of risk.
This is, simply, the old risk/reward relationship we all learned in our investment induction training. However, if we knew that the two assets were destined to remain negatively correlated, then diversification within the portfolio between the two assets would produce returns on a par with the average of the individual returns but portfolio volatility which – as the performances generally moved in opposite directions at any given time – is much lower than the individual volatility of each asset.
You should now be in a position to understand fully the implications of the grid at the start of this article and, in future weeks, the grids we will progressively build up.
Consider this concept further for a few minutes. It forms the springboard to our movement to completion of the cross correlation of all our asset and sector classes, over the next few weeks.