When constructing portfolios you need to recognize that two factors combine to provide the diversification effect. The primary factor is understanding how the returns on a new security correlate with the returns of other securities already in the portfolio.

Correlations can range from +1.00, which represents perfect positive correlation to -1.00, which defines two securities with perfect negative correlation. A high positive correlation implies that two securities tend to move in the same direction at the same time, which adds nothing in terms of the diversification effect.

The second factor that contributes to the diversification effect is the weight accorded to new assets or securities that are being added to the portfolio. Having said that, the weight applied to a new asset is irrelevant if the new asset has a high correlation with other assets in the portfolio.

Optimum portfolios balance these influences to attain the best diversification effect from the least amount of securities. With that in mind, let’s turn to the issue of how fixed income securities correlate with each other, and with other asset classes.

Correlations

We looked at correlations in the fixed income market based on ten years of data. Annualized returns, standard deviations, and correlation coefficients were calculated on 19 different fixed-income benchmarks, including government bonds, income trusts, foreign bonds, mortgage bonds, and high-yield bonds. Selected correlation results are shown in Table 1:

Table 1 Fixed-Income Correlations

This tells us that correlations for all of these benchmarks are positive and quite high, which means that the returns on all these fixed-income securities move together. Again, the most important thing is what this means at the portfolio level. The conclusion is that if you are going to have a fixed-income portion in your portfolio (and that’s usually a very good idea), it really does not matter all that much, from a correlation point of view, which fixed-income vehicle you use among these particular types.

The one caveat to this is the foreign bond component which has exhibited a high long term correlation with Canadian fixed income securities. However, given the strength in the Canadian dollar over the last eighteen months, the correlations between foreign and domestic bonds has dropped significantly. Having said that, we put more emphasis on the longer term numbers, if for no other reason, than a recognition that all assets are susceptible to mean reversion.

If you buy our position on this, then management style does not alter the role of bonds in a portfolio. It does not matter, for example, that a very skilled manager of a mortgage bond fund uses some elaborate and proprietary technique, and works hard to outperform all other mortgage bond funds.

At the end of the day, you can’t fool Mother Nature. If it walks like a duck and talks like a duck, then it’s a duck. And, in a portfolio correlation sense, a Canadian bond fund is a foreign bond fund is a mortgage fund is a short-term bond fund, because each is highly correlated with the others.

Negative Correlations

Does this mean that all income assets are positively correlated with each other? Not exactly. When we look at Canadian income trusts versus foreign bonds, the correlation is not very positive. Between 1992 and 1993, foreign bonds showed decreasing returns, and income trusts showed increased returns. From 1997 to 1998, the opposite was the case: Foreign bonds had increasing returns and income trusts had decreasing returns. In fact, in five of the ten years we looked at (1992, 1995, 1997, 1998 and 2000), their respective returns moved in opposite directions. These two securities, then, are great candidates for diversifying each other (see table 2)

Table 2 Fixed-Income Correlations

The actual correlation coefficient is –0.0894. The negative number indicates negative correlation. While the absolute value of the number is quite small, which means a minor diversification effect, we can take comfort in two things:

• We do not have to allocate equal dollar amounts to each of these securities. By altering the asset allocation between these two securities from 50/50 to a smaller allocation to the foreign component and relatively heavier allocation to the income trust component, the risk reduction properties can be greatly enhanced beyond what is initially indicated by the low correlation. Exploiting these efficiencies is the province of a professional portfolio manager, but you should know that the possibility exists.
• Although the number is only slightly negative, this correlation is much better than one that is slightly positive, and miles better than the correlations we saw between Canadian short-term bonds and Canadian mortgage bonds, for example (correlation of +0.977).

In the first point above, we are getting to the essence of portfolio management. Foreign bonds had a 6.6% annual compounded return over the 10 years covered by our research, and a standard deviation of ±25.1%. Income trusts had a 6.5% return over the same period, with a standard deviation of ±25.3%. An equally weighted portfolio of these two securities would return approximately 6.55% but would have a standard deviation of only about ±12%, which is a great improvement on the risk-reward ratio, all because of that negative correlation.