What effect does switching between bonds and equities have?

Most investors aim to maximise return. Actively switching, between equities and bonds, is one way to achieve this if it generates both higher return and lower risk.

This article examines this theory and concludes that active switching can enhance return and trim risk. As an extreme example we start by assessing switching between 100 per cent equities and 100 per cent bonds, even though this is not realistic.

When more realistic allocations are tested, the benefits of active switching fall and the conclusions are much the same as in the case of high equity allocation where investors can pursue extra return.

However they effectively have to double risk, meaning that a 75-25 per cent bond-equity allocation remains very hard to beat, when risk is taken into account.

In other words, investors should reduce the effort involved in asset switching and utilise this model, focusing instead on maximising returns in the individual asset classes.

Basic assumptions of the model

VIEW ALL

View all

Let us explain the underlying rationale for a simple switching model.

Tactical Asset Allocation is a favourite subject of many asset managers, and there may be some evidence to support its benefits. In this article we present a simple model of asset-class switching.

Specifically, a basic assumption of this model is that equity returns are auto-correlated. Before you run for hills, the term ‘auto-correlation’ explains a very basic idea.

Essentially, it means that each observation is related to the prior, in the sense that if equities are going up in price, they tend to do that price appreciation in a series of increases, which many have termed a ‘bull’ market.

Similar behaviour occurs when prices go down, or in ‘bear’ markets. All this occurs because the market is anticipating information, and it tends to anticipate information in one way or another.

Markets move in packs, and the individual assessments of many become expressed in the gradual anticipation of data, in markets moving continuously up, or continuously down.

If markets were not auto-correlated, then trends in equity markets would not occur. Yet, as we observe in Graph 1, markets do trend.

Notice in particular that the actual index performance tends to be very directional, staying away from the 200-day moving average most times.

Explanation of the basic model

Here is a brief overview of the model:

• First, we use the market-accepted benchmark of the 200-day moving average on the Australian All Ord (Acc.) index, and we develop a switching model that goes long and short equities.
• Second, when the model is negative on equities, as indicated by a negative 1, we select bonds, and use the daily performance of the composite bond index.
• Last, the overlay does badly when the market bunches around the 200-day moving average, as is the case now.

Signals for the model are described in Graph 2. Where the left-hand axis has a positive 1, the model prefers equities, and where the signal is negative 1, then the model prefers bonds.

Note how the model gives stable readings in stable ‘bull’ markets, and not that well when the markets are not decisive, and fail to trend away from the 200-day moving average.

Extreme

All you extremists out there…get ready.

We explore if you can beat the returns on any allocation by either being 100 per cent long equities, or long 100 per cent bonds.

While such an allocation is untenable to most investors, such an extreme allocation is used to explore the potential risk and return in such a case.

As Table 1 indicates, we generate a trading model that goes long equities when the annualised actual return is above the 200-day moving average of 12-month returns, and long bonds when the opposite is the case.

Roughly, this approach is long equities 70 per cent of the time, and long bonds 30 per cent of the time, if you add all the separate long days and short days up, and compare these to the total data set.

In Table 1, we compare the results of the above model to just being long equities.

Before trading costs the returns using the model are 12.27 per cent, yet after trading costs the returns are 11.07 per cent, with annual risk of 9.92 per cent. We calculate risk by annualising the standard deviation of the daily data.

Compare that to the long equity position, where we get less annual return, of 9.40 per cent, and much higher risk, of 14.82 per cent.

Despite the results, there is one very large problem with this approach: asset diversity.

Having 100 per cent equities or 100 per cent bonds in any investment portfolio is not a tenable position, as the benefits of asset diversification, where the negative correlation of return between bonds and equities cuts investment risk, are completely absent at all times.

Static results

Now, we can compare this extreme switching to a static allocation, where the allocation remains unchanged throughout the complete data set.

If an investor allocates 75 per cent to equities and 25 per cent to bonds (where in reality the equity allocation is much higher than the results summarised in Table 2, where return comes in somewhat under the active switching strategy, at 8.91 per cent – we can call this allocation ‘A’.

However, there is a problem – risk, which is still very high at 10.78 per cent.

This means that return will vary within any one year by around 11 per cent, so investors should expect a rough return between negative 2 per cent and positive 19 per cent.

Now, look at the static bond allocation strategy, with a 75 per cent allocation to bonds and 25 per cent allocation to equities, with only slightly lower return, yet still 8.61 per cent. We can call this allocation ‘B’.

Yet, notice the risk is much lower at 4.54 per cent.

Hence, in any one year, investors should expect returns to vary between circa 4 and 13 per cent. Notice the way the static return is 2.46 per cent lower than the active switching strategy in Table 1, while risk is roughly half.

Something more realistic

OK, we have the extreme model results, and the static results. What about the middle ground?

Here, we develop the following switching model for more realistic investment practice, by using more realistic asset allocations.

Instead of being long equities 100 per cent of the time, we adopt a less aggressive allocation to equities – for example 75-25 per cent equities to bonds.

In the alternative, instead of being long bonds, we can adopt a conservative allocation, 75 per cent bonds and 25 per cent equities. Details are summarised in Table 3.

As Table 3 indicates, the returns are good before and after trading costs, coming in around 8.75 per cent, with a risk of 8.01 per cent.

Yet, we wonder how all this trading and switching has really benefited the portfolio, when compared to the bond allocation in Table 2 above.

Here, risk is still high at 8.01 per cent, compared to a static allocation under allocation B where it is 4.54 per cent and return only 14 bps better.

Specifically, allocation B provides a return of 8.61 per cent, compared to the return of 8.75 per cent for the active switching model using realistic allocations.

The lower risk of static asset allocation B, where the portfolio consists of 75 per cent bonds and 25 per cent equities, while providing a marginally lower return, dramatically cuts risk and therefore improves reliability of return, a feature particularly attractive to those investors approaching or in retirement.

Conclusion

This article looked at a basic allocation model, which allocated between bonds and equities on the basis of returns being higher, or lower, than the 200-day moving average, where some auto-correlation in equity returns was the basic assumption of the model.

Evidence of the fact that this approach is relevant is the number of times the model trades per year; a little over six times a year. If equities were not auto-correlated, then that number would be much, much higher.

Active switching, using plausible asset allocations, not the extreme allocation of 100 per cent long equities, or 100 per cent long bonds, only generates a slightly higher return than the static allocations, but at what cost?

Risk is the answer here, as the risk of the active switching model is almost double the static allocation to a 75-25 per cent bond-equity allocation.

In other words, to get that extra few basis points in return, you need to accept double the risk. ‘Not a good trade-off’ for investors.

Active switching places the client with an allocation to equities of around 60 per cent on average, and this allocation does not derive the large benefits from bonds that the 75 per cent bond-25 per cent equity allocation provides.

Investors that actively allocate between equities and bonds assume they can beat the market in return and risk, but this article shows that it takes a lot of effort for a lot more risk to earn a small increased return.

Dr Stephen Nash is director for strategy and market development at Fixed Income Investment Group (FIIG).