Introduction
A few weeks ago, after producing a graph showing shares outperform cash and bonds over the long term, I was asked a question along the lines of “if shares outperform other asset classes over the long term, how come over the last decade equity-dominated balanced funds (which returned 4.5% per annum [pa]) have underperformed cash (which returned 5.4% pa)?”. The same issue was alluded to in a recent Bloomberg observation that in the US, bonds have beaten shares over the last 30 years. While one can quibble over the details, given these observations it is natural to think maybe it’s time to give up on stocks and switch to cash and bonds.
Stocks do outperform over the long term
The first point to note is that over the very long term, shares have provided higher returns than cash or bonds. The next chart is the one referred to earlier and shows the total returns from Australian shares, bonds and cash from 1900. Despite numerous disasters along the way, such as the World Wars, the Great Depression, the stagflation of the 1970s, the 1987 share crash, a major Australian financial crisis in the early 1990s - A$1 invested in Australian shares in 1900 would have risen to A$287,087 by last month with a compound return of 11.9% pa. By contrast, the compound returns of 4.6% pa and 6% pa for cash and bonds would have seen A$1 invested in these assets rise to only a fraction of this.
Shares beat cash and bonds over the long term - Australia
Source: Global Financial Data, Bloomberg, AMP Capital Investors
It’s been a similar story in other comparable countries. Following is the same chart for the US. Not quite as impressive but still the same story.
Shares beat cash and bonds over the long term - US
Source: Global Financial Data, AMP Capital Investors
The long-term outperformance of stocks over bonds and cash is as would be expected – the greater riskiness of shares is rewarded with higher long-term returns.
However, even in the long term there is a cycle
The following chart shows a real accumulation index for US stocks since 1900. The trend line represents a real rate of return of 6.2% pa.
Whenever the index is rising faster than the trend line, stocks are providing above trend returns. Vice versa when it falls relative to the trend line.
Long-term bull and bear phases in US shares
Source: Global Financial Data, AMP Capital Investors
Long-term bull and bear phases are evident, and the bear phase over the last decade is not unusual. This pattern also exists in other countries. The following chart shows the rolling 10-year return difference between shares and bonds. Every so often shares have lengthy phases where they underperform bonds, e.g. in the 1930s, 1970s and more recently.
Shares periodically go through a decade or so where they underperform bonds
Source: Global Financial Data, AMP Capital Investors
This suggests that at any point in time, the experience of the past 10 to 20 years is no guide to the long term. An investor in US stocks at the end of the 1960s would have been wrong to project the above average returns of the 1960s into the 1970s (when actual real returns averaged -0.7% pa). Likewise the bad 1970s were no guide to the 1980s (when real returns averaged +11% pa). In other words 10 to 20 years is not the long term when it comes to shares. So the fact that US shares have underperformed bonds over the last decade doesn’t mean they will over the next.
In fact, what’s evident is mean reversion. 10 to 20-year periods with above-trend returns and above-average returns relative to bonds and cash tend to be followed by weak 10 to 20-year periods where returns are below trend. The table below shows the top performing asset classes (out of equities, bonds, cash and property) for each decade over the past century in the case of the US, the world and Australia.
Top performing asset classes by decade
Source: Global Financial Data, Dimson et al, AMP Capital Investors
The 1982-2007 bull market in Australian shares arguably spoilt investors and we have simply forgotten that the superior longterm performance of shares comes with a cost, which is that there are sometimes lengthy periods during which shares can perform poorly.
The 10 to 20-year return cycle in shares reflects fundamentals.
It’s no guide to the ‘long term’. The 10 to 20-year secular cycle in shares appears to reflect a combination of factors including:
- Starting point valuations – US share prices were high relative to trend earnings (i.e. the price-to-earnings ratio) in 1929, the late 1960s and in early 2000 (after which followed the secular bear markets of the 1930s, 1970s and 2000s) and low in 1949 and 1982 (after which followed two decades of strong returns);
- Underlying economic developments – depression in the 1930s and inflation in the 1970s were bad for shares, whereas solid economic growth, disinflation, economic rationalism, globalisation etc. in the 1980s and 1990s were great for shares. Right now it’s deleveraging in the private and public sectors in the US and Europe which is proving to be bad for stocks.
- Technological innovation – rapid technological innovation helped push stock returns above trend in the 1920s (electricity, mass production), 1950s (petrochemicals, electronics) and 1990s (IT).
Perhaps the most important point is that the starting point matters. Ten years ago US stocks were offering a dividend yield of just 1.5%, but the 10-year bond yield was 4.6%. This made it much easier for bonds to outperform shares as indeed they have over the last decade. But it’s now going to be harder for bonds to outperform over the decade ahead as their yield has fallen to less than 2% whereas the dividend yield has increased to 2.2%. This is still low, but even if share prices do nothing over the decade ahead, shares will outperform bonds. Likewise in Australia, 10-years ago bond yields were 5.6% and dividend yields were just 4.3% so it was comparatively easy for bonds to do well. Today though, bond yields are 4.2% and the grossed up dividend yield is 6.8%. In other words, it’s currently easier for shares to outperform bonds over the decade ahead as bond yields are quite low relative to dividend yields. This is also highlighted in Australia with the dividend yield grossed up for franking credits now running well above bank term deposit rates which are now falling. In fact, on this basis the grossed up dividend yield of 6.8% compared to term deposit rates of around 5.5% imply shares are paying out 1.3% more cash per annum than term deposits.
The Australian dividend yield is up, deposit rates are down
Source: RBA, Bloomberg, AMP Capital Investors
Concluding comments
The historical record suggests:
- Over the very long term stocks do outperform most other asset classes;
- However, there are 10 to 20-year periods over which this is not necessarily the case. In this context the recent experience in share markets is not unusual; and
- The outlook at any point in time in part depends on the starting point. After a decade or so of above-average returns a period of slower returns is likely, and vice versa.
The long-term cycle in equity markets should clearly be allowed for when setting investment strategy for individual investors. While 10 years might not seem long for me, it is very long for my mother. So, as discussed in a recent note, an outcome or absolute return investment approach may be appropriate for those with a short-term investment horizon or specific investment needs. However, for those with a longer-term investment horizon it’s worth bearing in mind that in an historical context, the turbulence in share markets in recent years is not unusual and doesn’t tell us shares won’t provide superior long-term returns going forward. This is particularly so with dividend yields on shares rising at a time when yields on bonds, cash and term deposits are falling.
Dr Shane Oliver
Head of Investment Strategy and Chief Economist
AMP Capital Investors