We read an interesting article during the week that made a bold claim: the best way to predict stock market returns is to look at how much money investors are allocating to equities. (For example, a 50% equity allocation means that the average investor’s portfolio is 50% equities and 50% bonds/cash). The article is about ten years old, so we decided to update the data and see for ourselves if it worked.
The intuition is simple: as investors allocate more money to equities they will, of necessity, have to buy more shares for their portfolio, and sell bonds/cash (the other two key assets) to do so. This will, when done by all investors and across long periods of time, tend to push equity prices up because the average investor is now a net buyer of equities.
Eventually, this process will result in very high allocations to equities and is somewhat self-reinforcing – rising share prices feed investor enthusiasm for equities. As we will see in the chart below, high allocations to equities tend to coincide with the ends of bull markets.
The process will then begin to reverse itself as demand has been sated and the potential returns for equities at peak allocation levels will look poor. This means either that investors begin to sell equities in exchange for bonds or cash, or simply buy lower amounts of equities. In either case, this low demand (or even selling) over long periods of time leads to equities becoming cheap – either because prices have fallen, or because they have not kept up with fundamentals.
We can now begin to see the relationship: at periods of peak equity allocation, equity prices have been rising to high levels. Valuations are likely to be high as a result, and so subsequent returns for equities are likely to be low. At periods of trough equity allocation, subsequent returns are likely to be high because investors will be attracted by low equity prices (and thus high earnings yields) and will begin to increase their allocation to equities.
The chart above shows this relationship. There’s quite a bit in the chart, but it is well worth the effort to understand.
The blue line shows how much of the average investor’s portfolio is allocated to equities, and the axis is on the left-hand side. The orange line shows future 10-year annualised total returns for the S&P 500. The axis is on the right-hand side and is inverted (because the two series have an inverse relationship – higher allocations lead to lower returns and vice versa). The data has been sourced from the Federal Reserve and Robert Shiller, a Yale economist.
So, for example, we can read the chart as follows, using the points inside the green circle: the blue line shows that investors allocated, on average, 27% of their portfolio to equities in 2009. This makes sense because the global financial crisis had just occurred, share prices had collapsed, and investors were pessimistic. The orange line shows that investors earned an average annualised total return of 16% from equities over the next ten years (2009-2019). In other words, a low starting allocation led to very attractive returns.
What is amazing about the chart is how closely the two measures track each other. It’s highly unusual in finance to get such a clear pattern. The correlation between the two is -93%, suggesting a very strong negative relationship. A negative relationship means that as one variable increases (allocation to equities), the other decreases (10-year returns).
Looking to the future, the implications are clear to us: investor allocations to equities in the US are still high at 44.9% as of the latest reading in December 2022. Although this is down from a high of 52% in 2021, it is still in the top decile of allocations since 1945, so a very high level. Given the strong historic relationship between equity allocations and subsequent returns, we are forced to conclude that returns over the next 10 years are likely to be poor.
The table opposite shows 10-year S&P 500 returns for various allocations. Clearly, the lower the allocation, the higher the subsequent return. When equity allocations have been as high as they are currently (44.9%), subsequent ten-year returns on the S&P 500 have been 0.7-5.4% compound per annum. We can also use a very simple linear regression model, which suggests annualised returns of 3.3% from here – consistent with our first estimate.
We have been saying for some time that the key S&P 500 Index is richly valued relative to historic averages, and high starting valuations lead to poor subsequent returns. This new indicator adds a second way of looking at things, and confirms what we have been saying to date: there is an above-average probability of poor returns on US equities from here on a 5-10 year view. A sharp bear market, of course, could quickly change that.