Submitted by F.F.Wiley of Cyniconomics blog [6],
If you’ve ever sought advice from a financial advisor, you probably asked the question: “How much of my portfolio should I hold in stocks?”
Somewhere in the answer, you were probably offered long-term return estimates.
These estimates probably placed stock returns at approximately inflation plus 5 or 6%.
But what if standard estimates are too optimistic, as they were in the 1990s when advisors typically predicted double-digit long-term returns? Shouldn’t this change your investment allocations?
We’ll argue that the usual estimates are overoptimistic, and that investment allocations should be based on more realistic expectations.
Worse still, the discrepancy has reached enormous proportions. By projecting earnings forward over the next decade, we’ll show that stocks are now priced to barely outpace inflation at best.
Here’s a chart that demonstrates our approach:
[7]
The black line shows over four decades of S&P 500 (SPY) earnings, expressed in 2013 dollars. The colored lines are exponential trend lines calculated over the last one, two, three, four and five earnings cycles.
To define an earnings cycle, we begin at the exact point in a recovery when earnings busted through their prior peak. Therefore, cycles extend from the first all-time high in a particular earnings recovery to the first all-time high in the next recovery. This approach mitigates the problem of trend lines being sensitive to beginning and ending points. It’s far better to start and end a trend line midway through earnings recoveries than to calculate it from, say, an earnings peak to an earnings trough.
Conveniently, inflation-adjusted earnings likely breached their 2007 peak at the end of 2013, marking Q4 as the beginning of the next cycle and adding the period from 2007 to 2013 as the final cycle in our trend line calculations.
Estimating 10 year returns
As shown in the chart, we extend each trend line forward to estimate earnings in 2023. We can then estimate 10 year returns with suitable assumptions for the price-to-earnings (P/E) multiple.
Specifically, we use median P/Es from the same time periods that we used to estimate trend lines. Even though P/Es in 2023 will surely be higher or lower than historic medians, we have no way of knowing which of these possibilities will play out so far into the future. Just like the earnings trends above, our approach to estimating P/Es is the most neutral (not cherry picked to produce the “right” conclusions) that we could come up with.
Here are our calculations:
[8]
As shown in the right-hand column, all but one of the real return estimates are negative. What’s more, the picture looks even worse with more history. Here’s another chart that adds trend lines covering six, seven, eight, nine and ten earnings cycles, followed by a second set of calculations:
[9]
[10]
While the results speak for themselves, we’ll share some thoughts on specific time periods:
- Within the period shown in the first table, productivity grew most strongly [11] in the latter half of the 1990s and early 2000s. This is probably the biggest reason for differences between the “last 3 cycles” return estimate (which covers the period from 1988 to 2013) and the other estimates.
- Although our future could certainly include another productivity-based profits boom, other factors tell us to be cautious. Consider that earnings were boosted by declining interest rates in each of the last three cycles. Consider also the ratio of corporate profits to GDP, which reached all-time highs [12] in 2013. While these developments are baked into the 1988 to 2013 trend line, they’ll eventually come to an end and even reverse direction. Interest rates can only fall so far, while profit shares can only rise so far.
- Households, businesses and the government hold much more debt as a percent of the economy than they did forty years ago. Borrowing in all of these sectors has helped boost earnings in ways that can’t continue forever. In other words, even the worst results from the first table – covering five cycles from 1973 to 2013 – fail to convey the risks of our debt addiction.
Overall, our research couldn’t be further from the financial industry’s conventional thinking. Conventional estimates call for stocks to outpace inflation by 5 or 6%. The results above, on the other hand, show mostly negative real returns over the next decade.
Needless to say, we recommend questioning any advice that’s based on standard estimates.
Better yet, send our charts and tables to your advisor. Request an explanation for how stocks outpace inflation from today’s prices. Will earnings climb even further above established trends? Will P/Es never fall again?
Once you’ve established your advisor’s assumptions, weigh them against history. Challenge him to explain why this time is different.
Bonus chart
If you have to choose just one chart to show that current earnings (and especially 2014 forecast earnings) are out of line with established trends, we recommend our last chart below. We calculated trend lines for every combination of three or more consecutive earnings cycles (36 in all). The results leave little doubt about the discrepancy between current earnings and historic precedents.
[13]
More info
Technical notes for this post can be found here [14]. Also, we discussed the importance of long-term earnings trends in: “Why Stock Prices Are More Stretched than You Think: A Tale of 3 P/E Multiples [15]” and “P/E Multiples, Deleveraging and the Big Experiment: Sizing Up the Next Bear Market [16].”