The Demographics of Stock Market Returns Part I

Context
Many articles and academic papers have been written on the link between demographics and stock market returns. For example, a widely-circulated paper of 2011 from economists at the Federal Reserve Bank of San Francisco suggests a price to earnings ratio of 8.4 in 2025 for the US, which would mean a soon-to-be 50% stock market plunge that never recovers.
 
There are 2 multi-trillion dollar questions:

1. Is this true/credible/probable?
2. If it is, WHEN will this correction happen and what form will it take?

 
First, if it is true, the implication is that stocks should plunge approximately 50% and essentially never really recover, on average. Second, it means that we are really going to fall off a cliff soon, and it would be wise to get out of stocks now and hold low-return cash or quasi-cash such as AAA bonds for a decade. Let’s look into this a bit, shall we?

The idea of the model
The general approach to all demography-based models for stock market forecasting is simple and intuitive: all else equal, the more “buyers” there are relative to “sellers”, the more it is good for stock prices and stock indices. Since retired people tend to be “sellers” of stock, a decreasing ratio of buyers-to-sellers due to an aging population would put inexorably strong downward pressure on stock markets.
 
This makes a lot of sense: the most likely buyers of stock are people in their 40s that are established in their career and want to prepare for retirement, and the sellers are people in their 60s who want to get away from stock market risk and hold their wealth in less-risky forms.
 
To capture this effect, the researchers have taken an interesting approach: they approximate the “buy group” as the age group 40-49 and the sell group as the age group 60-69. They then make a strong case for the relation between the proportion of buyers per seller and the price to earnings ratio, with historical data from the 1950s until 2011: their “buyers-to-sellers” ratio explains the general trends of bull and bear markets. According to this metric, we are currently in a secular bear market that should hit bottom around 2025-2030.
 
Scenarios and predictions
Many people have used demographics to call incredibly low and high values for stock indices that were never reached. Others (or the same ones!) are now using this same demographic approach to call a decade of bear markets everywhere, with some empirical proof to this claim.
 
Fact 1: China, the USA, and all industrialized economies have recently tipped into a 20-year trend of decreasing ratio of working age population (15-64) relative to total population.
 
Fact 2: The ratio of 65+ relative to total population will increase for the next decade and more in China, the USA, and most industrialized economies.
 
Fact 3: Japan and Germany will have a significant absolute contraction (not just relative to total population) in the working age population for the next decade.
 
Since I am an economist who has an honour’s bachelor’s degree in math and loves analyzing data, I tried other scenarios. We’ll try a few scenarios in part II next week, but here is a simple one: I took the number of employed aged 18+ (including 65+) as an approximation for stock buyers and the number of non-working retired as an approximation for sellers of stock.
 
This captures the fact that there will be an increasing proportion of 65+ individuals who will continue working, and may be buyers (or holders) of stock even if they are older. I used the BLS and Fed projections for unemployment rates and participation rates by age cohort for the 2015-2030 period.
 
I essentially got the same result: my measure of “buyers-to-sellers” ratio contracts from 3.26 now to 2.26 in 2030: we lose one full buyer per seller over the next 15 years, suggesting a general secular bear market for the next 10-15 years.
 
This got me thinking: should we all get out of stocks and essentially wait a decade to come back? Will all this be priced in during the next stock market crash, marking the end of stocks for a decade?
 
The history of extreme predictions
The end of time has been predicted before… for centuries and millennia! These predictions never came to be. Yet the predictors in question kept saying “it will happen, it is a matter of time”… and we are still waiting, 5, 50, 500 years later!
 
Recession calls and extreme market calls are serious business, because if people listen to these calls, they allocate their funds based on these “convincing” calls for a DOW at 40 000 or a crash to 5000 that would never recover. If you are a “buy and hold” type and you read a convincing report that calls for a 10-15 year bear market, you get out now and move to the low-yield bond market.
 
Many doom predictors have called for the end of stock markets, the end of the USD, or an oil price at 300$ per barrel every year for the past 15 years. Recently, since 2012, there are calls of the crash of the US stock market that will be the mother of all crashes, even worse than the Great Depression, and more doom for the USD.
 
These people get lots of media attention, sell lots of books, and they are really convincing. They don’t tell you that they called extreme stock index values (or oil prices or USD value) before and were totally wrong or “off by a few (5 or 10) years” – they focus on the ONE call they got right (if ever) and tell you that this is “proof” that they are good at this stuff. They don’t tell you that their analyses are flawed and would not pass a basic smell test.
 
Of course, at one point they ARE indeed right: a recession happens, the stock market plunges and they get to say “you see: I told you this would happen, I called this last year and it is now happening”… and they did indeed! The problem is that they called it every year for 5 or 10 years, and if portfolio managers had listened to them, they would have parked all their client’s savings in low-yield bonds and sacrificed incredible annual returns, forcing retired clients to live in poverty!
 
Market projections based on demographics can be off for years or decades
Don’t get me wrong: demographic trend predictions are credible. The problem for financial markets is that predicting financial market trends and dynamics is orders of magnitude more complex and “timing” is significantly more important in financial planning relative to demographic analysis.
 
If you expect a bear market to start in 2016 and last a decade, and it never happens OR it starts 5 or 10 years later, your investment decisions are going to be way off, as well as your financial advice. Of course if you see the future with rosy glasses and you are blind to warning signals, you will also make very costly decisions in your investment strategies!
 
Bad decisions are costly and have very real consequences, and demography-based market projections can be totally off in both timing and orders of magnitude. I will demonstrate this next week, so be sure to read next week’s blog post.
 
The devil is in the details
You can’t pick one aggregate measure such as the working age population to total population ratio or even the “buyers to sellers” ratio, then use statistical correlations that track stock market trends, and use this as a call to predict the future of stock markets.
 
This is a severe oversimplification of reality, and it omits profoundly important details. Things can change significantly over a decade and small shifts in age group behavior, labor market conditions, regulation, global markets and growth, or estimated parameters can have HUGE impacts when aggregated, rendering the extreme doom scenarios groundless and costly.
 
First, the use of long term time series is risky, because the global economy changed dramatically since the 1990s, with an acceleration of technology adoption, complex inter connections between markets, global supply chains, globalization of capital and markets, widespread use of QE and other market-changing policies, massive carry trades, the greatest decrease in global poverty in human History, and incredible growth in emerging markets.
 
“Return-to-mean” type of models such as error correction models and the like can drive you to serious investment errors with even small calibration issues, identification errors, or omission to consider structural breaks. This is not theoretical academia where we are curious and we discuss a paper and its errors or possible improvements during a seminar – it is real money.
 
Long-dated time series do not capture structural breaks in the patterns of macro variables, from home bias and risk aversion estimation to technology, global growth, and MANY other structural developments that are fundamental since the 1990s. These omissions and long-dated statistical analyses can cause severe “identification errors” in models, which lead to a wrong interpretation of results and its consequences: bad decisions, bad policies, and a sub-optimal world.
 
Second, there are so many details that could cause things to turn out differently that it is arrogant (and potentially costly) to call a secular bear market using demographics (or any other metric). What important “small details” could shift a few parameters and change the entire story upside down just as convincingly? This is such a big question that it will be covered in the next blog post, next week in part II.
 
Conclusion
What to make of all this? Well, in the next stock market downturn, if markets price in all the doom and gloom we could read from demographics, it should plunge by 50% and never recover. If markets price in the possible counteracting forces, we should have a drop followed by a few years of positive gains, just like every other time since the 1920s, and have good positive returns on average, in the long run.
 
The bears and the bulls have plenty of material to convincingly defend their cases. Just don’t automatically buy a story just because it is convincing! The takeaway is that doom scenarios are just as possible as bull runs.
 
Stick to medium term projections (0-4 years) with a long run goal of maximizing returns according to your risk-return preferences, always keep the “structural aspects” in mind, and allocate your investments based on most-probable outcomes and rigorous research by competent people.
 
Steer clear of drama seekers looking for viewers and hype-driven ebook sales who have been calling the mother of all crashes for years and keep saying that they are right, but that it is “the system” that is messes up. These people have a broken record of saying that “things would be as they say if central banks and governments didn’t keep fiddling with markets.” Does that help you manage your money?
 
Don’t get me wrong: I think market analysis MUST integrate structural, long-term trend considerations to have a clear picture of the “long run canvas” on which to paint the short run analysis. But calls beyond 2 years, calls that are of the type “well it didn’t happen this time, but it is just a matter of time” are worthless, empty hand waiving… and costly. Cognitive bias is always a loser’s game, whether you are a bull or a bear. Be sure to read the follow-up story next week.
 
I hope you enjoyed this post and that things are a bit clearer now! Feel free to comment, share, and contact me. Cheers!
 
YourPersonalEconomist
contact@yourpersonaleconomist.com