Break your clients of market prediction addiction
After the wild ride 2020 took us on, it’s only natural that clients want to know what to expect from the stock market in 2021. When people ask how I think stocks and bonds will perform next year, I answer that if I could predict future market performance, I’d be richer than Jeff Bezos.
Since I’m not, I tell them about my philosophy when it comes to predicting stock movement. It’s one that’s served me in years of relative calm and in years of crisis — one that may not yield instant gratification and the rush of optimism we all crave after such a rocky year but one that can provide a realistic sense of investing and risks.
Understanding the past — looking ahead
This year of the pandemic is one I’d love to forget as unemployment surged, the economy tanked, the budget deficit mushroomed to over $3 trillion and social unrest increased as our nation’s political divide grew wider than the Grand Canyon.
As often happens amid such worldwide turmoil, stocks plunged — nearly 35% in the 33 days between Feb. 19 and March 23. But then the unexpected happened. Despite consistently bleak economic news, the U.S. stock market soared by 20.82% for the year, as measured by the total return (including dividends) of the Wilshire 5000 Total Stock Index.
I give my clients two explanations for this phenomenon. The stock market valuation is based on the future cash flows discounted for an expected rate of return. Clearly the near-term future cash flows of the aggregate of all stocks decreased and, while there were some winners (online retailers) and losers (brick-and-mortar retailers), COVID-19 didn’t help overall future long-term cash flows. But interest rates plunged to near zero and the discount rate used is the risk-free rate plus the market risk premium. Because the risk-free rate declined by more than the cash flow projections, valuations increased.
But the second, and by far the strongest, of the two explanations is that markets regularly fool us. Though this is less emotionally satisfying, it happens to be true.
The key lesson I draw from 2020 is: If we can’t explain the past, just think of the futility of predicting the future.
Appealing — if dubious — forecasts
As I’m only human, I like looking at optimistic U.S. stock forecasts such as this summary of strategists’ predictions of the S&P 500’s 2021 movement, published Dec. 17 by Yahoo Finance. Here are some selected predictions:
Piper Sandler +14%
JP Morgan +19%
LPL Financial +2.6%
Deutsche Bank +8%
These and more forecasts predict an up stock market in 2021. These predictions were just of the S&P 500 index (carving out dividends) and were made before the end of the year, but I have enough information to make my point. We all want precise, bullish stock predictions by top analysts and these are especially comforting because bonds are yielding close to zero with the possibility of negative rates as we are seeing in parts of Europe.
Still, I’d counsel skepticism for two reasons. First, there is a tendency for recency bias to make us predict the recent past of strong market returns. Second, big financial institutions tend to predict strong stock returns, which in turn argue for greater equity exposure now that bonds yield so little. It’s pretty hard to charge clients a lot of money to earn a gross return (before fees) of about 1% on high-quality bonds. Even the lowest stock prediction of a 2.6% return is better than the current yield on bonds.
So, when it comes to market forecasts, I tell clients the more appealing they seem, the more dangerous they typically are.
So what do I tell my clients?
While not nearly as emotionally compelling as point forecasts, probabilistic forecasts, which typically give a midpoint as well as an upper and lower band of returns in a reasonable best- and worst-case scenario from Monte Carlo simulations, are far more useful. My favorite probabilistic forecast for the coming year happens to be Vanguard’s 10-year return in its Economic and Market Outlook for 2021.
In this illustration, note how the firm gives a midpoint expectation and ranges with a 50% and a 90% likelihood of occurring. And rather than predicting a return for one particular year, it gives a longer term 10-year annualized projected return. So, for example, Vanguard is predicting an annualized return for U.S. stocks of 4.7%. The firm believes there is a 50% probability it will earn between 2.0% and 7.4% annually and a 90% probability U.S. stocks have returns of -2.1% to +11.6%. There is a 5% probability of losing more than 2.1% or gaining more than 11.6%.
Vanguard clearly thinks there is a value premium. Yet what really sticks with me are the probabilistic returns of international compared to U.S. stocks. Vanguard is predicting a midpoint return of 8%, which is 3.3 percentage points annually more than U.S. stocks. Vanguard says the sources of this excess return over U.S. stocks are lower valuation, higher international dividend yield, and a weaker dollar, partially offset by lower international earnings growth (more value).
However, Vanguard’s downside forecast for international stocks is a positive 1.6% annually, or a 3.7 percentage points higher return, than the downside of U.S. stocks. What I understand that to mean is that international has a higher expected return than U.S. with far less risk, at least at the 5% downside probability.
If I believed this, I’d say markets are pretty darn inefficient (weak-form inefficiency) and be recommending only international stocks to clients. I’m more than a bit skeptical, however.
A Vanguard spokesperson disagreed with my assessment, saying “The U.S. and International equity box and whisker charts are independent from one another, in that we are summarizing the probability distribution of each set of simulations. There is no surefire way to calculate the probability of U.S. outperformance based on that visual alone, due to the presence of correlation between the two-return series.”
Vanguard’s prediction of returns for intermediate-term bonds was pretty close to current yields, which I think is the best predictor of future returns. It’s certainly much better than the historic track record of economists’ dismal predictions of rates.
While I may not completely agree with Vanguard’s forecasts, the probabilistic nature, combined with both logic and a longer term view, are more useful and create dialogue you can use with your clients.
My predictions for 2021
Before I start, let me say I agree with Vanguard that making longer-term forecasts are generally more useful in understanding what may happen to a portfolio over time. But there are some benefits to understanding what may happen over a given year.
To explain this, let’s look at U.S. stocks. Vanguard predicts a 10-year expected annualized geometric return of 4.7% with a -2.1% downside at the five-percentile probability level.
However, in any given year, the downside could be much worse. According to my estimates, using Vanguard’s annual standard deviations, the 4.7% geometric return translates to roughly a 6% annualized arithmetic return (the higher the standard deviation, the greater the variation between the annual arithmetic return and the long-run geometric return). Using a normalized distribution, this translates to roughly a one-year return between a loss of 22% and gain of 34% using a 90% confidence level.
The reason it’s important is that the client may not stay the course if stocks lose 22% or more in one year. And I think risks are even higher.
So here is what I’m telling clients might happen over the next year. In the accompanying chart, I’m predicting one-year returns as well as annualized standard deviations. I’m using higher-standard deviations than the past for stocks because I believe volatility is here to stay. Finally, I’m using two standard deviations to get a 95% confidence level rather than the 90% Vanguard used.
Though I’m agreeing with Vanguard that international should outperform U.S., it’s not a free lunch, just compensation for taking on more risk. But I make it clear to the client that they should be prepared to lose 33% in U.S. stocks and 36% in international. Perhaps even more since a 95% confidence means performance should be outside of these ranges about once every 20 years. Further, I’m using a normal distribution and sometimes markets have fat tails.
My predictions won’t land me on CNBC with the talking heads but long-run predictions like Vanguard’s and short-term predictions such as mine give a more realistic sense of expectations and risks. Then we can talk about consequences of bad outcomes in terms of what it means to their lifestyle and dreams. If your client wants more precision, educate them on the dangers of the all-too-human addiction to predictions.