The Biggest White Lie Financial Professionals Tell
If you watch any financial media outlet for more than five minutes, you will hear the line, “The economy is expanding so it’s time to buy stocks.” It’s often said so much it’s considered a throwaway line that receives no pushback.
There’s just one problem – it’s not true, at least not in an actionable manner.
To prove my point, here’s a regression of annual returns of the S&P 500 and GDP growth from 1930 to 2017[i].
A quick and dirty explanation of the variables:
R-squared: Is often referred to as an accuracy measurement of the regression.
Prob > F (aka F-statistic): Explains the percentage the total regression output that was determined merely by chance.
P>|t| (aka p-value): Determines the percentage of the variable output that was determined merely by chance. Since this regression has one independent variable this figure will be the same as the Prob >F figure.
95% confidence interval: The upper and lower values at the 95% confidence level (p-value at 5%). Notice that zero is included in this interval, which means you cannot reject the null hypothesis that GDP growth has no significant relationship on S&P 500 returns.
The gold standard for F-statistic and P-values is 5% or less with a high R-squared; considering the first two are approximately 20% and the latter is less than 2%, this regression portends a weak relationship with limited explanatory value.
Perhaps there’s a very weak correlation between GDP growth and S&P 500 returns but nothing definitive enough to dictate investment policy.
Using the eyeball test
Another way to look at the data is to list the top ten years by GDP growth and S&P 500 returns and use the eyeball test.
As the chart below shows, investing in the top ten years of GDP growth would have given you a 70% chance of beating the median S&P 500 annual return. At the extremes -- the lowest GDP growth was 8.0% -- it appears there is a positive correlation with S&P 500 returns. Is this actionable in a world where 3.5% to 4% GDP growth is considered strong? Probably not.
More interesting is the top 10 years for S&P 500 returns. In seven out of the ten years GDP growth was lower than the annual median – five of which were years in which full-year GDP decreased.
In the long run, it is widely believed equity returns will trend GDP but changes in valuations and equity yield (defined as total dividends and net share buybacks) are larger drivers of short-term returns. Unless you’re planning on holding stocks for decades (you should) it's unlikely GDP returns will highly correlate to stock returns. Moreover, it's likely financial professionals are not referencing long-term returns, so it’s best to ignore statements that correlate equity returns to a growing economy.
i: Using quarterly returns would result in a different result, but may be more susceptible to lagging data – as in stock returns in period T+1 being affected by GDP growth in T. Using yearly figures with four GDP readings lowers this somewhat but still could suffer from lagging data. Fwiw, I did a vector autoregression at four lags and saw even less explanatory power. I didn’t include this to make the post as short as possible -- reach out if you want that data.