Analysis of historical data casts doubt on generally accepted correlation
There’s a reason why stock-picking investors consider earnings announcements as red-letter dates: since a company’s valuation represents its discounted future cash flows, any hint about its prospects for future revenue will help determine the fair price for that company’s stock.
At least, that’s how the theory goes.
In an article published by the CFA Institute, FactorResearch Managing Director Nicolas Rabener shared an analysis examining the correlation between U.S. stock-market returns and earnings data going back more than a century, which was obtained from Robert J. Shiller at Yale University.
“From 1904 to 2020, earnings growth and stock returns moved in tandem over certain time periods,” Rabener said, noting results based on five-year rolling returns calculated for both time series. “[H]owever, there were decades when they completely diverged, as highlighted by a low correlation of 0.2.”
The analysis was repeated with rolling-return windows of one year and 10 years; other attempts looked at real rather than nominal stock-market prices and earnings. In all cases, Rabener said the correlation between U.S. stock market returns and earnings growth was “essentially zero” over the last century.
Testing the possibility that stock prices hinged on expected rather than actual returns, Rabener described another analysis focusing on earnings growth for the succeeding 12 months, which assumed investors were perfect forecasters of U.S. stock earnings.
“[K]nowing the earnings growth rate in advance would not have helped these superinvestors time the stock market,” he said. “Returns were only negative in the worst decile of forward earnings growth percentiles. Otherwise, whether the earnings growth rate was positive or negative had little bearing on stock returns.”
Rabener extended the analysis to earnings growth and P/E ratios, with the expectation of “a strong positive correlation as investors reward high-growth stocks with high multiples and penalize low-growth stocks with low ones” – an explanation often offered for the extreme valuations of high-flying tech stocks.
But that relationship did not bear out in the data. Expected earnings growth over the following 12 months showed no relationship to the average P/E ratio, and higher forward growth actually resulted in P/E multiples slightly below the average.
“If the focus was current earnings, our explanation might be that an increase in earnings leads to an automatic reduction in the P/E ratio,” Rabener said. “But with forward earnings, these results are less intuitive.”
Those results, Rabener said, support the view that investors are irrational, and animal spirits matter at least as much as fundamentals. Many investors already ignore earnings data, he added, though he made a distinction between the recent millennial-fuelled surge in GameStop and momentum strategies pursued by hedge fund managers.
“[W]hile the former hardly seems like sound investing, the latter is a perfectly acceptable strategy that does not require any earnings data,” he said.