The Irony of Growth
How investors systematically err on growth forecasts
Robert Shiller won the Nobel Prize in part for his research documenting the fact that stock prices moved far more than fundamentals could justify: the dividend discount model was inadequate to explain the observed volatility of the stock market.
His observation inspired years of debate among academics as to the cause of this excess volatility. Some academics argued that the volatility could be explained by rational changes in the discount rate investors were applying to the market. Others argued that this volatility was a result of behavioral biases and limits to arbitrage.
But we have always been convinced by a simpler explanation: Market prices likely reflect investor predictions about future growth, and volatility is the natural consequence of the inevitable forecast errors. This explanation was first hypothesized by Mordecai Kurz at Stanford with a set of elegant mathematical theories, but the theory has largely failed to gain traction among empirically minded academics—until now.
A new set of papers by Pedro Bordalo, Nicola Gennaioli, Rafael La Porta, and Andrei Shleifer has identified strong empirical support for the idea that changing forecasts of future earnings growth, not wildly oscillating risk appetite or investor behavioral bias, explains the excess volatility puzzle.
Using analyst estimates, the authors create a synthetic, expectation-based index that explains a considerable degree of the excess volatility found by Shiller. The index assumes a constant discount rate and uses monthly data on stock market analyst forecasts for S&P 500 firms to explain departures from what could be rationally explained by a dividend discount model. The forecast inputs include expectations for annual dividends per share, earnings per share, and long-term earnings growth. Figure 1 shows this index (red) relative to the market (green) and the rational price (blue). Not only is the synthetic index sufficiently volatile, but it exhibits a high correlation with the market. The synthetic index matches both the frequency of the market and its direction. The data suggest that observed beliefs derived from analyst expectations may proxy the market as a whole.
Figure 1: Analyst-Derived Expectations vs. S&P 500 and Rational Price
Source: Expectations of Fundamentals and Stock Market Puzzles
This dynamic also holds for individual stocks. Previously, the same authors examined the returns to portfolios formed on long-term growth expectations and found that analyst forecasts of firms’ long-term earnings growth overreact to news about firm-level performance. Figure 2 shows the annual returns to decile portfolios based on these expectations. Notably, the decile comprising stocks with the lowest long-term growth prospects earned the highest average return of 15%. The decile comprising stocks with the highest long-term growth prospects earned the lowest average return of just 3%. The data shows that betting against analyst exuberance has been a good idea.
Figure 2: Returns to Portfolios Formed on Expectations for Long-Term Growth (1981–2015)
Source: Diagnostic Expectations and Stock Returns
Perhaps most importantly for investors, these growth predictions turn out to be wrong in predictable ways. Today we may not know what long-term growth rates are achievable, but we do know that there is a cyclicality to market expectations around realized and predicted growth. The data show that investors are ebullient when they should be bearish, and vice versa. Figure 3 shows how analysts tend to over-extrapolate their forecasts for long-term growth. The red line represents the predicted one-year change in long-term earnings growth, and the green line plots the error realized in those forecasts over a subsequent five-year period.
Figure 3: Forecasted Earnings Growth vs. Realized Forecast Error
Source: Expectations of Fundamentals and Stock Market Puzzles
Over the last 10 years, however, investing based on these insights would not have led to much success. Contrary to long-term analyses, stocks with exceptionally bullish long-term analyst prospects have outperformed their less prized peers. Tesla, Snap, Zoom, Peloton, DocuSign are examples of companies that rank in the top 25 for these variables today.
Historically, markets demonstrate a cyclical fallibility. Investors tend to over-extend when heading into bad times and exhibit excessive caution when heading into good times. This dynamic manifests as volatility and can be seen at the market level and the individual stock level. Analyst estimates for long-term growth serve as a reasonable sentiment barometer, and when market prospects appear too good to be true, they likely are.