The interaction between stock prices and corporate earnings has been extensively researched, notably through over a century of data analyzed by economist Robert Shiller. This relationship is critical for investors who seek to comprehend market trends over the long term. While the connection between these two elements is evident in the long run, examining fluctuations in their correlation can provide insights into potential future market performance.
This analysis will explore the strength of the relationship between earnings and stock prices, assess variations in correlation over time, and evaluate whether these fluctuations can serve as reliable indicators for predicting future stock market returns.
The long-term correlation between earnings and stock prices
To understand the long-term dynamics between stock prices and corporate earnings, a long-term horizon is defined as a period exceeding ten years. This timeframe is particularly significant for retirement planning and strategic asset allocation decisions. By calculating the correlation between earnings and prices during this period, we can assess how well earnings reflect market behavior.
Analyzing historical data
The analysis utilized monthly averages of the S&P Composite’s earnings-per-share alongside corresponding stock prices. This dataset, derived from Shiller’s extensive records spanning from 1871 to December, reveals a consistently high correlation across various timeframes. Notably, periods such as the aftermath of the 1940 Investors Act were examined to determine if this regulatory shift significantly altered the relationship between earnings and stock prices. Findings indicate that the introduction of investor protections did not result in a notable difference.
The examination of rolling periods, including ten- and twenty-year spans, highlighted that correlations tend to be robust over extended durations. However, as the timeframes shorten to five or ten years, the correlation fluctuates significantly, suggesting increased volatility in shorter intervals.
Variability in correlation and its implications
Throughout history, the correlation between earnings and stock prices has not remained static. For instance, during the tumultuous early decades of the 20th century, the correlation dipped to its lowest recorded level of 0.6. This period was marked by significant global events, including two world wars and the Great Depression, which undoubtedly influenced market stability.
Correlations in shorter timeframes
In contrast, when analyzing shorter rolling periods, correlations exhibited considerable variability. For example, during the rolling twenty-year period, correlations fell below 0.50 for an entire decade from 1918 to 1928, and again briefly in 1948. The ten-year rolling correlation even dipped below zero during critical events, such as the conclusion of World War I and the inflationary crisis of the late 1970s.
Short-term fluctuations became even more pronounced in the five-year rolling correlations, which often displayed extreme volatility and frequent shifts. This indicates that while earnings provide a solid foundation for long-term market behavior, they do not always correlate closely in shorter spans.
Does correlation variability signal future returns?
To explore whether variations in the earnings-price correlation can predict future stock returns, a series of regressions was conducted. The results indicated that while the long-term relationship between earnings and prices is robust, the predictive capability of correlation fluctuations is limited. For the rolling ten-year and five-year windows, the correlation values approached zero, indicating minimal predictive power.
Even for the rolling fifty-year period, which exhibited a stronger R² value of 0.53, the insights gained remain modest. Consequently, this analysis suggests that while earnings significantly explain long-term market behavior, they do not effectively serve as market-timing tools.
This analysis will explore the strength of the relationship between earnings and stock prices, assess variations in correlation over time, and evaluate whether these fluctuations can serve as reliable indicators for predicting future stock market returns.0