Inventory costs and company earnings transfer intently collectively over lengthy horizons, a relationship confirmed by greater than a century of information compiled by Robert Shiller. This evaluation examines the power of that long-term linkage and checks whether or not adjustments within the earnings–value correlation provide perception into future inventory market returns.
The outcomes present that whereas earnings assist clarify market conduct over time, fluctuations within the correlation itself don’t present a helpful foundation for forecasting returns. The sections that comply with doc empirical patterns throughout a number of rolling intervals and assess the bounds of utilizing correlation measures as market-timing instruments. The findings may assist monetary advisors body long-term market conduct for shoppers in a grounded and intuitive manner.
What This Evaluation Goals to Make clear
I look at the long-term relationship between inventory costs and company earnings for 2 principal causes.
First, the findings provide an easy option to clarify inventory market conduct over lengthy funding horizons. I outline an extended horizon as greater than 10 years, which is a helpful minimal timeframe for retirement planning and for making asset allocation selections.
Second, after calculating the correlations between costs and earnings, I examined whether or not adjustments within the correlation over time may function a number one indicator of future returns. Particularly, I requested whether or not intervals of unusually low historic correlation had been adopted by stronger or weaker subsequent inventory market efficiency.
Correlation Outcomes
The evaluation makes use of month-to-month averages of the S&P Composite earnings-per-share and the S&P Composite value. The reported month-to-month earnings, inventory value, and returns information for the S&P Composite firms are based mostly on Shiller’s information from 1871 by December 2024.
Throughout a number of time intervals, the correlations between earnings and costs had been persistently excessive.
| Time Interval | Correlation |
| Full information set (01/1871 – 12/2024) | 0.977 |
| 100 years (01/1925 – 12/2024) | 0.974 |
| Publish-1940 Traders Act (08/1940 – 04/2024) | 0.973 |
| 50 Years (01/1975 – 12/2024) | 0.963 |
I selected widespread time intervals to look at the info and be aware the next:
- One place to begin is the 1940 Traders Act, used to check whether or not outcomes differed after investor protections and extra uniform accounting requirements had been launched. The distinction seems negligible.
- The previous 10- and 20-year intervals had been included to mirror what is usually thought of a typical retirement-planning horizon.
Correlation Modifications Over Time
The correlation between earnings and inventory costs does fluctuate over time, notably throughout shorter horizons such because the five-, 10-, and 20-year home windows. The rolling 50-year correlations additionally range, although inside a a lot narrower vary.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
The bottom rolling 50-year correlation occurred through the first half of the twentieth century, when the info collection reached 0.6. Given the backdrop of two world wars, the Nice Despair, and restricted market regulation previous to 1940, it’s notable that the correlation didn’t fall additional.

Variability elevated because the time horizon shortened. Within the rolling 20-year collection, correlations fell beneath 0.50 for a full decade between February 1918 and December 1928, and once more briefly in December 1948.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
The rolling 10-year correlations fell beneath zero throughout three intervals: on the finish of World Struggle I and World Struggle II, and through the excessive inflation period of the late Nineteen Seventies and early Eighties.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
Rolling five-year correlations naturally confirmed probably the most volatility, with deeper drops and extra frequent swings, together with a number of intervals of damaging correlation. Each the typical and median rolling five-year correlations had been decrease than these noticed over longer horizons.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
Does the Variability in Correlations Correspond with Returns?
To check whether or not variation within the earnings–value correlation has any predictive worth for inventory returns, we ran regressions of correlation ranges in opposition to subsequent annualized returns.
The R² between S&P Composite earnings and value from 1871 by 2024 could be very excessive at 0.95. Given the power of this long-term relationship—and the relative rarity of low-correlation intervals—it’s affordable to ask whether or not these intervals may operate as purchase or promote alerts. In different phrases, does variation within the earnings–value correlation assist predict future returns?
I evaluated this query throughout a number of rolling time horizons. The ensuing R² values — linking correlation ranges to subsequent annualized returns — had been far decrease than the R² between earnings and value themselves. For the rolling 10-year and five-year home windows, the R² fell near zero, indicating nearly no predictive relationship.
The rolling 50-year interval confirmed the strongest relationship with a R2 of 0.53.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
For the rolling 20-year home windows, the R² was 0.24, reflecting significantly extra variability.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
Variability elevated additional within the rolling 10-year collection, the place the R² fell to 0.06.

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
The rolling five-year intervals present no constant sample. R2 is sort of 0.0 (precise: 1.27E-07).

Supply: Robert J. Shiller S&P information; Archer Bay Capital LLC
General, I discovered no proof that adjustments within the earnings–value correlation predict future annualized returns. The information present that the 2 measures don’t transfer collectively in any significant manner for horizons shorter than 50 years.
Predictive Energy of Correlation
The sturdy long-term relationship between earnings and costs presents a transparent clarification for the rise and fall of inventory markets over prolonged intervals. It offers a easy and intuitive framework for understanding long-run fairness traits.
Nevertheless, the second aim – figuring out whether or not adjustments within the correlation may function a predictive measure for annualized returns – was not achieved. The proof means that different components past the earnings–value relationship drive the speed of change in annualized returns, though the 2 collection transfer intently collectively over lengthy horizons.
Key Takeaways
- Earnings and inventory costs transfer intently collectively over lengthy horizons. Greater than 150 years of Shiller information present a persistently sturdy relationship between the 2 collection.
- Shorter home windows introduce substantial noise. Correlations fluctuate meaningfully over five-, 10-, and 20-year intervals, reflecting wars, inflation shocks, and structural adjustments.
- Correlation power doesn’t indicate predictive energy. Shifts within the earnings–value correlation have little potential to forecast subsequent returns at horizons related to most traders.
- Solely the longest home windows present restricted explanatory energy. Even the 50-year regressions, with an R² of 0.53, provide solely modest perception, whereas shorter horizons fall near zero.
Earnings assist clarify long-term market conduct, however they don’t assist time the market.
The writer is a Registered Funding Advisor consultant of Archer Bay Capital LLC/Built-in Advisors Community – a SEC Registered Funding Adviser. The data contained herein represents Campbell’s unbiased view or analysis and doesn’t symbolize solicitation, promoting, or analysis from Built-in Advisors Community or Archer Bay Capital LLC. It has been obtained from or is predicated upon sources believed to be dependable, however its accuracy and completeness will not be assured. This isn’t meant to be a proposal to purchase, promote, or maintain any securities.
