Featured monthlies / September 2026
When Bonds Stop Diversifying
Bonds are supposed to cushion equity risk. A century of history shows that their ability to do so depends on the inflation environment.
Bonds are a “fair weather” diversifier.
Across the full study, stock-bond correlation changes sign when inflation crosses 2.5%. Switch between environments to see how returns and diversification change.
Many investors remain committed to bonds in the name of diversification. The familiar logic is simple: equities drive returns, while high-quality bonds provide a cushion when equities struggle. That cushion can help preserve purchasing power and support compounding.
Our study uses 10-year Treasuries to examine that relationship. Across a century of observations, bonds delivered positive real returns and negative stock-bond correlation in the low-inflation group. In the high-inflation group, both characteristics changed.
Start with the full history, then compare the bond-diversification era investors remember with the current high-inflation contrast.
Average real returns
Stock / bond correlation
Source: Recurrent research; figures reproduced from “Stock Bond Diversification Monthly.” Real returns are inflation-adjusted. 2026 is a partial year. See methodology and source tables below.
1998–2020 / The era investors remember
The bond hedge investors remember was the exception.
For more than two decades, low and anchored inflation coincided with unusually negative stock-bond correlation and strong real bond returns. That experience became the portfolio rule of thumb.
From 1998 through 2020, bonds were an unusually effective diversifier during a period overlapping with much of today’s allocators’ professional experience. Stock-bond correlation was strongly negative in both inflation groups, and bonds delivered positive average real returns when equities were negative. The longer history is much less consistent.
The post-2020 period therefore looks less like a permanent relationship breaking and more like a return to a pattern that has appeared repeatedly when inflation is elevated: higher stock-bond correlation and weaker real bond returns.
The exception became the expectation
1998–2020 taught investors to expect a bond hedge that history does not guarantee.
Average real bond returns when equities were negative in the high-inflation group illustrate how unusually supportive that period was.
1998–2020 / Low inflation
−0.491998–2020 / High inflation
−0.95Stock-bond correlation was strongly negative in both inflation groups.
| Measure | Low inflation CPI <2.5% | High inflation CPI ≥2.5% |
|---|---|---|
| Years included | 15 | 8 |
| Average real bond return | 4.7% | 0.9% |
| Real bond return when equities negative | 7.0% | 12.7% |
| Stock-bond correlation | −0.49 | −0.95 |
Source: Recurrent research. Inflation group definitions and figures follow the supplied study.
When protection
fails.
In the high-inflation portion of the century-long study, bonds frequently lost purchasing power. Natural resources exhibited a different pattern.
Frequency of negative real bond returns.
All observations when CPI ≥2.5%.Frequency of negative real bond returns when real equity returns were also negative.
All observations when CPI ≥2.5% and real equity returns were negative.Frequency of negative real natural resources returns.
All observations when CPI ≥2.5%.1927–YTD 2026 · High inflation defined as CPI ≥2.5%. The conditional bond statistic uses a different denominator from the other two statistics.
A different inflation environment.
In the 2021–2026 YTD sample, all six observations fall in the high-inflation group. Stock-bond correlation is +0.81.
Since the start of 2021, no completed calendar year has recorded December-to-December inflation below 2.5%, according to the study. Even on a monthly basis, CPI exceeded 2.5% in 88% of the observations since the end of 2020.
While equities delivered strong returns, bonds produced a weaker return stream and lost purchasing power on average. Natural resources delivered strong nominal and inflation-adjusted returns over the same period.
2021–2026 YTD
| Equities | +10.9% |
|---|---|
| 10-year Treasuries | −6.2% |
| Natural resources | +21.8% |
Six observations, including partial-year 2026. Source: Recurrent research.
Diversification across asset classes is not necessarily diversification across economic risks.
When inflation makes stocks and bonds behave similarly, exposure to a different economic return driver matters. The appropriate allocation is an investor decision. The historical case for considering natural resources is clear.
Explore the source tables
Methodology & interpretation
This page presents summary statistics from Recurrent’s September 17, 2026 research document. The controls select published groups; they do not recalculate the underlying study. Low inflation is CPI below 2.5%; high inflation is CPI at or above 2.5%. Real returns are inflation-adjusted; nominal returns are not. Average returns are the averages reported in the source, not compound annual growth rates.
The study uses broad equities, 10-year Treasuries and natural resources equities. A negative correlation indicates a tendency to move in opposite directions; it does not guarantee protection. A positive correlation does not mean two assets always move together. Natural resources equities can incur substantial losses.
The source labels observations as years and includes partial-year 2026. The 2.5% CPI split measures the inflation level; it does not independently measure whether inflation expectations are anchored. Different historical windows contain different numbers of observations.
The 1998–2020 exhibit focuses on real equity and bond returns. “N/A” reproduces the source designation for conditional results in 2021–2026.
Sources credited in the original document: Recurrent research; St. Louis FRED for BLS data and 10-year Treasury returns; Kenneth R. French Data Library, Dartmouth College, for equity returns. Summary figures follow the revised research document and Recurrent’s supplemental century-long nominal bond returns.