The Hidden Tax on Safety: US 3-Month Treasury Bills Lost to Inflation in 38% of Months Since 1948
In 37.7% of all months since 1948, the US 3-month Treasury Bill returned less than inflation — and in the 2010s, that figure reached 87%.
A complete 78-year dataset of rolled 3-month Treasury Bill real returns, classified into four real-rate regimes. The step-by-step breakdown is in our framework on Central Banks and Monetary Policy: Institutions, Rate Cycles, and Market…. Since 1948, the “risk-free” rate has delivered negative real returns in more than one-third of all months, with two distinct concentrations: the 1940s–50s post-war financial repression period and the 2010s zero-rate era.
The US 3-month Treasury Bill has yielded a monthly average of 3.75% in nominal terms and approximately −0.12% in real terms since January 1948. Over 938 monthly observations, the real rate has been negative in 354 months, or 37.7% of the sample. This page provides the complete rolled-return history, a four-regime classification of real-rate conditions, and a decade-by-decade decomposition of T-Bill performance against inflation.
A dollar rolled continuously in US 3-month Treasury Bills since January 1947 is worth approximately $1.47 today in real terms — a cumulative real return of +47% over 79 years. But the path was not linear: 37.7% of all months delivered negative real returns, the longest consecutive stretch was 62 months (November 2009 – December 2014), and the worst 10-year rolling real return was −17.6%, ending June 2022. Note: this dataset measures the real return of rolled T-Bills, not the total-wealth impact of holding cash, which depends on alternative asset returns during the same periods (see Methodology and Limitations).
Nominal T-Bill rate
CPI YoY inflation
Real rate
Regime (56th percentile)
- In 37.7% of all months since 1948 (354 of 938 observations), the US 3-month Treasury Bill yielded less than trailing 12-month CPI inflation. In the 2010s, this figure reached 87% — higher than the 67% recorded during the 1970s inflation decade.
- Contrary to the conventional framing, the most consistently negative-real-rate decade on record was not the 1970s but the 2010s, when average real T-Bill rates fell to −1.20% and only 13 months out of 120 delivered positive real returns.
- The longest uninterrupted negative-real-rate regime in the post-war record ran 62 consecutive months from November 2009 to December 2014, spanning the Fed’s zero-rate and QE policy stance. During this window, the purchasing power of rolled T-Bills declined by approximately 7.7%.
- The worst 10-year rolling real return was −17.6%, recorded for the decade ending June 2022 — the combined effect of post-GFC zero rates and the post-COVID inflation surge of 2021–2022.
- The full sample includes 938 monthly observations spanning four distinct real-rate regimes, eight decades, and twelve NBER-dated recessions. The dataset includes a real-rate regime classification, multiple cumulative real return indices (base years 1947, 1954, 1971, 1981), and forward CPI YoY for reproducible real-rate computations.
950 observations · Monthly · Jan 1947 – Feb 2026 · CC BY 4.0 ·
Methodology ·
Cite this dataset
Monthly obs. with real rate
Months with negative real rate
Longest neg. run (months)
Peak nominal (May 1981)
Most negative real (Jan 1948)
Cumul. real return since 1947
Chart: Cumulative Real Return of Rolled US 3-Month Treasury Bills, 1947–2026
Purchasing power of $100 invested in rolled T-Bills, January 1947 = 100
Real returns accumulated unevenly: negative through 1948–1951, flat through the 1950s–1970s, sharply positive during the Volcker era, and back to negative through the 2010s ZIRP regime and 2021–2022 inflation surge.
A dollar rolled in T-Bills since 1947 has generated a +47% cumulative real return, but two-thirds of that gain was generated during the single 21-year Volcker-to-Greenspan window of 1981–2002. Outside this period, T-Bill real returns have clustered around zero or below.
Sources: Federal Reserve (TB3MS), Bureau of Labor Statistics (CPIAUCSL). Chart: Eco3min Research.
How to Read This Chart
The vertical axis shows the cumulative purchasing power of $100 invested in January 1947, assuming monthly rolling at the prevailing 3-month T-Bill secondary-market rate and deflated by CPI-U. The scale is logarithmic, which means equal vertical distances represent equal percentage changes — appropriate given the data range spans from roughly 85 to 180 over the sample.
Background colored bands classify each month into one of four real-rate regimes: deeply negative (<−2%), negative (−2% to 0%), neutral (0% to +1%), and positive (>+1%). The visual concentration of red and amber bands during 1948–1952, 1973–1980, and 2010–2022 makes the episodic nature of real-return destruction immediately visible. For the companion analysis of yield-curve regime conditions during these periods, see our US real interest rates history study.
The Function of T-Bills and the Limits of “Risk-Free”
The dominant framing of US Treasury Bills in textbooks and institutional practice treats them as the “risk-free rate” — the baseline against which all other asset returns are benchmarked. This framing is accurate in a narrow technical sense: T-Bills issued by the US Treasury face no credit risk under the going-concern assumption Demand at this short end now reflects a new structural buyer cohort — see how stablecoin issuers became marginal buyers at the short end of the curve., and their short duration eliminates interest-rate risk on any single held-to-maturity position.
The data reveals a different picture at the level of realized purchasing power. Over the full sample of 938 months from January 1948 to February 2026, the monthly average real return on rolled T-Bills is −0.12%. The distribution is bimodal: in 355 months (37.8%), real returns were negative; in 583 months (62.2%), they were positive. The cumulative real return from January 1947 stands at +47.3% as of February 2026 — an annualized compound real rate of approximately 0.49% per year.
T-Bills are not a long-horizon investment vehicle. The “rolled T-Bill” convention used in this dataset is an analytical construct for measuring the real return of a cash-equivalent strategy; it is not a description of how T-Bills are actually used by households, corporations, or institutional investors. In practice, T-Bills function as short-term cash management vehicles for liquidity buffers, collateral pools, and bridge financing — not as multi-decade wealth accumulation instruments.
What this dataset does not measure. This analysis tracks the real return of a cash position rolled in T-Bills. It does not measure total-wealth outcomes relative to alternative strategies (equities, long-duration Treasuries, real estate, gold) during the same windows. Over the same 1947–2026 span, the S&P 500 total return index delivered a cumulative real return of more than 6,000% — two orders of magnitude above T-Bills.The theoretical foundation for why the T-Bill real return has hovered near zero was formalized by Reinhart and Sbrancia (2011) in “The Liquidation of Government Debt,” which documents financial repression as a deliberate policy regime characteristic of sovereign-debt-heavy periods.
The “risk-free rate” is risk-free with respect to nominal default and short-horizon price fluctuation. It is not risk-free with respect to the preservation of purchasing power — a distinction that is measurable in the historical record and material over multi-year horizons.
Two Periods of Negative-Real-Rate Concentration
Negative-real-rate months are not uniformly distributed across the sample. Two distinct concentrations account for the large majority of cumulative real erosion: the post-war financial repression regime of 1948–1951, and the post-GFC zero-rate-plus-inflation regime of 2009–2022.
Post-War Financial Repression (1948–1951)
The rolled-T-Bill purchasing power index fell from 100 in January 1947 to 85.6 in December 1951, a cumulative real loss of 14.3%. The mechanism was explicit: the Federal Reserve operated under the 1942 Treasury accord, which pinned short-term rates near 0.38% to manage the wartime debt stock, while inflation ran at double-digit rates in 1946–1948. The Treasury–Fed Accord of March 1951 ended this regime, but real returns remained mostly negative through 1951 as the prior capital loss had already accumulated.
ZIRP Plus Post-COVID Inflation (2009–2022)
The second concentration is more recent and is now the most consistent negative-real-rate period in the post-war record. Between November 2009 and December 2014, the T-Bill real rate was negative for 62 consecutive months — the longest such run in the sample. A second leg, from March 2020 to mid-2023, coincided with the COVID-era emergency-rate regime and the subsequent inflation surge that peaked at 9.1% CPI YoY in June 2022. The worst 10-year rolling real return in the entire sample, −17.6%, was recorded for the window ending June 2022.
A legitimate analytical qualification is that the 2010s negative-real-rate regime was materially shallower in magnitude than the 1970s episode. The average real rate during the 1970s was −0.80%, compared with −1.20% in the 2010s — but the 1970s produced several months of deeply negative real rates below −4%, while the 2010s real rates were persistent but typically clustered between 0% and −2%. The 2010s were more uniform in negativity; the 1970s were more severe when negative. Both patterns appear in the dataset, and depending on whether consistency or severity is emphasized, different decades emerge as the “worst” period for T-Bill real returns. This page presents both metrics to allow readers to judge for themselves (see decade decomposition below).
Decade-by-Decade Decomposition
Decomposing the full sample by decade reveals seven distinct real-rate regimes, each with identifiable macro drivers. Three decades produced strongly positive cumulative real returns (1960s, 1980s, 1990s); three produced cumulative real losses (1950s, 1970s, 2010s); and two produced near-flat results (2000s, 2020s partial).
| Decade | Avg. nominal T-Bill (%) | Avg. CPI YoY (%) | Avg. real rate (%) | % months neg. real | Decade cumulative real return (%) |
|---|---|---|---|---|---|
| 1950s | 2.00 | 2.07 | −0.07 | 41% | −2.0% |
| 1960s | 3.98 | 2.33 | +1.65 | 0% | +16.1% |
| 1970s | 6.29 | 7.09 | −0.80 | 67% | −8.2% |
| 1980s | 8.82 | 5.56 | +3.26 | 8% | +46.6% |
| 1990s | 4.85 | 3.01 | +1.84 | 6% | +21.4% |
| 2000s | 2.69 | 2.57 | +0.12 | 43% | +1.6% |
| 2010s | 0.57 | 1.77 | −1.20 | 87% | −11.1% |
| 2020s (through Feb 2026) | 2.78 | 3.91 | −1.14 | 54% | −6.3% |
Two measures of “worst decade” for T-Bill real returns yield different answers: by cumulative decade real return, the 2010s (−11.1%) was the worst full decade since 1950; by average real rate, the 2010s at −1.20% edges out the 1970s at −0.80%; by share of negative-real months, the 2010s at 87% decisively exceeds the 1970s at 67%. By severity of the worst individual months, the 1940s–1970s produced deeper negative readings.
The signature chart below displays all five metrics across decades simultaneously. The 2010s row is highlighted because it scores worst on the consistency metric (share of negative-real months) and the total-return metric; the 1980s row is, by the same criteria, the strongest decade on record. For a companion analysis of the Fed policy decisions that produced these regimes, see our Federal Funds Rate track record study.
Real-Rate Regime Classification: A Decade-by-Decade View
US 3-Month T-Bill Across Decades: Five Metrics, Eight Regimes
The 2010s produced more consistently negative real T-Bill returns than the 1970s did — 87% of months versus 67%.
Sources: Federal Reserve (TB3MS), Bureau of Labor Statistics (CPIAUCSL). Chart: Eco3min Research.
112 months (11.9%). Characteristic of wartime/post-war financial repression, early 1970s inflation, and 2021–2022 post-COVID inflation surge.
243 months (25.9%). The modal “mild repression” state. Dominant in late 1970s, early 2000s, most of the 2010s ZIRP era.
147 months (15.7%). Transitional regime. Characteristic of early 1960s and mid-2000s. Real returns marginal in either direction.
436 months (46.5%). The plurality state. Concentrated in 1960s, 1980s Volcker-to-Greenspan high-real-rate era, and late 1990s.
Horizon Sensitivity: Rolling Real Returns at 1, 5, 10, and 20 Years
The decade-level view aggregates by calendar periods. A complementary analytical lens is the distribution of rolling real returns at standard investment horizons. This addresses a legitimate question: if T-Bill real returns have been negative in 38% of single months, how often are they negative over horizons relevant to actual cash-management decisions?
| Horizon | n | Median real return | Worst | Best | % windows with negative return |
|---|---|---|---|---|---|
| 1 year | 927 | +0.95% | −7.90% | +7.31% | 36% |
| 5 years | 879 | +3.56% | −12.70% | +31.86% | 37% |
| 10 years | 819 | +10.55% | −17.58% | +50.11% | 31% |
| 20 years | 699 | +21.68% | −23.16% | +83.04% | 12% |
At the 20-year horizon, 88% of rolling windows produced positive real returns — the long-run probability of real preservation is meaningful. But 12% of 20-year windows still produced negative real returns, with a worst-case observation of −23.2%. The worst 1-year real return observed was −7.9%, the worst 5-year was −12.7%, and the worst 10-year was −17.6%.
Methodological note: Rolling windows overlap, which inflates the effective number of independent observations and should be interpreted accordingly. All windows are computed from the cumulative real return index; forward-looking windows from recent dates (where the window has not yet elapsed) are excluded. Percentile figures are empirical and not distributional.
Past distributions are not predictive of future outcomes. Horizon-conditional statistics describe historical patterns, not expected returns.
- ▸ Real rate at +1.17% (Feb 2026): a decline below 0% would reclassify the regime to “Negative.” The last transition from Positive to Negative occurred in March 2020; the last sustained return to Positive occurred in August 2023.
- ▸ T-Bill nominal at 3.60%: a sustained move below CPI YoY (currently 2.43%) would mark the resumption of mild financial repression. See our companion Federal Funds Rate history and Core CPI inflation datasets for the policy transmission channel.
- ▸ Next key release: CPI for March 2026 (BLS release). Historically, shifts in real-rate regime have been driven more by inflation surprises than by T-Bill rate changes, given the shorter response lag of the former.
Historical Turning Points
January 1948 — Deepest Monthly Negative Real Rate in Sample
In January 1948, the 3-month T-Bill yielded 0.97% against CPI YoY of 10.24%, producing the deepest negative real rate in the post-war sample: −9.27%. This episode reflects the lagged inflation pass-through from removed wartime price controls, while the Treasury–Fed Accord that would eventually allow short rates to rise was still three years away.
May 1981 — Peak Nominal T-Bill Rate, 16.30%
In May 1981, the nominal T-Bill rate reached 16.30% — the highest in the full series going back to 1934. Simultaneously, CPI YoY was 9.78%, producing a positive real rate of +6.52%. This was the second-highest real-rate reading of the sample (exceeded only briefly in August 1983, at +6.88%). The Volcker Fed’s willingness to maintain deeply positive real rates through 1985 inaugurated a 21-year regime during which T-Bills delivered strongly positive real returns.
November 2009 – December 2014 — Longest Negative-Real-Rate Run
For 62 consecutive months, the real T-Bill rate was below zero — the longest continuous negative-real-rate regime in the sample. The average nominal T-Bill rate during this window was 0.07%; the average CPI YoY was 1.83%; the average real rate was −1.76%. The mechanism was mechanical: the Fed’s zero-interest-rate policy held short rates at the zero lower bound while inflation, though subdued by historical standards, exceeded the policy rate every month.
June 2022 — Worst 10-Year Rolling Real Return, −17.6%
For the 10-year window ending June 2022, rolled T-Bills produced a cumulative real return of −17.6% — the worst 10-year outcome in the 78-year sample. The window spans the ZIRP era through to the peak of the post-COVID inflation surge. Headline CPI peaked at 9.06% YoY that same month, while the T-Bill rate was only 1.49%, producing a real rate of approximately −7.6%.
February 2026 — Current Observation
The T-Bill rate stands at 3.60%, CPI YoY at 2.43%, and the real rate at +1.17% — classified as “Positive” (above +1.0%). The current reading ranks at the 56th percentile of all monthly real-rate observations since 1948. The most recent regime transition was the Fed’s rate-cut cycle that began in September 2024; since then, the nominal T-Bill rate has declined from 5.20% to 3.60% while inflation has receded from 2.60% to 2.43%. The net effect on the real rate has been a decline from approximately +2.6% to +1.17%, but the real rate remains solidly positive. For a fuller view of the current monetary cycle, see our Net Liquidity Index analysis.
Methodology
The dataset merges two monthly series from primary public sources: the 3-month T-Bill secondary-market rate (discount basis) from the Federal Reserve H.15 release, and the Consumer Price Index for All Urban Consumers (CPIAUCSL, seasonally adjusted) from the Bureau of Labor Statistics. The CPI series was interpolated for October 2025, which was not published because of the government shutdown that disrupted BLS data collection; the value is estimated via linear interpolation between September 2025 and November 2025 readings.
Core Formulas
monthly_inflation[t] = cpi_index[t] / cpi_index[t-1] – 1
real_return_m[t] = (1 + monthly_nominal[t]) / (1 + monthly_inflation[t]) – 1
cumul_real[t] = 100 * ∏(k=0..t) (1 + real_return_m[k])
Real-Rate Regime Classification Algorithm
elif real_rate < 0.0% : regime = “Negative”
elif real_rate < +1.0%: regime = “Neutral”
else : regime = “Positive”
Threshold sensitivity: Shifting the “negative” threshold from 0% to −1% reduces the share of “negative or worse” months from 37.8% to 24.0%. Shifting it to +1% increases the share to 53.5%. The 2010s retain the highest negative-or-worse share across all three threshold variants (87% at 0% cutoff, 72% at −1%, 93% at +1%), confirming that the core finding about the 2010s is robust to threshold choice.
Literature anchor. The framework of multi-regime real-rate analysis in the US follows Reinhart and Sbrancia (2011) and Reinhart and Rogoff (2009), both of which document financial-repression regimes using similar real-rate definitions. This dataset extends their US analysis through 2026 and introduces the four-bucket regime classification as a reproducible alternative to subjective periodization.
Dataset Design
| Variable | Type | Unit | Source | Calculation |
|---|---|---|---|---|
| date | str | YYYY-MM-DD | — | first of month |
| tbill_3m_nominal | float | % | FRED TB3MS | direct |
| cpi_index | float | 1982-84=100 | FRED CPIAUCSL | direct (Oct 2025 interpolated) |
| cpi_yoy | float | % | derived | (cpi[t] / cpi[t-12]) – 1 |
| real_rate | float | % | derived | tbill_3m_nominal – cpi_yoy |
| cpi_mom | float | dec. fraction | derived | cpi[t] / cpi[t-1] – 1 |
| real_return_m | float | dec. fraction | derived | (1+nom_m)/(1+cpi_mom) – 1 |
| nominal_return_m | float | dec. fraction | derived | tbill_3m_nominal / 12 / 100 |
| cumul_nominal_1947 | float | index | derived | 100 × ∏ (1+nom_m), base Jan 1947 |
| cumul_real_1947 | float | index | derived | 100 × ∏ (1+real_m), base Jan 1947 |
| cumul_real_1954 | float | index | derived | 100 × ∏ (1+real_m), base Jan 1954 |
| cumul_real_1971 | float | index | derived | 100 × ∏ (1+real_m), base Jan 1971 |
| cumul_real_1981 | float | index | derived | 100 × ∏ (1+real_m), base Jan 1981 |
| real_rate_regime | str | category | derived | see algorithm above |
| negative_real_rate_flag | int | 0/1 | derived | 1 if real_rate < 0 |
| decade | int | year | derived | (year // 10) × 10 |
| recession_flag | int | 0/1 | NBER | 1 if in NBER recession |
Python Reproduction Code
# Reproduce this dataset from primary sources import pandas as pd from fredapi import Fred fred = Fred(api_key='YOUR_KEY') tb = fred.get_series('TB3MS').to_frame('tbill_3m_nominal') cpi = fred.get_series('CPIAUCSL').to_frame('cpi_index') df = tb.join(cpi, how='inner').sort_index() # Interpolate Oct 2025 (shutdown gap) df['cpi_index'] = df['cpi_index'].interpolate() # Derived columns df['cpi_yoy'] = df['cpi_index'].pct_change(12) * 100 df['real_rate'] = df['tbill_3m_nominal'] - df['cpi_yoy'] df['nom_m'] = df['tbill_3m_nominal'] / 12 / 100 df['cpi_mom'] = df['cpi_index'].pct_change(1) df['real_m'] = (1 + df['nom_m']) / (1 + df['cpi_mom']) - 1 # Cumulative real index, base Jan 1947 = 100 df['cumul_real_1947'] = 100 * (1 + df['real_m']).cumprod() # Regime classification def classify(r): if r < -2.0: return 'Deeply Negative' elif r < 0.0: return 'Negative' elif r < 1.0: return 'Neutral' else: return 'Positive' df['regime'] = df['real_rate'].apply(classify)
Dataset Download & Reproducibility
950 observations · Monthly · January 1947 – February 2026 · Licensed under CC BY 4.0.
Data Sources & References
- Primary Board of Governors of the Federal Reserve System (US), 3-Month Treasury Bill Secondary Market Rate, Discount Basis [TB3MS], retrieved from FRED, Federal Reserve Bank of St. Louis, April 2026.
- Primary U.S. Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers: All Items in U.S. City Average [CPIAUCSL], retrieved from FRED, Federal Reserve Bank of St. Louis, April 2026.
- Primary National Bureau of Economic Research, US Business Cycle Expansions and Contractions (recession dating reference), reference edition April 2026.
- Research Reinhart, Carmen M., and M. Belen Sbrancia (2011). “The Liquidation of Government Debt.” BIS Working Papers No. 363, March 2011.
- Research Reinhart, Carmen M., and Kenneth S. Rogoff (2009). This Time Is Different: Eight Centuries of Financial Folly. Princeton University Press.
- Research Ibbotson, Roger G., and Peng Chen (2003). “Long-Run Stock Returns: Participating in the Real Economy.” Financial Analysts Journal, 59(1), 88–98.
- Reference Federal Reserve Bank of St. Louis, FRED Blog: “The cost of owing” (2018). Background on T-Bill rate uses as a benchmark.
Methodological Limitations
- The rolled-T-Bill return convention assumes costless monthly rollover at the prevailing secondary-market rate. Actual T-Bill investors face bid-ask spreads and transaction costs, which would marginally reduce realized real returns.
- CPIAUCSL methodology has evolved significantly since 1947. Changes include the 1983 rental-equivalence revision for owner-occupied housing, the 1999 geometric-mean aggregation, and ongoing item-substitution adjustments. Pre-1983 CPI readings are thus not strictly comparable to post-1983 readings at the item level.
- Monthly real returns are computed using contemporaneous monthly inflation (cpi_mom), not trailing YoY inflation. This produces a more accurate realized return but introduces month-to-month noise. The regime classification uses YoY inflation (cpi_yoy), which is the market-standard convention.
- Rolling-horizon statistics (1Y, 5Y, 10Y, 20Y) use overlapping windows. Statistical significance tests on these distributions would need to correct for autocorrelation; the percentile and median figures reported here are descriptive.
- The October 2025 CPI reading was linearly interpolated between September and November 2025 because of the government shutdown gap. This affects approximately one observation of the real-rate series and is immaterial to the multi-decade findings.
- The sample starts in January 1947 because CPIAUCSL (seasonally adjusted) begins that month. A longer pre-1947 T-Bill series exists (TB3MS back to 1934), but cannot be deflated with comparable CPI data on the same basis.
Frequently Asked Questions
What is the current real rate on US 3-month Treasury Bills?
As of February 2026, the 3-month T-Bill rate is 3.60% and CPI YoY is 2.43%, producing a real rate of +1.17%. This is classified as “Positive” (above +1.0%) in the Eco3min four-regime system and ranks at the 56th percentile of all monthly real-rate observations since 1948.
How often has the 3-month T-Bill delivered negative real returns historically?
In 355 out of 938 monthly observations since January 1948 (37.8% of the sample), the T-Bill rate was below CPI YoY inflation. The distribution is bimodal: two distinct concentrations in the sample (1948–1951 and 2009–2022) account for most of the negative-real months. The longest continuous stretch was 62 months, from November 2009 to December 2014.
Were the 1970s or the 2010s worse for T-Bill real returns?
It depends on the metric used. By share of negative-real months, the 2010s (87%) clearly exceeded the 1970s (67%). By average real rate, the 2010s (−1.20%) was slightly worse than the 1970s (−0.80%). But by severity of the worst individual months, the 1970s produced deeper negative readings (several months below −4%) than the 2010s (which stayed mostly in the 0% to −2% range). By cumulative decade real return, the 2010s at −11.1% was modestly worse than the 1970s at −8.2%.
Does this analysis account for taxes or transaction costs?
No. The cumulative real return figures are gross of all taxes and transaction costs. Federal income tax on T-Bill interest would materially reduce after-tax real returns, particularly during high-nominal-rate periods like the late 1970s and early 1980s when pre-tax nominal rates above 10% produced substantial tax liabilities. State-tax treatment of T-Bill interest varies. Readers computing personalized after-tax figures should deduct their applicable federal and state marginal rates from the nominal-return series in the CSV.
Can this dataset be used to compare T-Bills against equities or gold as a long-term investment?
Not directly. This dataset measures the real return of a rolled-T-Bill cash position in isolation. It does not benchmark T-Bills against alternative strategies (equities, long-duration Treasuries, real estate, gold, or portfolios thereof) during the same periods. Such comparative analysis requires alternative-asset total return series deflated by the same CPI, which are not included here. As a reference point, the S&P 500 total return index delivered a cumulative real return of more than 6,000% over the same 1947–2026 span — roughly two orders of magnitude above rolled T-Bills. For an analytical framework on equity valuation and real rates, see our Real Interest Rates vs CAPE Ratio study.
How robust is the regime classification to the threshold choices?
The −2% / 0% / +1% thresholds in the four-regime classification were chosen to align with the empirical distribution of observations. Shifting the “negative” threshold from 0% to −1% reduces the share of “negative or worse” months from 37.8% to 24.0%; shifting it to +1% increases the share to 53.5%. Across all three variants, the 2010s retain the highest share of negative-or-worse months (87%, 72%, and 93% respectively), confirming that the core ranking is robust to threshold choice. The decade ranking on average real rate is also stable across these threshold variants.
Cite This Dataset
Related Eco3min Research
Last updated — 10 May 2026
Disclaimer – Financial Information: The analyses, commentary, and content published on eco3min.fr are provided for informational and educational purposes only. They do not constitute investment advice or a solicitation to buy or sell financial instruments. Past performance is not indicative of future results. All investment decisions involve risk and are the sole responsibility of the reader.