Regime Returns
Historical asset returns conditional on macro regime and trajectory — and where the current reading sits among its closest historical analogs.
What it is
This page extends the regime quadrant chart into a conditional return study. Each calendar month since March 1967 is assigned to one of four macro regimes using the same two-axis framework: CFNAI (growth relative to trend) on the x-axis, Core PCE inflation Z-scored against the 1991–2019 post-Volcker baseline on the y-axis. The distributions of subsequent 3-, 6-, and 12-month returns for US equities, long-term government bonds, T-bills, and commodities are tabulated across those regime states.
A second layer conditions on trajectory — the 6-month change in CFNAI and Core PCE Z-score defines a direction vector discretized into eight compass points plus a stationary category. The trajectory tables ask whether the same quadrant classification at different angles of approach produces meaningfully different forward return patterns.
Equity returns are from the Kenneth R. French Data Library (US value-weighted market total return, July 1926+). Bond and T-bill returns are from Welch & Goyal (2008), updated through 2024: long-term government bonds use Ibbotson SBBI data through 2021, Bloomberg indices from 2022. Commodity returns use the Bloomberg Commodity Index total return (Bloomberg export, February 1990+).
Why it matters
Most asset allocation frameworks start with the business cycle — expanding or contracting — and adjust expected returns accordingly. Adding the inflation axis makes explicit a distinction the business cycle does not encode. During a Disinflationary Expansion, long-duration bonds and equities can both perform well because the central bank is accommodative and real rates are declining. During a Boom, credit overextends and sets up the subsequent reversal. During Stagflation, neither traditional equity nor nominal fixed income reliably provides cover because the monetary authority faces conflicting objectives. The pattern is stable enough across cycles to be informative even if it is never predictive at the specific-date level.
The trajectory layer adds whether conditions are improving or deteriorating. A Boom reading arriving from Disinflationary Expansion (moving northeast on the quadrant chart) is structurally different from one departing toward Stagflation (moving northwest). These are not academic distinctions — the 1978–1980 episode and the 2005–2007 episode occupied similar quadrant positions but had different trajectories and very different outcomes for credit and equity. The goal here is not to predict the next 12 months but to identify which historical episodes most resemble the current configuration.
The explicit framing throughout is descriptive, not predictive. The tables present historical base rates. Thin cells are acknowledged as such. The honest use of this framework is as a conditioning variable for scenario analysis.
Historical returns by regime
Key findings
- Equity returns diverge sharply: Disinflationary Expansion averages 15.0% over 12 months versus 5.6% for Stagflation — a 9.4pp cross-regime spread that holds at every horizon from 3m through 12m.
- Stagflation real return: 5.6% nominal but only 2.7% real — concurrent inflation claims roughly half the apparent equity gain.
- Within Stagflation, trajectory direction matters as much as the quadrant label: months where inflation is still rising (N, n=10) return 5.5%; months where it has peaked and begun falling (S, n=16) return 17.6%.
- Commodities are not the stagflation hedge that conventional wisdom suggests. In the post-1990 sample, BCOM averaged -5.2% over 12 months in Stagflation — the only negative quadrant reading — with a 9.0pp cross-regime spread that nearly matches equities’ 9.4pp spread. The 1970s stagflation episodes, when commodities rose for years, are not in the post-1990 sample.
- Bond regime sensitivity is a duration choice: long-duration government bonds show a 1.2pp cross-regime spread at 12 months; the Bloomberg Aggregate shows 1.8pp, driven by shorter effective duration. See the bond comparison table below.
Average 12-month forward returns by regime across two sample windows. The post-1990 window is generally the more relevant reference for modern portfolio construction — the Volcker disinflation changed the stock-bond correlation from positive to negative, and pre-1982 data can mislead on diversification properties during Stagflation and Contraction states. The full history back to 1967 is included because it contains a larger sample of the Stagflation quadrant — the 1970s provide the only modern precedent for sustained above-trend inflation with below-trend growth in the post-Bretton-Woods era. BCOM is excluded from the full-history table; it is only available from 1991.
Equity forward returns by horizon (post-1990)
| Regime | N | Equity 3m (ann) | Equity 6m (ann) | Equity 12m |
|---|---|---|---|---|
| Disinflationary expansion | 112 | +14.9% | +16.5% | +15.0% |
| Boom (late cycle) | 111 | +13.2% | +11.0% | +13.4% |
| Stagflation | 87 | +5.7% | +5.6% | +5.6% |
| Contraction | 98 | +14.4% | +14.7% | +13.8% |
| Max − Min spread | +9.2% | +10.9% | +9.4% | |
| Unconditional mean | +12.3% | +12.2% | +12.3% |
3m and 6m returns are annualized for comparability with 12m. The Stagflation signal is consistent across all three horizons.
Nominal and real 12-month returns — all assets (post-1990)
| Regime | N | Equity nom | LT Govt Bond nom | Agg Bond nom | T-Bill nom | Equity real | LT Govt Bond real | Agg Bond real | T-Bill real |
|---|---|---|---|---|---|---|---|---|---|
| Disinflationary expansion | 112 | +15.0% | +6.8% | +4.4% | +1.9% | +13.5% | +5.3% | +2.9% | +0.4% |
| Boom (late cycle) | 111 | +13.4% | +7.9% | +5.2% | +3.8% | +10.6% | +5.1% | +2.4% | +1.0% |
| Stagflation | 87 | +5.6% | +7.6% | +6.2% | +3.6% | +2.7% | +4.7% | +3.2% | +0.7% |
| Contraction | 98 | +13.8% | +6.7% | +5.2% | +1.1% | +12.3% | +5.2% | +3.8% | -0.4% |
| Max − Min spread | +9.4% | +1.2% | +1.8% | +2.6% | +10.9% | +0.6% | +1.3% | +1.4% | |
| Unconditional mean | +12.3% | +7.2% | +5.2% | +2.6% | +10.1% | +5.1% | +3.0% | +0.4% |
Real returns deflated by concurrent monthly Core PCE inflation; results substantively unchanged using forward inflation.
Long-term government bond vs. Bloomberg Aggregate, 3m / 6m / 12m (post-1990)
| Regime | N | LT Govt 3m | LT Govt 6m | LT Govt 12m | Agg 3m | Agg 6m | Agg 12m |
|---|---|---|---|---|---|---|---|
| Disinflationary expansion | 112 | +1.8% | +3.2% | +6.8% | +1.1% | +2.1% | +4.4% |
| Boom (late cycle) | 111 | +1.4% | +3.4% | +7.9% | +1.1% | +2.3% | +5.2% |
| Stagflation | 87 | +2.7% | +4.5% | +7.6% | +1.7% | +3.2% | +6.2% |
| Contraction | 98 | +1.4% | +3.4% | +6.7% | +1.4% | +2.9% | +5.2% |
| Max − Min spread | +1.3% | +1.2% | +1.2% | +0.6% | +1.0% | +1.8% | |
| Unconditional mean | +1.8% | +3.6% | +7.2% | +1.3% | +2.6% | +5.2% |
LT Govt Bond: Welch & Goyal (2008) long-term government bond series (Ibbotson SBBI through 2021, Bloomberg from 2022). Agg Bond: Bloomberg US Aggregate Index (January 1990+).
Commodities — Bloomberg Commodity Index (post-1990)
Bloomberg Commodity Index total return data is available from February 1990, covering the 1990 oil spike, the 2008 commodity peak, and the 2022 surge.
| Regime | N | BCOM 3m | BCOM 6m | BCOM 12m |
|---|---|---|---|---|
| Disinflationary expansion | 112 | +0.6% | +1.2% | +2.7% |
| Boom (late cycle) | 111 | +1.4% | +2.6% | +3.8% |
| Stagflation | 87 | -1.5% | -2.7% | -5.2% |
| Contraction | 98 | +0.5% | +1.6% | +3.7% |
| Max − Min spread | +2.9% | +5.2% | +9.0% | |
| Unconditional mean | +0.4% | +0.9% | +1.6% |
The BCOM Stagflation result is counterintuitive. Conventional intuition holds that commodities are a stagflation hedge — real assets that should benefit from inflation. The data shows the opposite: post-1990 stagflation episodes produced negative nominal 12-month forward commodity returns, the only negative quadrant reading in the table. The reason is timing: post-1990 stagflation episodes (1990 oil spike, 2008 commodity peak, 2022 surge) were each defined by commodity prices having already moved. The 12-month forward window measures what came after — typically the commodity reversal. The 1970s stagflation episodes, where commodity prices rose for years rather than spiking and falling, are not in the post-1990 sample.
Full history — equity, bonds, T-bills from 1967 (nominal only)
| Regime | N | Equity 3m | Equity 6m | Equity 12m | Bonds 3m | Bonds 6m | Bonds 12m | T-Bill 3m | T-Bill 6m | T-Bill 12m |
|---|---|---|---|---|---|---|---|---|---|---|
| Disinflationary expansion | 112 | +3.5% | +7.9% | +15.0% | +1.8% | +3.2% | +6.8% | +0.5% | +0.9% | +1.9% |
| Boom (late cycle) | 287 | +2.8% | +4.7% | +9.9% | +1.5% | +3.3% | +6.8% | +1.3% | +2.7% | +5.8% |
| Stagflation | 212 | +2.4% | +5.6% | +12.1% | +2.4% | +4.6% | +9.5% | +1.5% | +3.0% | +6.0% |
| Contraction | 98 | +3.4% | +7.1% | +13.8% | +1.4% | +3.4% | +6.7% | +0.3% | +0.6% | +1.1% |
Trajectory matters
The tables below break down 12-month forward equity and bond returns by trajectory direction within each quadrant, using the post-1990 window. The 6-month change in CFNAI (growth axis) and 6-month change in Core PCE Z-score (inflation axis) define a vector; its angle is discretized into eight compass directions, with a stationary category for low-magnitude changes (vector magnitude below 0.3). Direction labels: E = growth improving with stable inflation; N = inflation rising with stable growth; NW = growth slowing while inflation rises; S = disinflation with stable growth; SE = recovery (growth improving, inflation falling); and so on.
Many cells are thin. Eight directions across four quadrants gives thirty-two sub-cells from roughly 420 post-1990 months. Cells with three or four observations are shown with their count; cells with fewer than three are suppressed. Treat the trajectory table as a directional indicator, not a point estimate.
Disinflationary expansion
| Trajectory direction | N | Equity 12m | Bond 12m |
|---|---|---|---|
| E — growth improving | 23 | +18.8% | +6.5% |
| NE — both rising | 13 | +17.7% | +4.0% |
| N — inflation rising | 9 | +10.6% | +11.9% |
| NW — growth slowing, inflation rising | 5 | +0.4% | +6.0% |
| W — growth slowing | 3 | +16.1% (n=3) | -0.4% (n=3) |
| SW — both declining | 7 | +19.9% | +8.0% |
| S — inflation falling | 15 | +10.9% | +9.2% |
| SE — recovery | 13 | +17.5% | +5.8% |
| Stationary | 24 | +14.4% | +6.4% |
Boom (late cycle)
| Trajectory direction | N | Equity 12m | Bond 12m |
|---|---|---|---|
| E — growth improving | 17 | +17.2% | +9.7% |
| NE — both rising | 8 | +12.1% | +13.3% |
| N — inflation rising | 25 | +3.1% | +1.8% |
| NW — growth slowing, inflation rising | 4 | +12.1% (n=4) | +4.1% (n=4) |
| W — growth slowing | 8 | +19.0% | +2.7% |
| SW — both declining | 3 | +19.2% (n=3) | +13.3% (n=3) |
| S — inflation falling | 15 | +18.1% | +9.1% |
| SE — recovery | 17 | +9.3% | +10.0% |
| Stationary | 22 | +20.0% | +11.2% |
Stagflation
| Trajectory direction | N | Equity 12m | Bond 12m |
|---|---|---|---|
| N — inflation rising | 10 | +5.5% | +8.8% |
| NW — growth slowing, inflation rising | 18 | +2.6% | +8.2% |
| W — growth slowing | 15 | +6.4% | +10.2% |
| SW — both declining | 18 | +3.3% | +9.0% |
| S — inflation falling | 16 | +17.6% | +6.4% |
| SE — recovery | 7 | +14.9% | +10.5% |
| Stationary | 20 | +2.1% | +3.4% |
Contraction
| Trajectory direction | N | Equity 12m | Bond 12m |
|---|---|---|---|
| E — growth improving | 3 | +1.9% (n=3) | +15.4% (n=3) |
| NE — both rising | 3 | +18.4% (n=3) | +1.3% (n=3) |
| N — inflation rising | 11 | +18.7% | +3.9% |
| NW — growth slowing, inflation rising | 9 | +4.1% | +8.9% |
| W — growth slowing | 11 | +20.3% | +7.1% |
| SW — both declining | 24 | +13.3% | +5.8% |
| S — inflation falling | 14 | +16.7% | +6.2% |
| SE — recovery | 6 | +6.2% | +7.2% |
| Stationary | 17 | +13.8% | +7.9% |
Where we are now
The five closest historical analogs to the current reading, found by Euclidean distance in normalized (CFNAI, Core PCE Z, ΔCFNAI 6m, ΔPCE-Z 6m) space, weighted 70% on position and 30% on trajectory (both components normalized to unit variance). The past 24 months are excluded from the candidate set, keeping the analog set disjoint from the 24-month trail visible on the Regime chart. For each analog, realized 3-, 6-, and 12-month forward equity and 12-month bond returns are shown where available. A shorter distance indicates a closer configuration in the growth-inflation-momentum space.
Inflation is rising 1.5× faster than growth is improving (+0.94 σ vs. +0.63 σ over the trailing six months) — inflation dynamics are dominant in the current trajectory.
Current reading — 2026-04-01
Quadrant: Boom (late cycle) | Trajectory: NE
CFNAI: 0.14 | Core PCE Z: 2.47 σ
| Date | Quadrant | Trajectory | CFNAI | PCE Z | Distance | Eq 3m | Eq 6m | Eq 12m | Bd 12m |
|---|---|---|---|---|---|---|---|---|---|
| 1987-09-01 | Boom (late cycle) | N | 0.47 | 2.75 σ | 0.224 | -23.2% | -17.1% | -11.4% | +16.3% |
| 2006-08-01 | Boom (late cycle) | N | 0.18 | 1.39 σ | 0.236 | +8.2% | +9.9% | +15.4% | +5.8% |
| 1992-04-01 | Boom (late cycle) | E | 0.23 | 2.7 σ | 0.251 | +2.6% | +3.1% | +10.8% | +20.3% |
| 1990-03-01 | Boom (late cycle) | NE | 0.17 | 3.68 σ | 0.262 | +5.7% | -10.7% | +13.5% | +12.9% |
| 1987-11-01 | Boom (late cycle) | N | 0.26 | 2.9 σ | 0.275 | +17.9% | +17.1% | +23.8% | +10.3% |
How to read it
The return tables are descriptive statistics on a sample of roughly 420 post-1990 months. Many quadrant cells contain 60–120 observations; trajectory sub-cells typically 8–30. This is not an excuse to dismiss the patterns — conditional return differences of 5–10 percentage points per year are economically meaningful even without formal significance tests — but it is a reason to treat the numbers as base rates for scenario planning rather than point forecasts. No p-values are reported because many cells are too thin to support inference and the honest framing is descriptive. Average regime persistence post-1990 is 2.2 months; most months are classified during transitions rather than during stable regimes. Trajectory tables should be read as conditioning on the configuration at time t, not as describing sustained regime states.
The Contraction quadrant's strong forward returns reflect a real-time identification problem, not an investment opportunity. CFNAI is a concurrent indicator with a one-month publication lag; deep contraction readings (CFNAI below −2) appear at or after market troughs. By the time the Contraction signal fires, most of the drawdown has occurred. The forward returns measure what happened to investors who bought at the trough — a counterfactual exercise, not an actionable signal. The other three quadrants do not have this identification problem.
T-bill real returns in the Contraction regime are slightly negative (−0.4%) because the Contraction-classified months in the post-1990 sample concentrate in zero-interest-rate periods (post-2008, post-COVID). Even with low concurrent inflation, T-bill nominal yields floored near zero produce mildly negative real returns.
The pre-/post-Volcker comparison requires an interpretive judgment. The full-history table uses data from 1967, spanning monetary regimes structurally different from the post-1990 era. In the pre-Volcker decade (roughly 1967–1982), the stock-bond correlation was positive — meaning Stagflation and Contraction states did not offer the bond diversification that post-1990 investors experienced. Readers using the full-history table for allocation decisions should apply that discount.
The 1991–2019 baseline for Core PCE Z-scoring embeds the assumption that the post-Volcker Federal Reserve operated in a stable low-inflation regime — one in which above-trend credit growth did not persistently destabilize the price level. The trajectory tables compare the current reading against historical periods that shared that post-Volcker context. Whether the post-2022 episode marks the end of that regime or a recoverable shock is the framing question. The analog matching is not a prediction; it is a reminder of what happened the last time the data looked approximately like this.
The quadrant labels reflect a particular theory: that credit expansion drives above-trend growth, monetary expansion drives persistent inflation, and the natural resting point is the boundary between quadrants. Readers from the New Keynesian or demand-management traditions may prefer different boundaries or different labels for the same data. The underlying CFNAI and Core PCE series are neutral on these questions.
Data sources for return series: Equity returns from the Kenneth R. French Data Library (US value-weighted total return, July 1926+). Long-term government bond returns from Welch & Goyal (2008) updated through 2024 (Ibbotson SBBI data through 2021, Bloomberg indices from 2022). Aggregate bond returns from the Bloomberg US Aggregate Index (January 1990+, sourced from user-provided Bloomberg export). T-bill returns from Welch & Goyal Rfree column (monthly holding-period return, 1871+). Commodity returns from the Bloomberg Commodity Index total return (January 1990+; replaces prior DJP ETN proxy, which carried roll-yield drag of approximately 1pp per year).