THE NEUTRAL RATE (r∗) as an Economic Signal and Policy Anchor

I. Introduction: The Ghost in the Machine

The neutral rate of interest — r∗, spoken aloud as “r-star” — is defined as the real short-term policy rate consistent with full employment and stable inflation over the medium term. It is the gravitational centre of modern monetary policy: the rate at which the central bank is neither pressing the accelerator nor the brake on an economy running at potential. Yet it is unobservable. It cannot be looked up in a table or read off a market screen. It must be inferred, modelled, estimated — and perpetually argued over.

In early 2026, r∗ has re-emerged as the single most consequential macroeconomic variable confronting the G7. After nearly fifteen years defined by secular disinflation, suppressed equilibrium rates, and the “lower-for-longer” monetary orthodoxy, advanced economies are confronting what appears to be a structural repricing of capital. What was once assumed to be a near-zero equilibrium real rate is now being reassessed upward across the United States, the euro area, the United Kingdom, Canada, and Japan.

The implication is profound. If r∗ has shifted durably higher — as a growing body of evidence and market pricing suggests — then the entire architecture of fiscal sustainability, financial asset valuations, debt dynamics, exchange rate equilibrium, and central bank reaction functions must be recalibrated. The era of costless capital, the era in which governments could borrow on the long run at a real cost near zero, may be drawing to a close. How policymakers locate, communicate, and act upon this moving star will define the macroeconomic decade that follows.

“Our models are telescopes, not truths. They point toward the stars, but they do not create them.”

II. Historical Trajectory: From Secular Stagnation to Structural Repricing

1. The Era of “Lower for Longer” (2008–2021)

Following the 2008 Global Financial Crisis, equilibrium real rates declined sharply across advanced economies. Estimates from central bank research teams and multilateral institutions suggested that r∗ in the United States had fallen from roughly 2 per cent in the 1990s to near zero by the mid-2010s. The pioneer empirical framework — the Laubach-Williams (2003) model, subsequently extended by Holston, Laubach and Williams (2017) to a multi-country context — traced this secular compression through a combination of demographic, productivity, and risk-preference forces that played out over decades rather than years.

Key structural forces drove this decline: a global savings surplus, particularly from East Asian exporters and commodity-producing nations; demographic ageing, raising precautionary savings and compressing the marginal efficiency of capital; weak total factor productivity growth; post-crisis corporate and household deleveraging; and heightened risk aversion suppressing private investment demand. Together these forces produced what Larry Summers, reviving a Hansonian concept, labelled “secular stagnation” — a chronic shortfall of investment demand relative to saving at any positive real rate.

Central banks responded with quantitative easing, forward guidance, and in some cases negative policy rates. The Federal Reserve, European Central Bank, Bank of England, and Bank of Japan all sought to converge market rates toward an inferred equilibrium that appeared close to zero or even below. The consequence was a regime in which r∗ functioned as a floor: any attempt to raise rates meaningfully above it was perceived as contractionary. Financial conditions were persistently accommodative, asset valuations expanded, and fiscal authorities benefited from historically low real borrowing costs.

2. The Pandemic Disruption and the Inflation Shock (2020–2024)

The COVID-19 shock initially reinforced the low-rate paradigm. However, the combination of unprecedented fiscal transfers, supply-chain fragmentation, energy market dislocations driven by the war in Ukraine, labour reallocation, and geopolitical deglobalisation triggered the highest inflation cycle in four decades across the G7.

As central banks executed the fastest tightening cycle since the early 1980s, an unexpected pattern emerged: growth proved more resilient than models calibrated on post-2008 data had predicted. In the United States, real GDP expanded robustly through 2023 and 2024 even as the federal funds rate rose above 5 per cent. Labour markets remained historically tight. Corporate investment, particularly in artificial intelligence infrastructure, surged. This resilience was not what one would expect from an economy operating well above its equilibrium rate — it suggested, powerfully, that the equilibrium rate itself may already have been higher than assumed.

3. The Current Re-rating (2025–2026)

By early 2026, the narrative has shifted decisively. Advanced economies no longer confront a structurally excess savings problem. They face instead a structural surge in capital demand of historically unusual breadth: massive investment in artificial intelligence infrastructure, with major U.S. hyperscalers collectively on track to invest approximately $400 billion in AI capital expenditure in 2025 alone; electrification and grid modernisation under net-zero commitments; defence spending increases following geopolitical fragmentation; semiconductor reshoring and strategic industrial policy; and fiscal deficits that remain elevated relative to pre-pandemic norms.

The result is a capital absorption dynamic that differs fundamentally from the 2010s. Consensus estimates that once placed U.S. real r∗ between 0.5 and 1.0 per cent are increasingly clustering in the 1.5 to 2.5 per cent range. European estimates have similarly shifted, though with slightly lower medians given demographic drag and weaker productivity momentum in the euro area.

III. Intellectual Lineage: Wicksell, Keynes, and Friedman Revisited

The debate over r∗ is not merely technical. It reflects competing philosophies of macroeconomic equilibrium that trace their lineage back more than a century.

Wicksell’s Natural Rate: The Original Formulation

The intellectual origin of r∗ lies with the Swedish economist Knut Wicksell, who in 1898 defined the natural rate of interest as “the rate of interest on loans that is neutral with respect to commodity prices, tending neither to raise nor to lower them.” Wicksell argued that if the market rate of interest were held below the natural rate, credit expansion would drive prices upward; if held above it, deflation would follow. The concept anticipated the Natural Rate Hypothesis by seventy years and provided the essential scaffolding on which modern inflation-targeting central banking is built. The Federal Reserve’s Taylor Rule — the near-universal benchmark for relating the policy rate to inflation and output gaps — implicitly incorporates an estimate of r∗ as its long-run real intercept.

The Keynesian Perspective: Animal Spirits and the Marginal Efficiency of Capital

For John Maynard Keynes, equilibrium rates of interest reflect the “marginal efficiency of capital” — the expected rate of return on investment — which is shaped by psychology, uncertainty, and what he called “animal spirits.” Under this view, r∗ is socially and psychologically contingent rather than determined purely by structural fundamentals. Expectations, once depressed, can become self-fulfilling, trapping an economy in a low-rate, low-investment equilibrium even when the structural fundamentals for higher rates are present.

The post-2008 period was interpreted by Keynesian economists as precisely such a liquidity trap: even zero interest rates failed to stimulate adequate private investment because expectations of future returns remained depressed. From this perspective, today’s upward reassessment of r∗ reflects a genuine revival of investment optimism — driven by AI, the green transition, and defence — that has lifted the marginal efficiency of capital toward levels not seen since the late 1990s technology boom.

The Monetarist and Natural Rate Perspective: Structural Forces

For Milton Friedman, and subsequently Edmund Phelps, the natural rate of interest — like the natural rate of unemployment — is determined by structural economic forces: productivity growth, demographic trends, and intertemporal preferences. Under this framework, if central banks set rates persistently below the natural rate, inflation accelerates; if above it, output contracts. The inflation surge of 2021 to 2023, across all G7 economies simultaneously, offered the most striking empirical validation of the Natural Rate Hypothesis since the stagflation of the 1970s.

IV. The Pirandello Analogy: A Necessary Fiction

The Italian playwright Luigi Pirandello, in his 1921 masterwork Six Characters in Search of an Author, presented characters who possess an authentic inner reality but cannot find the author capable of giving it fixed, external form. r∗ occupies an analogous epistemological position within macroeconomics.

It is conceptually necessary for equilibrium analysis — without a neutral anchor, policymakers cannot distinguish between a restrictive and an accommodative stance. Yet it is empirically invisible: no instrument can directly observe the rate at which an economy is in perfect balance. It is dependent on the “author” — the econometric model — that interprets the available data through a particular theoretical lens. And different models, interpreting the same economic constellation, produce materially different coordinates for the same star.

As of early 2026, the Federal Reserve Bank of New York’s Holston-Laubach-Williams estimate places the U.S. real neutral rate at approximately 0.84 per cent (Q2 2025 data). The Federal Reserve Bank of Richmond’s model estimates it above 2.2 per cent. The Cleveland Fed’s Zaman model places the medium-run nominal neutral rate at 3.7 per cent, and estimates a 77 per cent probability that current policy is in restrictive territory. Market-implied forward rates suggest a long-run nominal neutral near 3 per cent. Meanwhile, RSM’s chief economist places real r∗ at 1.75 to 2 per cent.

Whatever r-star is — whether it’s the New York Fed’s estimate of 1.13%, or the Richmond Fed’s view of 2.2%, or our estimate of 1.75% to 2% — the rate has moved higher, and that is already having a significant impact on America’s real economy. — RSM US, 2025

Yet abandoning the concept would deprive policymakers of any shared reference point for calibrating the stance of policy. The dispersion of estimates is not a reason to discard r∗ but rather a sobering reminder of the uncertainty that must always accompany its application. r∗ is a flawed necessity — but it is a necessity nonetheless.

V. Competing Frameworks Guiding G7 Policy

1. The Productivity-Trend Model (Holston–Laubach–Williams)

Now maintained and published by the Federal Reserve Bank of New York, the HLW framework links the neutral rate to trend potential growth and demographic factors through a semi-structural state-space model. Its current readings — a real neutral rate of approximately 0.84 per cent as of mid-2025 — suggest only a modest rise in r∗ from post-2008 lows, arguing that secular stagnation forces remain structurally operative.

Implication: Real neutral rates in the U.S. are near 1.0–1.3 per cent, lower in the euro area.

Risk: Underestimation of the structural capital demand impulse from AI, defence, and the energy transition. The model’s demographic inputs were calibrated on pre-AI, pre-geopolitical-fragmentation data and may not adequately capture the investment-demand shock now unfolding.

2. The Empirical-Response Model (Lubik–Matthes / Richmond Fed)

This framework infers the equilibrium rate from the economy’s revealed resilience in the face of rate hikes. Given the durability of U.S. labour markets and robust corporate investment in 2024 and 2025 despite elevated policy rates, the model concludes that the equilibrium must be significantly higher — placing real r∗ above 2.2 per cent.

Implication: Policy may not have been as restrictive as believed during the tightening cycle, helping to explain the growth resilience.

Risk: Overestimation if elevated fiscal stimulus artificially masked the contractionary effect of higher rates.

3. The Cleveland Fed Zaman Model

A richer framework conditioning on a broader information set, the Zaman model estimated as of Q2 2025 a medium-run nominal neutral rate of 3.7 per cent, with a 68 per cent confidence band of 2.9 to 4.5 per cent. Given the then-prevailing federal funds rate of 4.25 to 4.5 per cent, the model assessed with 77 per cent probability that policy was in restrictive territory — providing the most formally probabilistic of the Fed system’s published estimates.

4. Market-Implied Forward Rates (5y5y Real Forwards)

Bond market pricing — particularly five-year forward rates five years ahead — provides a market-based estimate of long-run neutral. Federal Reserve Board staff analysis (Covitz and Engstrom, FEDS Notes, February 2026) finds that far-forward real risk premia have increased substantially, driven by elevated fiscal deficits, persistent capital expenditure cycles, and heightened geopolitical risk. The FOMC’s December 2025 dot plot shows a median long-run federal funds rate of 3.0 per cent, but individual projections range from 2.6 to 3.9 per cent nominal — an extraordinary 130 basis point spread within the same policy committee.

5. The Nonmonetary Forces Framework (Governor Miran)

A notable contribution by Federal Reserve Governor Stephen Miran (September 2025) explicitly incorporates nonmonetary factors — border policy, trade renegotiation, and fiscal changes — into the assessment of r∗. Miran argues that the sharp contraction in immigration under 2025 border policy may be putting significant downward pressure on the neutral rate, drawing on research estimating that a one percentage point drop in annual population growth reduces r∗ by approximately 0.6 percentage points — implying a nearly 0.4 percentage point downward drag from the immigration slowdown alone. This introduces a significant and underappreciated policy-driven demographic shock into the neutral rate debate, one that lagged models will not capture for quarters or years.

VI. Bayesian Scenario Analysis: Locating r∗ Under Uncertainty

The wide dispersion of model-based estimates — spanning more than 130 basis points in real terms across frameworks simultaneously maintained by Federal Reserve system researchers — is not a failure of econometrics. It is an honest acknowledgement of structural uncertainty about the forces shaping equilibrium. Bayesian inference offers a natural and disciplined language for navigating that uncertainty: instead of searching for a single “true” value of r∗, one maintains a probability distribution over scenarios, updates it as new data arrive, and uses the resulting posterior to bound policy decisions.

The Framework: Priors, Signals, and Posteriors

The Bayesian approach to r∗ begins with prior probabilities assigned to competing structural scenarios, reflecting our baseline beliefs before conditioning on the most recent data. These priors are informed by the long-run empirical literature — Laubach-Williams, demographic projections, trend productivity estimates — as well as by the theoretical frameworks of Wicksell, Keynes, and Friedman discussed above.

The prior is then updated through a likelihood function — essentially asking: given each structural scenario, how probable are the observable signals we actually see in 2025 and 2026? Observable signals include the resilience of U.S. GDP growth despite elevated policy rates; the magnitude of AI capital expenditure; the behaviour of long-run inflation expectations; fiscal deficit trajectories; and labour market data conditioned on the immigration policy shift. The resulting posterior probability distribution provides a richer and more honest basis for policy calibration than any single point estimate.

Formally, if we denote the four structural scenarios as S₁ through S₄ with prior probabilities P(Sᵢ), and a vector of observable signals as Ω — encompassing AI CapEx realisation, fiscal trajectory, demographic signals, and market pricing — then the posterior probability of each scenario is given by Bayes’ theorem:

P(Sᵢ | Ω)  ∝  P(Ω | Sᵢ) × P(Sᵢ)

The posterior distribution over r∗ is then the probability-weighted average of the scenario-specific r∗ estimates, with the weights determined by the updated scenario probabilities. This framework does not eliminate uncertainty — it quantifies and structures it. It also makes explicit the signals that would cause a rational policymaker to revise their r∗ estimate up or down, in real time, as new data arrive.

The Four Scenarios

Scenario I: Secular Stagnation Persistence (Prior: 20%)

Scenario I — Secular Stagnation Persistence   [Prior probability: 20%]

AI realisation: Disappoints: narrow gains, limited economy-wide diffusion.

Fiscal path: Consolidation: deficits compress toward 4% of GDP.

Demographics / immigration: Immigration collapse lowers labour supply and investment demand.

Posterior real r∗ (U.S.): 0.50 – 0.90% real

In this scenario, the structural forces that drove r∗ toward zero in the 2010s remain dominant. AI investment disappoints relative to headline CapEx figures: the spending is real, but productivity gains are narrow, concentrated in a small number of sectors, and fail to diffuse economy-wide. The demographic drag from declining immigration reinforces the Miran channel, reducing labour supply growth and dampening investment demand. Fiscal consolidation, driven by bond market pressure and political compromise, reduces the sovereign borrowing impulse. r∗ drifts back toward its post-2008 equilibrium in the 0.50 to 0.90 per cent real range, consistent with the HLW model’s current estimate of 0.84 per cent.

This scenario’s prior weight of 20 per cent reflects the fact that HLW’s semi-structural framework has the most rigorous empirical pedigree, but its demographic inputs may be poorly calibrated to the AI-era investment surge. It would imply that much of the tightening cycle was excessive, and that deferred effects are still working through the system with long and variable lags.

Scenario II: Moderate Structural Shift — Base Case (Prior: 45%)

Scenario II — Moderate Structural Shift — Base Case   [Prior probability: 45%]

AI realisation: Partial: genuine productivity gains in leading sectors, incomplete economy-wide diffusion over 3–5 years.

Fiscal path: Deficits persist at around 6% of GDP; no dramatic escalation.

Demographics / immigration: Modest immigration decline; demographic drag partially contained.

Posterior real r∗ (U.S.): 1.25 – 1.75% real

The base case assigns the highest prior probability to a moderate but durable upward shift in r∗. AI investment delivers genuine but incomplete productivity gains. Fiscal deficits persist at around 6 per cent of GDP, maintaining sovereign borrowing pressure, but do not escalate dramatically. The immigration contraction creates a modest negative drag that partially offsets the AI and fiscal impulses. Real r∗ in the United States stabilises in the 1.25 to 1.75 per cent range — consistent with the RSM estimate, the lower end of the Richmond Fed’s implied range, and the real rate implied by the FOMC’s December 2025 median long-run dot of 3.0 per cent nominal against a 2 per cent inflation target.

Under this scenario, current policy at 4.25 to 4.5 per cent nominal is moderately restrictive, consistent with the Cleveland Fed’s Zaman model assessment, validating a patient, data-dependent path toward 3.0 to 3.5 per cent over 18 to 24 months.

Scenario III: High Neutral / New Paradigm (Prior: 25%)

Scenario III — High Neutral / New Paradigm   [Prior probability: 25%]

AI realisation: Transformative: economy-wide TFP uplift of 1.5 percentage points or more.

Fiscal path: Deficits remain elevated; defence and green CapEx accelerate simultaneously.

Demographics / immigration: Immigration stabilises; labour market absorbs the AI transition.

Posterior real r∗ (U.S.): 2.00 – 2.50% real

This scenario envisions a genuine structural break in the equilibrium rate, driven by a convergence of investment demand forces not seen since the postwar reconstruction era. AI delivers a durable total factor productivity boost economy-wide, lifting both potential output and the marginal efficiency of capital. Defence and green CapEx accelerate simultaneously, as Germany’s fiscal loosening triggers a wider European investment push. Sovereign fiscal deficits remain elevated and are partially absorbed through a structural widening of term premia.

Under Scenario III, current nominal policy at 4.25 to 4.5 per cent is only modestly above neutral, and the Fed’s December 2025 median projection of 3.0 per cent long-run nominal would need to be revised upward to 4.0 to 4.5 per cent — a major recalibration of the entire rate-setting framework. The prior of 25 per cent reflects the genuine plausibility of the AI investment case, while acknowledging the historical tendency for technology-investment booms to produce shorter-than-expected r∗ uplift before adoption diffuses and the CapEx cycle matures.

Scenario IV: Fiscal Dominance Break — Tail Risk (Prior: 10%)

Scenario IV — Fiscal Dominance Break — Tail Risk   [Prior probability: 10%]

AI realisation: Irrelevant: fiscal stress dominates asset pricing entirely.

Fiscal path: Loss of bond market confidence; term premium surge; sovereign yields de-anchor.

Demographics / immigration: Political pressure suppresses real rates below any meaningful neutral level.

Posterior real r∗ (U.S.): Nominally elevated; severely negative in real terms (concept becomes inoperative)

The fiscal dominance scenario is qualitatively distinct from the others. It does not predict a particular level of r∗ so much as a breakdown in the mechanism by which r∗ is expressed in market pricing and policy rates. Sustained fiscal deficits at or above 6 per cent of GDP, combined with political pressure on monetary independence — illustrated most acutely by the DoJ subpoena of the Federal Reserve in January 2026 — produce a loss of bond market confidence in the long-run fiscal trajectory of the United States. Term premia surge, and the yield curve steepens in a way that central bank policy rates cannot contain.

The paradox of fiscal dominance is that it can produce nominally high interest rates while simultaneously suppressing real rates through elevated inflation expectations: a modern echo of the financial repression that Reinhart and Sbrancia documented as the primary debt-reduction mechanism in the post-WWII period. Real r∗ becomes effectively indeterminate — the concept loses its operational meaning as a policy anchor precisely when it is most needed. The 10 per cent prior reflects the view that U.S. institutions remain sufficiently robust to resist full fiscal dominance in the medium term, but the January 2026 episode raised the tail probability non-trivially from near-zero to a level that responsible risk management cannot ignore.

Observable Signals and Posterior Updating

The following nine signals, each observable as of February 16, 2026, constitute the primary inputs to the likelihood function. Read together, they reveal a balance of evidence that points modestly but meaningfully away from Scenario I and toward the moderate structural shift of Scenario II, with genuine but bounded probability mass remaining on Scenarios III and IV.

U.S. AI CapEx (hyperscalers): ~$400 billion in 2025 — ↑ r*: Raises investment demand well above historical norms; shifts prior weight toward Scenario III.

FOMC December 2025 long-run dot: 3.0% nominal median — ↑ r* (moderate): Supports the base case; committee median is consistent with approximately 1.0% real neutral.

HLW model estimate: 0.84% real (Q2 2025) — ↓ r* signal: Strengthens the Scenario I prior; secular demographic and productivity forces not yet exhausted in this framework.

Richmond Fed model estimate: above 2.2% real — ↑ r* signal: Supports Scenario III; resilience-based inference implies neutral has risen substantially.

U.S. fiscal deficit: above 6% of GDP through 2026–2027 — ↑ r*: Raises sovereign borrowing demand; crowds private capital and widens term premia.

Immigration contraction under 2025 border policy — ↓ r*: Reduces labour supply growth; the Miran channel implies approximately −0.4 percentage points on r*.

Germany fiscal expansion: investment up approximately 20% in 2026 — ↑ r* (euro area): Partially offsets demographic drag in the euro area; adds upward pressure to European capital demand.

BofA Fund Manager Survey: 53% cite AI stocks as bubble — Ambiguous: If correct, AI CapEx overshoots rational returns; an eventual correction would lower the real neutral rate.

DoJ subpoena of the Federal Reserve (January 2026) — ↓ r* (tail risk): Raises Scenario IV probability incrementally; political pressure for premature easing constitutes a fiscal dominance signal.

Posterior Estimates and Policy Implications

Weighting the scenario-specific r∗ estimates by their prior probabilities — and conditioning qualitatively on the signal evidence above — yields a probability-weighted posterior expectation for U.S. real r∗ of approximately 1.40 to 1.65 per cent. This sits comfortably within the moderate structural shift range, somewhat below the Richmond Fed’s point estimate and well above the HLW model’s current reading. In summary: Scenario I (secular stagnation persistence, 20% prior, 0.50–0.90% real r*); Scenario II (moderate structural shift, 45% prior, 1.25–1.75% real r*); Scenario III (high neutral / new paradigm, 25% prior, 2.00–2.50% real r*); and Scenario IV (fiscal dominance break, 10% prior, indeterminate in real terms). The signals from Table 1 would shift weight modestly from Scenario I toward Scenarios II and III relative to these priors.

Crucially, the framework is dynamic. The signals that would cause the largest single revision in posterior r∗ are a sustained AI productivity breakthrough visible in aggregate TFP data — which would sharply raise the probability weight on Scenario III; a credible medium-term fiscal consolidation plan — which would lower the Scenario IV tail probability and reduce term premia; and a resolution of the Federal Reserve independence crisis confirming institutional credibility — which would reduce the dispersion of the posterior distribution and lower the premium that markets embed in long-duration U.S. assets.

For central banks navigating under Bayesian uncertainty, the posterior distribution matters more than the posterior mean. The optimal Bayesian policy in early 2026 is one of data-conditional gradualism: moving rates toward 3.0 to 3.5 per cent nominal over 2026, while continuously updating the scenario weights in response to incoming productivity data, fiscal developments, and labour market signals. This is not indecision. It is rational inference under structural uncertainty.

The question is not what r-star is today. The question is what distribution over r-star values should govern policy decisions — and what observable evidence would cause that distribution to shift. That framing converts an abstract philosophical puzzle into an actionable decision science.

VII. The G7 in Comparative Perspective: Divergence Beneath the Surface

The United States

The U.S. presents the strongest case for a durably higher r∗. A relatively youthful demographic profile by G7 standards, the global centre of AI investment, the world’s reserve currency status enabling deficit financing at scale, and aggressive fiscal expansion through the OBBBA all point toward elevated capital demand. The federal deficit is projected to remain above 6 per cent of GDP through the current presidential term, generating structural sovereign borrowing that competes with private capital across the yield curve. The 10-year Treasury yield is expected to drift toward 4.5 per cent by year-end 2026, reflecting both structural fiscal pressures and elevated term premia.

The Euro Area

The ECB held its deposit facility rate at 2.0 per cent through the end of 2025 and the opening of 2026 — a level that Vanguard characterises as neutral for the euro area. Staff projections put euro area growth at 1.4 per cent in 2025 and 1.2 per cent in 2026, with inflation revised to 1.9 per cent for 2026. Demographic drag and weaker productivity growth partially explain the lower euro area neutral, but Germany’s fiscal loosening and pan-European defence commitments are adding an upward impulse to capital demand that was absent in the 2010s. In Bayesian terms, the euro area scenario weights sit closer to Scenarios I and II, with considerably less probability mass on Scenario III than the U.S.

The United Kingdom

The Bank of England cut Bank Rate to 3.75 per cent in December 2025, its fourth reduction of the year. The February 2026 Monetary Policy Report projects CPI inflation falling back to 2.0 per cent by mid-2026. The U.K.’s structural neutral rate appears somewhat higher than the euro area but meaningfully lower than the U.S., reflecting a combination of fiscal tightening — the 2025 Autumn Budget raised taxes to their highest share of GDP in post-war history — and weaker private-sector investment dynamism. Services inflation remains elevated and productivity growth has been, in the Bank’s own assessment, exceptionally weak on average since 2023.

Japan

Japan remains the outlier among G7 central banks. The Bank of Japan’s gradual exit from yield curve control and negative interest rate policy has moved the policy rate modestly positive but remains well below other G7 levels. Japan’s structural r∗ — constrained by extreme demographic ageing, entrenched deflationary psychology, and very high public debt ratios — is likely still very low in real terms, even as nominal rates rise from deeply negative starting points. Japan’s experience remains the sharpest cautionary tale of what Scenario I looks like when fully entrenched over a 30-year period, and the most powerful reminder of what is at stake if any G7 economy mislocates its neutral rate and remains too accommodative for too long.

VIII. Leadership at the Federal Reserve: Transition, Uncertainty, and Independence

The Warsh Nomination

On January 30, 2026, President Trump formally nominated former Federal Reserve Governor Kevin Warsh to succeed Jerome Powell as Chair, pending Senate confirmation. Warsh has been a vocal critic of excessive reliance on econometric equilibrium estimates. His publicly articulated approach emphasises greater weight on market-based signals, faster balance-sheet normalisation, scepticism toward mechanical forward guidance, and a rules-based monetary regime. Analysts have described his proposed framework as “Productive Dovishness”: an inclination to ease in 2026 driven by the view that AI-driven productivity gains could boost economic growth without reviving inflation. Deutsche Bank has cautioned, however, that it does not view Warsh as structurally dovish over the medium term.

From a Bayesian perspective, a Warsh-led Federal Reserve would likely place greater weight on market-implied signals — the 5y5y forwards, the term structure — as inputs to the likelihood function, and less on the HLW model’s demographic-trend outputs. This methodological shift would, all else equal, move the Fed’s implicit scenario weights toward Scenario III, and accelerate any upward revision of the long-run dot from the December 2025 median of 3.0 per cent nominal.

The Independence Crisis

On January 11, 2026, Federal Reserve Chair Jerome Powell publicly disclosed that the Department of Justice had served the Fed with grand jury subpoenas related to his June 2025 Senate testimony. Powell stated directly: “The threat of criminal charges is a consequence of the Federal Reserve setting interest rates based on our best assessment of what will serve the public, rather than following the preferences of the President.”

“This is about whether the Fed will be able to continue to set interest rates based on evidence and economic conditions — or whether instead monetary policy will be directed by political pressure or intimidation.” — Jerome Powell, January 11, 2026

A bipartisan group of former Fed Chairs and leading economists immediately compared the administration’s actions to moves made in more impoverished countries. JPMorgan Chase CEO Jamie Dimon stated that everyone he knows believes in Fed independence. The episode illustrates, with unusual clarity, the Scenario IV risk embedded in the current political environment. When a president believes rates are far too high, that belief is implicitly a claim about the neutral rate — that current policy is far above neutral and therefore needlessly contractionary. In the Bayesian framework, every such episode raises the tail probability on Scenario IV incrementally, and shifts the loss function confronting independent central bankers asymmetrically.

IX. Strategic Risks of Mislocating or Abandoning the Neutral Anchor

1. Communication Breakdown

If central banks lose confidence in their r∗ estimates — or abandon the concept altogether — forward guidance loses coherence. Bond markets, which price duration risk against a view of long-run equilibrium, would face structurally elevated volatility. The current FOMC — where a 130 basis point spread in long-run rate projections already exists among 19 voting and non-voting members — provides a preview of what incoherent signalling looks like in practice. A Bayesian framework makes the uncertainty explicit and manageable; its absence leaves markets to price it implicitly, at higher cost.

2. Fiscal Dominance Risk

Without an equilibrium anchor, the political pressure to suppress rates for debt-sustainability purposes becomes harder to resist. As G7 sovereign debt-to-GDP ratios are projected to reach 137 per cent by 2030, and the U.S. deficit is expected to remain above 6 per cent of GDP through the Trump administration’s second term, Scenario IV is not hypothetical. Reinhart and Sbrancia documented that governments facing high debt with inflationary environments tend to respond with financial repression when institutional constraints are weak. The January 2026 episode raised the Scenario IV probability from negligible to meaningful.

3. Stop–Go Cycles

Absent a neutral reference, monetary policy risks oscillating between over-tightening and under-tightening in ways that amplify rather than dampen the business cycle — recreating the volatility of the 1970s. The inflation surge of 2021 to 2023 and the subsequent rapid tightening cycle may already be read as a first episode of such a stop-go dynamic, enabled by an excessively low r∗ anchor that permitted rates to remain too low for too long.

4. Asset Price Misallocation

If the structural upward migration of r∗ is real but policymakers hold rates below the new equilibrium for political or communication reasons, capital misallocation follows. Long-duration assets — technology equities, commercial real estate, long-dated bonds — will be priced against a neutral rate that does not reflect reality. The Bank of America’s November 2025 Global Fund Manager Survey, in which 53 per cent of respondents characterised AI stocks as having reached bubble proportions, suggests that markets themselves are not fully confident that AI investment will generate returns commensurate with the implied cost of capital — consistent with the risk of Scenario III being over-weighted by markets relative to the true posterior.

X. Structural Implications for the G7

A structurally higher r∗ carries implications that cascade across every domain of macroeconomic policy and financial market structure.

Permanently higher real borrowing costs compress the set of viable public investment projects and force a genuine reckoning with fiscal sustainability. Infrastructure, social programmes, and pension systems designed on the implicit assumption of near-zero real rates face a fundamental viability challenge. The zombie companies problem — firms that survive only because ultra-low borrowing costs allow them to roll over debt indefinitely — will be resolved through restructuring and default for a significant tail of enterprises.

Equity valuation multiples face secular compression. The S&P 500, trading at approximately 22 times forward earnings heading into 2026, is partly priced on an assumption of low real discount rates. A durable upward shift in r∗ of 1 to 1.5 percentage points implies a fair-value reduction in equity multiples of similar magnitude, absent compensating earnings growth. The AI productivity case is the key offset: if AI delivers the transformative scenario, higher potential output validates higher multiples. If it disappoints, the higher cost of capital falls on a narrower earnings base, with correspondingly larger equity correction.

Housing markets face structural repricing. A generation of homeowners and developers who financed at the rates prevailing from 2010 to 2021 faces a permanently higher carry cost on new purchase and refinancing decisions. The refinancing wall projected for 2026 to 2027 represents the moment when the structural repricing of capital becomes most concretely visible at the household and corporate level.

The importance of productivity-enhancing reform is elevated. In a world of higher r∗, the premium on policies that raise the marginal efficiency of capital — labour market flexibility, regulatory streamlining, competition policy, innovation incentives — is correspondingly higher. Growth cannot rely as heavily on financial leverage and cheap credit. This places AI at the centre not just of the capital demand story but of the supply story: if AI genuinely raises total factor productivity, it simultaneously elevates r∗ through investment demand and validates the higher rate through improved potential output.

XI. Conclusion: The Moving Star

The neutral rate has not vanished. It has migrated.

The era in which capital was structurally abundant and its real price near zero appears to be ending. The combination of AI-driven investment demand, energy transition capital expenditure, defence rearmament, and persistently elevated sovereign borrowing has altered the equilibrium price of money in ways that demographic forces and productivity pessimism alone cannot easily offset.

The Bayesian framework developed in Section VI provides a disciplined structure for navigating this uncertainty. It assigns prior probabilities to four structurally distinct scenarios, updates them against the observable signals of early 2026, and yields a posterior expectation for U.S. real r∗ of approximately 1.40 to 1.65 per cent — consistent with the base case of moderate structural shift, but with meaningful probability mass assigned to both the secular stagnation persistence and the high neutral scenarios. The tail risk of fiscal dominance, while carrying only a 10 per cent prior, cannot be dismissed in a political environment where the executive has directly challenged central bank independence.

The appropriate policy response to this posterior distribution is data-conditional gradualism: moving rates deliberately but not precipitously toward a revised estimate of neutral, while continuously updating the scenario weights in response to AI productivity data, fiscal developments, immigration signals, and the resolution — or escalation — of the Federal Reserve independence crisis. This is not indecision. It is rational inference under structural uncertainty, applied to the most consequential macroeconomic variable of the decade.

The central challenge for 2026 is not whether r∗ exists. It is whether policymakers can correctly locate it — and defend their estimate from those who would set it by political fiat.

The integrity of that navigation determines not just the path of interest rates. It determines the credibility of the institutions that anchor the modern monetary order. Our models are telescopes, not truths. They point toward the stars, but they do not create them. The stars, however, do move. And 2026 will test whether the telescopes — and the institutions that operate them — are equal to the task.

Note on Sources and Methodology

All empirical references in this essay draw on publicly available research and data accessible as of February 16, 2026. Key sources include: Federal Reserve Bank of New York r∗ tracker (HLW and LW models); Federal Reserve Bank of Cleveland Economic Commentary (Horn and Zaman, 2025); Federal Reserve Board FEDS Notes (Covitz and Engstrom, February 2026); Federal Reserve Board Speech (Governor Miran, September 22, 2025); ECB December 2025 policy decision and Survey of Professional Forecasters; Bank of England Monetary Policy Reports (November 2025, February 2026); FOMC December 2025 Summary of Economic Projections; Vanguard Economic and Market Outlook 2026; BlackRock Investment Institute 2026 Outlook; RSM US Economic Outlook 2025–2026; LPL Research 2026 Fixed Income Outlook; CFR analysis of Warsh nomination (February 2026); and contemporaneous reporting on the Federal Reserve independence controversy of January 2026. The Bayesian scenario framework in Section VI is an analytical construct developed in this essay; the scenario-specific r∗ bounds are grounded in the published model estimates cited above and do not represent the views of any single institution.