Guide

r-star neutral rate explained

Harbor Credit Union's ALM committee labeled fed funds “deeply restrictive” through most of 2024 because its internal model anchored on a 2.5% real neutral rate from a 2019 staff memo. When the Fed held at 5.25% while core PCE fell toward 2.8%, the desk stayed short duration — betting on imminent cuts that did not arrive for another year. The error was not missing one CPI print; it was treating r-star as a fixed constant while productivity, demographics, and fiscal supply had shifted the underlying equilibrium.

r-star (r*, the neutral rate of interest) is the real short-term policy rate consistent with an economy at full employment and stable inflation at target. It is unobservable — you infer it from models, market prices, and central-bank research. Yet it anchors almost every debate about whether policy is tight or loose. After Harbor replaced its static 2.5% anchor with a time-varying Holston-Laubach-Williams estimate and a market-implied cross-check, stance labels matched FOMC communication more closely and false cut signals dropped 28%. This guide covers what r-star is, real versus nominal neutral rates, estimation methods, secular decline drivers, policy stance mapping, the Harbor refactor, a technique decision table versus Taylor-rule-only overlays, pitfalls, and a checklist — alongside our guides on the Taylor rule, monetary policy, and real interest rates.

What r-star is

The neutral rate of interest is the real (inflation-adjusted) short-term rate that neither stimulates nor restrains aggregate demand when output is at potential and inflation is at target. At this rate, the economy can stay in equilibrium without the central bank leaning on the accelerator or the brake.

r-star is distinct from:

The current policy rate — what the Fed actually sets today; it may be above, at, or below neutral.
The terminal rate — where markets expect policy to settle at the end of a hiking or cutting cycle; terminal can overshoot neutral during disinflation campaigns.
Long-term bond yields — they embed term premium, inflation expectations, and credit risk, not just the short-run neutral rate.
NAIRU — the unemployment rate consistent with stable inflation; r-star is the interest-rate analogue in the labor-inflation framework.

Central bankers reference r-star when explaining why the same nominal rate can be accommodative in one decade and restrictive in another. A 2% fed funds rate meant loose policy in 2004 and tight policy in 2022 because inflation, r-star, and real rates had moved.

Real r-star versus nominal neutral

Economists usually define r-star in real terms: the policy rate minus expected inflation. The nominal neutral rate is approximately r-star plus the inflation target (often 2% in the U.S.), but this shortcut breaks when inflation expectations are de-anchored or when the target itself is in flux.

Policy stance is often summarized as:

Real policy rate = nominal fed funds − expected inflation (survey or breakeven proxy)
Stance gap = real policy rate − r-star estimate

A positive stance gap suggests restrictive policy; negative suggests accommodation. The magnitude matters for duration positioning: a +150 bp gap historically correlates with slower growth and eventual easing, but the speed of transmission depends on financial conditions, fiscal policy, and credit channels beyond the policy rate alone.

How r-star is estimated

Structural models (HLW / Laubach-Williams)

The Holston-Laubach-Williams (HLW) framework and its predecessor Laubach-Williams use a small macro model with unobserved components: potential output, the output gap, trend growth, and the time-varying natural rate. Fed staff publish updated HLW estimates quarterly. These models smooth through noise but revise materially after GDP benchmark revisions or when inflation dynamics shift.

Market-implied approaches

Forward curves, inflation swaps, and TIPS breakevens can imply a market r-star when combined with term-structure models. Market methods react faster than structural filters but embed risk premia and can be distorted during liquidity stress or QE periods when term premium is elevated.

Survey and rules-based proxies

The New York Fed's Survey of Primary Dealers asks respondents for long-run neutral rate estimates. The Taylor rule treats r-star as a parameter in a reaction function — useful as a benchmark but not a direct measurement. Simple historical averages of real rates over long windows are seductive and usually wrong after structural breaks.

Why estimates disagree

Published r-star ranges often span 0.5% to 2.5% in real terms for the U.S. Disagreement reflects different model priors, treatment of the post-2008 period, and whether fiscal deficits and safe-asset demand are modeled as permanent shifts. Treat r-star as a distribution, not a point forecast.

Drivers of secular decline and recent reversals

From the 1980s through the 2010s, estimated r-star trended lower across advanced economies. Common explanations include:

Slower productivity growth — lower trend GDP growth reduces the return on capital and equilibrium real rates.
Demographics — aging populations save more and invest less, pushing down r-star (the “global savings glut” thesis).
Inequality and precautionary saving — higher income shares to high savers can increase desired saving at the margin.
Safe-asset demand — emerging-market reserve accumulation and post-crisis regulation increased demand for government bonds, compressing yields.
Lower risk appetite / higher risk premia on capital — after financial crises, desired investment may fall even when rates are low.

The 2022–2024 inflation shock revived debate about whether r-star had risen: fiscal expansion, reshoring investment, defense spending, and AI-driven capital expenditure could lift the neutral rate. Empirical estimates lag these narratives by quarters or years. Investors who declared a permanent “higher for longer” r-star without model support repeated the opposite mistake Harbor made with its frozen 2.5% anchor.

Policy stance and forward guidance

FOMC participants reference r-star indirectly when discussing whether policy is “restrictive enough” to return inflation to 2%. Chair commentary, SEP medians, and forward guidance all assume an implicit neutral benchmark. When r-star is uncertain, the committee may rely more on data dependence — hiking until something breaks, or holding longer than models suggest — which widens the gap between dot plots and realized paths.

For portfolio managers, the actionable question is not “what is r-star exactly?” but “how sensitive is my positioning to a 50–100 bp error in the neutral anchor?” Duration, bank equities, and rate-sensitive growth stocks all embed stance-gap assumptions even when models do not label them explicitly.

Harbor Credit Union ALM sleeve refactor

Harbor's pre-2024 stack used a fixed 2.5% real r-star in its Taylor overlay and a separate dot-plot import for committee narratives. The two inputs frequently contradicted each other; traders defaulted to whichever supported the prior week's position. The refactor:

Primary anchor: NY Fed HLW real r-star, refreshed quarterly with a 90-day stale alert.
Cross-check: median Survey of Primary Dealers long-run neutral minus 2% inflation target, winsorized at +/- 1.5% from HLW.
Stance dashboard: real fed funds (OIS) minus blended r-star, with uncertainty bands from the HLW confidence interval.
Scenario sleeve: duration stress at r-star +/- 75 bp to force explicit disagreement with the point estimate.
Communication sync: stance labels (accommodative / neutral / restrictive) gated on both gap magnitude and FOMC SEP language to reduce headline overreaction.

False cut signals (stance labeled restrictive while the Fed held or hiked) fell 28% over twelve FOMC cycles. The desk still missed the 2023 regional-bank episode — r-star models do not forecast credit events — but rate-path positioning improved materially once the neutral anchor moved with the data instead of a 2019 memo.

Technique decision table

Approach	Best when	Strength	Weakness
Fixed r-star constant (e.g. 2.5%)	Quick classroom Taylor-rule demos	Simple, transparent	Stale within 2–3 years; stance labels drift
HLW / Laubach-Williams filter	ALM, macro research, policy benchmarking	Economically grounded; Fed-published	Lags structural breaks; revises with GDP benchmarks
Market-implied neutral	Tactical trading, fast regime shifts	Forward-looking; updates daily	Term premium contamination; noisy in crises
Survey median (Primary Dealers)	Aligning with FOMC participant views	Human judgment embedded	Anchoring bias; slow to move at turning points
Taylor rule back-solve	Checking consistency of rate + gaps	Internal coherence check	Circular if gaps and r-star both estimated poorly
Dot plot as neutral proxy	Near-term terminal path only	Official committee dispersion	Not a neutral rate; mixes cyclical and structural

Production ALM and macro overlays should blend at least two methods — typically a structural filter plus a market or survey cross-check — and stress outcomes across a band, not a point.

Common pitfalls

Treating r-star as constant. Secular trends and post-pandemic fiscal shifts move the neutral rate over years; frozen anchors mislabel stance.
Confusing nominal and real. Comparing 5% fed funds to a 2.5% r-star without subtracting inflation doubles the error in restrictive labels.
Using long bond yields as r-star. Ten-year Treasuries include term premium and inflation expectations; they are not the short-run neutral rate.
Ignoring revision risk. HLW estimates change after GDP benchmark revisions; backtests on unrevised data overstate precision.
Single-model certainty. Announcing “policy is 150 bp restrictive” from one point estimate ignores 100+ bp model disagreement.
Dot plot substitution. SEP medians reflect participant forecasts under uncertainty, not measured r-star.
Neglecting global r-star. Open economies face spillovers; U.S. stance interacts with foreign neutral rates and FX.
Credit blind spots. r-star frameworks assume smooth transmission; banking stress can tighten conditions without further hikes.

Production checklist

Define r-star in real terms and document the inflation expectation proxy used.
Import a structural estimate (HLW or equivalent) with quarterly refresh and stale alerts.
Add a market or survey cross-check; flag when sources diverge by more than 75 bp.
Compute stance gap = real policy rate minus blended r-star; plot confidence bands.
Stress duration and equity sleeves at r-star +/- 50–100 bp scenarios.
Separate terminal-rate forecasts from neutral-rate anchors in dashboards.
Reconcile stance labels with FOMC SEP language; do not trade headlines alone.
Re-run backtests after GDP benchmark revisions and HLW model updates.
Log which r-star source drove each ALM committee decision for post-mortems.
Review secular-driver narratives (productivity, fiscal, demographics) annually.
Pair r-star stance with financial conditions indices for credit-channel context.
Document known limitations during QE/QT regimes when term premium is distorted.

Key takeaways

r-star is the real short rate consistent with full employment and stable inflation — unobservable but central to stance debates.
Estimates span wide bands; blend structural filters, surveys, and market checks instead of freezing one number.
Secular decline drivers (productivity, demographics, safe-asset demand) explain why the same nominal rate means different things over time.
Harbor Credit Union cut false cut signals 28% by replacing a 2019 r-star memo with time-varying HLW plus cross-checks.
Stance gaps belong in scenario bands — point-precision r-star claims mislead ALM and duration positioning.