Guide

Forward guidance explained

Harbor Credit Union's ALM desk read the September Summary of Economic Projections dot plot as a firm calendar: three 25-basis-point cuts in the following year, median fed funds at 4.6%. The treasury team added two years of duration to the investment book that afternoon. At the press conference, the Chair stressed that dots were forecasts, not promises and that policy would stay restrictive until inflation was clearly on a sustainable path to 2%. Two-year Treasury yields rose 18 basis points before the statement PDF finished circulating. Harbor's mark-to-market loss on the new sleeve exceeded a quarter's net interest margin — not because rates moved, but because forward guidance was repriced faster than the balance sheet could hedge.

Forward guidance is how central banks communicate the likely future path of policy rates — through speeches, FOMC statements, press conferences, and the quarterly dot plot. It is a policy tool distinct from today's hike or cut: by shaping inflation expectations and the yield curve, guidance can tighten or loosen financial conditions before the next meeting. After Harbor rebuilt its rate-path overlay with a guidance-type classifier (Delphic vs Odyssean, calendar vs outcome-based) layered on top of the Taylor rule baseline, drawdowns around FOMC weeks fell materially in backtests. This guide covers guidance taxonomy, transmission channels, the dot plot and SEP, credibility constraints, the Harbor Credit Union refactor, a technique decision table versus rate-hike-only models, pitfalls, and an investor checklist.

What forward guidance is — and what it is not

Monetary policy has two levers at every meeting: the current policy rate and the expected future path. Forward guidance operates on the second lever. A central bank that holds rates steady but signals “higher for longer” can push up two- and ten-year yields as much as a 25-basis-point hike — because bond prices embed the entire expected path, not only today's setting.

Guidance is not a binding contract. Unlike an Odyssean pledge tied to observable thresholds (discussed below), most FOMC communication is conditional: policy “will depend on the totality of incoming data.” Markets still treat it as a forecast with credibility weight — the gap between legal flexibility and market interpretation is where ALM desks win or lose.

Delphic guidance — describes the central bank's baseline forecast (“we expect to cut rates three times next year if inflation moderates”). Markets can disagree; the bank can revise without breaking a promise.
Odyssean guidance — ties future policy to observable outcomes (“rates will stay at this level until core PCE averages 2% for three months”). Stronger commitment; higher reputation cost if abandoned.
Calendar-based — pins action to dates (“no hikes before year-end”). Powerful at the zero lower bound; dangerous if inflation surprises.
Outcome-based (state-contingent) — pins action to inflation, unemployment, or financial-stability triggers. More flexible but harder for markets to price precisely.

Post-2022 tightening cycles leaned Delphic and data-dependent: the Fed retired explicit calendar promises from the zero-lower-bound era and emphasized “higher for longer” until expectations re-anchored. That shift increased volatility around SEP releases because dots and speeches could diverge within the same afternoon.

Transmission: how words move the curve

Forward guidance works through three overlapping channels:

Expectations channel — if households and firms believe the policy rate will stay elevated, they discount future cash flows at higher rates, cool hiring and capex, and moderate wage demands before hikes fully transmit.
Term premium and curve channel — short-rate expectations feed directly into two- through five-year yields; long-end moves also reflect growth and QE/QT supply. A hawkish shift in guidance steepens or flattens the curve depending on whether the market reprices cuts or hikes at the long end.
Financial conditions channel — equity risk premia, credit spreads, and FX react to the implied path. Tighter guidance can substitute for rate hikes when the bank wants to slow the economy without surprising fragile funding markets.

The effect is largest when guidance is credible and surprising. A widely expected dot plot produces little move; a press conference that contradicts the dots — as Harbor learned — can dominate the statement's rate decision. Research on the effective lower bound era showed Odyssean calendar guidance had measurable macro impact; in positive-rate regimes the marginal effect is smaller but still tradable around FOMC weeks.

The dot plot, SEP, and reading FOMC communication

The Federal Reserve's quarterly Summary of Economic Projections (SEP) publishes participant medians for GDP, unemployment, inflation, and the appropriate fed funds rate at year-end horizons. The famous dot plot shows each participant's rate dot anonymously — not a committee vote, not a commitment.

Practical reading rules:

Median vs dispersion — a tight cluster of dots signals consensus; wide vertical spread means markets should weight the press conference over the median.
Horizon mismatch — dots show year-end levels, not meeting-by-meeting paths. Interpolating a cut count from the median is fragile when the Chair emphasizes data dependence.
SEP inflation path — if the median core PCE projection rises while the rate median is unchanged, the implied real policy stance has eased even without a cut — a subtle form of guidance.
Statement vs dots vs Q&A — the FOMC statement is the official policy record; the Chair's Q&A often clarifies or softens it. Harbor now scores each layer separately instead of treating the dot plot as the single source of truth.

Other central banks publish variants: the ECB staff projections, Bank of England MPC minutes with vote splits, RBNZ OCR tracks. The taxonomy in this guide transfers; only the release calendar and legal mandate differ.

Credibility, time consistency, and the reversal problem

Forward guidance only works if markets believe the central bank will follow through — or pay a visible cost for deviating. Time inconsistency arises when a bank promises low rates to support employment today but would prefer to tighten tomorrow once inflation rises. Without reputation capital, Odyssean pledges are ignored.

Three credibility anchors matter in practice:

Inflation targeting history — decades of near-2% outcomes make “higher for longer” believable; a country with repeated de-anchoring needs stronger Odyssean thresholds.
Fiscal dominance risk — if markets suspect the treasury will pressure the central bank, guidance on real rates loses weight regardless of speeches.
Balance sheet alignment — QT that drains reserves while guidance promises easy money sends mixed signals; investors price the net stance, not the rhetoric alone.

Guidance reversals — walking back calendar cuts or extending a pause — are not always errors. When inflation shocks, Delphic guidance should update. The trading mistake is treating yesterday's dots as sticky when the mandate clearly requires a pivot.

Harbor Credit Union ALM sleeve refactor

Harbor's pre-refactor model fed the SEP median fed funds path directly into a Nelson-Siegel curve fit for duration targeting. It ignored guidance type and same-day communication overrides. After the September drawdown, ALM added three layers:

Guidance classifier — tags each FOMC cycle as predominantly Delphic or Odyssean based on statement language (explicit thresholds vs conditional forecasts).
Statement-dot divergence score — natural-language comparison of the Chair's press conference against the median dot shift; triggers hedge overlays when divergence exceeds a calibrated threshold.
Taylor overlay — blends SEP medians with a Taylor-rule-implied path so extreme dot outliers do not dominate positioning when macro slack and inflation gaps disagree.

Backtests from 2019 through 2025 showed lower peak drawdown in the two-week FOMC window and only a modest reduction in average carry — acceptable tradeoff for a mutual whose members are rate-sensitive borrowers. The refactor did not predict the dots; it sized risk for the possibility that guidance would be repriced intraday.

Technique decision table

Signal source	Best for	Update frequency	Main weakness
Current policy rate only	Overnight funding, pass-through pricing	Per meeting	Misses curve moves from guidance
SEP median dot plot	Baseline year-end rate forecasts	Quarterly	Not a commitment; intraday overrides
FOMC statement text	Official stance, threshold language	Per meeting	Lags press conference nuance
Chair press conference	Same-day repricing, divergence trades	Per meeting	Interpretation risk; no verbatim binding force
Market-implied OIS path	Continuous hedge ratios, RV vs dots	Continuous	Embeds risk premia, not pure policy forecast
Blended guidance index	ALM duration, multi-quarter NII forecasting	Per FOMC + weekly OIS	Requires calibration and regime rules

Use the realized policy rate for today's funding cost; use forward guidance and implied paths for duration and hedge ratios. Treat dots as one input among several, not the entire story.

Common pitfalls

Counting cuts from dot medians — year-end levels do not map cleanly to meeting cadence; use OIS for implied meeting paths.
Ignoring press conference overrides — the statement and dots can be hawkish while Q&A sounds dovish, or the reverse.
Treating all guidance as Odyssean — most modern Fed communication is Delphic; betting on binding calendar paths invites whipsaws.
Static Taylor benchmarks — r-star moves; a Taylor overlay needs refreshed neutral rate estimates, not 2019 parameters.
SEP lookahead in backtests — dots release at 2:00 p.m. ET with the statement; do not use them in pre-FOMC simulations.
Mixing QE with rate guidance — balance sheet policy shifts the term premium; a unchanged dot plot can still ease conditions via portfolio channels.
Single-country extrapolation — BOJ yield-curve control was Odyssean with different legal tools; do not paste Fed logic onto every central bank.
Neglecting fiscal news — large deficit revisions move long yields independently of FOMC dots; decompose before blaming guidance misses.

Production checklist

Classify each FOMC cycle as Delphic or Odyssean from statement language.
Track median dot shifts and vertical dispersion, not only the headline median.
Score statement vs press conference divergence within two hours of release.
Compare SEP inflation medians to rate medians for implicit stance changes.
Overlay market OIS paths to detect when investors disagree with dots.
Blend Taylor-rule baseline with SEP so outliers do not dominate ALM.
Hedge FOMC-week duration risk when divergence score exceeds threshold.
Archive statement PDFs and transcripts for reproducible NLP backtests.
Separate term-premium moves from pure policy-path repricing in attribution.
Monitor foreign central bank guidance when FX hedges embed rate differentials.
Stress-test guidance reversal scenarios (fewer cuts than dotted) quarterly.
Document which communication layer drove each post-meeting P&L move.

Key takeaways

Forward guidance shapes the expected policy path — and can move the yield curve as much as a rate change when it surprises.
Delphic forecasts differ from Odyssean commitments; most modern Fed guidance is conditional and revisable.
The dot plot is a forecast distribution, not a vote or promise — press conferences often override the median.
Harbor Credit Union cut FOMC-week drawdowns by layering a guidance classifier and divergence score on top of SEP medians.
Credible guidance anchors inflation expectations; mixed signals from QT, fiscal news, or reversed dots punish static duration bets.