Skip to content

← Macro intelligence

NY Fed Recession Probability

Twelve-month-ahead U.S. recession probability derived from the 10-Year minus 3-Month Treasury yield curve spread, using the public Estrella & Hardouvelis 1991 probit methodology calibrated on 1959-2009 historical data. Inversions historically precede effective recessions within 12 months.

Tier proMacro intelligencemacro-cyclesrecession-probabilityyield-curve10y3mny-fedestrella-hardouvelis

What is it?

This metric computes the 12-month-ahead recession probability using the public Estrella & Hardouvelis 1991 methodology published by the Federal Reserve Bank of New York. The formula is P(recession_12m_ahead) = Φ(α + β × spread_T10Y3M), where α ≈ -0.535 and β ≈ -0.604 are public FRB-NY coefficients calibrated on the historical period 1959-2009. The T10Y3MM spread (10-year Treasury minus 3-month Treasury, from FRED) is the input - inversion (negative spread) historically precedes recessions within 12 months. The reading is bounded between 0 and 100 percent.

How to read

The horizontal axis is time, the vertical axis is the recession probability as a percentage from 0 to 100. Horizontal threshold lines mark the low probability ceiling (10%, dashed green) and the recession-likely zone entry (30%, solid red). Colored zones reinforce the regimes : green band 0-10% (low risk) and red band 30-100% (elevated probability). Historically, the methodology has correctly preceded the 1990, 2001, 2008, and 2020 U.S. recessions with peaks above 30% in the 12 months prior. The enriched tooltip shows the current probability and the underlying T10Y3M spread.

Key zones

• Below 10 percent: low recession probability, normal yield curve (10Y above 3M significantly) • 10-30 percent: moderate probability, progressively flattening curve • 30-50 percent: elevated probability, moderately inverted curve (slightly negative T10Y3M) • Above 50 percent: very high probability, severe inversion historically preceding recession within 12 months across multiple documented occurrences since 1981 (1990, 2001, 2008, 2020)

What to observe

• Crossings of the 30 percent line as recession-likely zone entry markers (historically precede effective recessions within 12 months) • The duration of the inversion as a severity indicator - short inversions under 6 months are less reliable than sustained inversions over 12 months • The exit from inversion (probability moving from above 30 percent to below 30 percent) often coincides with recession start - this is a documented FRB-NY observation: de-inversion precedes recession rather than inversion itself

Historical context

The Estrella & Hardouvelis 1991 methodology is observable since 1981 (T10Y3MM full series in FRED). Historical recessions preceded by probability above 30 percent include the 1990-1991 recession with probability peak late 1989, the 2001 dotcom recession with peak late 2000, the 2008-2009 Great Recession with peak late 2006-2007, and the 2020 COVID recession with peak late 2019. The methodology has a solid track record across approximately 5 documented recessions over the 1981-2024 period. The full pattern of inversion plus de-inversion plus effective recession is generally spread over 18-24 cumulative months.

Expert notes

The Estrella & Hardouvelis 1991 methodology is a simple probit regression on the T10Y3M spread - it does not capture all possible recession regimes, notably exogenous shock-induced recessions like COVID 2020 where prediction was partially shifted in time. Complement with cross-source analysis (PMI, employment, retail sales) for a holistic view. The α and β FRB-NY coefficients are calibrated on 1959-2009 - possible future re-calibration per structural economic evolutions post-2009 such as ZIRP rate floors and permanent QE. **Cross-rubric reading** - Recession probability is intrinsically a tradFi macro reading. No on-chain counterpart is semantically applicable, since the metric captures a macro-economic dynamic, not a blockchain primitive.

Common mistakes to avoid

Do not infer an exact recession timing from the probability - the 12-month window is a historical average, actual timing varies from 6 to 18 months depending on the cycle. Do not confuse recession probability (a statistical model output) with official recession (NBER retrospective dating). Do not use as a sole investment timing indicator - couple with PMI, employment, and leading economic indicators for cross-source validation.

Programmatic access

REST API

curl -sS \
  'https://api.trinityinsights.io/api/v1/macro-intelligence/macro-v2-ny-fed-recession-probability/history?days=90' \
  -H 'X-API-Key: $TRINITY_API_KEY'

MCP server

{
  "tool": "get_chart_value",
  "metric_id": "macro-v2-ny-fed-recession-probability",
  "timeframe": "1y"
}

Required tier: pro. See the pricing grid for the tier list and the MCP documentation for multi-client configuration.

Related metrics

Institutional disclaimer

Trinity Insights is an educational and analytical tool. The metric above does not constitute investment advice. Trinity Insights is not a Crypto-Asset Service Provider (CASP) registered under MiCA Regulation (EU) 2023/1114. See the full disclaimer.