Rates tight, credit loose: conditions sit 0.13 standard deviations easier than normal
Are overall financial conditions tightening or easing, and what is driving it?
In brief
Are financial conditions tightening or easing, and what is moving them?- Rates are tight (+0.40σ) and credit is loose (−0.54σ) in Sep 2026; on balance conditions are easier than normal.→ Exhibit 1
- The index is −0.13 standard deviations from its 1991–2019 average in Sep 2026: conditions are easier than normal.→ Exhibit
- Conditions were little changed over three months (−0.02σ, under half the 2015–19 standard deviation of three-month moves) and tightened by 0.00σ over twelve.
- The weighting choice does not change the sign of the reading: equal weights −0.13, principal component −0.50.
- The index barely predicts output: a tighter reading preceded industrial growth 2.3 points lower a year on, R² 0.08.→ Exhibit
| What would change this view | Trigger | Today | Decided by |
|---|---|---|---|
| Conditions move | equal-weight index outside −0.20 to −0.06 (half a typical three-month move either way) | −0.13 (Sep 2026)not crossed | the next three monthly readings, the first for Oct 2026 |
The reasoning behind each point
- SplitRates are tight (+0.40σ) and credit is loose (−0.54σ) in Sep 2026; on balance conditions are easier than normal.
Each is the index made from its own inputs alone: rates from the real policy rate, the 10-year yield and the curve; credit from the corporate spread, the excess bond premium and banks' lending standards. Market inputs alone read −0.13σ. The equal-weight index of all seven is −0.13σ from its 1991–2019 average.
- LevelThe index is −0.13 standard deviations from its 1991–2019 average in Sep 2026: conditions are easier than normal.
42nd percentile since 1991. The principal-component version reads −0.50; the seven leave-one-out versions range from −0.31 to −0.03. Tightest inputs: the yield curve (+1.0σ) and the 10-year yield (+0.3σ); easiest: the excess bond premium (−0.7σ).
- DirectionConditions were little changed over three months (−0.02σ, under half the 2015–19 standard deviation of three-month moves) and tightened by 0.00σ over twelve.
Over three months the largest contributions came from bank lending standards (−0.05σ) and the 10-year yield (+0.04σ); over twelve months from the 10-year yield (+0.07σ) and bank lending standards (−0.06σ).
- RobustnessThe weighting choice does not change the sign of the reading: equal weights −0.13, principal component −0.50.
The first principal component explains 40% of the inputs' variance and weights the corporate spread most (23%). Market inputs alone read −0.13; rate inputs alone +0.40; credit inputs alone −0.54.
- Does it lead?The index barely predicts output: a tighter reading preceded industrial growth 2.3 points lower a year on, R² 0.08.
Estimated on 1991–2009, per standard deviation of the index. In 1991–2019 the index is most closely correlated with production growth 3 months later (correlation −0.40). Out of sample the twelve-month-ahead forecast error is 114% of a constant-mean forecast's in 2010–2019 and 112% in 2020–2025 (below 100% = the index helps). The index is a description of conditions, not a forecast; the 2020–21 recession and recovery were driven by the pandemic, not by financial conditions.
Next: Policy rates — the rate behind them all
Conditions against their own normal
Are conditions tighter or easier than normal, on this index?Financial conditions index since 1991, under equal and principal-component weights
About this exhibit
Conditions index levels and one-, three- and twelve-month changes by weighting
| Series | Level | 1m | 3m | 1y | Pct since 1991 | Level, recent | As of |
|---|---|---|---|---|---|---|---|
| The index | |||||||
| Financial conditions index, equal weights | -0.13 sd | +0.03 sd | −0.02 sd | ±0.00 sd | 42 | Sep 2026 | |
| Financial conditions index, first principal component | -0.50 sd | −0.09 sd | −0.11 sd | −0.11 sd | 22 | Sep 2026 | |
- Range
Seven leave-one-out versions, each dropping one input: −0.31 to −0.03. This is specification uncertainty, not a confidence interval.
About this exhibit
The seven inputs, and which way each pulls
Which of the seven inputs are tightening conditions, and by how much?Exhibit 1 · Each input's contribution to the financial conditions index
About this exhibit
Input contributions to the conditions index over one, three and twelve months
| Input | 1 month | 3 months | 12 months | Level (today) |
|---|---|---|---|---|
| Real policy rate | ±0.00 | ±0.00 | −0.06 | −0.02 |
| 10-year yield | +0.03 | +0.04 | +0.07 | +0.05 |
| Curve (2y−10y) | +0.02 | ±0.00 | +0.03 | +0.14 |
| Corporate spread | ±0.00 | +0.02 | +0.02 | −0.10 |
| Excess bond premium | −0.04 | −0.02 | −0.02 | −0.11 |
| Dollar, 12m | +0.02 | −0.01 | +0.02 | −0.07 |
| Bank standards | ±0.00 | −0.05 | −0.06 | −0.02 |
| Index | +0.03 | −0.02 | ±0.00 | −0.13 |
About this exhibit
The index inputs with their latest values and norms, z-scores and weights
| Input | Latest | Last observed | Mean | SD | z | Equal weight | PCA weight |
|---|---|---|---|---|---|---|---|
| Real policy rate: federal funds less core PCE inflationfederal funds monthly average less PCE ex food and energy, y/y | 0.63 % | Sep 2026 | 0.85 | 2.00 | −0.11 | 0.143 | −0.136 |
| 10-year Treasury yieldH.15 constant maturity, monthly average | 4.99 % | Sep 2026 | 4.37 | 1.80 | +0.34 | 0.143 | −0.124 |
| Yield curve: 2-year less 10-year Treasuryinverted = positive = tighter | -0.34 pp | Sep 2026 | -1.19 | 0.88 | +0.97 | 0.143 | −0.123 |
| Corporate spread: HQM 10-year par less Treasuryinvestment grade | 78.00 bp | Sep 2026 | 124.06 | 64.49 | −0.71 | 0.143 | +0.229 |
| Excess bond premiumrisk appetite; Gilchrist–Zakrajšek | -39.19 bp | Sep 2026 | 8.29 | 63.78 | −0.75 | 0.143 | +0.203 |
| Broad dollar, 12-month changeappreciation = tighter | -0.68 % | Sep 2026 | 2.20 | 6.02 | −0.48 | 0.143 | +0.020 |
| Banks tightening C&I standards (large and middle-market firms)net percentage, latest survey carried until the next | 0.00 % | Jul 2026 | 3.48 | 21.53 | −0.16 | 0.143 | +0.166 |
First principal component: eigenvalue 2.78 of 7, 40% of variance; second 2.18.
About this exhibit
Each index input in standard deviations from its 1991–2019 average
About this exhibit
The index inputs and their one-, three- and twelve-month changes
| Series | Level | 1m | 3m | 1y | Pct since 1991 | Level, recent | As of |
|---|---|---|---|---|---|---|---|
| Inputs (signed: higher = tighter) | |||||||
| Real policy rate: federal funds less core PCE inflationfederal funds monthly average less PCE ex food and energy, y/y | 0.63% | +3bp | +4bp | −77bp | 57 | Sep 2026 | |
| 10-year Treasury yieldH.15 constant maturity, monthly average | 4.99% | +31bp | +52bp | +87bp | 69 | Sep 2026 | |
| Yield curve: 2-year less 10-year Treasuryinverted = positive = tighter | -0.34 pp | +0.12pp | +0.02pp | +0.21pp | 71 | Sep 2026 | |
| Corporate spread: HQM 10-year par less Treasuryinvestment grade | 78bp | −1bp | +7bp | +9bp | 20 | Sep 2026 | |
| Excess bond premiumrisk appetite; Gilchrist–Zakrajšek | -39bp | −17bp | −11bp | −11bp | 13 | Sep 2026 | |
| Broad dollar, 12-month changeappreciation = tighter | -0.7% | +0.75pp | −0.26pp | +0.84pp | 34 | Sep 2026 | |
| Banks tightening C&I standards (large and middle-market firms)net percentage, latest survey carried until the next | 0.0% | ±0.00pp | −8.10pp | −9.50pp | 52 | Sep 2026 | |
About this exhibit
Does the weighting matter?
Would another weighting change the reading, and has the index led output?Conditions index from all inputs and from each input group
About this exhibit
Versions of the conditions index, latest and twelve-month change
| Version | Latest | 12m chg | Corr. |
|---|---|---|---|
| Equal weights | −0.13 | ±0.00 | 1.00 |
| First principal component | −0.50 | −0.11 | 0.23 |
| Leave out: the real policy rate | −0.13 | +0.07 | 0.97 |
| Leave out: the 10-year yield | −0.21 | −0.08 | 0.97 |
| Leave out: the yield curve | −0.31 | −0.04 | 0.96 |
| Leave out: the corporate spread | −0.03 | −0.02 | 0.97 |
| Leave out: the excess bond premium | −0.03 | +0.03 | 0.98 |
| Leave out: the dollar | −0.07 | −0.02 | 0.97 |
| Leave out: bank lending standards | −0.12 | +0.07 | 0.97 |
| Market inputs only | −0.13 | +0.17 | 0.95 |
| Credit inputs only (spread, EBP, standards) | −0.54 | −0.16 | 0.67 |
| Rate inputs only (real rate, 10-year, curve) | +0.40 | +0.11 | 0.63 |
About this exhibit
Conditions index correlation with later industrial production growth, by lead
About this exhibit
Year-ahead industrial production forecasts from the index against a constant forecast
| Sample | Months | Model RMSE | Constant RMSE | Ratio | Corr. |
|---|---|---|---|---|---|
| Estimation 1991–2009 (b -2.34, R² 0.08) | 228 | – | – | – | – |
| Test 2010–2019 | 120 | 4.66 | 4.08 | 114% | -0.56 |
| Test 2020–2025 | 69 | 3.74 | 3.35 | 112% | -0.04 |
Ratio below 100%: the index beats a constant forecast. The z-score window runs to 2019, so only the second test period is fully out of sample.
About this exhibit
Bank lending standards and demand
Are banks tightening the terms they offer, and is loan demand holding up?Net share of banks tightening lending standards, by loan type
About this exhibit
Net share of banks reporting stronger loan demand, by loan type
About this exhibit
Standards and demand by loan type, Q3 2026 survey
| Series | Net % | 1 survey | 4 surveys | Pct since 1990 | Level, 6 years | As of |
|---|---|---|---|---|---|---|
| Standards: net % of banks tightening | ||||||
| C&I, large and middle-market firms | ±0.0% | −8.1pp | −9.5pp | 51 | Q3 2026 | |
| C&I, small firms | +1.8% | −4.8pp | −6.4pp | 53 | Q3 2026 | |
| CRE, construction and land development | −3.7% | −8.6pp | −13.4pp | 17 | Q3 2026 | |
| CRE, nonfarm nonresidential | −11.3% | −8.0pp | −22.8pp | 4 | Q3 2026 | |
| CRE, multifamily | −5.7% | −5.7pp | −10.5pp | 15 | Q3 2026 | |
| Mortgages, GSE-eligible | ±0.0% | +3.7pp | ±0.0pp | 77 | Q3 2026 | |
| Home equity lines | −4.0% | −5.8pp | −9.2pp | 25 | Q3 2026 | |
| Credit cards | +6.7% | +4.7pp | −3.7pp | 54 | Q3 2026 | |
| Auto loans | −4.5% | −2.5pp | −0.7pp | 45 | Q3 2026 | |
| Other consumer loans | ±0.0% | −5.5pp | −3.8pp | 56 | Q3 2026 | |
| Demand: net % of banks reporting stronger demand | ||||||
| C&I, large and middle-market firms | +16.1% | +11.3pp | +44.7pp | 78 | Q3 2026 | |
| C&I, small firms | +3.6% | +3.6pp | +31.5pp | 59 | Q3 2026 | |
Appendix
Specification, inputs, exclusions, sensitivity and the test30 checks, all pass
Specification and caveats
- ModelStandardised, then averaged.
Seven published inputs, each signed so that higher means tighter, standardised with its mean and standard deviation over 1991–2019 (fixed, so a new month never rescales history). Equal-weight index: the average z-score. PCA index: weights from the first principal component of the inputs' correlation matrix over the same window, scaled so the absolute weights sum to one, with the sign set so that a wider corporate spread reads tighter. Because spreads widen when rates fall in downturns, the first component can put negative weight on the rate inputs: it is a credit-stress factor rather than a policy-stance factor. Units: standard deviations; zero = the 1991–2019 average.
- InputsAll official data.
Real policy rate: H.15 federal funds (monthly average) less the twelve-month change in BEA's PCE price index excluding food and energy. 10-year Treasury yield and the 2-year less 10-year curve: H.15 constant maturities. Corporate spread: Treasury HQM 10-year par yield less the 10-year Treasury. Excess bond premium: the Board's FEDS Notes file. Dollar: twelve-month change in the Board's nominal broad dollar index, spliced before January 2006 onto the goods-only broad index by its level ratio in January 2006. Bank lending standards: SLOOS net percentage of domestic banks tightening C&I standards for large and middle-market firms, assigned to the survey month and carried until the next survey.
- ExcludedWhat is not in it.
Not used: equity prices and volatility (the only equity source available is licence-restricted), mortgage rates (no source with display rights confirmed), and the Chicago Fed NFCI and other published indexes (licence). Leaving equities out makes the index slower to register market stress than indexes that include them.
- Ragged edgeLatest month.
Inputs arrive at different lags. An input's latest value is carried forward at most 2 months to complete the latest month; carried values are listed under gaps and the index month is the last month in which every input is observed or carried.
- SensitivitySpecification, not sampling, uncertainty.
Sensitivity: the principal-component weights, seven leave-one-out indexes (the shaded range), and indexes from market, rate and credit inputs alone. The spread between versions is specification uncertainty, not sampling error; no statistical confidence band is claimed.
- Out of sampleDoes it lead output?
Out-of-sample note: industrial production growth twelve months ahead regressed on the equal-weight index over 1991–2009, then evaluated on 2010–2019 and 2020–2025 against a constant forecast (the estimation-period mean). The z-score window (to 2019) overlaps the first test period, so only the second is fully out of sample. Payroll growth is not used because BLS files require a contact header this page's data requests do not carry.
- SLOOSNet percentages.
SLOOS: net percentage = share of banks reporting tightening (considerably or somewhat) less the share reporting easing, among domestic respondents; demand: stronger less weaker. Surveys are dated by the Board to the quarter in which they are conducted (January, April, July, October) and released a few weeks later.
- ChecksInputs, arithmetic, weights.
Inputs are matched to their publishers (H.15 monthly averages, BEA inflation, HQM par yields, H.10 averages, SLOOS bank-group bounds); z-scores, contributions and PCA weights are verified to reproduce the index. 30 checks run at every update.
Sources
- publicBoard of Governors of the Federal Reserve System, H.15 Selected Interest Rates: federal funds (daily and monthly), Treasury constant maturities 2 and 10 years (monthly)
Source: Board of Governors of the Federal Reserve System · https://www.federalreserve.gov/releases/h15/
- publicU.S. Bureau of Economic Analysis, NIPA Table 2.8.4 and 2.8.11: PCE excluding food and energy price index (NipaDataM.txt)
Source: U.S. Bureau of Economic Analysis · https://apps.bea.gov/national/Release/TXT/NipaDataM.txt
- publicU.S. Department of the Treasury, High Quality Market (HQM) Corporate Bond Yield Curve, par yields (hqm_qh_pars.xls)
Source: U.S. Department of the Treasury · https://home.treasury.gov/data/treasury-coupon-issues-and-corporate-bond-yield-curve/corporate-bond-yield-curve
- publicBoard of Governors of the Federal Reserve System, FEDS Notes excess bond premium file (ebp_csv.csv)
Source: Board of Governors of the Federal Reserve System; Favara, Gilchrist, Lewis and Zakrajšek (2016), updated · https://www.federalreserve.gov/econres/notes/feds-notes/ebp_csv.csv
- publicBoard of Governors of the Federal Reserve System, H.10 Foreign Exchange Rates: nominal broad dollar index (daily and monthly; goods-only broad index before 2006)
Source: Board of Governors of the Federal Reserve System · https://www.federalreserve.gov/releases/h10/
- publicBoard of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey on Bank Lending Practices (Data Download Program)
Source: Board of Governors of the Federal Reserve System · https://www.federalreserve.gov/data/sloos.htm
- publicBoard of Governors of the Federal Reserve System, G.17 Industrial Production (out-of-sample evaluation only)
Source: Board of Governors of the Federal Reserve System · https://www.federalreserve.gov/releases/g17/
Reproduction checks
| Check | Published | Computed | Tolerance | Result |
|---|---|---|---|---|
| H.15 monthly federal funds = calendar-day average of the daily rate (weekends and holidays carry the previous business day), Jan 1990–Sep 2026 (441 months; months differing by more than 0.005) | 0 | 0 | 0 | pass |
| Core PCE inflation: twelve-month change of BEA's price index (ex food and energy) = BEA's published percent change from a year earlier, Jan 1991–Sep 2026 (429 months; max abs error, pp) | 0 | 0.05 | 0.051 | pass |
| HQM 10-year par yield used in the credit input = par yield from Treasury's HQM spot curve, every month since Jan 1991 (429 months; max abs error, pp) | 0 | 0.005 | 0.005 | pass |
| H.10 monthly broad dollar index = average of the daily index, Jan 2006–Sep 2026 (249 months; months differing by more than 0.005) | 0 | 0 | 0 | pass |
| Dollar splice: monthly changes of the goods-only broad index (before Jan 2006) and the current broad index move together over their 167-month overlap (correlation ≥ 0.9) | 1 | 0.996 | 0.1 | pass |
| SLOOS: the all-domestic-bank net percentage lies between the large-bank and other-bank net percentages (a respondent-weighted average), 20 standards and demand series, every survey (1,561 surveys; violations beyond 0.05) | 0 | 0 | 0 | pass |
| Inputs complete over the standardisation window 1991-01–2019-12 (missing input-months) | 0 | 0 | 0 | pass |
| Z-scores: every input has mean 0 and standard deviation 1 over the window (max abs deviation) | 0 | 0 | 0 | pass |
| Equal-weight index = average of the seven z-scores, every month (max abs gap) | 0 | 0 | 0 | pass |
| Principal component: corr(z) × w = λ × w for the first eigenvector (λ = 2.78, 40% of variance; max abs residual) | 0 | 0 | 0 | pass |
| PCA absolute weights sum to 1 and reproduce the PCA index as Σ w × z, every month (max abs gap) | 0 | 0 | 0 | pass |
| Equal-weight index: input contributions add to the change over 1 month to Sep 2026 | 0.025 | 0.025 | 0 | pass |
| Equal-weight index: input contributions add to the change over 3 months to Sep 2026 | -0.021 | -0.021 | 0 | pass |
| Equal-weight index: input contributions add to the change over 12 months to Sep 2026 | 0.001 | 0.001 | 0 | pass |
| PCA index: input contributions add to the change over 1 month to Sep 2026 | -0.094 | -0.094 | 0 | pass |
| PCA index: input contributions add to the change over 3 months to Sep 2026 | -0.114 | -0.114 | 0 | pass |
| PCA index: input contributions add to the change over 12 months to Sep 2026 | -0.108 | -0.108 | 0 | pass |
| Equal-weight index lies inside the range of the seven leave-one-out indexes every month (months outside) | 0 | 0 | 0 | pass |
| Nber.month_text_equals_month_number.peaks_max_days | 0 | 0 | 0 | pass |
| Nber.month_text_equals_month_number.troughs_max_days | 0 | 0 | 0 | pass |
| Nber.contraction_equals_trough_minus_peak.max_months | 0 | 0 | 0 | pass |
| Nber.stated.april_2020.row_found | 1 | 1 | 0 | pass |
| Nber.stated.february_2020.band_from_days_off | 0 | 0 | 0 | pass |
| Nber.stated.april_2020.band_to_days_off | 0 | 0 | 0 | pass |
| Nber.stated.june_2009.row_found | 1 | 1 | 0 | pass |
| Nber.stated.december_2007.band_from_days_off | 0 | 0 | 0 | pass |
| Nber.stated.june_2009.band_to_days_off | 0 | 0 | 0 | pass |
| Nber.json_equals_excel.rows_differing | 0 | 0 | 0 | pass |
| Nber.page_table_equals_excel.rows_differing | 0 | 0 | 0 | pass |
| Nber.page_states_the_band_definition | 1 | 1 | 0 | pass |
- Board of Governors of the Federal Reserve System, H.15 Selected Interest Rates: federal funds (daily and monthly), Treasury constant maturities 2 and 10 years (monthly) — Source: Board of Governors of the Federal Reserve System
- U.S. Bureau of Economic Analysis, NIPA Table 2.8.4 and 2.8.11: PCE excluding food and energy price index (NipaDataM.txt) — Source: U.S. Bureau of Economic Analysis
- U.S. Department of the Treasury, High Quality Market (HQM) Corporate Bond Yield Curve, par yields (hqm_qh_pars.xls) — Source: U.S. Department of the Treasury
- Board of Governors of the Federal Reserve System, FEDS Notes excess bond premium file (ebp_csv.csv) — Source: Board of Governors of the Federal Reserve System; Favara, Gilchrist, Lewis and Zakrajšek (2016), updated
- Board of Governors of the Federal Reserve System, H.10 Foreign Exchange Rates: nominal broad dollar index (daily and monthly; goods-only broad index before 2006) — Source: Board of Governors of the Federal Reserve System
- Board of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey on Bank Lending Practices (Data Download Program) — Source: Board of Governors of the Federal Reserve System
- Board of Governors of the Federal Reserve System, G.17 Industrial Production (out-of-sample evaluation only) — Source: Board of Governors of the Federal Reserve System