How tight is US credit?
One weekly number for how hard it is to borrow in the US, built from corporate bond spreads, commercial paper, bank business lending and the Fed's survey of loan officers, mortgage rates, the fed funds rate, the yield curve, stock-market volatility and drawdowns, and the dollar. Zero is the average since 2001; +1 means one standard deviation tighter than that. Five sub-indices beside it show where the tightness is: for companies, for home buyers, in the policy rate and the curve, in markets, or in private credit.
Day by day
The headline index recomputed every business day with the same weights: spreads, commercial paper, fed funds, the curve, the VIX, stocks and the dollar at each day's value; the loan officer survey, bank lending and the weekly mortgage rate at their latest release. Bars: what each group contributes (above zero tighter). Daily values are noisier than the weekly averages above.
The index since 1990
Weekly, Friday dates. Grey bands: US recessions (NBER). Dashed: the Chicago Fed's National Financial Conditions Index credit subindex for comparison (also 0 = average, positive = tighter; its own scale).
Where the tightness is
Each sub-index is the plain average of its inputs' tightness scores (standard deviations from the 2001–present average, oriented so higher = tighter). They can disagree: companies can borrow cheaply while mortgages are expensive.
What is moving the headline
The headline index split into its inputs' contributions, grouped. Bars above zero push it tighter, below zero looser; the line is the index.
Private credit
Private lending to companies (much of it now funding data-centre and AI build-outs) has no public price, so two windows stand in. Left: the listed lenders, business development companies, below their 52-week highs (total return; the dotted line is the BDC ETF); a fall that high-yield bonds do not share is stress specific to private credit. Right: commercial banks' loans to nonbank financial institutions, the bank credit lines behind private-credit funds (H.8; the January 2025 reclassification jump is linked out of the growth rate). These series start between 2005 and 2021, so they form their own sub-index and are not in the headline.
The inputs this week
Tightness score = standard deviations from the 2001–present average, oriented so a positive score is tighter (blue: looser than average, red: tighter). Weight = the input's loading in the headline index.
Market credit spreads (last three years)
ICE BofA option-adjusted spreads over Treasuries. FRED carries only the last three years of these series, too short to estimate weights, so they are shown here rather than inside the index; Moody's Baa and Aaa spreads carry the corporate-bond signal in the index back to the 1980s.
How the index is built
- Inputs (all from FRED, St. Louis Fed): Moody's Baa and Aaa corporate yields minus the 10-year Treasury; 3-month AA nonfinancial and financial commercial paper minus the 3-month bill; year-on-year growth of commercial paper outstanding, of bank commercial & industrial loans (H.8) and of nonfinancial corporate debt (Financial Accounts, Z.1); the Senior Loan Officer Opinion Survey's net share of banks tightening business-loan standards (large & middle-market and small firms) and net share reporting stronger demand; the effective fed funds rate and the Freddie Mac 30-year mortgage rate, both real (minus core PCE inflation over the past year); the mortgage rate minus the 10-year Treasury; and the 10-year minus 2-year Treasury spread.
- Markets: the VIX (weekly average), the S&P 500's drawdown from its highest close of the past year, and the year-on-year change in the Fed's broad trade-weighted dollar (spliced to its predecessor before 2006). With them the index also registers market-led tightenings that loan officers and spreads see later: late 2018, 2022, April 2025.
- Orientation: every input is signed so that a higher value means tighter credit (slower loan and debt growth, weaker loan demand and a flatter curve count as tighter).
- Real rates: the fed funds and mortgage rates fell for forty years; in nominal terms that decline would dominate any statistical combination, so both enter net of core inflation.
- Weights: the headline is the first principal component of the standardised inputs over 2001–present (when every input exists), signed so wider corporate spreads count as tighter and scaled to one standard deviation. It captures the common credit cycle: loan-officer standards, corporate spreads and commercial paper carry most of the weight, while the real fed funds rate, loan growth and corporate debt growth barely load and the yield curve loads with the opposite sign (curves steepen in recessions as spreads widen). That is why the three sub-indices, which weight their inputs equally, sit beside it.
- No hindsight: each input enters a week only once it would have been published (bank loans about 9 days after the data week, core PCE about two months after the month, the loan officer survey about 50 days after its quarter date, Z.1 about five months after its quarter), and is carried forward to the next release. Daily series are weekly averages. Before 2001 the index uses the inputs that existed, re-weighted.
- Check: the index correlates with the Chicago Fed's NFCI credit subindex since 1990 and peaks in every credit event: 1998, 2001, 2008, 2020 and 2023.
Data: Federal Reserve Bank of St. Louis (FRED); Board of Governors of the Federal Reserve System (H.8, H.15, Commercial Paper, Senior Loan Officer Opinion Survey, Financial Accounts Z.1); Moody's; Freddie Mac; ICE Data Indices (ICE BofA spreads, via FRED); Chicago Fed (NFCI); NBER recession dates. Not investment advice.