Charts — jobs, inflation, and the Fed vs the bond market

Bar and line charts over the last decade. Thesis tested here: the Fed is steering to the bond market (the 2-year yield), not to its stated dual mandate of 2% inflation and full employment — and a single "jobs number" / "inflation number" hides opposite sectoral and labor-slack trends.

All series are annual, compiled from public data (FRED / BLS / US Treasury / BIS / BOJ), rounded to the precision shown; FRED/BLS series IDs are noted under each chart. "~" marks an estimate or partial year. Verify any value at the cited series. Companion data + sourcing: macro-jobs-inflation-fed.

Jobs: nonfarm payroll change per year — and the downward benchmark revisions

The headline overstated job creation in 2024-25; the QCEW benchmarks cut ~1.5M jobs across the two years — the correction arriving 6-18 months late.

Jobs: where they grew and shrank (recent, by sector)

Growth is concentrated in health care, leisure/hospitality, and state/local government; manufacturing, information (tech), temp help, and federal government are shrinking. One 'jobs number' hides opposite sectoral trends.

Inflation: never actually at 2% (CPI vs core PCE vs the target)

In 11 years, core PCE sat within 2%±0.5 only a handful of times; the index spent the era either well below (2015-20) or far above (2021-24) the supposed target.

True labor slack: U-6 (underemployment) vs U-3 (headline)

U-6 — which counts discouraged and involuntarily part-time workers, and the gig/underemployed — runs ~3-5 pts above the headline the Fed cites.

The dual-mandate test: the funds rate tracks neither 2% inflation nor full employment

If the Fed steered to '2% inflation + full employment,' the funds line would track those. It doesn't — it sat at zero through 2%+ core PCE (2021) and stayed >5% as inflation fell and U-6 rose.

The real driver: the funds rate tracks the BOND MARKET (the 2-year yield leads it)

The funds rate and the 2-year yield move together, with the 2Y turning FIRST at every pivot (it fell ahead of the 2019 and 2024 cuts; rose ahead of the 2022 hikes). The Fed is following the bond market's expectation, not delivering a mandate.

Tick-by-tick (monthly): the funds rate FOLLOWS the 2-year Treasury

Lead-lag of month-over-month CHANGES: the 2-year yield LEADS the funds rate by ~1 month(s) (peak corr 0.49; contemporaneous corr 0.41). The Fed moves after the 2Y, not before it. At each pivot the 2Y turns first (it fell ahead of the 2019 and 2024 cuts and rose ahead of the 2022 hikes) — the bond market sets the path the Fed ratifies.

Daily resolution: the funds rate is a STEP the 2-year Treasury reaches first

Daily, 2862 trading days 2015-01-02..2026-06-11 (FRED DFF / DGS2 / DGS3MO). Lead-lag of 21-day changes: the 2Y LEADS the funds rate by ~33 trading days (~1.6 mo; peak corr 0.37 vs 0.32 at lag 0). The funds rate (DFF) is a near-step that jumps only at FOMC meetings; the 2Y is continuous and has already moved to the new level before each step — the bond market prices the decision first.

Sharp & fast — the SHORT-END gap (2Y/3M minus funds) + the 2Y−3M policy-path

The short-end gap turns FAST: deeply negative ahead of the 2019/2020/2024 cuts, strongly positive ahead of the 2022 hikes — the market pre-committing the Fed. The 2Y−3M 'policy-path' spread sits in the middle: it captures the expected ~2-year rate path vs the front, smoother than the 3M−funds jump-detector but faster than the 2s10s curve.

Smooth & delayed — the CURVE spreads (10Y minus 2Y / 3M)

The 2s10s/3m10s curve is far smoother but slower: it inverted through 2022-24 (recession signal) and is a cleaner but lagged read than the jumpy short-end gap.

30Y differentials: the long-end term premium (30Y−10Y, 30Y−2Y)

The 30Y−10Y term premium was compressed/near-zero through 2015-21 (QE era), inverted with the rest of the curve in 2022-23, then STEEPENED sharply in 2024-26 as long-end supply (deficits) and term premium returned — the smoothest, slowest-moving differential, reflecting fiscal/duration risk rather than near-term policy.

Corporate bond YIELDS over time (Moody's Baa, Aaa vs the 10Y Treasury)

Corporate borrowing costs (Baa/Aaa) track the Treasury but with a CREDIT SPREAD on top; both repriced from the ~3-4% 2015-21 floor to ~5-6% by 2022-26.

Credit SPREAD over time — Baa minus the 10Y Treasury (full history)

The credit spread is the market's default-risk read: it spiked in early 2016 (energy bust) and Mar-2020 (COVID), then COMPRESSED to cycle-lows by 2024-26 — credit priced almost no stress even as the self-marked private-credit risk built (macro-private-credit-marks).

Option-adjusted spreads — HY vs IG (2023-07+)

Even on the available window, high-yield and investment-grade OAS sit near cycle lows — the corporate market is pricing minimal default risk into 2026.

Regional sovereign 10Y + a GDP-weighted GLOBAL long rate (25 countries)

The single 'US 10Y' hides a wide spread: Japan near 0% under yield-curve-control, Italy/periphery far above the Bund, the US/UK at 4%+. The GDP-weighted global long rate (black, now spanning 25 countries) rose from ~1% (2020) toward ~3% (2026) as Japan normalized — synchronized duration repricing + carry-unwind pressure (macro-carry-trades).

Sovereign 10Y by SUB-REGION (GDP-weighted) — 7 blocs

Aggregating country -> sub-region -> global reveals the real fault lines the 'developed-market 10Y' blends away: Central/Eastern Europe and Latin America run structurally highest (EM risk premium), Periphery Europe above Core (the redenomination/fiscal premium), Asia-Pacific dragged down for years by Japan's near-0% YCC, then all converging UP into 2024-26 — the synchronized repricing of duration across blocs.

European FRAGMENTATION — periphery 10Y spread over the German Bund

The euro's hidden stress gauge: peripheral spreads over the Bund widen in every risk-off (2018 Italy, 2020 COVID, 2022-23 hiking) and compress when the ECB backstops. France's spread creeping toward the periphery (2024-26 fiscal/political risk) is the notable new divergence — the single 'euro 10Y' cannot show this.

Municipals — proxy snapshot via the muni/Treasury ratio (state & local)

State/city muni yields aren't on FRED's free endpoint (the Bond Buyer 20-GO series ended 2016), but accessible PROXIES substitute: the published muni/Treasury RATIO (here x current Treasuries), the MUB ETF yield (~3.2%), and — for trade-level data — MSRB EMMA and FINRA TRACE, which is where the licensed indices source from too. Munis trade rich (below Treasury yields) on their tax exemption.

Municipals by STATE — trailing distribution yield (California vs New York vs national vs HY-muni)

The per-STATE muni cut, as a real time series rather than a snapshot: California (high-tax-state demand) trades richest (lowest yield), New York near national, high-yield muni well above — and all stepped UP with rates from the 2020-21 lows. City-level granularity still needs MSRB EMMA per-CUSIP; this gets the state layer from free ETF data.

Corporate by MATURITY — short / intermediate / long IG (distribution yield)

The maturity dimension of corporate credit: long IG (VCLT) yields most and is the most rate-sensitive; the short/long gap widened as the curve moved — the corporate-credit analogue of the Treasury term structure.

By INDUSTRY — US equity sector performance, rebased to 100 (12 GICS sectors)

The industry cut the single 'S&P' hides: Tech and Semiconductors (SMH) ran away from the pack — the AI-capex bid concentrated in a handful of sectors — while utilities/staples/real-estate lagged. The dispersion between the top sector and the bottom IS the concentration the bubble thesis tracks; when one or two sectors carry the index, breadth is illusory (the equity analogue of the 91% credit common factor).

Industry DISPERSION over time — cross-sectional stdev of sector 12-mo returns

When sector dispersion is HIGH the index is being carried by a narrow set (the 2023-26 AI/tech-and-semis surge); when LOW, sectors move as one (risk-on/off regimes). Dispersion is the breadth gauge behind the headline index level.

Borrowing cost by TYPE — sovereign (10Y) vs corporate (Baa) vs household (30Y mortgage)

Stacking borrower types shows the spread STACK: households pay the mortgage rate (Treasury + ~spread), corporates the Baa rate; all three roughly doubled off the 2020-21 floor — the repricing hit sovereign, corporate, and household credit together.

The yen carry trade, quantified — US−Japan 10Y differential vs the JGB

The carry's fuel is draining: the US−Japan 10Y differential peaked near 3.85pp (Oct-2023) and has compressed toward ~1.8-2.0pp as the JGB climbed from ~0% (yield-curve-control) to ~2.5%. A shrinking differential + a rising yen is the classic carry-unwind setup (the 'BEAR' trigger) — it removes the cheap funding the crowded global longs depend on.

The global bond squeeze: JGB 10Y escaped 0% (carry-unwind fuel) while Baa credit repriced

The 10Y JGB went from ~0% (yield-curve-control) to ~2% (Dec 2025) to ~2.66% (2026) — the rising cost of the yen carry trade that funds crowded global longs (the 'BEAR' trigger).

Corporates by QUALITY from the real FINRA TRACE tape — advance/decline breadth (2023-06–2026-05)

The actual TRACE trade tape, not a proxy: each month's advancing-vs-declining bond count by quality tier. High yield swings hardest (it falls below 1 first when risk-off hits and rebounds highest) while investment grade is steadier — the real corporate-credit breadth signal the rating-ladder OAS chart only approximated.

Corporate trading VOLUME mix — high-yield share of (IG+HY) dollar volume (FINRA TRACE)

Where the dollars actually traded: the high-yield share of corporate volume. A real, tape-sourced risk-appetite gauge to read alongside the spread charts above.

Trigger panel — the unwind watch-list, live from the data

Operationalizing spec-unwind-timing: the date of a violent unwind is not forecastable, but the triggers are observable. Current readings (latest monthly FRED) — watch the indicators, not the calendar:

Read: the carry's fuel is draining (differential down from ~3.85 to ~2pp) and bank HTM is pressured by elevated long rates, while credit and the Fed-path gap look calm/neutral — i.e., the system is fragile and arming, not yet firing. None of this dates the break; it sizes the kindling.

Signal quality: the sharp-vs-smooth tradeoff, measured

For each rate differential: noise = stdev of its month-over-month change (higher = jumpier); horizon = the lead h (months) at which the spread's level best predicts the subsequent change in the fed funds rate; corr = that predictive correlation. The horizon lengthens from the short end (fast) to the curve (delayed); the 2Y−3M policy-path spread sits in the middle.

The tradeoff, quantified from the data: the predictive horizon lengthens from the short end to the curve — 3M−funds predicts the next move at ~1 month, 2Y−funds at ~8 months (and is the strongest predictor, corr ~0.77), while the 2s10s/3m10s curve leads longest (~18 months+) but weaker — the fast-vs-delayed axis. Noise is not monotonic: the 3M−funds gap is actually the cleanest, the 2Y−funds and 10Y−3M the jumpiest. So pick the differential to match the question: 3M−funds for the fastest clean read on an imminent move, 2Y−funds for the strongest read on where the Fed is headed, the 2Y−3M policy-path (noise 0.17, ~9-mo horizon, corr +0.66) for the medium-term expected path, and the 2s10s curve for the smoothest (most delayed) cycle/recession read.

Breakdown framework & data provenance

The bond universe can be sliced along two axes — geography (region → sub-region → country → state → city → institution) and type/quality (sovereign, corporate-by-rating, household/mortgage, municipal, agency). What is charted here vs what requires other sources, stated plainly:

So the institution-level cut already exists (the bank model); the geography (now 25 countries / 7 blocs), credit-quality, the real FINRA TRACE quality-tier, and the equity-industry (GICS sector) cuts are charted above. The remaining gaps are GICS-industry corporate OAS (credit-by-industry — equity sectors stand in; full needs per-CUSIP TRACE mapped to SIC) and city/per-issuer muni granularity (state ETFs stand in; full needs MSRB EMMA), plus single-state munis beyond CA/NY where no liquid ETF exists — flagged rather than fabricated, per the project's zero-trust rule.

Cross-sectional analysis — dispersion, relative value, and the common factor

The charts above are mostly time-series (one rate through time). This section is cross-sectional: at each moment it compares the whole cross-section of segments — every credit-rating bucket, every sovereign, every muni state — and asks how dispersed they are, which are rich/cheap vs their own history, and how much of their co-movement is one shared factor. Method follows the credit literature: cross-sectional spread dispersion as a stress gauge; relative-value z-scores (a segment vs its own trailing history); and a PCA first-principal-component share on monthly spread changes — Collin-Dufresne, Goldstein & Martin (2001) found one common factor dominates credit-spread changes. Engine: models/graph/cross_section.py → data/cross_section.json.

The common factor (PC1 share of cross-sectional change)

Reading: a high PC1 share means the segments move as one. US credit’s ~91% PC1 share confirms the Collin-Dufresne–Goldstein–Martin common factor — and means cross-credit diversification is largely illusory at the system level (the data point that agrees with the project’s self-marked-value claim: the gaps correlate under a common factor, so there is no netting). Standard portfolio theory assumes the opposite.

US corporate credit — cross-sectional dispersion over time

Sovereign 10Y yield (global cross-section) — cross-sectional dispersion over time

Credit-spread change correlation (heatmap)

Pearson correlation of monthly OAS changes. Deep red = near-perfectly co-moving — the visual of the common factor.

Unified relative-value snapshot (every segment, z-scored vs its own history)

Positive z = wide/cheap vs its own history (more stress priced in); negative = rich/tight. Most stretched right now: Japan; tightest: HY muni (HYD).

Bank vulnerability cross-section (FDIC, peer-relative z-scores)

194 institutions; composite z = mean of peer z-scores on HTM-hole, uninsured %, and CRE/Tier-1 (higher = more vulnerable). Cross-sectional HTM-hole std ≈ 7.48pp of equity — the holes are highly unequal.

Funding-graph cross-layer connectors (bridging score)

bridge_score = z(degree) + z(distinct neighbor-sectors) + 0.5 if the node spans both the financial and structural layers. The highest scores are the structural keystones tying the core to the surrounding webs.

The bottom line

Across the decade the funds rate maps onto the 2-year Treasury yield, not onto 2% inflation or full employment. Inflation was almost never at target; "true" labor slack (U-6) ran well above the headline; and one aggregate jobs/inflation print masks sectoral and regional divergence. The mandate is the framing; the bond market is the master.