Duration, curve, Factors and crowding — the rates book. Charts live in the chartbook.
G7 Fixed Income · Factors
What is driving the 10-year
Stacked in-sample OLS of the 10-year or 30-year. Layers plus the starting fitted value plus the residual equal the tenor on every date — an accounting identity, not identified shock shares. Fiscal risk is the 10s30s spread, dominance is the Fed’s share of the debt, foreign official is monthly TIC holdings, and debasement is 40% gold + 20% silver + 40% bitcoin. Click a factor off to refit.
10-year now
5.18%
nominal CMT
Fitted
5.05%
185 overlapping sessions
Residual
0.13 pp
unexplained
Δ 10-year
+100 bp
window
Δ 10s30s
-44 bp
fiscal price
Δ debasement
-3.90%
40/20/40 basket
Δ foreign official
-2.70%
TIC monthly
Δ KW term premium
+38.9 bp
staff model
Chart 1
10-year — stacked factor contributions
Each shaded band is that factor’s contribution since 2025-12-31. Bands stack: starting fitted + factors + residual = the 10-year. Hover for the add-up in pp. Identity error < 1e-6.
- Starting fitted
- Crowding out
- Yen carry
- Debasement
- Foreign official
- Fiscal dominance
- Fiscal risk
- Fed credibility
- Monetary policy
- Growth resilience
- Oil
- Residual
- 10-year
Chart 2
Residuals and Fiscal driving the last 2 days
Day-on-day 10-year change, last 30 sessions. A +4 bp print is four cubes. Title updates from the last two sessions. Residual is what the toggled set did not move.
2026-09-24 +7 bp (Fiscal risk +3 · Residual +4)
Last two 10-year sessions: 09-23 +15 bp (Fiscal risk +2 · Residual +13) · 09-24 +7 bp (Fiscal risk +3 · Residual +4). A cube is 1 bp of that day’s net move, coloured by the factor that added it. Residual is unexplained by the toggled set.
| Factor | Instrument | Grade | β (pp per unit) | Note |
|---|---|---|---|---|
| Oil | WTI | Fair instrument | -0.0001 | Oil-price shock. Collinear with breakevens when the tape is an inflation scare. |
| Growth resilience | NY Fed Weekly Economic Index (else industrial production) | Fair instrument | -0.0232 | Weekly activity, forward-filled. IP is the fallback. Not the S&P. |
| Monetary policy | Kim–Wright expected short-rate path (else funds) | Fair instrument | 0.8917 | ACM-style expected-path piece of the 10-year. Do not use the 2-year — it is almost the 10-year. |
| Fed credibility | 5y5y inflation compensation | Proxy — not the factor itself | 0.1639 | Unanchoring proxy, not a survey of credibility. Moves with oil and breakevens. |
| Fiscal risk | 10s30s Treasury spread (DGS30 − DGS10) | Fair instrument | -0.4562 | Daily market price of long-duration fiscal risk. Steepen is the scare; flatten is not. Collinear with the 30-year — click off if the fit splits. |
| Fiscal dominance | Fed Treasury holdings / federal debt | Proxy — not the factor itself | 0.0486 | How much of the Treasury market the Fed still owns. Updates weekly. A falling share is QT, not “fiscal dominance”. |
| Foreign official | TIC foreign official Treasury holdings (monthly) | Proxy — not the factor itself | 0.0000 | Treasuries held by foreign official institutions (central banks), monthly. A fall is official selling. Beta should be negative if selling is lifting yields. |
| Debasement | 40% gold + 20% silver + 40% bitcoin | Proxy — not the factor itself | 0.0004 | Rebased to 100 at the first common print. Missing legs renormalize. Rejected when yields rise with a firm dollar. |
| Yen carry | USD/JPY | Fair instrument | -0.0004 | Carry / global-duration proxy. A yen-unwind driver needs USDJPY down as yields rise. |
| Crowding out | US IG OAS | Overlaps another fiscal factor | 0.3256 | Private credit spread, not coupon supply. Shares a duration factor with fiscal risk. |
Chart 3
10s30s — day-on-day
Daily Δ of DGS30−DGS10 in bp. A + print is a steepen (scare). Window flatten is the sum of these prints — Chart 2 fiscal cubes are today’s Δ × β, not the YTD level.
Chart 4
Fed vs foreign governments — who owns the debt
Neither series prints every trading day. Teal bars: the weekly change in the Fed’s slice of US federal debt, in basis points. Purple dots: the monthly change in Treasuries held by foreign official institutions (TIC — mostly other central banks), in percent. Blank calendar space is a week with no new report, not a missing bar.
- Fed share of the debt (weekly, bp)
- Foreign official holdings (monthly, %)
Chart 5
ICE US Dollar Index (DXY) vs debasement basket
Weekdays only (bitcoin prints weekends; DXY does not). 40% gold + 20% silver + 40% bitcoin. ICE DXY is the live dollar. Yields up with a firm dollar is the rejection.
- ICE US Dollar Index (DXY)
- Debasement
| Series | Source | Code | Start | End | Δ |
|---|---|---|---|---|---|
| 2-year Treasury yield | MakroGnosis | US.GT2.D.NSA | 2025-12-31 | 2026-09-23 | +138 bp |
| 5-year Treasury yield | MakroGnosis | US.GT5.D.NSA | 2025-12-31 | 2026-09-23 | +126 bp |
| 10-year Treasury yield | MakroGnosis | US.GT10.D.NSA | 2025-12-31 | 2026-09-24 | +100 bp |
| 30-year Treasury yield | MakroGnosis | US.GT30.D.NSA | 2025-12-31 | 2026-09-23 | +56 bp |
| 5-year TIPS real yield | MakroGnosis | US.RY5.D.NSA | 2025-12-31 | 2026-09-24 | +123 bp |
| 10-year TIPS real yield | MakroGnosis | US.RY10.D.NSA | 2025-12-31 | 2026-09-24 | +92 bp |
| 5-year breakeven inflation | rmAPI | US_BKEVN_5Y_D | 2025-12-31 | 2026-09-24 | +7 bp |
| 10-year breakeven inflation | rmAPI | US_BKEVN_10Y_D | 2025-12-31 | 2026-09-24 | +8 bp |
| 5y5y forward inflation | rmAPI | US_5Y5Y_INFEXP_D | 2025-12-31 | 2026-09-24 | +9 bp |
| WTI crude | MakroGnosis | US.OILWTI.D.NSA | 2025-12-31 | 2026-09-22 | +68.37% |
| Broad dollar index | MakroGnosis | US.DXYBROAD.D.NSA | 2025-12-31 | 2026-09-18 | -0.19% |
| USD/JPY (yen per dollar) | MakroGnosis | US.USDJPY.D.NSA | 2025-12-31 | 2026-09-18 | +0.04% |
| Effective federal funds rate | MakroGnosis | US.FFR.D.NSA | 2025-12-31 | 2026-09-23 | +24 bp |
| Industrial production | MakroGnosis | US.IP.M.SA | 2025-12-01 | 2026-08-01 | +1.55% |
| NY Fed Weekly Economic Index | FRED | WEI | 2025-12-27 | 2026-09-19 | 0.52 |
| Gold | LBMA | gold_pm | 2025-12-30 | 2026-09-24 | -2.32% |
| S&P 500 · not official | rmAPI | US_SP500_D | 2025-12-31 | 2026-09-24 | +12.54% |
| Kim–Wright 10-year term premium · not official · staff model | FRED | THREEFYTP10 | 2025-12-31 | 2026-09-18 | +38.9 bp |
| Kim–Wright 10-year expected short rate · not official · staff model | FRED | THREEFY10 | 2025-12-31 | 2026-09-18 | +76.8 bp |
| Federal debt outstanding | rmAPI | US_FED_DEBT_TOTAL_Q | 2025-10-01 | 2026-01-01 | +1.43% |
| Fed Treasury holdings | rmAPI | US_FED_TREAS_W | 2025-12-31 | 2026-09-23 | +7.82% |
| US IG credit OAS · not official | rmAPI | US_IG_OAS_D | 2025-12-31 | 2026-09-23 | -2 bp |
| Silver · not official | rmAPI | US_SILVER_D | 2025-12-31 | 2026-09-23 | -7.48% |
| Bitcoin · not official | FRED | CBBTCUSD | 2025-12-31 | 2026-09-24 | -3.68% |
| Foreign official Treasury holdings | FRED | FORTREASPOS99990 | 2025-12-01 | 2026-07-01 | -2.70% |
| ICE US Dollar Index (DXY) · not official | Yahoo | DX-Y.NYB | 2025-12-31 | 2026-09-24 | +3.06% |
| 10s30s Treasury spread | Derived | DGS30−DGS10 | 2025-12-31 | 2026-09-24 | -44 bp |
| Fed share of federal debt | Derived | 100×TREAST/GFDEBTN | 2025-12-31 | 2026-09-23 | +69.1 bp |
| Debasement basket · not official | Derived | 0.4·gold+0.2·silver+0.4·BTC | 2025-12-31 | 2026-09-24 | -3.90% |
The stacked chart is an in-sample OLS of the selected tenor on the toggled instruments: intercept + β·x, plus residual. Layers are β_i·(x_i,t − x_i,start) so they plus the starting fitted value plus the residual equal the tenor on every date. That is an accounting identity of the fit, not a structural SVAR. Fiscal risk is the 10s30s spread (daily), not the quarterly debt stock. Fiscal dominance is the Fed’s Treasury holdings as a share of federal debt — QT is a falling share. Foreign official demand is monthly TIC holdings (FORTREASPOS99990); weekly H.4.1 custody is dead on FRED and WSHOTSL now aliases SOMA Treasuries. Debasement is 40% gold + 20% silver + 40% bitcoin, rebased to 100 at the first common print; missing legs renormalize. Kim–Wright is a Fed Board staff model, not an FOMC estimate. Toggle a factor to refit, as on /fx/flows. This board does not replace /stealth/yield-factors.

