On by default: it starts after your first click on the page, and you can turn it off at any time.Playing now. You can turn it off at any time; this browser will remember your choice.Music is off, and this browser will remember. Press Play to hear it again.
Background music produced and licensed by DingSong.
Issue 1 · Late October 2026
US inflation → US 10-yr yield
Common explanation (Direct mechanism · same direction): When inflation rises, bond investors demand more inflation compensation and the 10-year yield goes up.
This issue: US inflation at 64 (warm), US 10-yr yield at 91 (hot) — both above normal, in line with the common explanation.
Hot and cold describe position, not good or bad: warm does not mean a fall is coming, and an extreme reading does not mean a quick reversal. We describe history only, with no buy or sell advice.
Hover over a pendulum or tick to see its name; click to open that indicator.Tap a pendulum or tick to see its name; tap again to open that indicator.
Where it stands now
Where the two indicators stand this issue
US inflation: 64 (0.41 percentage points above its 10-year average). US 10-yr yield: 91 (2.09 percentage points above its 10-year average).
Tests
Does this link hold up?
Common explanationWhen inflation rises, bond investors demand more inflation compensation and the 10-year yield goes upSource: Federal Reserve Bank of San Francisco, “A New Conundrum in the Bond Market?” (2017). The article looks at how yields react on the day inflation data are released (the surprise part only); the 10-year yield mainly reflects expected inflation over the next ten years, and near-term inflation expectations matter very little.
Tested against our dataBy position (each compared with its own normal — do their highs and lows move together?): agreement between US inflation and US 10-yr yield (same month) is +0.50; across every 5-year period (shifted one year at a time), the two moved in that direction in 40 of 49.By change (how far each position moved over the past 12 months — do their swings move together?): agreement +0.32; the two moved in that direction in 44 of 48. The two measures disagree: their long-run highs and lows move together, but their swings within a year often do not (by change, only “weak”).
By position40/49
By change44/48
1973–19782021–2026
Each cell is a five-year period (shifted one year at a time): filled = in step with the common explanation, outlined = out of step.●●○Supported by one measure
Periods when it did not holdAcross the 5-year periods (shifted one year at a time, so they overlap), the periods out of step cover — by position (49 periods): 1982–1989, 1990–1996, 2009–2016 and 2021–2026; by change (48 periods): 1981–1987 and 2002–2008.
Does it predict?Prediction test (a historical test, unrelated to “Where it stands now” above and not this site’s forecast): does adding the past 12 months of US inflation make US 10-yr yield easier to predict? Result: No predictive link found (p = 0.619; the smaller, the less likely by chance). Forecasting a recent stretch of data not used for estimation, the error is 2.4% higher.○○○No link found
History
The two lines, overlaid
Data: US: Bureau of Labor Statistics, via FRED (Federal Reserve Bank of St. Louis); Japan: World Bank Global Economic Monitor (CC BY 4.0); others: OECD; Federal Reserve H.15, via FRED (Federal Reserve Bank of St. Louis). Positions: our calculation.
US inflationUS 10-yr yield
Time lag
At what lag do they match best?
Longer bar = moves more in the direction of the common explanation (the number is the agreement).
same month+0.50
3-month lag+0.51
6-month lag+0.52
9-month lag+0.54
12-month lag+0.55
18-month lag+0.51
Limits
What this card cannot tell you
Moving together does not prove that one causes the other.
Results for every period are listed in full, including periods that did not agree.