These are 12-month changes in oil and the 10Y yield rescaled onto the same axis. Below is the 5-year rolling correlation. Before the mid-1980s, oil was fixed at a posted price (so there was nothing to correlate). Once oil started trading freely, the correlation turned steadily positive and climbed to about 0.4-0.6 after 2008.

Each bar is the correlation between monthly oil returns and yield changes within one decade. The nominal 10Y yield comes from 0.00 in the 60s to about 0.4 in the 2010s and 2020s. Once you can split the yield from 2003, breakevens run 0.49-0.57 while TIPS stays between -0.14 and 0.12. The nominal link looks like its coming from inflation expectations.

These are rolling 1Y and 3Y correlations of weekly oil returns with each yield. The breakeven link has held near 0.5 since 2012, and the nominal yield sits between 0.2 and 0.3 while TIPS sits around 0. At a weekly frequency, how much of the yield is inflation compensation determines how much oil is involved in price action.

This shows how many bps each yield moves for a 10% oil move over 3Y rolling windows. For breakevens it got as high as 15 bps in 2012-14 and sits at about half of that today. The real yield is recently positive. This means oil moves are starting to show up in real yields as well, not just inflation expectations.

Correlating price levels gives numbers between -0.4 and 0.9 with little economic rationale to support. Both series drift, so correlating weekly changes gives a stable relationship that appears to have strengthened by era.

Apart from being the coolest graphs ever, these heatmaps show where in time and over what length of the cycle oil and each yield move together. Brighter outlined areas are statistically significant and it is clear here that oil and breakevens move together over cycles of roughly 6M to 2Y for the sample of data I collected. Most of this is in sync too, with arrows pointing rightward. Oil and real yields seem to only align in random patches, showing that breakeven links hold over medium-term cycles rather than week over week.

These bars test if oil moves predict yield moves in later weeks, or the inverse. Granger causality is separated for each period, and it seems like all of the relationship happens within the same week. Looks like neither market leads oil, and inflation markets reprice oil news quickly. This clusters in crises when markets fall out of sync.

Oil's moves are split into parts that track real yields, which is a widely-used proxy for growth and demand, and supply representing the rest. Real yields don't help explain oil's movement, but the supply part correlates about 0.5 with breakevens while the demand part sits near zero. The inflation market reprices oil no matter why it moves, so there is a direct pass-through from oil prices to expected inflation in this data alongside the economic logic.