WMB daily-return beta to Brent conditional on 10-year yield shock
For a midstream name like WMB, the crude link is supposed to be the story—until bonds start moving. The hypothesis: when the 10-year Treasury jumps 50 basis points in a month, income buyers take over, repricing pipeline equities as rate-sensitive vehicles and knocking the daily-return beta to Brent toward zero. Across 701 trading days, only 13 qualify as yield-shock days. That is a small window, but enough to look.
The evidence goes the other way. On shock days, the estimated Brent beta was 0.109—modestly higher than the 0.062 on normal days, not lower. The interaction p-value of 0.77 says that gap is indistinguishable from noise. The crude linkage gets noisier rather than clearly vanishing, and the point estimate runs against the decoupling thesis. The full model, methodology, and robustness checks are laid out in the analysis below.
Over the past ~3 years, does WMB's daily-return beta to Brent crude fall when the 10-year Treasury yield has risen more than 50 basis points over the prior month? I expect pipeline/midstream equities to trade as rate-sensitive income vehicles during bond-market shock, breaking their crude linkage and dragging the beta toward zero as income buyers demand higher yields.
How this was measured
Resampled WMB minute bars to daily closes and computed close-to-close returns. Brent returns use brent_daily_df reindexed to WMB trading days with forward-fill; the 10-year Treasury yield change is the 21-trading-day difference from treasury_10y_df, thresholded at +50 basis points to flag yield-shock days. We estimated WMB's daily-return beta to Brent via a single OLS model with an interaction term: beta_normal (no yield shock) and beta_shock (yield shock) are recovered from that model. Robust HC3 standard errors are used. Window analyzed: 2023-08-30 to 2026-07-31.
The key numbers
Reading the numbers
Over 701 trading days, only 13 were yield-shock days. Brent beta was 0.062 normally and 0.109 on shock days—slightly higher, not lower—and the p-value of 0.77 means the difference is plausibly random, so there's no evidence of crude decoupling.
The charts
The bar on the right, for yield-shock days, is taller than the normal-regime bar: 0.109 versus 0.062. That is the opposite of the expected drop toward zero. With only 13 shock days, this apparent increase is not statistically meaningful, so the chart undercuts rather than supports the rate-sensitive decoupling story.
The 60-day rolling beta spends its time bouncing around a mean of about 0.11, ranging from roughly -0.23 to 0.55. It starts near 0.10 and ends near 0.08, so it never settles into a sustained period near zero. The wide swings show how noisy short-window beta estimates are, which is exactly why the 13-day shock-regime estimate should be read cautiously.
Each dot is one normal-regime day pairing Brent's return with WMB's return. The cloud tilts only slightly upward, matching the 0.062 beta, meaning WMB moves with Brent on average but the relationship is loose. WMB daily returns cluster tightly around their 0.12% mean, so most of a typical day's move is driven by something other than crude.
This scatter uses only the 13 yield-shock days, a very small cloud: Brent returns span about -5% to +3.7% while WMB returns span -1.7% to +4.1%. The estimated shock-day beta of 0.109 is driven by this handful of points and has a p-value of 0.48, meaning it is statistically indistinguishable from zero. The takeaway is that 13 days simply cannot support the claim that WMB breaks its crude linkage during bond-market shocks.
Regime summary
| Regime | N | Beta | p-value (beta != 0) | Mean WMB return | Mean 10y monthly change |
|---|---|---|---|---|---|
| Normal | 688 | 0.0624 | 0.0151 | 0.0012 | 0 |
| Yield shock | 13 | 0.1087 | 0.484 | 0.0029 | 0.554 |
| Full sample | 701 | 0.0634 | 0.0043 | 0.0013 | 0.01 |
The takeaway
The short answer is no: WMB's beta to Brent did not fall during yield-shock days, so the rate-sensitive decoupling you hypothesized does not show up in this sample. Across 701 trading days, only 13 met the +50bp 21-day Treasury move, and on those days the estimated Brent beta was 0.109 — actually modestly higher than the 0.062 beta on normal days. That gap is far from meaningful: the interaction p-value is 0.77, meaning there is roughly a 77% chance such a difference would appear even if the true betas were identical. On normal days the 0.062 beta is distinguishable from zero (p=0.015), but on the 13 shock days the beta is effectively indistinguishable from zero (p=0.48) — so the oil linkage gets noisier rather than clearly vanishing. With just 13 shock days, this is essentially inconclusive on the decoupling question, and the point estimate actually runs the wrong way. Practical takeaway: there is no evidence yet that a Treasury shock breaks WMB's crude link; if that was the thesis, the daily-return data over the past three years do not support it.
The fine print
- Only 13 yield-shock days in 701; the shock events cluster and overlap, so the comparison is fragile.
- The +50bp threshold is arbitrary; results could shift with a different yield-shock definition.
- Beta is measured on same-day close-to-close returns, so it misses delayed or intraday reactions.
- This is one midstream name; the outcome may not generalize to the broader pipeline/midstream group.