XOM monthly excess return during CPI-surprise months (CPI MoM > trailing 12-mo avg)
An old inflation thesis says energy stocks should catch a bid when consumer prices move hotter than trend. Testing it on XOM versus SPY over the past three years produces a clear answer: not on this evidence. Across 21 overlapping months, 13 qualified as CPI-surprise months, and XOM outperformed SPY only about half the time — 54%, a coin flip.
The mean excess return in surprise months was +0.38%, versus -1.69% in non-surprise months. That gap sounds supportive, but it collapses under a Welch t-test: p = 0.54. With just 13 surprise months, the spread is statistically indistinguishable from luck. The full analysis below shows the monthly distribution, the test setup, and why the coincident CPI lag matters for interpreting any edge. The takeaway: this three-year window doesn't give the inflationary-surprise thesis much weight.
For XOM over the past ~3 years, does its monthly return consistently beat SPY in months when the month-over-month CPI change is above its trailing twelve-month average? I expect XOM to outperform during inflationary surprises because commodity-linked revenues benefit from rising prices.
How this was measured
Calendar-month close-to-close returns for XOM and SPY. CPI month-over-month percent change computed from the Consumer Price Index (monthly, 1982-84=100). For each month, the trailing 12-month average of CPI MoM changes is calculated. A month is flagged as a 'CPI-surprise' month when the current MoM CPI change exceeds that trailing average. Excess return is defined as XOM return minus SPY return in the same month. The distribution of excess returns in CPI-surprise months is compared to that in non-surprise months using a Welch t-test. This is a coincident analysis, not point-in-time: CPI data for a given month is reported with a 2-week lag, so the market may have already priced in the inflation surprise by the time it is officially published.
The key numbers
Reading the numbers
Across the 21 overlapping months, XOM beat SPY by an average 0.38% in the 13 CPI-surprise months but lagged by 1.69% in the other 8. That gap looks supportive, but the p-value of 0.54 means the difference could easily be chance.
The charts
This box plot compares XOM's monthly return minus SPY's in CPI-surprise months versus other months. In surprise months the average excess return is slightly positive (0.38%), while in other months it is clearly negative (-1.69%). The detail worth your eye is the huge overlap: surprise months range all the way from -17.9% to +16.3%, so the positive average comes from a few strong months, not a steady beat. For your question, the chart is suggestive but far from proof — XOM does better on average in inflationary surprise months, but the pattern is too scattered to call consistent.
Recent CPI-surprise months (most recent first, max 20)
| Month | XOM return | SPY return | Excess (XOM-SPY) | CPI MoM | CPI trailing 12m avg |
|---|---|---|---|---|---|
| 2026-05 | -0.0582 | 0.0499 | -0.1081 | 0.0063 | 0.0041 |
| 2026-04 | -0.0748 | 0.1037 | -0.1785 | 0.0085 | 0.0037 |
| 2026-03 | 0.1001 | -0.0434 | 0.1435 | 0.0105 | 0.0034 |
| 2026-02 | 0.0851 | -0.0099 | 0.095 | 0.0047 | 0.0031 |
| 2026-01 | 0.1751 | 0.0117 | 0.1634 | 0.0037 | 0.0027 |
| 2025-09 | -0.0111 | 0.0333 | -0.0445 | 0.0025 | 0.0025 |
| 2025-08 | 0.0296 | 0.0221 | 0.0075 | 0.0029 | 0.0024 |
| 2025-06 | 0.0568 | 0.0529 | 0.0039 | 0.0034 | 0.0022 |
| 2025-05 | -0.0188 | 0.0521 | -0.0708 | 0.0021 | 0.0019 |
| 2025-04 | -0.1187 | 0.004 | -0.1227 | 0.0031 | 0.0019 |
| 2025-03 | 0.0709 | -0.0599 | 0.1308 | 0.0022 | 0.002 |
| 2025-02 | 0.0494 | -0.0077 | 0.0571 | 0.0044 | 0.0023 |
| 2025-01 | -0.006 | 0.0213 | -0.0273 | 0.0065 | 0.0025 |
The takeaway
Short answer: no — the data don't show XOM consistently beating SPY in CPI-surprise months, and any edge is basically a coin flip. Across the 21 overlapping months, 13 were CPI-surprise months, and XOM beat SPY in only 7 of those (54%). The mean excess return was +0.38% in surprise months versus -1.69% in non-surprise months, but the t-test says p = 0.54, meaning that gap could easily be luck. With just 13 surprise months, the sample is too thin to call this a real signal. The recent months show why: January and March 2026 had large positive excess returns (+16.3 and +14.4 points), but April and May were deeply negative (-17.9 and -10.8 points). So the average is carried by a few big swings, not by consistent monthly outperformance. The practical takeaway: this three-year window does not support the inflationary-surprise thesis, and you'd need more data and a more forward-looking inflation measure before putting weight on it.
The fine print
- Only 21 months of overlap, with just 13 surprise months — the test is underpowered and can't separate signal from noise.
- CPI for month M is released in mid-M+1, so this uses the coincident print, not a true point-in-time surprise; markets may have priced it in.
- The trailing 12-month average is a simple benchmark; survey-based inflation expectations could classify surprise months differently.
- XOM's moves may reflect the whole energy sector rather than XOM-specific commodity exposure.