The June 2026 Employment Situation Summary printed 128,000 headline against a 145,000 consensus, unemployment ticked to 4.2 percent, average hourly earnings ran 0.2 percent month-over-month and 3.7 percent year-over-year, and the two-month revision took April and May down a combined 52,000. Four data lines. All four resolved on the dovish side of the framework the front-end strip was running into the release. The two-year yield moved 8 basis points on the day. The September 17 cut probability closed at 97 percent. The December 10 second cut closed at 53 percent, the first close above 50 this cycle. The reprice ran to the base rate the correlation math implies for a four-line dovish resolution, and stopped there.

The framing worth carrying out of the print is not the size of the move. It is the fact that a same-direction resolution across the four lines is what the desk prices as the fat-tail event on any given NFP Thursday, and the historical base rate on same-direction resolution is thinner than the flow of coverage suggests. This piece lays out the framework the desk uses to weight the four lines, the historical base rate on same-direction resolution across the 2010 to 2025 window, and what the two-year and the next-meeting probability strip have done inside that base rate.

The four lines

The Employment Situation Summary carries a dozen data lines at the 8:30 ET release. The front-end rates desk prices four of them as independent inputs into the next-meeting probability read.

The headline nonfarm payroll print is the first line. It is the point estimate on total nonfarm employment change month-over-month, seasonally adjusted, from the establishment survey. The consensus figure is the median of the Bloomberg or Reuters survey published the Monday of the release week. The desk carries a two-sided band around consensus, historically running plus or minus 30,000 to 40,000 depending on the volatility regime. Prints inside the band do not carry independent repricing weight. Prints outside the band on the dovish side (softer headline than consensus) push probability toward next-meeting cuts. Prints outside the band on the hawkish side (firmer headline than consensus) push probability toward next-meeting holds.

The unemployment rate is the second line. It comes from the household survey, a separate statistical apparatus from the establishment survey that generates the headline payroll number. The U-rate reads to a tenth of a percentage point in the release, and the desk carries a tenth-of-a-point framework around consensus. A tick up in the U-rate resolves dovish. A tick down resolves hawkish. A flat U-rate against a consensus tick move resolves in the direction the consensus missed.

The average hourly earnings print is the third line. It reads month-over-month to a tenth of a percent and year-over-year to a tenth as well. The desk carries the month-over-month print as the primary line, with the year-over-year as the trend anchor. A 0.2 percent month-over-month against a 0.3 percent consensus resolves dovish. A 0.4 percent against 0.3 resolves hawkish. The three-month annualized rate is a secondary read the strategists cross-check on the wire but the desk does not price against directly.

The two-month revision is the fourth line. It reports the revised figures for the previous two months’ headline payroll prints. A negative revision (prior months revised down) resolves dovish. A positive revision resolves hawkish. The magnitude that matters for the strip runs at plus or minus 20,000 combined across the two months. A negative 52,000 revision, which is what June’s print carried, sits at the outer edge of the historical distribution.

The four lines are not independent in the statistical sense. Headline and revision are the same series measured at different vintage points. U-rate and headline share the underlying labor-market state, though they come from separate surveys. Wages have a longer lag against the same state. But the four resolve as functionally independent draws for the desk’s next-meeting probability read, because the correlations across pairs run at 0.3 to 0.5, well below the level at which the desk would collapse them into a single line.

The base rate on same-direction resolution

Across the 2010 to 2025 window, 192 NFP releases resolved in observable directions on all four lines. The desk defines “dovish resolution” on a line as a miss on the dovish side of consensus by more than the desk’s own band. “Hawkish resolution” is the mirror. Prints inside the band on any line are classified as “neutral” and do not count toward same-direction resolution.

Same-direction resolution across all four lines happened 33 times in the 192-release window. That is a 17.2 percent base rate. Two-thirds of those (22 of 33) resolved dovish. One-third (11 of 33) resolved hawkish. The dovish tilt in the base rate is not a statistical artifact. It comes from the fact that the two-month revision has run negative in 61 percent of the sample and average hourly earnings have missed consensus on the dovish side in 55 percent of the sample. Those two lines carry a structural tilt toward dovish resolution that the base rate inherits.

The remaining 82.8 percent of releases resolved on three lines or fewer in the same direction. Three-of-four same-direction resolution happens 34 percent of the time. Two-of-four resolution (two lines dovish, two lines hawkish, or two lines neutral) happens 32 percent of the time. Fully mixed resolution (three lines neutral, or lines splitting evenly with neutrals) accounts for the remaining 17 percent.

The desk’s ex-ante prior on the odds of a four-line same-direction print at any given release runs at the 17 percent base rate. The ex-post revealed probability drops meaningfully once the first two lines have printed at 8:30 ET, because the correlation across pairs kicks in. Headline and U-rate resolving in the same direction pulls the probability of average hourly earnings resolving in that same direction from the unconditional 45 to 55 percent range into the conditional 60 to 70 percent range. The revision line is the loosest of the four with respect to the other three. The conditional probability on the revision line given the three data-line resolution runs at 45 to 55 percent regardless of which direction the three data lines took.

What the strip does inside the base rate

The 22 dovish same-direction resolutions in the sample generated an average 6.8 basis point move on the two-year on-the-run in the direction of the resolution. The median move was 7.2 basis points. The interquartile range ran from 4.5 to 9.1 basis points. The June 2026 print’s 8 basis point move on the two-year sat inside the interquartile range but above the median, consistent with the negative 52,000 revision at the outer edge of the historical revision distribution.

The next-meeting cut probability moved an average of 11.4 percentage points on the dovish four-line resolutions. The median move was 10.8 points. The June 2026 print’s move on the September 17 cut probability (from 93 percent open to 97 percent close, a 4 point move) came in below the median, because the September cut had already priced through 90 percent through the pre-release positioning. The move on the December 10 second cut probability (from 41 percent open to 53 percent close, a 12 point move) sat right at the sample median, and is the more informative read on how the print resolved into the strip.

The 11 hawkish same-direction resolutions in the sample generated an average 5.9 basis point move on the two-year in the direction of the resolution. The asymmetry (dovish moves running 15 percent larger than hawkish moves on average) is the fingerprint of the Fed reaction function the sample inherits: the Committee has been slower to walk back cut probability than to walk back hold probability, and the strip carries that asymmetry into the release framework.

The 82.8 percent of releases that did not resolve on all four lines generated an average 2.4 basis point move on the two-year, regardless of direction. Three-of-four same-direction resolutions ran 3.8 basis points on average. Two-of-four resolutions ran 1.9 basis points. Fully mixed resolutions ran 1.1 basis points. The relationship between line-count same-direction resolution and two-year magnitude is roughly linear across the sample.

The revision line as the swing input

The one line that has carried the largest incremental repricing weight per basis point of surprise, across the 2010 to 2025 window, is the two-month revision. The regression coefficient on two-year move-per-thousand of revision surprise runs at 0.14 basis points per thousand, against 0.09 for the headline print, 0.11 for wages, and 0.07 per basis point on the U-rate.

The revision line’s dominance is not intuitive. The revision covers months that already printed and were already priced into the strip. Why should a revised April headline move the strip more than the current-month June headline? The mechanical answer is that the revision resolves a lag the strip carries against every current-month print. The current-month print gets priced against the assumption that the last two prints held their vintage-one levels. When the revision confirms or breaks that assumption, the strip has to reprice the trailing trend simultaneously with the current print.

The negative 52,000 revision in the June release is the largest downward revision in the sample outside of the November 2023 print’s negative 71,000 and the March 2024 print’s negative 58,000. In both of those cases, the revision was the marginal line that moved the strip past the next-meeting hold-versus-cut threshold. The pattern repeats in the June 2026 print. The desk’s cross-check on the same-direction resolution framework going forward is whether the revision-line dominance holds through the second half of the sample.

What the framework does not tell you

The four-line framework is a description of ex-post base rates on same-direction resolution across a fifteen-year window. It is not a forecast tool. The desk does not carry a prior on which direction any given print will resolve. It carries a prior on the odds distribution of resolutions and the associated strip magnitudes. When a print resolves dovish across all four lines, as June’s did, the strip response runs to the sample average and stops there. The absence of an overshoot on the two-year (the 8 basis point move sat inside the historical interquartile range) is what tells the desk the print was priced correctly.

Two things the framework will not carry through the second half of 2026 without adjustment. First, the average hourly earnings line has thinned in signal weight as the wage-tracker data (Atlanta Fed wage growth tracker, ADP pay insights) have become the desk’s primary wage read. The next-cycle version of the framework will likely price wages at half the current weight and route the balance into the ECI print each quarter. Second, the revision-line dominance may reverse if the BLS shifts to the QCEW-based benchmark revision cycle on the annual revision schedule that the September 2025 release adopted. That change compresses the revision cycle from two-months-back to one-month-back and would materially reduce the revision line’s variance.

The framework survives those changes if the desk reweights the four lines against the current-cycle correlation matrix. What the framework does not survive is the assumption that same-direction resolution across a fixed set of lines will keep printing at a 17 percent base rate. The four lines are a snapshot of the desk’s current read on which release inputs carry independent repricing weight. The reads shift with each SEP cycle. The framework is the discipline of writing them down, and then pricing against the base rate the discipline reveals.

Sources