Some releases share a minute. The market moves once, and which report caused it cannot be separated - which is why the cross-release pages leave those dates out, while the release pages keep them, marked. This page compares the two: what the exclusion is protecting against.
| release | lands with | on its own | sharing a minute | ratio | does sharing change it? |
|---|---|---|---|---|---|
| Retail sales | PPI, Payrolls, CPI | 0.26% (29) | 0.44% (10) | 1.67x | Moves much more |
| GDP | Core PCE | 0.27% (24) | 0.40% (11) | 1.49x | Moves much more |
| Core PCE | GDP | 0.39% (26) | 0.40% (11) | 1.03x | No real difference |
| PPI | Retail sales | 0.44% (31) | 0.40% (7) | 0.91x | No real difference |
Read the quiet releases first. GDP and retail sales travel noticeably further on the dates they share with a bigger report (the ratio column), which is what you would expect if much of the move came from the release beside them, though this cannot say how much. That is the argument for leaving these dates out of the cross-release pages, and it is why those rows are marked rather than deleted on each release page. The shared samples are small, so these are directions, not measurements.
Some releases print at exactly the same minute as another. GDP and Core PCE sometimes share their 08:30 (11 times on this record); retail sales sometimes lands with PPI, and on rare mornings with the jobs report or CPI. The market moves once on those mornings and the move belongs to both reports, with no way to split it, which is why the cross-release pages leave shared minutes out (the release pages keep them, marked). This page compares those shared days, on their own, with the same release's solo days, so the exclusion is a visible decision rather than a hidden one.
GDP on its own moves the Nasdaq 100 a median of 0.27%. On the days it shares a minute with Core PCE the median is 0.40%, 1.49 times as much. Core PCE on the same shared mornings shows no change at all, 1.03 times its own solo median. The likeliest reading is that most of those mornings' move came from Core PCE, though nothing here can split it, and counting it under GDP as well would make GDP look like a release it is not. Retail sales tells the same story at 1.67 times; it mostly shares with PPI, now and then with the jobs report or CPI.
PPI is the one row below one, at 0.91, and its partner is retail sales, the quieter of the two. That is consistent with the same reading from the other side: when the louder release is the one sharing the minute, sharing does not add to it.
On the board, a release that shares its minute is marked rather than hidden, and its reaction on that date is still recorded and counted on its release page, just not in the cross-release medians. When I see a shared minute coming I size for the louder partner and I read the reaction as the louder partner's, whatever the quiet one printed. I treat a GDP morning that is also a Core PCE morning as mainly a Core PCE morning.
The shared samples are small: 11 shared mornings for GDP, 7 for PPI, 10 for retail sales. These are directions, not measurements, and one unusual morning would change any of the ratios. The attribution is also an inference. The data cannot see which report the market was reading; it can only see that the quiet release looks louder in company and the loud one looks the same, and the simplest explanation of that is the one given above.
Releases: Core Personal Consumption Expenditures (Core PCE) Consumer Price Index (CPI) Durable Goods Orders Fed Rate Decision (FOMC) Gross Domestic Product (GDP) Housing Starts and Building Permits Import and Export Prices Job Openings and Labor Turnover (JOLTS) New Home Sales Nonfarm Payrolls (NFP) Producer Price Index (PPI) Productivity and Unit Labor Costs Retail Sales Trade Balance (International Trade)