Do Online Job Postings Capture Job Vacancies? -- by Michael Dalton, Lisa B. Kahn, Andreas I. Mueller

URL SCAN: Do Online Job Postings Capture Job Vacancies? -- by Michael Dalton, Lisa B. Kahn, Andreas I. Mueller
FIRST LINE: We ask whether online job postings capture U.S. job vacancies, matching near-universal Burning Glass (BG) postings to the representative Job Openings and Labor Turnover Survey (JOLTS).

The Dissection

This is a measurement-validation paper. It tests whether Burning Glass postings can function as a proxy for vacancies by comparing them with JOLTS, then reweighting the data by establishment characteristics. Its real contribution is to improve the labor market’s dashboard. It does not establish that the labor market itself remains structurally healthy, or that human labor remains indispensable. The supplied excerpt ends mid-sentence at “the lev…”, so claims beyond that fragment are unavailable.

The Core Fallacy

The central category error under the Discontinuity Thesis is confusing accurate vacancy measurement with durable human economic necessity. If BG converges with JOLTS, that validates the sensor—not the organism. A vacancy is an observed opening, not proof that the role cannot be automated, that it will remain open, or that mass employment can continue to sustain the wage-consumption circuit.

The paper can therefore refine the measurement of the pre-discontinuity labor market while saying almost nothing about P1–P3. Better counting does not defeat cognitive automation, coordination failure, or the collapse of productive participation.

Hidden Assumptions

• JOLTS is treated as the representative benchmark rather than another instrument with its own blind spots.
• Reweighting on observables captures the relevant selection effects; unobserved differences in skill, urgency, duplication, evergreen postings, and firm strategy may remain.
• The 2007–2022 relationship between postings and openings remains useful during an AI-driven structural break.
• “Job vacancy” is implicitly treated as a meaningful unit of future human production, although a vacancy can be temporary, redundant, or technically automatable.
• Historical representativeness is allowed to carry analytical weight in a period when the object being measured may itself be disappearing.

Social Function

Primary classification: partial truth. Secondary classifications: prestige signaling and transition management.

The paper likely establishes a legitimate and useful empirical correction: online postings, especially from large establishments, may capture a substantial share of openings. But institutional validation can also anesthetize the reader into believing that a better labor-market series means a more governable labor-market future. It upgrades the instruments used to count the exposed population; it does not give that population bargaining power, ownership of AI capital, or productive indispensability.

The Verdict

Useful calibration, strategically insufficient. This paper may prove that online postings captured vacancies during the measured period. It does not prove that vacancies survive the automation transition. Under DT logic, it is a sharper dashboard for the corpse—not evidence that the corpse is alive.