Labour-force surveys have long asked whether a person held more than one job during a reference period. The resulting multiple-jobholding rate is a familiar official statistic. Research published over the past decade has asked a narrower question: how much secondary work those questions miss when the second activity is informal, intermittent, online or not recognised by the respondent as a “job” in the questionnaire’s sense.
Alternative work arrangements and the Contingent Worker Supplement
Lawrence Katz and Alan Krueger examined alternative work arrangements using the Contingent Worker Supplement (CWS) to the Current Population Survey and related sources. Their widely cited estimates of the share of workers in alternative arrangements sparked both follow-up research and critique. A later revision by the same authors, responding to new CWS fielding and to commentary on their earlier extrapolations, lowered some of the growth claims associated with the mid-2010s estimates. The episode is relevant here not as a verdict on any single number, but as an illustration that the measured size of nontraditional work is sensitive to survey design and to how researchers fill gaps between survey years.
What “second job” questions capture
Standard multiple-jobholding items typically ask about additional jobs held last week or during a reference week. That framing captures concurrent employment that the respondent reports as a job. It is less suited to work that is occasional, unpaid in a given week, or organised as a series of tasks rather than a standing position. Katharine Abraham and Ashley Amaya, among others, have discussed how questionnaire language and the structure of employment questions affect whether secondary activities are reported at all. When secondary work is online and episodic, the probability of being counted in a weekly multiple-jobholding measure can fall even when annual tax or bank records show activity.
Administrative comparisons
Comparisons between survey reports and administrative earnings records show that surveys and tax files disagree about who has secondary earnings and how large those earnings are. Disagreement runs in both directions: some survey reports of secondary work lack a matching administrative trail at the thresholds researchers examine, and some administrative secondary earnings are not reported as second jobs in the survey week. Researchers treating either source as complete without stating the other source’s coverage will overstate certainty.
Limits that remain
None of the studies summarised here claims that a single corrected multiple-jobholding rate replaces the official series. Katz and Krueger’s work, the CWS documentation from the Bureau of Labor Statistics, and the questionnaire research associated with Abraham and colleagues each describe a different cut of the same underlying phenomenon. Official rates remain useful for what they measure — reported concurrent jobs in a household survey reference week. They are a weaker guide to the full annual incidence of secondary platform or task-based work unless that incidence is measured with instruments designed for it.
This note does not advise anyone about holding more than one job. It summarises how published research describes the measurement of that arrangement in labour statistics.
Sources named in the text · September 2026