What tax filings reveal about platform earnings that surveys miss

Household surveys ask people to recall work and earnings. Tax information returns ask platforms and other payers to report amounts paid. When Brett Collins, Andrew Garin, Emilie Jackson, Dmitri Koustas, and Mark Payne examined Internal Revenue Service (IRS) microdata on Forms 1099-K, they documented substantial online platform activity that standard labor-force questions did not fully mirror. The finding is not that surveys are useless; it is that administrative records observe a different slice of the earnings map.

Form 1099-K reporting thresholds and platform coverage have changed over time, so year-to-year counts require careful reading of filing rules. Even with those caveats, tax-based series tend to show more people receiving platform-mediated payments than CPS-style measures of electronically mediated employment in overlapping periods. Researchers attribute part of the gap to underreporting of small or intermittent platform work in surveys, and part to definitional mismatch: a payment reported on an information return is not identical to a self-identified “job.”

Katz and Krueger’s work on alternative work arrangements, and subsequent debates about Contingent Worker Supplement design, sit in the same terrain. Survey modules that ask about electronic intermediaries can miss respondents who think of platform tasks as casual selling rather than employment. Tax data, conversely, can include payments that are not labor income in an economic sense (reimbursements, mixed business receipts) unless researchers apply filters.

The literature also stresses selection. People who file returns with platform income are not a random draw of all adults. Intensity varies widely: many accounts show small annual totals, while a thinner tail accounts for a large share of dollars. Summaries that quote only the headcount of 1099-K recipients without the earnings distribution can overstate how typical high platform earnings are.

For editors covering “side income,” the practical implication is triangulation. A responsible briefing pairs (1) survey definitions and their known undercount risks, (2) administrative payment counts with threshold notes, and (3) qualitative research on how workers label platform activity. Cyan Atlas presents those layers as contours on one cyanotype—not as a single elevation that claims to be the true surface.

This article does not interpret any reader’s tax position and does not recommend filing strategies. It summarizes how published research uses IRS information returns to illuminate platform earnings that surveys may under-detect, and why methodological footnotes belong next to the headline numbers.