Working paper

Calculating and projecting Supplemental Poverty Measure thresholds

The paper reconstructs national Supplemental Poverty Measure thresholds from public Consumer Expenditure microdata. Every tenure and year from 2019 through 2025 lands within 2.8 percent of the published value, and most within 2 percent. It then compares four rules for projecting thresholds: over 2020 to 2025, adjusting the last published threshold by CPI-U gives a mean absolute error of 2.29 percent, against 0.51 percent for replicated consumption growth. A 2025 nowcast committed to before BLS published, an equal blend of replicated growth and a composite price index, landed within 1.17 percent of the published thresholds, where CPI-U adjustment was off by 2.58 percent. Embedded below is the manuscript rendered from commit 6937830.

Revision 6937830 · 2026-09-14 · Max Ghenis, PolicyEngine