Abstract
National labor-market statistics can be correctly centered while concealing substantial geographic variation. I ask how much observed variation is lost when state unemployment and vacancy conditions are summarized by twelve Federal Reserve district means and evaluated against an independently constructed national input. Using monthly data from 2001 through 2025, I separate variation averaged away within districts, variation preserved between them, and the national input’s centering relative to the state-supported mean. Centering accounts for only 0.69 percent of state-to-national squared distance, while 53.99 percent is lost within districts and 45.32 percent remains between them. At the institutionally relevant scale of twelve regions, three illustrative procedures retain median labor-force-weighted shares of only 44–51 percent. Under an illustrative monetary-rule calibration and on the common 237-month support, district distances of at least 25 basis points expose 33.2 percent of the labor force on average; subtracting each district’s strictly prior 60-month mean lowers exposure to 20.6 percent. The results do not prescribe regional interest rates or measure welfare losses; they show that evaluating an aggregate requires examining both its center and the regional distribution suppressed beneath it.
The aggregation puzzle
The Federal Reserve pursues a national employment objective with one policy rate, but the labor-market conditions summarized by national statistics are geographically distributed. Closeness to the national number therefore does not reveal how much local variation was pooled before that comparison was made. The paper makes that hidden compression visible by separating three objects: variation averaged away within districts, variation preserved between district means, and the distance between the state-supported mean and the independently constructed national input.
Measurement approach
The analysis constructs monthly state labor-market gaps from BLS unemployment and vacancy data for January 2001 through December 2025. County assignments allocate states that cross Federal Reserve district boundaries, and labor-force weights aggregate the resulting cells to twelve district means. An exact squared-distance identity then tracks what the district map suppresses, what remains visible between districts, and where the national input sits relative to that distribution. Scale and zoning exercises compare the institutional map with other contiguous regionalizations without treating any alternative as an optimal redesign.
Interpretation
The national input is well centered, but that fact answers only one part of the measurement question. Across three illustrative procedures, twelve-region summaries retain median labor-force-weighted shares of 44.1 to 51.2 percent of state dispersion. The estimates do not prescribe regional interest rates, redraw Federal Reserve districts, or identify welfare losses. They establish a narrower point: a national statistic can represent the center of a distribution accurately while concealing substantial regional variation. When the cross-state distribution matters, state-level or comparably disaggregated information is needed alongside the national aggregate.
Citation
Pusateri, Nicholas R. 2026. “One Nation, Under Aggregation: Twelve Federal Reserve Districts and the National Labor-Market Summary.” Working Paper. URL: https://nicpusateri.com/one-rule.
@article{pusateri2026one,
title={One Nation, Under Aggregation: Twelve Federal Reserve Districts and the National Labor-Market Summary},
author={Pusateri, Nicholas R.},
journal={Working Paper},
year={2026},
url={https://nicpusateri.com/one-rule},
}