The Local Labor Market

The national unemployment rate is an average of places that are not average. This takes it apart: 6 labor-market measures across 79 geographies — census regions and divisions, every state, and a set of large metropolitan areas — monthly from 1976-01 to 2026-05.

247,438 observations across 422 series. Because the panel carries the participation rate and the employment-to-population ratio alongside the unemployment rate, you can ask the question that matters: when a state’s unemployment rate falls, is that people finding work, or people leaving the labor force?

Freshness, stated plainly

The panel above is the fresh tier: its last real observation is 2026-05. Two further tiers ship with this product, and both are older. They are separate files with their own labels, and nothing on this page silently blends them with the fresh panel.

• Counties — 534 counties in all, 191,553 observations, last real observation 2025-06. Those are three different numbers and they stay separate: only 277 of those counties are published monthly, while 532 are published annually — so a monthly county chart can cover far less ground than the headline county count suggests. Either way this is a minority of the roughly 3,143 U.S. counties, and the fresh panel above has no county tier at all.
• Payrolls by area and industry — 1,400,362 observations covering 323 industries across 259 areas in 48 states, last real observation 2025-06. Areas and states are counted separately and deliberately: this tier's geography is mostly metropolitan areas and New England city and town areas rather than whole states, and reporting the area count as a state count would overstate its geographic reach roughly fivefold.

• The county figures cover a minority of the roughly 3,143 U.S. counties, and they reach us second-hand rather than as a direct pull from the Bureau, so they are materially older than the main panel. The fresh panel above carries no county figures at all.
• County coverage is far thinner month by month than year by year: most counties here are published only once a year. The two reporting frequencies are kept apart and must never be drawn as a single line.
• The payrolls-by-industry figures come from the same second-hand source and carry the same age. Their geography is mostly metropolitan areas rather than whole states, which is why areas and states are counted separately.
• The main panel's metropolitan coverage is a handful of large metropolitan areas, not every metropolitan area the Bureau publishes.
• There is no household-survey microdata here, so breakdowns by demography crossed with geography are limited to what the Bureau already publishes.