Employment, US and the world

Shaded vertical bands mark US recessions as dated by the National Bureau of Economic Research (1981-82, 1990-91, 2001, 2007-09, 2020).

Definitional gaps across agencies

The four agency sources do not share unemployment definitions. BLS CPS U-3 covers persons age 16+ who are either actively job-searching in the prior 4 weeks or on temporary layoff. OECD's harmonised unemployment rate (HUR) follows the ILO ICLS standard at age 15+ (16+ in some countries) and pools country-specific labour-force surveys on a calendar-month basis. Eurostat's une_rt_m applies the same ILO ICLS standard to age 15-74 across EU + EFTA. The World Bank WDI series uses the ILO modelled estimate, which for countries with weak labour-force-survey infrastructure is a model output rather than direct measurement. The result: a 5.0% US unemployment rate and a 5.0% Spanish rate read the same on the choropleth but reflect different underlying populations. Differences of 0.3-1.0 pp between sources are common.

JOLTS: declining response rates

BLS's Job Openings and Labor Turnover Survey (~21,000 establishments monthly) has seen its response rate fall from ~70% in 2014 to ~33% in 2024 (BLS, Issues in JOLTS Methodology, August 2024). Standard errors on monthly JOLTS estimates have widened ~30% over the same window. The headline "job openings" series also counts unfilled vacancies on the last business day of the month with no age adjustment, so a stale posting counts the same as a new one. First-release vs second-release revisions on national openings average ~50-100k.

LAUS: model-based state estimates

State-level unemployment rates come from BLS's Local Area Unemployment Statistics program. These are model-based estimates, not direct measurement (Handbook of Methods, Chapter 4): a signal-extraction model combining CPS, CES, and UI claims data per state. Monthly standard errors are ~0.2-0.4 pp at 90% confidence per state. The annual benchmark revision (typically February release) can move state-level series by ±0.3 pp.

YoY column units

The per-country snapshot table renders YoY change in percentage points for rate metrics (unemployment, LFPR, employment ratio, quits-rate, layoffs-rate) and in percent change for level metrics (job openings, hires, in thousands). The column header switches accordingly. For rates, pp is derived algebraically from the stored relative-change figure: pp = value × yoy / (100 + yoy).

US national · Bureau of Labor Statistics

• LNS14000000 Civilian unemployment rate (16+, seasonally adjusted)
• LNS11300000 Labor force participation rate
• LNS12300000 Employment population ratio
• JTS000000000000000JOL Job openings, total nonfarm (JOLTS)
• JTS000000000000000HIL Hires, total nonfarm
• JTS000000000000000QUR Quits rate, total nonfarm
• JTS000000000000000LDR Layoffs and discharges rate, total nonfarm

US states · BLS LAUS

• LAUST<FIPS>0000000000003 State-level unemployment rate (seasonally adjusted)

OECD member countries

• DSD_KEI@DF_KEI Harmonised unemployment rate, monthly, % of labour force (UNEMP / PT_LF)

EU + EFTA · Eurostat

• une_rt_m Unemployment rate, monthly (% of active population)
• lfsi_emp_q · EMP_LFS Employment rate, quarterly (% of working-age population employed)
• lfsi_emp_q · ACT Labour-force participation, quarterly (activity rate)

Rest of world · World Bank WDI

• SL.UEM.TOTL.ZS Unemployment, total (% of total labour force), annual

Daily refresh across every source. BLS releases monthly values on a fixed calendar (first Friday for jobs, JOLTS lags by ~6 weeks); Eurostat and OECD update on a rolling basis; World Bank WDI ships annual values, so its lines look stepwise. Re-pulls during each agency's revision window refresh any updated values via composite-id upsert. The cross-country snapshot prefers a country's most recent monthly source over its annual WDI line whenever both exist.