iron26 / uk-ashe-earnings
dataset · UK earnings

UK Earnings by Region & Occupation

UK employee gross earnings and paid hours by region and occupation (2-digit SOC), 2016 to 2023, as one clean .parquet file with every ONS code carrying a readable label. Built from the Annual Survey of Hours and Earnings (ASHE Table 3). Free, under the Open Government Licence.

Download Parquet (26.0 MB) Download CSV (gzip, 42.9 MB)

4,447,872 rows · 13 UK regions/countries · 36 occupation groups · mean, median & 10th–90th percentiles · hourly / weekly / annual pay & hours · by sex and full-/part-time · 2016–2023.

Preview — median gross annual pay by region, 2023

Median gross annual pay, all employees (all occupations, both sexes, full- and part-time), 2023. The full file has every occupation group, every percentile, hourly/weekly/annual pay and hours, split by sex and working pattern, back to 2016.

Region / countryMedian gross annual pay
London£38,943
South East£31,100
Scotland£29,675
United Kingdom£29,669
East£28,883
North West£28,506
West Midlands£28,192
South West£28,059
Wales£27,571
Yorkshire and The Humber£27,528
Northern Ireland£27,102
East Midlands£27,062
North East£26,576

Showing 13 UK regions/countries with a published median for 2023. Source figures are ONS estimates; small cells are suppressed by the ONS and appear as null in the file.

What's in the file

One row per observation. Every dimension ships as BOTH a stable ONS code (filter on it) and a readable label (show it):

  • value — the observation (float; null where the ONS suppresses a small cell)
  • cv — coefficient of variation, the ONS quality measure for the estimate
  • data_marking — ONS suppression / quality flag where present
  • year — survey year, typed int (2016–2023)
  • administrative_geography_code / _label — e.g. E12000007 / London
  • soc_code / soc_label — 2-digit Standard Occupational Classification group
  • averages_and_percentiles_code / _label — mean, median, 10th…90th percentile
  • hours_and_earnings_code / _label — e.g. annual gross pay, hourly pay excl. overtime, paid hours
  • sex_code / _label · working_pattern_code / _label — all / male / female · all / full-time / part-time
Why Parquet. The ONS ships ASHE Table 3 as a 727 MB row-oriented CSV in its internal “v4” shape. This file is the same data, typed and columnar — open it directly in pandas, polars, DuckDB or R in one line, at a 26.0 MB download (96% smaller).

Why we built this

ASHE is the ONS's authoritative survey of UK pay, but the bulk download is a 727 MB CSV in an internal “v4” layout — observation value, quality columns, then a code+label column for every dimension — that you have to understand before you can load it, and that is too large to open casually. We did that parse once, deterministically, typed the values, and wrote a 26.0 MB columnar Parquet (plus a CSV twin). The numbers are the ONS's, unchanged; we only made them small and ready to query. We refresh on the ONS's annual cadence.

Source & licence

Source: Office for National Statistics, Annual Survey of Hours and Earnings, ASHE Table 3 (region by occupation, 2-digit SOC), via the ONS dataset API. Contains public sector information licensed under the Open Government Licence v3.0. Attribution: “Source: ONS, Annual Survey of Hours and Earnings, licensed under the Open Government Licence v3.0. Normalized by iron26.” Last refreshed from source: 2026-06-14.