Portland decline: data behind the charts Each file is plain CSV with a header row. Source ids refer to the source list on the page. Nothing is interpolated or estimated. A year is absent when the source has no value for it. city-vs-suburbs.csv Rows: 60 Columns: area_key, area: the place. year. jobs: all jobs counted at workplaces in the area. index_2019_100: jobs as a percent of the 2019 count. Sources: census-lodes-wac, census-lodes-xwalk How it was computed: Jobs in census blocks (LODES8 workplace file, all jobs, JT00, total C000) summed by area. A block belongs to the City of Portland when the crosswalk lists its place as Portland city, at 2020 boundaries, in every year. Multnomah County outside Portland is the rest of that county. Washington and Clackamas counties exclude the small parts of Portland that lie in them. Years 2010 to 2024. downtown-rings-jobs.csv Rows: 5 Columns: ring, label: distance band from the downtown center. jobs_2017, jobs_2019, jobs_2024: all jobs. change and pct columns: 2024 less the earlier year, as a count and a percent. Sources: census-lodes-wac, census-lodes-xwalk How it was computed: City of Portland blocks only. Distance is the great-circle distance in miles from latitude 45.5189, longitude -122.6793 to the internal point of each block given in the crosswalk. The bands do not overlap: within 0.5 mile; over 0.5 to 0.75; over 0.75 to 1.25; over 1.25 to 2; over 2. Adding the first two bands gives the 0.75 mile figures used on the page. downtown-by-sector.csv Rows: 5 Columns: sector_key, sector: the group. jobs_2019, jobs_2024. change: 2024 less 2019. pct: percent change. Sources: census-lodes-wac, census-lodes-xwalk How it was computed: City of Portland blocks within 0.75 mile of the downtown center. Office-type is LODES sectors CNS09 to CNS13 (information; finance and insurance; real estate; professional, scientific and technical services; management of companies). Restaurants, hotels and shops is CNS07, CNS17, CNS18 and CNS19 (retail; arts, entertainment and recreation; accommodation and food services; other services). Government is CNS20 (public administration). Health and education is CNS15 and CNS16. Other is every remaining sector. The five groups add to the total. office-by-distance.csv Rows: 30 Columns: area_key, area. year. office_jobs: jobs in LODES sectors CNS09 to CNS13. index_2019_100. Sources: census-lodes-wac, census-lodes-xwalk How it was computed: Same blocks, distance rule and office-type sectors as downtown-by-sector.csv. Downtown is within 0.75 mile of the center; the other series is City of Portland blocks more than 2 miles from it. Years 2010 to 2024. center-share-long.csv Rows: 162 Columns: region_key, central_county, region: the counties compared. year. central_share_pct: the central county jobs as a percent of the region jobs. bea_file: which BEA file the year comes from. Sources: bea-caemp25s, bea-caemp25n How it was computed: BEA total employment (line 10, full-time and part-time jobs by place of work, proprietors included) for the central county divided by the sum for the listed counties. 1969 to 2000 come from the SIC file and 2001 to 2022 from the NAICS file; total employment is defined the same way in both. Broomfield County, Colorado has no figures before 2001. The BEA series ends in 2022. central-county-rank.csv Rows: 50 Columns: rank: 1 is the largest gain. metro. central_county, county_fips. jobs_2019, jobs_2025: annual average employment, all ownerships. pct_change. Sources: bls-qcew-annual, census-pep-cbsa-2025 How it was computed: The 50 metropolitan areas with the most residents in the 2020 estimate (July 2023 delineation). The central county is the county in the metro with the most jobs in 2019. That is not the historic core everywhere: it is Middlesex for Boston, Oakland for Detroit and Alameda for San Francisco. office-type-jobs.csv Rows: 33 Columns: county_key, county. year. office_type_jobs: private annual average employment in three supersectors. index_2019_100. Sources: bls-qcew-annual How it was computed: Private employment in information (1022), financial activities (1023) and professional and business services (1024), added. Hennepin and Ramsey counties are added together. King County (Seattle) is omitted because of a 2022 industry recoding: its trade, transportation and utilities jobs fell by about 50,000 between 2021 and 2022 while this group rose by about 89,000, so its series is not comparable across that year. Years 2015 to 2025. moves.csv Rows: 12 Columns: irs_years, second_year: the pair of filing years. arrivals and departures: tax returns moving between Multnomah County and other states. index columns: the same count with the average of 2017-18 and 2018-19 set to 100, for Multnomah and for the median of nine comparison counties. leaver_income columns: adjusted gross income per departing return (all US destinations) as a percent of income per return that did not move. Sources: irs-soi-migration How it was computed: IRS county inflow and outflow files, rows "Total Migration-Different State" for returns to and from other states, "Total Migration-US" for income of leavers, and the county non-migrant row for stayers. The comparison counties are King, Denver, Hennepin, Franklin, San Diego, Maricopa, Salt Lake, Travis and Wake. The index is used because county sizes differ; counts are given for Multnomah. The 2014-15 and 2016-17 files are known to be irregular and are marked. pop-gap-parts.csv Rows: 3 Columns: part, label: component of population change. County columns: the five-year sum, July 2020 to July 2025, per 1,000 residents in 2020. gap_to_median_of_three: Multnomah less the median of King, Denver, and Hennepin and Ramsey together. share_of_gap_pct: that gap as a percent of the sum of the three gaps. Sources: census-pep-county-2025 How it was computed: Census Vintage 2025 county components of change (NATURALCHG, INTERNATIONALMIG, DOMESTICMIG) summed over the years ending July 1, 2021 to 2025 and divided by the July 1, 2020 estimate. County international migration is modeled by the Census Bureau, not counted. state-outside-metro.csv Rows: 4 Columns: state_key, label. jobs_2019, jobs_2025: annual average employment, all ownerships. pct_change. Sources: bls-qcew-annual, census-pep-cbsa-2025 How it was computed: Statewide employment less the employment of the counties of the largest metro area that lie in that state: Portland (Multnomah, Washington, Clackamas, Yamhill, Columbia), Seattle (King, Snohomish, Pierce), Denver (ten counties) and Minneapolis-St. Paul (the Minnesota counties). Metro county lists are the July 2023 delineation. downturns.csv Rows: 10 Columns: year. multnomah_pct, multnomah_jobs: change in Multnomah County jobs from the year before. rest_of_region_pct, rest_of_region_jobs: the same for Washington, Clackamas and Clark counties together. Sources: bea-caemp25s, bea-caemp25n How it was computed: Every year from 1981 to 2022 in which BEA total employment in Multnomah County was lower than the year before. 2001 is not compared with 2000 because the two years come from different BEA files. event-order.csv Rows: 29 Columns: series_key, series. year. portland, washington_clackamas: the count on each side. ratio_index_2019_100: Portland divided by the suburbs, with 2019 set to 100. Sources: census-lodes-wac, census-lodes-xwalk, census-zbp How it was computed: Jobs: LODES8 blocks in the City of Portland against blocks in Washington and Clackamas counties outside Portland, with the sector groups of downtown-by-sector.csv, 2015 to 2024. Establishments: Census ZIP Code Business Patterns, all sectors, for a set of Portland ZIP codes against Washington and Clackamas county ZIP codes, 2015 to 2023. Neither source has a 2025 value. downtown-blocks.csv Rows: 89 Columns: block: the 2020 census block number. lat, lon: the block's internal point. distance_miles: distance from the downtown center. jobs_2019, jobs_2024: all jobs counted at workplaces on the block. change: 2024 less 2019. building: the building or address on the block, where it was looked up. note: the largest changes by sector and what is known about the block. Sources: census-lodes-wac, census-lodes-xwalk, portlandmaps-taxlots, census-tigerweb-blocks How it was computed: Every City of Portland census block within 1.25 miles of latitude 45.5189, longitude -122.6793 whose job count (LODES8 workplace file, all jobs, JT00, total C000) changed by 150 or more between 2019 and 2024, largest loss first. The building is named for the fifteen largest losing blocks and the 25 largest gaining blocks, from the tax lots that fall inside the block outline. A tax lot shows the owner of a property, not its tenants. The other blocks have no building name. This file is not built by build-data.py; it comes from the block research behind src/lib/portland-decline/blocks.ts. employer-decisions.csv Rows: 33 Columns: employer. building. action: what the employer or building owner did, and how any job count was matched to it. date. lodes_jobs_2019, lodes_jobs_2024: jobs in the sector named in the action, for every employer on the block; blank where no match was made. destination. stated_reason: the employer's or owner's own words, or blank. reason_source: who said it and when. source_ids, source_urls: the sources for the row. provisional: yes where the row rests on a secondhand source or on an inferred match between a block and an employer. Sources: listed in each row; the ids refer to the source list on the page. How it was computed: One row per decision documented by a saved news report, company filing or public record. The job records give counts by block and industry, never by employer, so a count beside an employer's name is the count for that industry on that block. Nothing is entered as a reason unless the employer or owner said it. This file is not built by build-data.py; it comes from the block research behind src/lib/portland-decline/blocks.ts. police-officers-response.csv Rows: 10 Columns: year. sworn_officers_oct31: sworn officers employed by the Portland Police Bureau on October 31. high_priority_calls_dispatched: calls for service the Bureau classes as high priority. mean and median minutes from the call being received to an officer arriving. The mean is also split into minutes waiting for an officer to be assigned and minutes travelling. Sources: fbi-cde-police-employees, ppb-dispatched-calls How it was computed: Officer counts are taken as published. Call figures are computed from the Bureau's dispatched calls file for calls with a recorded response time. Years 2016 to 2025. downtown-by-industry.csv Rows: 17 Columns: industry. jobs_2010, jobs_2019, jobs_2024: all jobs. change_2010_2019, change_2019_2024: differences in jobs. pct_2019_2024: percent change. Sources: census-lodes-wac, census-lodes-xwalk How it was computed: City of Portland blocks within 0.75 mile of the downtown center (latitude 45.5189, longitude -122.6793), the same blocks as downtown-by-sector.csv. Each row is one LODES industry sector (CNS04, CNS06, CNS07, CNS09 to CNS20). Utilities, manufacturing, transportation, farming and mining are added together as "All other industries". The last row is all jobs. LODES reports these 20 sectors and nothing finer. county-industry-change.csv Rows: 56 Columns: period (2019-2025 or 2015-2019). group, industry. One column per place with the percent change in jobs in its central county. median_three_most_similar: the median of Seattle, Denver and Minneapolis-St. Paul. median_nine: the median of the nine places other than Portland. portland_rank_1_is_largest_fall and places_ranked. multnomah_jobs_start and multnomah_job_change: jobs in Multnomah County. Sources: bls-qcew-county-detail How it was computed: BLS QCEW annual average employment, private employers (ownership 5), except "All jobs" (all ownerships) and "Government, all levels" (federal, state and local). Industries are NAICS codes 5112 or 5132 (software publishers), 5415, 334, 5411, 5412, 5413, 5416, 5418, 5417, 522, 524, 523, 551, 5613, 31-33, 23, 42, 48-49, 44-45, 722, 721, 71, 622, 621 and 6113. Minneapolis-St. Paul is Hennepin and Ramsey counties added together. A blank cell has no comparable figure: the Bureau withheld it, the place had fewer than 300 jobs in the industry, or jobs were reclassified (retail and head offices in King County, head offices and manufacturing in Travis County, computer manufacturing in Wake County). county-industry-by-year.csv Rows: 3360 Columns: place, central_county, industry, year, jobs. Sources: bls-qcew-county-detail How it was computed: The same industries and rules as county-industry-change.csv, as annual average jobs for every year from 2014 to 2025. Blank where there is no comparable figure.