Doc: AMR Texas oil county income data dictionary
Status: Fixed 2024 baseline, November 14 2026 analytical edition
Owner: American Mineral Registry
Last verified: 2026-11-14
Depends on: BEA CAINC1 pinned archive; RRC April 1, 2024 preliminary January production table; analysis.py and figures.py

# Source and selection

BEA primary archive: https://apps.bea.gov/regional/zip/CAINC1.zip

Source table: CAINC1 county personal income summary. Source footnotes state last updated February 5, 2026, new 2024 statistics and revised 2020 through 2023 statistics. All historical observations use this same pinned vintage. Monetary values are current dollars, without inflation adjustment. The file's personal income line is in thousands of dollars; PCPI is dollars; population is persons.

RRC primary selection: https://www.rrc.texas.gov/news/040124-texas-oil-and-gas-production-statistics-for-january-2024/

The group is the ten counties in Table 3, ranked by preliminary January 2024 crude oil production, published April 1, 2024: Midland, Martin, Upton, Loving, Howard, Karnes, Reeves, Reagan, Andrews and Glasscock. Oil from oil leases excludes separately reported condensate. This is not a current ranking, annual production ranking or income selected group.

# Core files

| CSV | Rows | Observation and scope |
| --- | --- | --- |
| texas-all-counties-2010-2024.csv | 3,810 | Every one of 254 Texas counties in each year 2010 through 2024 |
| fixed-oil-cohort-2010-2024.csv | 150 | Same ten named counties, fifteen years |
| texas-state-2010-2024.csv | 15 | Published Texas state totals, retained separately from county rows |
| cohort-weighted-and-unweighted.csv | 15 | Combined cohort totals and weighting comparisons, one per year |
| fixed-cohort-2024.csv | 10 | All selected counties, sorted by 2024 statewide PCPI rank |
| fixed-cohort-growth-2010-2024.csv | 10 | Matched endpoint growth factors and statewide ranks |

# Core field definitions

`fips`: five digit BEA geographic code, retained as text. Texas is 48000; county codes begin 48. Literal surrounding spaces and quotes in the source GeoFIPS are removed when reading.

`county`: source geographic name with the Texas suffix and source footnote asterisk removed for display. Texas state rows retain the Texas name.

`year`: calendar year of income and source resident population.

`personal_income_thousand_usd`: source CAINC1 line 1, thousands of current dollars. Multiply by 1,000 for dollars before dividing by persons. It is resident personal income across the source's full components, not oil industry revenue, production value or private royalty receipts.

`population`: source CAINC1 line 2, persons. BEA specifies Census Bureau midyear resident population estimates. It is neither households nor employment.

`reported_pcpi_usd`: source CAINC1 line 3, published whole dollar per capita personal income. County ranks, county mean/median and county to state ratios use this published field.

`calculated_income_per_resident_usd`: line 1 times 1,000 divided by line 2, with Decimal precision. Small differences from reported whole dollar PCPI are rounding. Growth calculations use this unrounded quotient.

`in_fixed_oil_cohort`: yes for the fixed ten names, no for other Texas counties. The same membership applies in every year.

`texas_rank_competition`: one plus the number of the 254 Texas counties with strictly higher reported PCPI in that year. Ties share a rank and leave the following gap. State rows are excluded from ranking.

`ratio_to_texas_reported_pcpi`: reported county PCPI divided by the same year's published Texas PCPI. Unit is a multiple, not a percentage point change.

`source_url`: original authority, not an AMR measured survey.

# Combined annual summary

`cohort_count`: 10 every year.

`personal_income_thousand_usd` and `population`: sums across the fixed ten county observations in the same year.

`population_weighted_income_per_resident_usd`: sum of personal income in thousands times 1,000 divided by sum of population. Calculated directly from underlying totals rather than from rounded county PCPIs.

`unweighted_median_county_pcpi_usd`: average of the fifth and sixth ordered reported county PCPI values, every county one observation. It is not a median of people or households.

`unweighted_mean_county_pcpi_usd`: arithmetic mean of the ten reported county PCPI values, no population weights.

`texas_reported_pcpi_usd`: published Texas CAINC1 line 3 for the year.

`weighted_ratio_to_texas`: combined cohort quotient divided by published Texas PCPI for the same year.

`texas_population_share_percent`: sum of the ten county populations divided by published Texas population, times 100.

`texas_personal_income_share_percent`: sum of the ten county personal income totals divided by published Texas personal income, times 100.

`counties_above_texas`: count of selected counties with reported PCPI strictly greater than published Texas PCPI in the same year.

# Growth table

`period`: 2010-2024. This literal date range does not imply monthly observations.

`income_factor`: 2024 personal income divided by 2010 personal income.

`population_factor`: 2024 population divided by 2010 population.

`income_per_resident_factor`: calculated 2024 income per resident divided by calculated 2010 income per resident. It equals income factor divided by population factor.

`income_growth_percent`, `population_growth_percent`, `calculated_income_per_resident_growth_percent`: corresponding factor minus one, times 100. Negative population growth is an observed decrease, not missingness. Personal income growth and income per resident growth are nominal, without an inflation adjustment.

`rank_2010`, `rank_2024`: competition ranks among all 254 Texas counties using published PCPI in each year.

# Additional complete comparison views

The following four files are generated by `figures.py` after the core analysis. They are supporting views of the same source observations and are not additional surveys.

`rrc-fixed-cohort-selection.csv`: ten exact RRC Table 3 entries, preliminary January 2024 oil lease crude volumes in barrels, published rank, county, scope, period and primary URL.

`figure-income-history-relative-to-texas.csv`: fifteen annual rows containing the combined cohort quotient, unweighted county median, Texas PCPI, their respective ratios to Texas and count above Texas. The median ratio is county median divided by published Texas PCPI.

`cohort-county-concentration-2024.csv`: ten rows. Each county's population and income are divided by the respective total of the same fixed ten counties. Both shares are percentages and use different denominators.

`figure-summary-measures-2024.csv`: four rows. Combined cohort quotient, unweighted county median, unweighted county mean and published Texas PCPI, with each weighting denominator named.

# Precision, missingness and interpretation

The essential panel is complete: 254 counties plus Texas, three measures, fifteen years, or 11,475 source numeric cells. Missing values are not imputed, converted to zero or treated as evidence of absence. Source files and exact checks are in the reproduction bundle. The panel covers Texas and does not establish a complete national ranking.

Decimal arithmetic uses 34 digits in the shared analysis. Table and prose dollars are normally rounded to whole dollars; ratios in the first article table use three decimal places; shares use four decimal places; growth percentages use one decimal place. The CSVs retain calculation precision. A displayed zero in a rounded derivative, if any, would not imply a missing source value.

Resident personal income is not local GDP, oil production revenue, median household income, wealth, take home pay or private mineral royalty income. No causal oil production effect is estimated. The source does not identify how gains are distributed among residents or whether the same people reside in a county at both endpoints.

AMR's calculation, compilation and original figures may be reused with attribution under CC BY 4.0, preserving the original authorities, income definitions, cohort rule, periods, units and denominators. Underlying producer material retains its original terms.
