[FEATURE] Estimates of Self Employed Individuals

[GregorSchroeder]

Is your feature request related to a problem? Please describe

The Census LEHD LODES and BLS QCEW datasets used to generate employment estimates do not include certain types of non-UI covered employment. Specifically, this excludes certain types of self-employments. This is a non-trivial amount of employment and has been historically estimated by SANDAG in forecasts prior to the Series 15 forecast as well as in employment estimates SANDAG produced historically.

Describe the solution you'd like

1. Get ACS 5-year B24080 at block group level
2. Assign block group level self-employment to MGRAs. This could be area-based, housing unit based, point-level EDD based, land-use based, or area-based. To be discussed and determined.
3. Control to county-level self-employment (see discussion below)

The only source we are currently aware of for publicly available small area self-employment data is the 5-year American Community Survey Table B24080. Unfortunately, the 5-year ACS for a given year is typically released late in the following calendar year. See the 2023 ACS release schedule and the 2024 ACS release schedule.

Fortunately, this is a common issue in SANDAG's Annual Estimates Program and is typically handled by controlling to a more aggregate dataset that is on an earlier release schedule. Unfortunately, investigation into common sources of employment data have not proven fruitful. The California Employment Development Department does not produce any estimates of self-employment, the Bureau of Economic Analysis no longer produces county-level data, and the Bureau of Labor Statistics Quarterly Census of Employment and Wages does not include self-employment. At this time, the only source of more aggregate self-employment data on an earlier release schedule we are aware of would be the ACS 1-year Table B24080, but that is only released a couple months prior to the 5-year ACS and does not align with SANDAG's Spring/Summer release schedule for Annual Estimates. The 1-year ACS table could be used for controlling and simply hold the last available year's value constant until a different source of data presents itself.

Describe alternatives you've considered

Need to continue investigating other sources for aggregate control data.
Could potentially use trends from previous years data to extrapolate missing control data.

Additional context

This query could be used to grab 1-year control totals at the county-level. A similar query would be used to grab the 5-year data at the block group level.

SELECT
    [year],
    SUM([value]) AS [value]
FROM [detailed].[values]
INNER JOIN [detailed].[variables]
    ON [values].[table_id] = [variables].[table_id]
    AND [variables].[variable] = [values].[variable]
INNER JOIN [detailed].[tables]
    ON [values].[table_id] = [tables].[table_id]
WHERE
    [tables].[name] = 'B24080'
    -- there is no 1-year data release for 2020
    AND ([tables].[product] = '1y' OR ([tables].[year] = 2020 AND [tables].[product] = '5y'))
    -- TODO: currently excluding "self-employed in own incorporated business workers", is this correct?
    AND REPLACE([variables].[label], ':', '') IN (
        'Estimate!!Total!!Male!!Self-employed in own not incorporated business workers',
        'Estimate!!Total!!Female!!Self-employed in own not incorporated business workers'
    )
GROUP BY [year]