How we estimated Oakland city worker residency: the methodology
The scientific and statistical process we used to estimate how many City of Oakland employees actually live in Oakland and comparable California cities.
Executive summary
This report explains the step-by-step scientific and statistical process we used to estimate how many City of Oakland employees actually live in Oakland and comparable California cities. Because cities are not required to publicize this information (some do so voluntarily), and California state law prevents local governments from forcing employees to live in the cities they serve, most city HR departments do not publish lists of employee ZIP codes.
To bridge this data transparency gap, we used established scientific and statistical methods—the same kinds used by economists and public policy researchers—plus available data from Oakland and other cities to build a reasonably accurate estimate. By combining information in federal databases with verified data from comparable cities (our “model groups”), we corrected the natural errors found in survey data. Ultimately, this statistical modeling estimates that about 32% of Oakland’s city workforce lives in Oakland, with a mathematically defensible range falling between 22% and 42%.
Here is how the methodology works, and how we adjusted our models to reflect real-world data and evidence.
1. Statistical triangulation
Since the City of Oakland does not publish lists of its employees’ ZIP codes, we employed a statistical method that analyzed large datasets from the federal government using a method called triangulation—i.e., looking at a problem from multiple angles using different sources to see where the data overlaps.
Person-Based Data: We used the U.S. Census Bureau’s American Community Survey (ACS), filtering for people who marked their job as “local government.”
Jobs-Based Data: We used the Census Bureau’s commuting database (LEHD), which tracks cell phone and tax data to map where people in “public administration” live versus where they work.
By combining these two datasets, we built an initial, rough estimate based on the statistical data they contain.
2. Calibration
An early challenge with our initial Census estimates is that federal data groups all “local government” workers together. If we pull Census data for Oakland, it doesn’t just count City of Oakland employees; it also counts public school teachers, transit workers, and regional utility staff who happen to work inside Oakland boundaries.
Because the Census-only measurement tool is slightly blurry or biased, we employed other data to help calibrate it. To do this, we used a statistical method based on Bayesian principles that compared the initial, blurry data against a “model group” to derive a more exact error rate.
How we did it: A few local government jurisdictions do publish their exact, official residency numbers. San Francisco officially reports a 41.9% residency rate, Los Angeles reports 36.3%, and Alameda Unified School District 57%. We ran our blurry Census query for these jurisdictions to see what the Census estimates the rate was.
Why we did it: By comparing the Census guess to these cities’ actual numbers, we discovered exactly how “off” the Census data usually is (it typically misses the mark by 3 to 6 percentage points). We then applied that same mathematical correction to our Oakland data to fix the blur. This adjustment gave us our reasonably accurate baseline estimate of 32% for Oakland, with a confidence interval (margin of error) of approximately +/-10%.
3. Confidence interval
Because we are using survey data and making statistical adjustments, it is mathematically imprecise to give one single, perfect percentage. All surveys have a margin of error.
To account for this, we used a statistical method that runs thousands of simulated scenarios to calculate a confidence interval. This tells us that while 32% is our best, middle-ground estimate, the true number could realistically sit anywhere between 22% and 42%. When estimating Oakland’s residency rate, we looked at this entire range, knowing the true number is highly likely to fall within those boundaries.
4. Real-world adjustments
A flat 32% average doesn’t tell the whole story. Different jobs have different lifestyles, shift schedules, and rules. We adjusted the model for different municipal unions based on hard data and operational realities:
Where we have the actual data (overriding the model)
Police Officers (OPOA): We did not have to estimate this at all. Actual Oakland Police Department staffing reports explicitly state that only 8% to 9% of sworn officers live in Oakland. Because the hard, official data exists, we threw out the Census estimate entirely and used the city’s exact numbers.
Where we adjusted for shift schedules (cross-referencing)
Firefighters (IAFF Local 55): The raw Census data overstates the statistically-derived percentage of resident firefighters because of their unique schedules. Firefighters work 48 hours straight, then get 96 hours off. Because they commute only about 10 times a month (instead of 20+ like an office worker), they can easily live in further distant cities— and copious anecdotal information suggests that the vast majority of Oakland firefighters do live outside of town.
How we adjusted it: We looked at official data from San Francisco and Los Angeles. In those cities, public safety residency is routinely less than half the rate of the civilian workforce. We applied this known public-safety drop-off to Oakland, reasonably adjusting the firefighter estimate down to 15% to 20% to match the real-world impact of their shift structure.
Where we adjusted for contract rules
Civilian and Office Workers (SEIU 1021): We adjusted this group’s estimate higher (35% to 48%) because Oakland gives a 5-point bonus on hiring exams to local residents, which statistically increases local hiring for these entry-level and service roles.
Engineers and Management (IFPTE 21): We adjusted this group lower (22% to 35%) because these are highly specialized jobs. For example, the city recruits and employs engineers— a highly specialized and in-demand profession— from all over the Bay Area, meaning many already own homes in other cities before they are hired, have greater means to absorb the expense of longer commutes, are more likely to have work-from-home options, and are less likely to relocate.
5. Adjusting the peer cities
When we applied these statistical methods to other California cities that don’t publish their ZIP code data, we similarly adjusted the estimates to account for the unique math of their local jobs and housing markets:
Sacramento (estimated 43%): We adjusted Sacramento’s numbers upward in part because they have historically greater wages-to-housing parity, and a strict proximity rule: emergency workers must live within a 35-air-mile radius of the city.
Fremont (estimated 18%): We adjusted Fremont’s numbers downward using an economic housing metric. Fremont has 1.65 jobs for every housing unit in the city, a relatively high jobs-to-housing ratio for the region. Because the jobs substantially outnumber the homes, thus creating upward pressures on the housing market in that city, it is a statistical likelihood that more of its workers commute from out of town.
San Jose (estimated 24%): We adjusted San Jose downward based on the extreme disparity between its local real estate prices and municipal salaries, a dynamic that pushes city staff into less-expensive housing markets located within commuting distance, such as the Central Valley.
Conclusion
The above-described methodology resulted in a reasonably accurate and staisticallky defensible estimate of Oakland city worker residency, with a confidence interval of ±10% .
The ideal situation would be for the city of Oakland to voluntarily produce its own concrete data about worker residency, as the cities of San Francisco and Los Angeles, and Alameda Unified School District have done— perhaps through an independent report produced by the Office of the Oakland City Auditor, which presumably would have access to the city’s personnel data for the purpose.
In the absence of official data and reporting from the City of Oakland, and because no other studies of city worker residency yet exist, we assert that this analysis is currently the best and most accurate estimate available— and further note that its findings are roughly consistent with the known data produced by other local government jurisdictions.
If the City of Oakland produces a report based on actual employee residency data in the future— and we encourage the city to do so— then we will update and republish this analysis accordingly.
References
Oakland Police Department. Quarterly Staffing Reports. This provided the exact 8-9% numbers for sworn officers, overriding the statistical estimates. https://www.oaklandca.gov/files/assets/city/v/1/city-administrator/documents/informational-memos/2022/quarterly-policy-staffing-report-q4-2021.pdf
Port of Oakland and SEIU Local 1021. Union Contract (MOU). This provided the local residency hiring points used to adjust civilian models upward. https://www.portofoakland.com/wp-content/uploads/2024/03/SEIU-Local-1021-and-The-Port-of-Oakland-MOU-July-1-2022-September-30-2025-signed.pdf
City and County of San Francisco. 2023 Workforce Report. Along with concrete data provided by the city of Los Angeles and Alameda Unified School District, this provided our “model group” data to calibrate the estimate. https://media.api.sf.gov/documents/DHR-Workforce-Report-2023_3lZNFRh.pdf
City and County of San Francisco. 2020 Annual Workforce Report, Phase I.
https://sfdhr.org/sites/default/files/documents/Reports/annual-workforce-report-2020.pdf
City and County of San Francisco. 2023 Workforce Report. https://media.api.sf.gov/documents/DHR-Workforce-Report-2023_3lZNFRh.pdf
San Francisco Chronicle. Here’s where San Francisco government workers live.
https://www.sfchronicle.com/bayarea/article/sf-city-government-employee-20329218.php
City and County of Los Angeles. 2022 City Controller Payroll Analysis. This additional data provided the 36.3% anchor data for cross-referencing public safety drops. https://controller.lacity.gov/landings/2022-employee-residence-analysis
City of Alameda. 2020 Housing Affordability and Displacement Report. This additional data provided the 57% anchor data for Alameda Unified School District. https://www.alamedaca.gov/files/assets/public/v/3/departments/alameda/econ-dev-amp-comm-services/city-of-alameda-housing-affordability-and-displacement-report.pdf
Port of Oakland. 61% local hiring reached at Port of Oakland’s Army Base development. https://www.portofoakland.com/press-release-406
City of Sacramento. Pre-employment Conditions. This provided the 35-mile emergency worker rule used to adjust our estimate for that city. https://www.cityofsacramento.gov/content/dam/portal/hr/Divisions/ECD/PreEmploymentConditions.pdf
City of Sacramento. Pre-employment Conditions.
https://www.cityofsacramento.gov/content/dam/portal/hr/Divisions/ECD/PreEmploymentConditions.pdf
City of Sacramento. Employee Handbook.
City of Sacramento. Rules and Regulations of the Civil Service Board.
U.S. Census Bureau. American Community Survey (ACS) Public Use Microdata. Used to gather person-based data and filter for “Class of Worker.” https://www.census.gov/programs-surveys/acs/microdata.html
U.S. Census Bureau. Longitudinal Employer-Household Dynamics (LEHD). Used to track commute flows for public administration workers. https://lehd.ces.census.gov/data/lodes/LODES8/LODESTechDoc8.3.pdf
City of Fremont. Draft 2023-2031 Housing Element.
https://www.hcd.ca.gov/housing-elements/docs/fremont-6th-draft082522.pdf
Community Scale. Fremont, CA - Housing Forecast - CommunityScale.
https://app.communityscale.io/dashboard/municipality/0626000/Fremont-CA
Association of Bay Area Governments (ABAG). Jobs-Housing Balance Report. https://abag.ca.gov/sites/default/files/factor_j3_jobs-housing_balance_v2.pdf
U.S. Census Bureau. American Community Survey Microdata. https://www.census.gov/programs-surveys/acs/microdata.html
U.S. Census Bureau. Data - Longitudinal Employer-Household Dynamics.
U.S. Census Bureau. OnTheMap. https://onthemap.ces.census.gov/index.html
U.S. Census Bureau. LODES Technical Document Version 8.3. https://lehd.ces.census.gov/data/lodes/LODES8/LODESTechDoc8.3.pdf
U.S. Census Bureau. Census Geocoder Documentation. https://www.census.gov/programs-surveys/geography/technical-documentation/complete-technical-documentation/census-geocoder.html
U.S. Census Bureau. Geocoding Services API. https://geocoding.geo.census.gov/geocoder/Geocoding_Services_API.html