How to Build a Labor-Shed Answer a Site Selector Can Verify

Workforce availability is the dimension where tier-3 cities are most tempted to overclaim, and the one where the overclaim is easiest to check. The pattern repeats itself: a metro population figure presented as a labor pool, a community college's enrollment presented as a pipeline, an OEWS employment estimate covering a multi-county published area presented as available workers at a specific site.

Devin Hillsdon-Smith of Hyphen Strategies described the result in a 2025 Area Development piece: a proposal claiming 200,000 workers when the defensible figure was closer to 120,000. The 2026 Guild Pulse Check found 51% of 49 respondents listing workforce availability and quality among current elimination factors (respondents could cite multiple factors; 82% identified as industrial specialists). How many of those eliminations resulted from inflated data rather than genuine scarcity is not measured. But a director who presents an indefensible number gets eliminated on credibility before the labor market itself is evaluated.

The common inflation patterns are identifiable. Counting the entire OEWS estimation area when only part of it falls inside the commute shed. Labeling OEWS employment estimates as "available workers" when OEWS does not survey availability. Summing figures from OEWS, OnTheMap, and ACS into a single number without disclosing that the sources cover different universes, geographies, and vintages. Presenting training-program completions as guaranteed hires. Each produces a number larger than what the data support, and each is checkable by a selector with access to the same public sources.

A passing answer has four components: a commute boundary derived from the prospect's operating conditions, occupation-level employment data with its survey mechanics disclosed, a documented crosswalk from the prospect's staffing plan to published classification codes, and a presentation format the selector can independently verify.

What follows is the procedure for building each one. It is written so workforce board staff and community college administrators who compile supporting data can act on individual sections without additional context. Issue #1's workforce architecture established the occupation-level framework and the principle that OEWS does not prove availability. This is the mechanical build.

Step 1. Define the Commute Boundary From the Prospect's Operating Profile

The commute boundary is a geography shaped by the prospect's shift schedule, offered wages, and the road network connecting the site to residential concentrations of relevant workers.

No reviewed public source establishes a universal 30-, 45-, or 60-minute manufacturing commute radius. Iowa Workforce Development's 2025 statewide labor-shed study found workers with transferable skills commuting an average of 14 miles one way. Respondents willing to change employment said they would travel an average of 28 miles. Those are Iowa survey results. They describe one state's observed patterns, not a transferable rule.

Inputs that determine the boundary:

• Shift start and end times. A 6:00 a.m. start eliminates workers who depend on transit systems beginning service at 6:30. A rotating 12-hour shift compresses the pool differently than a standard 8-hour day. ACS PUMS files include departure-time and arrival-time variables showing when manufacturing workers in the area actually travel.

• Offered wage band. Workers commute farther for higher wages. OEWS publishes occupation-specific wage percentiles for each published area. If the prospect offers a machinist wage at the 75th percentile of the local OEWS distribution, the realistic commute shed is wider than if the offer sits at the 25th.

• Road network from the site. ACS table B08126 reports transportation mode by broad industry. In most tier-3 markets, manufacturing workers overwhelmingly drive. The boundary should follow the actual road network, not a radius drawn on a flat map. A site with interstate access in two directions produces a different shed shape than one served by two-lane state highways.

• Observed commuting patterns. Census OnTheMap shows where workers currently employed in a selected area live, using LEHD origin-destination data at census-block level. It accepts user-drawn areas and shapefiles, so the analysis can center on the industrial park rather than the county. Standard distance bands: under 10 miles, 10–24 miles, 25–50 miles, over 50 miles. These describe existing home-work relationships for current employers. They do not predict who will apply for a new position.

What OnTheMap will not give you: its "Goods Producing" industry segment combines agriculture, mining, construction, and manufacturing, and it publishes no detailed SOC occupations. Use it for commute-pattern evidence, not occupation-specific counts.

Producing the boundary: The method for converting these inputs into a defined geography is drive-time isochrone analysis, a travel-time polygon generated from the site coordinates using the actual road network at a specified time of day. GIS platforms and several free web tools produce isochrones. Run the analysis at the prospect's shift-start time rather than midday, because congestion patterns and travel times differ materially. If the prospect operates multiple shifts, produce a separate isochrone for each start time.

On the wage-commute relationship, the reviewed sources do not publish a national elasticity. As a working rule, a wage offer above the local OEWS median for the target occupation supports extending the boundary beyond observed commute distances, and a below-median offer supports contracting it. That is practitioner judgment, not an empirical formula, and it should be labeled as such in the documentation.

What to document:

• Site coordinates and the boundary used
• Prospect shifts, start/end times, days of operation
• Offered wage band for each occupation, with the OEWS percentile comparison
• Travel-time method: software, transportation modes, time of day modeled
• OnTheMap vintage, job type, industry filter, selected geography
• Written rationale when the boundary differs by occupation or shift

Step 2. Understand What OEWS Actually Measures Before You Cite It

The May 2025 OEWS release, published May 15, 2026, is not a May 2025 headcount.

Survey design: OEWS is an employer survey of occupational employment and wages for nonfarm wage-and-salary workers. It excludes the self-employed, owners of unincorporated firms, private-household workers, and unpaid family workers. Two panels of roughly 186,000–189,000 establishments are surveyed each year, one in May and one in November. The May 2025 estimates combine six panels spanning November 2022 through May 2025, covering approximately 1.1 million establishments at a 66.2% response rate.

BLS uses a model-based method (MB3) to produce estimates. Responding establishments' data predict staffing and wages for nonrespondents and unsampled establishments. Older wage data are adjusted to the May 2025 reference date. Older staffing patterns carry forward under an assumption that industry staffing patterns change slowly.

A plant opening, closure, or automation conversion that occurred in 2024 enters the estimates gradually, because each release retains three years of panels.

Geographic granularity: OEWS publishes estimates for approximately 530 metropolitan and nonmetropolitan areas. Metropolitan areas follow OMB Bulletin 23-01 delineations. Nonmetropolitan areas are OEWS-specific constructions established with state workforce agencies. A single published nonmetropolitan area may contain many counties at very different distances from your site.

BLS publishes an area-definitions spreadsheet showing which counties fall in which estimation area. Those county links do not create county-level estimates.

The boundary-crossing problem: Your prospect-specific commute shed will almost certainly span parts of multiple OEWS estimation areas. BLS publishes no method for combining parts of several areas into a custom estimate with a new reliability measure.

Three workable approaches:

1. Show each overlapping OEWS area separately. Preserve the published estimates and their reliability measures. Note which portion of the commute shed falls within each area.
2. Contact state workforce agencies. BLS's FAQ directs users seeking substate detail beyond the public tables to state agencies, which may maintain unpublished estimates for smaller geographies.
3. Use a different dataset for the geographic dimension. OnTheMap for block-level origin-destination patterns, ACS PUMS for custom commute-time tabulations at PUMA scale (areas of at least 100,000 people).

Do not merge an OEWS occupation count, a LODES worker-flow count, and an ACS household estimate into one unlabeled "available labor" figure. The sources have different universes, geographies, vintages, and units.

Reliability fields: OEWS publishes a relative standard error (RSE) for each employment and mean-wage estimate. An employment estimate of 5,000 with a 2% RSE has an approximate 90% confidence interval of 4,840 to 5,160. RSEs are not published for median or percentile wages. An absent occupation row does not prove zero employment; it may reflect quality thresholds or disclosure restrictions.

How to label it: Every OEWS figure in your presentation should carry: "May 2025 OEWS estimates, released May 15, 2026; six survey panels from November 2022 through May 2025." The wording of the claim matters as much as the label. "OEWS estimates 1,240 workers employed as machinists across the Springfield MSA" is traceable. "1,240 machinists are available" adds a meaning OEWS does not survey.

Step 3. Build the SOC Crosswalk From the Prospect's Staffing Plan

The prospect gives you job titles. OEWS reports Standard Occupational Classification codes. Most labor-availability answers go wrong in the translation between the two, because the translation requires judgment and the judgment usually goes undocumented, which means the selector cannot check it.

Current system: The 2018 SOC contains 867 detailed occupations organized into 459 broad occupations, 98 minor groups, and 23 major groups. OEWS publishes about 830 categories. Full definitions, coding structure, illustrative examples, and "includes" and "excludes" statements are downloadable from the BLS 2018 SOC site.

Classification rules that affect your crosswalk:

SOC classifies by work performed, not by the employer's title. Two workers with the same title can receive different SOC codes if their duties differ. Each worker is assigned to one detailed occupation. When a job fits multiple occupations, SOC assigns it to the one requiring the highest skill level; if skill requirements are equivalent, it goes to the occupation where the worker spends the most time.

Two situations you will encounter in every industrial staffing plan:

• One title, multiple duty sets. "Maintenance technician" may cover mechanical, electrical, controls, and facilities positions. Divide the headcount by actual duties before selecting SOC codes. Ask the prospect for duty descriptions, not titles alone.
• One worker, mixed duties. A production worker who also performs quality checks still receives one SOC code under the highest-skill/most-time rule. Document which code you selected and why.

Workers spending at least 80% of the job on supervision are coded to the first-line supervisor occupation. Below 80%, they are coded with the workers they supervise.

For each position in the staffing plan, document:

• Prospect title and duty description
• Headcount by shift, at startup and at full ramp
• Required experience, credentials, and employer-defined skills
• Selected 2018 SOC code and the definition text supporting it
• Alternative SOC code considered and reason for the final choice
• Whether OEWS publishes the detailed occupation, a broad occupation, or an OEWS-specific aggregation for that code

Issue #4's demand-envelope framework showed that similarly sized projects can require materially different occupation mixes. The crosswalk is where that difference becomes specific enough to evaluate against published data.

Connecting SOC to training completions via CIP:

The joint NCES-BLS 2020 CIP to 2018 SOC Crosswalk relates six-digit Classification of Instructional Programs codes to six-digit SOC occupations. The relationship can be one-to-many or many-to-many. NCES states explicitly that the crosswalk is based on code definitions and expert judgment, not empirical placement data. A matched CIP identifies a potentially related program. It does not establish that the institution's completers entered the occupation, remain in the labor shed, hold the prospect's preferred certification, or are available for its shift.

For each training connection, document:

• Institution name and IPEDS Unit ID
• Six-digit CIP code, program title, award level, reporting period
• Reported completions (kept separate from unique individuals and local placements)
• CIP-to-SOC relationship and alternative occupations in the federal crosswalk
• Institution-confirmed current seats, instructors, equipment, and first-completion date when available

One CIP maps to several occupations. You cannot allocate a program's graduates among occupations, and you cannot assume that multiple award levels represent multiple different people. State both limits plainly when presenting the data.

Step 4. Assemble the Presentation for Independent Verification

Hillsdon-Smith described the diligence practice: asking which region was counted, who was excluded, requesting underlying sources, validating figures against public or third-party data. The Guild's 2024 Fall Forum emphasized full data transparency, objective methods, and data integrity. Build the answer assuming the selector will check every figure.

Structure the answer in four separable components:

1. Commute-shed definition. Map, method, inputs, rationale. If the boundary differs by occupation or shift, show each version.

2. OEWS evidence table. For each SOC code in the crosswalk:

• Complete BLS area name and the counties it contains
• SOC code and occupation title
• Employment estimate and employment RSE
• Median wage and selected percentile wages
• Location quotient where useful
• Full vintage label

If the commute shed crosses multiple OEWS areas, show each area separately. Do not sum them into a single figure without disclosing that the sum covers territory outside the commute polygon.

3. SOC crosswalk record. The full translation from prospect titles to SOC codes, with duty descriptions and coding rationale. This lets the selector trace your occupation counts back to the staffing plan.

4. Training evidence table. Completions by institution, CIP, award level, and year, with the CIP-SOC relationship and its limitations stated. Separate observed past completions from future institutional commitments. A signed letter from a community college president committing to launch a program is a different category of evidence than three years of IPEDS completion data. Issue #3 laid out the progression from claim to commitment; training evidence follows the same ladder.

Supplementary evidence: If you have additional data — an Iowa-style labor-shed survey measuring willingness to change employment and required wage, or employer-reported residential ZIP codes establishing observed commute patterns — present it as a separate layer with its own source, vintage, sample size, and margin of error. Supplementary evidence gains value when clearly labeled. It loses value when blended into the OEWS figure.

What to hand your partners:

This methodology requires data from institutions the ED office does not control. The workforce board or state LMI unit handles OEWS retrieval, area construction, RSE disclosure, and SOC coding support. The community college handles IPEDS completions, CIP-coded program inventory, current capacity, and institutional commitment letters. Issue #2 mapped these counterparty relationships in detail.

Give each partner the specific section of this methodology that applies to their contribution, along with the prospect's staffing plan or as much of it as the NDA permits. A workforce board analyst who receives a list of prospect job titles without duty descriptions cannot build a defensible SOC crosswalk. A community college administrator who receives SOC codes without the CIP crosswalk cannot identify which programs to report.

The finished answer should contain no figure the selector cannot independently pull from the same public sources. In diligence, they will.