National Job Projections vs. Local Openings: What Should You Compare?

What exactly is being compared: future projections or present openings?

In the United States, national and state occupational projections describe modeled employment conditions over a future period; current vacancies describe positions employers are recruiting to fill now. A projected annual opening is another forecast measure, not a live vacancy count. Before comparing any figures, match the occupation definition, geography, projection horizon, release vintage, and unit. Otherwise, apparent disagreement may come from comparing different things.

| Measure | What it describes | What it does not establish | | --- | --- | --- | | National occupational projection | A modeled outlook for employment in an occupation across the United States over a stated projection period. | Hiring at a particular employer or the number of jobs available in one region today. | | State or regional occupational projection | A modeled outlook for employment in an occupation within the stated state or substate area and period. | That the forecast is a live vacancy count, or necessarily more accurate than the national estimate. | | Projected annual occupational openings | Average annual opportunities attributed to projected employment change and occupational separations during the projection period. | Distinct, currently advertised positions or a promise that a particular worker can obtain one. | | Current vacancies | Positions that employers are actively seeking to fill at a specified point or during a defined collection window. | Long-run employment growth, or a complete count unless the source and coverage support that interpretation. |

The U.S. Bureau of Labor Statistics’ “Relationship with state-level projections” explains that state projection programs are developed by state labor-market information offices, using national projections as needed. The national and state products therefore have related inputs, but state methods and the geographic and time detail available can vary. A state forecast can include useful local industrial context; its regional granularity does not turn it into evidence of current recruiting. Check the issuing office, area covered, occupation code or title, forecast years, and publication vintage on the specific table rather than assuming every state's series is built or updated identically.

The word “openings” needs particular care. In the BLS “Occupational Separations Methodology Frequently Asked Questions,” projected annual occupational openings combine projected employment change with occupational separations. Separations represent workers leaving an occupation or the labor force, so projected opportunities can exist even when the occupation’s total employment is flat or declining. The published figure is an average over a projection period. It does not mean that the same number of unfilled jobs will appear each year, that those positions are advertised simultaneously, or that they are accessible to a reader with a particular background.

In ordinary job-search language, an opening usually means a position currently being recruited for. Statistical vacancy measures can impose a precise definition and reference date; a posting search instead records advertisements found within its coverage and collection window. Neither should be relabeled as projected annual openings. For a U.S. worker, compare a national projection with the matching state product as forecasts, then use present-tense evidence separately to investigate recruitment. Outside the United States, begin with the relevant national statistical agency’s definitions and geographic products; similar labels across countries may not denote identical measures.

Sources: Relationship with state-level projections; Occupational Separations Methodology Frequently Asked Questions

Does a regional forecast deserve more trust?

No general rule says the more local occupational projection is more accurate. The Cleveland Fed’s “Estimating Occupational Employment: A Comparison of National, State-Produced, and Trend-Based Employment Projections” tested the assumption against realized employment growth from 2014 through 2024. Its comparison covered 647 occupations across 36 states and evaluated national projections, projections produced by states, and a trend-based extrapolation. The report says national and state-produced forecasts had similar average absolute errors; both performed better than the trend-based estimates, but errors for all approaches were substantial.

That result is a reason to resist choosing a forecast by geographic detail alone. State-produced estimates did not show a broad average accuracy advantage in this historical comparison, while the national estimate was not error-free simply because it came from a common national program. The report also notes that accuracy varied across states. An average comparison across many occupations and states cannot tell a worker which estimate will be closer for a particular occupation, area, release, or future period.

The test concerns forecast performance against realized employment change over the stated historical period. It does not test how useful a projection is for every decision, establish why a particular forecast missed, or estimate current recruiting. Nor does a historical backtest certify how the next projection cycle will perform. Forecast calibration and local relevance are separate qualities: a locally produced series may be more informative about the economy a worker can actually access, even where the study provides no general evidence that it will be more accurate on average.

When the national and state figures diverge, treat the gap as a prompt to inspect the products, not as a contest in which “local” automatically wins. Verify that both use the same occupational classification and comparable projection years. Check the release date and assumptions, and look for state-specific industry structure that could plausibly explain a difference. If the products cover different horizons or vintages, label that mismatch before interpreting their values; a newer or more detailed table is not directly comparable merely because both use the same occupation name.

The practical conclusion is conditional. If you cannot move, the state estimate may deserve more weight in planning because it describes the geography tied to your feasible employers, but that is a relevance judgment, not a claim of superior forecast accuracy. If your target market spans several states or remote work is realistic, national context may help frame the broader direction while state estimates locate variation. Either way, use the historical Cleveland Fed result to keep confidence modest: similar average error does not mean identical forecasts, and a useful local context does not remove uncertainty about future employment.

Do not average two disagreeing estimates unless their definitions, periods, and methods make a combined number meaningful; the cited comparison does not validate a blended forecast for an individual decision. Keep both estimates labeled, investigate the source of the difference, and seek evidence suited to the decision at hand. A projection can inform a longer-term question about employment direction. It cannot settle whether employers in a reachable local market are recruiting now, a distinct question that requires present-tense evidence.

Sources: Estimating Occupational Employment: A Comparison of National, State-Produced, and Trend-Based Employment Projections

What can current vacancy and job-posting evidence show?

For a present-tense check, pair a statistically defined vacancy measure with a small, documented sample of relevant advertisements. In the Bureau of Labor Statistics’ “JOLTS Data Definitions,” an opening is a position open and unfilled on the last business day of the month, with work available, a start possible within 30 days, and active external recruiting. That definition makes JOLTS useful for a consistent snapshot of vacancies, but it does not give a list of employers or identify which occupations those vacancies represent.

The scope matters especially when the worker is evaluating one occupation in one state. BLS “JOLTS State Estimates” cover total nonfarm jobs, not occupation-specific vacancies. A state-level total therefore cannot answer how many accountants, analysts, or other particular professionals are being recruited. It can provide broad context about vacancy conditions in the state; the worker still needs an occupation-level signal from another source. Official surveys have stronger statistical definitions and systematic collection, while their published geography and occupational detail may not line up with a specific career question.

Online advertisements can fill part of that detail gap. The OECD report “How well do online job postings match national sources in large English speaking countries?” compared Lightcast posting data with official vacancy and employment sources in Australia, Canada, the United Kingdom, and the United States. Its cross-country comparison found that posting coverage and its relationship to official measures vary across places, sectors, occupations, and periods. Postings can reveal the named tasks, credentials, disclosed pay, schedule, work location, and date that an aggregate vacancy rate leaves out; the benchmark does not establish that any hand-collected sample is complete or representative.

Use postings as a bounded pattern check, not a market census. Choose a feasible region and a precise role or adjacent title, then set a short observation window—for example, the next two weeks. Record each distinct employer and role, the posting date or date found, on-site, hybrid, or remote terms, stated pay, schedule, credentials, and recurring task requirements. Note likely duplicate ads separately and avoid counting a repost as independent demand. This log helps distinguish a repeated requirement from a single advertisement and makes the sample’s boundaries visible; it does not convert advertisements into an estimate of all openings.

A useful log also separates what an advertisement states from what the reader infers. “Bachelor’s preferred” is not the same as a required credential; a listed salary range is not necessarily the likely offer; and “remote” may carry geographic or schedule conditions. Preserve the wording and date rather than converting each listing into a yes-or-no signal. Group recurring requirements only after reviewing distinct employers, and note when a requirement appears in just one listing. This makes the sample useful for deciding what to verify next without implying that the pattern represents every employer in the area.

If few relevant advertisements appear, extend the window or check another source, such as employer career pages or a public workforce service, and record that the evidence remains sparse. Search terms, board coverage, duplicate ads, and employers’ posting practices can all affect what appears. No result in a short search means only that the search found none under those conditions; it does not show that no work exists. Conversely, repeated ads indicate observable recruiting activity, not a promise of continued hiring, an accessible offer, or a complete count. Read the postings alongside the survey’s broader snapshot, asking each source only the question its coverage can support.

Sources: JOLTS Data Definitions; JOLTS State Estimates; How well do online job postings match national sources in large English speaking countries?

What does local employment and pay add to the decision?

Local employment estimates describe the occupational base in an area; wage estimates provide a reference for that labor market and period. Neither is a count of active vacancies or a prediction of what a particular worker will earn. The BLS release “Occupational Employment and Wages in Sacramento-Roseville-Folsom — May 2025” estimates 13,170 accountants and auditors in the metro area and reports wage statistics from its Occupational Employment and Wage Statistics survey. The estimate is a May 2025 snapshot for a defined metropolitan geography, not a live count of recruiting employers.

As a worked illustration of what those measures can and cannot do, the Sacramento figure supplies a scale reference: accounting is an established occupation in the measured metro area. The reported mean wage supplies a second reference, describing an average across the covered workers and jobs under the survey’s definitions. A mean does not show the full wage distribution, tell where an individual offer would fall, or establish that a reader meets the requirements for a role. Nor does a metro-wide estimate tell whether the relevant positions are remote, within commuting distance, or compatible with a particular schedule. Because an average combines varied jobs and workers, it can sit above or below the pay available in a particular entry point, specialty, or employer. The published mean is therefore a comparison point to investigate, not a personal salary floor or an outcome forecast.

For a practical comparison, put the wage reference beside your own minimum workable income, then check the conditions that make a job feasible: commute or remote eligibility, hours, care responsibilities, health needs, credentials, and the time and cost of any training. These are personal constraints to test against specific opportunities, not facts inferred from the regional estimate. A local employment base can help identify where related employers and occupational pathways may exist, but it cannot establish current hiring or individual access to the mean wage. A smaller base may still include a niche employer or role; the estimate alone cannot rule one in or out.

Keep the dates and boundaries attached to both numbers. May 2025 employment and pay estimates should not be silently treated as current in a later decision, and a metro boundary may not match the area a worker can realistically reach. Use the count to understand local occupational scale and the wage statistic as context for a pay conversation; then compare actual, recent opportunities with the household floor and work arrangements the person can sustain. The decision is about feasibility across these constraints, not whether a published average can be personally claimed.

Sources: Occupational Employment and Wages in Sacramento-Roseville-Folsom — May 2025

How should AI exposure change the reading of a projection?

An AI-exposure category adds a structured signal about how occupational tasks may connect to AI capabilities and use; it does not tell you whether employment in your region will grow or how many employers are recruiting. The BLS page “Artificial Intelligence (AI) exposure categories” combines five external measures: some estimate theoretical task applicability, while others draw on observed AI use. BLS transforms these unlike inputs into relative occupational categories. Read the result as a comparative classification within that framework, not as a direct measurement of the share of your own work that a system can perform.

That construction matters when comparing a national or regional employment projection. A projection addresses a future employment scenario over a defined period. An exposure category describes a different dimension of work, using occupational mappings and inputs collected at particular times. Because the inputs are heterogeneous and some capability and usage measures are dated, the category can help identify where closer task review may be warranted, but it cannot establish that a nearby employer has adopted a tool, will redesign a role, or will reduce headcount. A high relative category cannot override evidence about local recruiting; a lower category cannot establish immunity from workflow change.

A relative ranking also changes what a reader should do with a label: it supports comparison inside the BLS framework, not a standalone threshold for choosing a career. Two occupations placed in different categories are not thereby ranked for local hiring prospects, pay, or the value of a worker’s experience. Nor does a category reveal which duties dominate a particular position. Job titles contain different task mixes across employers, so the occupational average needs to be translated back into the duties the worker actually performs before it can guide a learning choice.

BLS explicitly says its exposure categories do not forecast job growth, AI adoption, wages, productivity, job replacement, or whether AI will automate tasks or augment workers. Those exclusions are central to interpretation, not fine print. The category is neither a local vacancy series nor an individual worker’s displacement probability. It also does not say whether a task change will make a role more productive, shift its skill mix, or reduce demand. Those outcomes depend on employer choices and other conditions the classification does not measure. Keep occupational outlook and task exposure as separate entries in your decision: one informs a future demand question, the other points toward tasks worth examining.

For a worker, exposure is still useful because task changes may begin before they appear in aggregate employment figures. Use the signal as a prompt to list your recurring tasks and check which are digital, repeatable, judgment-heavy, dependent on relationships, or subject to accountability and verification. Then compare that task picture with current recruiting evidence for the feasible region. The free checker at /ai-job-risk-checker offers transparent task-level change-pressure signals and first actions; it does not produce a validated probability of displacement. Its result can help organize questions about your work, while regional hiring evidence must answer whether relevant opportunities are appearing locally.

Sources: Artificial Intelligence (AI) exposure categories

Which parts of the local role transfer from the worker's current task bundle?

Occupation labels group varied jobs, so a local opening becomes more informative when you compare its actual duties with work you can already demonstrate. O*NET’s “Accountants and Auditors” task descriptions include invoice processing, account reconciliation, checking records for accuracy, and analyzing financial information or controls. The taxonomy supplies a vocabulary for comparison; it does not report how common each task is, what any Sacramento employer asks applicants to do, or whether a local workplace uses AI.

Consider an explicitly hypothetical mapping, not a report about a real worker or employer. Imagine a posting labeled accounting or finance that mentions invoice coding, reconciliations, accuracy review, and control procedures. A worker who has handled similar responsibilities could mark the overlap, then describe the context: what information they reconciled, what discrepancies they resolved, what records they checked, and what approval or audit trail they maintained. The posting may use different wording, combine duties differently, or omit work that happens after hiring. This comparison is a question to verify, not evidence that the tasks or tools match in a particular job.

The useful distinction is between task familiarity and demonstrated transfer. Familiarity means you recognize the activity; transferable evidence shows what you did, at what level of responsibility, and how you handled exceptions. For example, a claim of reconciliation experience is more concrete when you can explain how you investigated mismatches and documented resolution. Accuracy review becomes stronger evidence when you can identify the checks you performed and how errors were escalated. These examples describe ways to present existing work; they do not imply that every accountant performs these activities or that a posting requires them.

Use each gap in the mapping to form a local verification question. If postings repeatedly mention a system, credential, reporting standard, client interaction, or approval responsibility you have not handled, ask a recruiter, hiring manager, workforce adviser, or someone doing comparable work whether it is essential on entry or can be learned after hiring. Ask which tasks occupy the role in practice, who owns final approval, and how unusual cases are handled. A posting can describe an idealized bundle, while O*NET is generalized across the occupation. Neither establishes a specific employer’s actual workflow, required skills, or AI adoption. Local conversations and the employer’s description are needed to test the fit.

Before investing in a course or making a larger move, write down a few recurring tasks from your current week, the result you are accountable for, and one example that demonstrates each relevant skill. Compare those examples with several current postings in the geography and work arrangement you could accept. Separate duties you have done, duties you could demonstrate through a work sample, and requirements you still need to verify. This task-to-posting map can show whether an upgrade builds on experience or whether an adjacent role requires a specific missing credential. It cannot by itself establish hiring demand or guarantee that an employer will credit your experience.

Sources: Accountants and Auditors

What should a worker do when national, regional, and local signals point in different directions?

Mixed signals should change the size and order of a move, not be collapsed into a personal odds estimate. The practical rule derived here is to make the most reversible move that is supported by both the work itself and the reachable market, then increase commitment only when the unresolved evidence improves. This is a decision aid, not a validated score: a projection, a posting sample, and a worker’s constraints do not share a common scale that can be added into one number.

When national and regional direction broadly align and distinct local employers repeatedly recruit for roles with recognizable task overlap, test an upgrade first. An upgrade keeps the occupation or field and adds a verified capability, credential, or responsibility that local employers actually request. Before paying for training, check whether the requirement is essential on entry, whether existing experience already covers part of it, and whether the likely pay range and work arrangement meet your minimums. If those conditions fit, a short course, supervised project, or work sample can test the gap with limited cost. Repeated recruitment is a reason to investigate fit, not proof that you will receive an offer.

When the regional projection looks favorable but current occupation-specific evidence is sparse, treat the forecast as a reason to investigate, not a reason to resign or enroll in an expensive program. The next step is to improve the evidence: widen the observation window, inspect employer career pages or public workforce services, and ask a local workforce adviser or people doing comparable work which recurring requirement matters in practice. If the possible skill has value in your current role too, a small learning experiment may be sensible while the market check continues. Sparse postings can reflect a narrow search or timing; they do not settle the outlook either way.

When pay, location, schedule, credential access, or task transfer does not fit, a favorable projection alone is weak grounds for commitment. An adjacent move may preserve more of your experience while changing the employer, specialty, or task mix; compare its entry requirements and feasible compensation with an upgrade in place. A larger change deserves a higher evidence threshold because it may require more training time, foregone income, relocation, or disruption to health and family routines. If those costs are not workable, reject that route for now even if its broad outlook appears stronger.

Use constraints to set the experiment’s ceiling. A low-cost project that can be done alongside work is more reversible than a degree, unpaid transition, or move to another region. A pathway that requires a salary below your household floor is not made feasible by a strong employment projection. Conversely, thin vacancy evidence does not mean all preparation must stop: durable learning that also improves current work may preserve options. Decide what evidence would justify the next commitment, and spend only enough to obtain it.

What is the next low-risk step?

Choose one target occupation and the geography and work arrangement you could realistically accept. Align the national and regional projections by occupation definition, period, and unit. Then collect one current signal for that same role, such as distinct recent postings or an employer’s own careers page, and verify one recurring requirement with an employer, worker in the field, or public workforce contact before spending materially. Keep a note of the date, source, exact role wording, and any location or pay condition, so a later search can show whether the picture changed. If no useful local signal is available, write down what is missing—current recruiting, entry pay, or whether a credential is essential—and ask a source positioned to answer that question. A small sample will not certify a market; it should identify the next fact that could change your choice. That keeps the decision open while turning a broad forecast into an answerable question. The free task checker can organize task-level change-pressure signals, not provide validated displacement odds. The paid career roadmap compares scenarios and constraints but guarantees neither employment nor income.

Questions readers ask