Job Market by City: Reading the Job Base Before You Move

Page 151 of the City of Rochester, Minnesota's financial report for 2024 lists Mayo Clinic at 50,385 employees, which the city works out to 40.81% of everyone employed there. The next line down is the public school district, at 2.40%. Nine years earlier the same schedule had Mayo at 28.65%. The town did not grow out of its dependence on one employer as it got bigger. It grew into it.

That page exists because an accounting standard makes it exist. Every city and county that publishes an audited financial report under Governmental Accounting Standards Board rules carries a schedule titled Principal Employers, and it is one of the few places where "who actually hires here" is written down by someone with no interest in selling you a house. It is also incomplete in ways that matter, which is why it is the first of five lookups here rather than the only one.

The question these lookups answer is not "is there a job for me." If you work remotely, you have one. The question is what the town's economy looks like on the day the arrangement ends, whether because a contract finishes or because the company decides that everyone within fifty miles of an office has to come in. That second scenario is exactly why getting your remote work state approval in writing belongs on the same sheet as the job base. One tells you whether you can stay employed and live there. The other tells you whether you can live there and find employment.

Every figure below was pulled from the source file and re-checked on 12 September 2026. The worked example is Ada County, Idaho, the county that holds Boise, chosen because its numbers show every trap at once. Each screen is one you can open for your own candidate county in an evening.

Every audited city budget has a page that names the ten largest employers

GASB Statement No. 44, issued in 2004, rewrote the statistical section that sits at the back of every Annual Comprehensive Financial Report. Among the tables it requires is a list of principal employers: up to the ten largest by headcount, shown for the current year and for nine years earlier. The document is usually a PDF of 150 to 300 pages on the finance department's website. Search for the government's name plus "ACFR" or "annual comprehensive financial report," open the table of contents, and look under the statistical section for a heading like Demographic and Economic Information.

Rochester's version, Schedule 16 of the 2024 report, is about as clean as these get: exact headcounts, ranks, percentages, two years side by side. The City of Boise's fiscal 2024 report, page 134, shows what the same table looks like when the data is harder to get.

Total Boise MSA employment on that page: 409,700. Boise publishes ranges rather than counts because, as the footnote says, the list includes "only employers that have given the Department permission to release employment range data." The source is the state labor department by way of the regional economic development partnership, not a survey the city ran.

Three things to check before you copy any of this onto a sheet.

First, the denominator. Rochester's footnote 2 says the percentage is computed against "Rochester MSA 2024 employment" from the Bureau of Labor Statistics. So 40.81% is Mayo's Rochester headcount divided by every job in the metropolitan area, not just the city. It is still an enormous number. It is just not the number the column heading implies.

Second, whether the two years use the same definitions. Boise's 2015 column lists "State of Idaho (includes BSU)" as the largest employer at 13,079 people and 4.6%. The 2024 column lists Boise State University and the Department of Health and Welfare as separate entries and has no State of Idaho line at all. The state did not shrink. The list was rebuilt on different rules, and the nine-year comparison the table invites is not one you can make from it.

Third, what the change in the top line means. Divide each of Rochester's Mayo counts by its own percentage and the schedule implies a metro of about 118,600 jobs in 2015 and about 123,500 in 2024. The metro added roughly 4,800 jobs in nine years. Mayo added 16,395. Those two figures may rest on slightly different denominators, so treat the arithmetic as a direction rather than a measurement, but the direction is unmistakable: everything in that metro that is not Mayo got smaller.

Write down three things from the schedule: the largest employer's share, the sum of the top ten, and whether the top share rose or fell over nine years. Then leave the document, because it cannot tell you what industries those employers are in, what they pay, or what happened last quarter.

The county file the BLS put out on 28 August is the census behind all of it

The Quarterly Census of Employment and Wages is not a survey. It is built from the unemployment insurance tax reports that every covered employer files with its state, which is why the BLS is comfortable calling it a census: in March 2026 it counted 154,771,941 jobs nationally, and the first-quarter 2026 news release, dated 28 August 2026, puts the national average weekly wage at $1,654. The release tables themselves only cover the 376 counties with 75,000 or more jobs. The full data behind them covers every county, and that is what you want.

Two ways in. The QCEW Data Viewer lets you pick a county, a quarter, an ownership (private, federal, state, local) and an industry level, and returns establishments, employment by month, total wages, the average weekly wage, the location quotient and the over-the-year change. Or skip the interface: the open data files sit at a predictable address, https://data.bls.gov/cew/data/api/2026/1/area/16001.csv for Ada County in the first quarter of 2026, where 16001 is the county's five-digit FIPS code. Swap in your county and the year and quarter you want. The file for Ada County in that quarter holds 2,097 data rows under the header.

Second quarter 2026 is scheduled for 2 December 2026. Annual averages for 2025 came out with the fourth quarter of 2025 and were revised with this release: Ada County averaged 295,234 jobs across 2025 with an average annual pay of $73,786.

Here is what the March 2026 file says about Ada County's private sector, sorted by size. The all-ownership total for the county is 295,338 jobs at 25,027 establishments, with an average weekly wage of $1,503.

Government is its own set of rows, and in a state capital it is not small: 15,725 state jobs, 17,799 local government jobs, 5,970 federal. Local education alone is 10,688 of those. A town whose second-largest employer block is the public sector behaves differently in a downturn from one where it is retail, and the ownership split is the only place that shows up.

Four columns, four different questions. The jobs column says how big an industry is. The wage column says what it pays, and it pays to notice that the biggest sector here pays $1,361 a week while manufacturing, less than half its size, pays $2,939. The change column says which way it is moving: construction up 10.1% in a year is a boom, and professional services down 1.9% in a metro that markets itself as a tech destination is worth a second look. The location quotient says what the town is for, and it deserves its own section.

A location quotient of 9.49 says what the town is for

The BLS explains the quotient in one sentence: it compares "the concentration of an industry within a specific area to the concentration of that industry nationwide." Local industry share divided by national industry share. A value of 1 means the industry holds the same share of local jobs as it does nationally. Las Vegas, the BLS's own example, runs well above 1 in leisure and hospitality. The same page notes the quotients are published in the Data Viewer, in the CSV files and in the state and county map application.

In Ada County's sector table above, nothing looks dramatic. Construction at 1.64 is the highest. Health care is 1.06, which is to say Boise has almost exactly the national share of health care jobs even though health care is its largest private sector. That is what a large sector looks like when it is large everywhere.

Drill one level down and the picture changes. NAICS 3344, semiconductor and other electronic component manufacturing: 6,852 jobs at 32 establishments, location quotient 9.49, average weekly wage $5,085. One level further, NAICS 334413, semiconductor and related device manufacturing: 6,673 jobs at 24 establishments, quotient 18.39, weekly wage $5,178. That wage is nearly three and a half times the county's all-industry average. Boise's manufacturing sector as a whole sits at a quotient of 0.78, below the national share. Its semiconductor slice sits at more than eighteen times it. The sector number hid the town's single most important economic fact.

Now look at the rows beside it. Two of the six-digit industries under 3344 show no employment at all and a disclosure code of N, each with one establishment. The BLS suppresses any cell that would identify a single employer's figures. Across Ada County's private sector, 331 of the 787 six-digit industry rows in this file are suppressed the same way. Suppression is not missing data. It is a signal: a high quotient at the four-digit level sitting above a set of blanked six-digit rows with one or two establishments each is the shape of an industry that is really one company.

Which brings back the city's own table. Boise's financial report lists Micron in a range of 5,000 to 5,999, as a fiscal 2024 average, from a list the company agreed to appear on. The QCEW shows 6,673 jobs in Micron's six-digit industry in March 2026, across 24 establishments, from tax filings the company had no say in. The figures differ by period, by definition (a firm versus an industry code) and by source, and neither is wrong. The mistake is writing either one down without the other two facts beside it, because someone will eventually ask which is right, and "both, and here is why" is only an answer if you kept the provenance.

The neighborhood demographics tables have the same property: the geography and the source decide the number more than the place does. Write the source and the read date next to every figure on this sheet and the disagreements stop being a problem.

What the same job pays there, from a survey released on 15 May

The QCEW knows what industries pay. It does not know what occupations pay, and an occupation is what you actually carry from one employer to the next. That is the Occupational Employment and Wage Statistics program. The May 2025 estimates were released on 15 May 2026, and the metropolitan area tables list, for each metro and each of roughly 800 occupations, the number employed and the hourly and annual wage at the 10th, 25th, 50th, 75th and 90th percentiles.

For the Boise City metro, three lines pulled from the May 2025 file:

A software developer whose remote contract ended in Boise would be looking at a local median about 3% below the national one, in a market with 3,240 such jobs. Nurses are the surprise: at $97,880 against a national $97,550, Boise pays them a shade more than the country does. The all-occupations median is 3.7% under the national figure, which is the discount the whole town works for. Whether that is good news depends entirely on the housing line of your sheet, which is the point: the wage cut and the rent cut have to be on the same page to mean anything, and the seven numbers that decide whether a move pays is where they meet.

Three cautions on reading OEWS, and the first one is where most of the damage happens. Take the median column, not the mean, and never let the two share a sentence. Boise's all-occupations mean is $64,530 against a median of $49,100: a gap of a third, because a thin layer of very high earners drags the average somewhere no typical worker stands. Inside a single occupation the gap narrows and can flip: software developers run $136,130 mean against $131,590 median, but nurses run $95,990 against $97,880, a compressed top end pulling the average below the middle. Which direction it leans is itself worth noting, and you only see it by pulling both columns. Quoting a mean as though it were the going rate is the most common way this table gets misread, and on the all-occupations line it flatters a town by fifteen thousand dollars.

The estimates also pool several survey panels collected over three years, so a metro whose wages moved sharply in the last eighteen months will lag. And the employment figure is the one people skip and should not. A median wage computed on 240 jobs describes a handful of employers; one computed on 3,240 describes a market. Pull both.

Unemployment at 3.3% is the least useful number here

It is the one everyone quotes. The BLS Local Area Unemployment Statistics program put the Boise metro at 3.3% in July 2026, down from 3.5% in June, and it is series LAUMT161426000000003, which you can open directly in the BLS database. It is a model-based estimate built to be revised, it says nothing about which industries are moving, and a low rate in a town that people are leaving looks identical to a low rate in a town that is hiring. Note it for direction and move on.

Three datasets do the job better, and each answers one thing.

The QCEW over-the-year columns already in your county file answer which industries are moving. Construction up 10.1% and professional services down 1.9% in the same county in the same year is a story about what kind of growth Boise is having, and it is a story the unemployment rate cannot tell.

The Census Bureau's Quarterly Workforce Indicators answer whether the movement is hiring or attrition. Built from the same state unemployment insurance records as the QCEW but linked at the worker level, QWI reports hires, separations and turnover by industry and by worker age for every county. The current release vintage is R2026Q2, issued in the April-to-June window of 2026 — a release label, not the quarter the data reaches, which differs by state. The program's data notices, stamped 16 June 2026, carry the caveats to read first: Alaska's series stops at the second quarter of 2016 and Michigan's at the fourth quarter of 2021, and the tabulations by firm age and firm size run well behind the rest. If your candidate is in either state, this tool is not available to you in any useful form. The companion OnTheMap application answers a narrower question, where the people who live in this county actually work, which is how you find out that a pleasant county is a bedroom community for a metro forty miles away.

County Business Patterns from the Census Bureau answers how many establishments there are in each size class, which is the check on single-employer dependence that the QCEW's suppressed cells can only hint at. It is annual and slow: as of 12 September 2026 the Census Bureau still lists 2023 as the latest CBP reference year, more than two years behind the QCEW quarter you just pulled. Use it for structure, not for the latest year.

Nine lines per candidate county

Each line is one lookup, and each lookup has one source. Fill them in this order, because the fourth line decides whether the remaining five are worth your evening.

Line 4 is the gate. If the sectors that pay well are shrinking and the ones growing are the ones that pay $508 a week, the rest of the sheet only tells you how bad the fallback is. If they are growing, lines 5 through 9 tell you whether the growth depends on one company.

What Ada County's sheet actually says

Read together, the nine lines describe a town that is not what either its boosters or its skeptics say. The largest employer on the city's own list holds 1.7% of metro employment, and the top five together hold about 7.4%. By the Rochester measure, this is one of the least employer-dependent metros you could pick. Nobody in Boise is one hospital board decision away from a regional recession.

But the wage structure is a different matter. The county's all-industry weekly wage is $1,503. The one industry that pays $5,178 is 24 establishments with a quotient of 18.39, and the blanked cells beside it say most of those jobs sit inside a single firm. The high end of Boise's pay scale is a specialization, and a specialization in one industry is one cycle away from being a problem regardless of how many establishments the county has in total. Construction at +10.1% is the growth story of the year, and construction is the sector that stops first when rates or migration turn.

For a remote software developer, the local fallback is 3,240 jobs at a median about 3% under national, in a professional services sector that shed 1.9% of its jobs in the year to March 2026. That is a real fallback. It is not a deep one, and it is not a growing one at the moment.

None of that is a verdict on the town. It is what the public record says as of 12 September 2026, and the same record will say something different on 2 December when the second quarter comes out. Open the Data Viewer, pick your candidate county and the first quarter of 2026, set the level to four-digit private industries, and sort by location quotient. The row at the top is the town's real business. Everything else on the sheet is context for that one row.