Cities Don't Stand Still

Detroit, thirty years early

Between 1946 and 1956, at the height of Detroit's auto industry, the Big Three automakers — General Motors, Ford and Chrysler — opened twenty-five new plants in metropolitan Detroit. Not one of them was inside the city.¹ Auto manufacturing employed just over 220,000 people inside the city limits in 1950; by 1970 that number had fallen by more than half. The plants had not vanished. They had moved to the suburbs.

The decline everyone associates with the 1970s and 1980s had been underway since the early 1950s — before foreign competition, before the oil shocks, before any of the usual explanations. And it was measurable the entire time.

That is the useful thing about a city's economic base: it changes slowly and legibly, and the change shows up in employment data long before it shows up in a skyline. Detroit is the famous case, but every metro is going through changes of its own. Austin was a government town thirty years ago. Charlotte made textiles before it made loans.

Pittsburgh shows how long the sequence can run: river trade, then coal and iron, then steel, then commerce and services, and today medicine and research. Five bases in two centuries, each built while the previous one still looked permanent. Cities work through a sequence of bases — and you can see the sequence in the data well ahead of the headlines.

Each of those transitions was visible in the data first. This paper explains how to read that data, and how to use MetroLQ to do it for any of roughly 430 U.S. metropolitan areas.

What a city sells to the outside world

Some of what a metro produces is sold beyond its borders, and some is consumed at home.

A distribution center that ships across four states, a bank headquarters, a university, a chip plant — all of them collect revenue from customers who don't live there. Economists call this the export base, or the basic sector. Everything else — the grocery stores, the dentists, the teachers — recirculates dollars already in the local economy.

This paper uses economic base and export base to mean the same thing. Strictly, the economic base also counts outside income that arrives without a job attached, such as retirees' investment returns, but the export sector is most of it, and it is the part employment data can measure.

Only export activity can make a metro bigger. A new supermarket does not cause anyone to eat more food; it redistributes local spending. A new headquarters brings in payroll from outside, and that payroll gets spent locally, supporting jobs that would not otherwise exist.

This is the employment multiplier, and it is larger than most people expect. The standard estimates put it at roughly two to four total jobs for every export job, and two and a half to nine residents for every export job.²

Follow one job through the chain. A semiconductor plant outside Austin hires an engineer, paid by customers in California and Taiwan. Her paycheck becomes a mortgage payment, a dentist's bill, groceries brought by a delivery driver, and property taxes that help pay a teacher. The dentist, the driver and the teacher have households of their own, and those households need somewhere to live. None of those jobs exists because anyone decided Austin needed them. They exist because a plant sells chips to people who don't live there.

A change in the export base runs through the multiplier and arrives as households. Households are what occupy buildings.

That chain — export base to jobs to households to rooftops — is why the base sits upstream of every demand assumption in a pro forma. It runs in reverse just as quickly. When General Motors ended SUV production at its Janesville, Wisconsin plant two days before Christmas 2008, it took the 1,200 jobs that remained there — down from 7,000 at the plant's 1970 peak — and about 3,000 more at area suppliers. Within a year the cuts had reached the day cares, restaurants and bars that served the plant.³

Headline job growth tells you a market's size is changing. It does not tell you what the market is built on.

The location quotient

Employment data divides the economy into ten major industry groups. Before comparing any metro to the country, it helps to see what the country looks like.

United States employment by industry, July 2026⁴

A location quotient asks a simple question: does this metro do more of a given activity than the country on average?

Take the metro's share of its employment in an industry and divide it by the nation's share of employment in that same industry. Leisure and hospitality accounts for about 302,000 of the 1.17 million jobs in Las Vegas — 25.8% of jobs. Nationally, as the table shows, leisure and hospitality is 11.1% of employment. Divide 25.8 by 11.1 and you get a location quotient of 2.3. MetroLQ computes it on unrounded figures and shows 2.31.

An LQ of 1.0 means the metro looks like the country. Above 1.0 means it does more of that thing than the country does, and the surplus is presumed to be serving customers outside the region — which makes it part of the export base.

Employment is the measure rather than output for two practical reasons: metro-level GDP by industry is not reliably published, and employment is what turns into households anyway. A firm that hires 400 people creates 400 paychecks, and paychecks become payments for housing, food, and discretionary spending.

What makes location quotients useful is charting them — across all of a metro's industries at once, against other metros, and over time. That is what MetroLQ is for, and the sections that follow show all three.

Reading one metro, and comparing two

Las Vegas and Nashville each support roughly 1.2 million jobs. On the metrics that appear at the top of a broker's deck — job growth, population growth, in-migration — they are close to twins.

They are not the same market.

As of July 2026, Las Vegas supports an export base of roughly 200,000 jobs, about 17% of total employment, and 86% of that base sits in a single sector: leisure and hospitality. The export base counts only each sector's jobs above the national share, which is why it is smaller than leisure employment itself. Nashville's export base is roughly 82,000 jobs, or 7% of total employment, spread across six sectors — professional services, leisure, information, financial activities, trade, and other services. ⁵

The sectors themselves also behave differently. In 2020, Las Vegas employment fell 26.3% from its pre-pandemic level at the trough. Washington, D.C., which is also a concentrated metro, fell 11.6%. The difference is what each is concentrated in: Washington’s largest position is government — roughly 670,000 jobs, a fifth of the metro — and government employment has historically moved less with the national cycle than any other major sector. Two markets can be equally concentrated and still behave nothing alike.

This is what the comparison view is for. It does not say which market is better — that depends on what an investor is trying to do, how long they intend to hold, and what they already own. It shows what each market is built on, how much of it there is, and which direction it has been moving.

Austin: a base that changed

The third thing MetroLQ does — and the most interesting — is show a metro over time. Austin is one of the clearest examples in the country.⁶

Between 1995 and 2026, Austin's employment grew from about 525,000 to about 1.42 million. Every single one of its ten major sectors grew. Every single one of them grew faster than its national counterpart. And three of their location quotients fell anyway.

Government is the case worth sitting with. Austin's government sector added 85,500 jobs, from 130,500 to 216,000, and grew faster than the national government sector. Nothing went wrong by any normal measure. Its location quotient fell from 1.51 to 1.02 regardless. Education and health added 263% more jobs and its LQ fell too.

A metro's own growth rate is the bar every sector has to clear.

Austin outgrew the country by a wide margin, and that margin became the bar: a sector holds its location quotient only if it outgrows its national counterpart by as much as the metro outgrows the nation. Government beat its counterpart — just not by that much. A falling location quotient does not mean a sector is shrinking; it can simply mean the metro found something else to grow into.

Which is what happened. While government's share was falling, information grew 254% during a period when the national information sector shrank by 2%. Austin added roughly 35,000 information jobs while the country was losing them. That is what specialization looks like.

For decades the standard economic-base classification listed Austin as a government town, and it was: the state capital and a flagship state university — both government employers in the data — with an LQ of 1.51.⁷ Today it is an information and professional-services town. But notice how the transition happened. Government did not leave. It got out-grown. That is often how base transitions work, and it is nearly invisible unless you are looking at the right measure.

See this for any metro: open MetroLQ, choose a market, and press play on the time view. Austin takes about half a minute.

Detroit: a base that wasn't replaced

Between 1995 and 2026, Detroit's metropolitan employment went from about 2.06 million to about 2.05 million. It did not grow, while the national economy added 35%. Every one of Detroit's ten sectors trailed its national counterpart.

And yet Detroit's manufacturing location quotient rose, from 1.23 to 1.45, while manufacturing employment fell from about 372,000 to about 235,000. Detroit lost more than a third of its manufacturing jobs and became more concentrated in manufacturing, relative to the country, at the same time. The reason is the same arithmetic in reverse: manufacturing was disappearing from the rest of the country even faster. Detroit's other industries did not shrink — non-manufacturing employment actually grew — so the change came from the national side of the ratio. Nationally, manufacturing's share of all jobs fell from 14.7% to 7.9%. Detroit's fell from 18.1% to 11.5% — a smaller drop — so its location quotient went up.

A rising LQ is not automatically good news. In Austin, information's rising LQ meant a new industry was being built. In Detroit, manufacturing's rising LQ meant the region was holding onto a shrinking industry.

Detroit's problem was never that it was good at building cars — it still is. The problem was that the base began dispersing in the 1950s and nothing was built to replace it. The plants moved to the suburbs, then to other states, then abroad. Each move was individually rational and locally small. Together they removed the thing the entire regional economy depended on, and inside the city the multiplier ran in reverse for decades.

The recent picture is more mixed than the reputation suggests: education and health employment in the metro has grown 54% since 1995, and the region has stopped shrinking. But nothing has yet grown fast enough to become the next base.

Austin and Detroit are the same arithmetic run in opposite directions. Austin set a very high bar, and three sectors that grew enormously still fell short. Detroit set a very low bar, and manufacturing cleared it only because manufacturing was shrinking faster everywhere else. Neither story is legible from a single year's numbers.

Using MetroLQ

MetroLQ applies this analysis to roughly 430 U.S. metropolitan areas and divisions using employment data from the Bureau of Labor Statistics. It does three things.

See one metro. Every major sector plotted at once, with its location quotient, its employment, and its growth. This is the profile view: what is this market actually in the business of, and how much of it is there.

Compare metros. Put two or three markets side by side and the differences that headline statistics hide become obvious — the size of each export base, how concentrated it is, and in what.

Watch it change. Chart the same metro over time and you see trend rather than position. This is where base transitions show up, and it is the kind of view that would have shown you Detroit in 1960 and Austin in 1998.

A five-minute routine for a market you are looking at: find the highest location quotients and note what the market is built on. Check whether each is rising or falling, and ask which side of the arithmetic moved — the sector, or the metro around it. Then ask the only question the data cannot answer for you: what do you believe about those industries, and about the companies inside them?

What it doesn't tell you

Four limits worth carrying. Employment counts activity, not value; two industries with the same headcount can contribute very differently. Some high-LQ sectors serve local residents rather than exporting. Remote work has scrambled recent data for knowledge industries, because where someone is employed and where they work are no longer the same place. And metropolitan boundaries are periodically redrawn, which can put a step in a long series that is not real economic change.

The larger limit is the honest one. Economic base analysis tells you how growth is transmitted through a region: it arrives through the export base and multiplies outward. It is also silent about growth that comes from replacing imports with local production — a real, if slower, second engine. It does not tell you what will happen to those industries. A location quotient is a statement about exposure, not a forecast of growth or decline.

Where this leaves you

A metro's economic base is the answer to a question most market analysis never asks: what does this place sell to the rest of the world, and who is buying? Everything downstream — jobs, households, absorption, rent — follows from that answer.

The base is not permanent. Cities work through a sequence of them, usually slowly and usually without announcing it. Austin's transition happened while every sector was growing. Detroit's happened while the plants were still open. Neither was visible in a headline, but both were visible in the data.

The job is not to find good markets. It is to understand the industries driving each market, and to decide whether you want to invest in its future. Go look at yours.

Works cited

Data

U.S. Bureau of Labor Statistics. State and Metro Area Employment, Hours, and Earnings, and Current Employment Statistics. All employees, not seasonally adjusted. Accessed August 2026. https://www.bls.gov/sae/

All location quotients, employment levels and growth rates in this paper are calculated by Clear Bay Capital from that source. Austin and Detroit figures span June 1995 to May 2026; Las Vegas, Nashville, Washington and the national table reflect July 2026, the most recent month as this paper was finalized. Metro and national figures are matched to the same calendar month so seasonal patterns do not distort the comparison. MetroLQ updates as new data is released, so figures in the tool will move past those cited here.

Detroit

Ozimek, Adam. “Myths and Lessons from a Century of American Automaking.” Economic Innovation Group, August 1, 2025. https://eig.org/myths-and-lessons-from-american-automaking/

Economic base theory and the multiplier

Geltner, David, Norman G. Miller, Jim Clayton, and Piet Eichholtz. Commercial Real Estate Analysis and Investments. 3rd ed. OnCourse Learning, 2014.

Ling, David C., and Wayne R. Archer. Real Estate Principles: A Value Approach. 4th ed. McGraw-Hill, 2013.

Walsh, Joseph. “Economic Base Analysis: Location Quotients.” University of Wisconsin–Madison, Wisconsin School of Business.

Janesville

Goldstein, Amy. Janesville: An American Story. Simon & Schuster, 2017.