I. The Problem of the Unit
North American settlement is routinely described at two scales and governed at a third. Analysts speak of metropolitan areas — the commuting-defined statistical units that agencies in all three countries produce — and of nations, provinces, and states. But a great deal of what actually happens economically happens at a scale between the two: a scale at which Baltimore and Philadelphia are not separate labor markets so much as neighboring rooms in one large house, at which a stamping plant in Guanajuato and an assembly line in Ontario share a single order book, at which a housing shortage in San Francisco is resolved by construction in Stockton.
This intermediate scale has no census category, no legislature, no tax base, and no ballot. It is nevertheless the scale at which freight moves, at which specialized labor circulates, at which water and power are allocated, and at which the largest infrastructure decisions either succeed or fail. The term of art for it is the megaregion.
The concept is old enough to have a lineage and young enough to remain contested. Jean Gottmann’s Megalopolis (1961) described the urbanized Northeastern seaboard as a new kind of settlement — not a big city and not a set of cities, but a continuous field of interpenetrating metropolitan influence. The idea lay relatively dormant in American planning until the early 2000s, when the Regional Plan Association’s America 2050 initiative revived it, naming eleven U.S. megaregions and using the frame to argue for intercity rail and coordinated infrastructure investment. Richard Florida and collaborators produced parallel work using nighttime-light and economic-output measures. Later analyses — most notably Garrett Nelson and Alasdair Rae’s 2016 use of census commuting flows to partition the country algorithmically — tested the delineations against actual behavior and found the boundaries both more numerous and less tidy than the advocacy maps suggested.
That tension is the honest starting point for this paper. The megaregion is simultaneously (a) a real feature of continental economic geography and (b) a category whose boundaries are analyst-dependent, whose most popular delineation was produced by an organization advocating for a particular investment program, and whose components differ enough from one another that a single label may obscure more than it reveals.
This paper does three things. It surveys the span of North American megaregions across the United States, Canada, and Mexico. It identifies the binding mechanisms that allow such regions to cohere and grow. And it proposes a typology — not a list of eleven or fifteen names, but a classification by binding agent and by morphology, on the argument that the megaregions of North America are not one kind of thing described at different sites but several distinct kinds of thing sharing a scale.
II. The Span
A. Orders of Magnitude
Continental population is roughly 500 million across the three USMCA states. Something on the order of three-quarters of that population lives within regions plausibly classified as megaregional, on a small fraction of the land area. This concentration is not a projection; it is the current condition. Growth over the coming decades is expected to be overwhelmingly internal to these regions or at their fringes, which is why the delineation question is not merely taxonomic.
Figures below are approximate and are given to establish orders of magnitude rather than to fix boundaries. Boundary-drawing is precisely the operation on which analysts disagree.
B. United States
The Northeast (BosWash). Roughly 450 to 500 miles from southern Maine to the Richmond area; on the order of 50 million people. The oldest and densest, and the only one where continuous passenger rail service actually functions as a megaregional circulatory system.
The Great Lakes. Chicago to Pittsburgh, with arms to Minneapolis, St. Louis, and — if one admits the international dimension — Toronto and the Windsor–Quebec corridor. Fifty to sixty million depending on where the Canadian line is drawn. The largest by land area and the most internally uneven, containing both the continent’s densest interior labor market and its most depopulated industrial towns.
The Texas Triangle. Dallas–Fort Worth, Houston, San Antonio, Austin: roughly 1,300 to 1,400 miles of triangle perimeter enclosing something over 21 million people. The fastest-growing large megaregion in absolute terms.
Southern California. Los Angeles, Orange County, the Inland Empire, San Diego, extending across the border to Tijuana. Twenty-four million and up, depending on binational treatment.
Northern California. Bay Area, Sacramento, the northern San Joaquin Valley — a housing-cost-driven expansion of a formerly compact metropolitan region into a two- and three-county commuter field.
Cascadia. Eugene to Vancouver, British Columbia — about 450 miles, ten to twelve million, of which roughly a quarter is Canadian.
Florida. The peninsula’s twin coasts plus the I-4 corridor; over 20 million.
Piedmont Atlantic. Birmingham–Atlanta–Charlotte–Raleigh, with a Nashville question mark; roughly 20 million.
Gulf Coast. Houston through the Louisiana petrochemical parishes to Mobile and beyond; roughly 15 million.
Arizona Sun Corridor. Phoenix to Tucson, with Nogales as a border terminus; roughly seven million.
Front Range. Cheyenne to Pueblo along I-25; roughly five million.
C. Canada
The Quebec City–Windsor Corridor. About 1,150 kilometers containing on the order of 18 to 19 million people — more than half of Canada’s population in a single linear band. Within it, the Greater Golden Horseshoe alone approaches 10 million and functions as a distinct sub-megaregion with its own provincial growth plan, which makes Canada one of the few jurisdictions on the continent to have given a megaregional unit an actual statutory planning instrument.
The Calgary–Edmonton Corridor. Roughly 300 kilometers, three and a half million or so; small by continental standards but holding the great majority of Alberta’s population and functioning as a coherent labor and services corridor.
The Lower Mainland, generally treated as the northern anchor of Cascadia rather than as a separate unit.
D. Mexico
Mexico’s megaregional geography is both the most demographically concentrated and, in one respect, the most institutionally advanced on the continent.
Central Mexico (la Megalópolis). Mexico City together with Toluca, Puebla, Cuernavaca, Pachuca, Tlaxcala, and Querétaro: on the order of 35 million people, roughly a quarter of Mexico’s population, within a radius that would fit comfortably inside the Texas Triangle. It is the only North American megaregion with a formally constituted multi-jurisdiction body named for the scale itself — the Comisión Ambiental de la Megalópolis, coordinating six federal entities on air quality and related matters.
The Bajío. Querétaro, Guanajuato (León, Irapuato, Celaya, Silao), Aguascalientes, and San Luis Potosí — the automotive and aerospace manufacturing belt, bound northward to Texas and eastward to the Gulf ports.
Monterrey–Saltillo–Nuevo Laredo. Northeastern Mexico’s industrial and logistical core, functionally continuous with the Texas Triangle through the Laredo crossing.
Guadalajara and the Occidente. A large primate metropolitan region with electronics manufacturing depth and a thinner corridor structure than the Bajío.
The Northern Border Strip. Tijuana, Mexicali, Ciudad Juárez, and the Reynosa–Matamoros complex: not a single megaregion but a set of paired seams, each bound more tightly to its U.S. counterpart than to the others.
The Yucatán/Caribbean corridor. Mérida–Cancún–Playa del Carmen: an emergent case, tourism- and migration-driven, with new rail infrastructure altering its internal geography.
E. The Binational Seams
Four crossings are dense enough to require treatment as units rather than as edges: San Diego–Tijuana, Paso del Norte (El Paso–Ciudad Juárez–Las Cruces–Sunland Park), Laredo–Nuevo Laredo, and Detroit–Windsor. These are the sites where two national megaregional systems touch, and they are the sites where the megaregion concept is put under the most stress, because their cohesion is directly and immediately a function of policy — of visa regimes, tariff schedules, inspection staffing, and bridge capacity — rather than of anything resembling natural growth.
III. What Makes Them Cohere
Megaregions are not merely large. They are integrated in specific ways, and the mechanisms of integration differ enough from one case to another that they can be used as classification criteria. Nine binding agents recur.
1. Overlapping labor sheds. The classical mechanism. Metropolitan areas are defined by commuting; megaregions form where the commuting fields of adjacent metropolitan areas overlap sufficiently that no clean partition exists. This is what makes Baltimore–Washington–Philadelphia one thing. It is also the mechanism whose empirical strength is most often overstated: commuting-flow analysis shows that many named megaregions are not, in fact, joined at the daily-commute level. Their integration runs through other channels.
2. Freight corridors and inland ports. I-95, I-35, I-5, I-75, and Mexican Federal Highways 57 and 85 are not simply roads between places; they are the spines along which warehousing, distribution, and light assembly locate. Intermodal facilities — Alliance in Fort Worth, Rickenbacker in Columbus, CentrePort in Winnipeg, Interpuerto in San Luis Potosí — create employment lattices that follow the corridor rather than the metropolis. The Gulf Coast and Great Lakes megaregions are more legible as freight geographies than as commuter geographies.
3. Multi-airport systems and short-haul air. A megaregion typically operates as a single air-service market with several access points. The Northeast is served by a dozen airports treated by travelers as substitutes; Southern California likewise. Where high-frequency short-haul air links exist — Dallas–Houston, Toronto–Montreal, Mexico City–Monterrey — they function as the business-travel equivalent of a commuter rail line, binding cities that share no daily labor market.
4. Grid, water, and pipeline systems. ERCOT is very nearly coterminous with the Texas Triangle plus its resource hinterland. Colorado River allocation binds Southern California, the Sun Corridor, and the Front Range into a single hydrological negotiation whether or not they share anything else. Mexico City’s Cutzamala system draws its water across state lines, making the Megalópolis a hydraulic unit before it is an administrative one. Petrochemical pipelines make the Gulf Coast a single plant distributed over 600 miles.
5. Research and training complexes. Firms cluster where firm-specific skills are reproduced. The Piedmont Atlantic’s Research Triangle, the Boston–New Haven biomedical band, Ontario’s engineering schools feeding the automotive corridor, Monterrey’s Tec system feeding the northeastern industrial belt: universities are the slowest-moving and most durable megaregional binding agent, because they anchor labor supply for decades.
6. Housing cost gradients. A significant portion of megaregional expansion is not chosen but forced. When housing supply is constrained at the core, the commuting field extends outward until it reaches affordability — producing the Central Valley extension of the Bay Area, the Inland Empire’s growth relative to coastal Los Angeles, the Poconos-to-New-York supercommute, and the exurban ring of the Golden Horseshoe. This mechanism generates megaregional form as a symptom of dysfunction rather than as a sign of health, and it is worth distinguishing on those grounds.
7. Production sharing under a common trade regime. The USMCA’s regional value content and labor value content rules for automobiles create a legal incentive for components to cross borders repeatedly before final assembly. The result is a manufacturing region — Ontario, Michigan, Ohio, Indiana, Tennessee, Coahuila, Nuevo León, Guanajuato, Querétaro — that is genuinely integrated at the level of the firm and the order book while being wholly unintegrated at the level of labor mobility. This is a distinct kind of cohesion and deserves its own typological slot.
8. Migration and kinship networks. Chain migration produces durable channels between specific sending and receiving regions: particular Mexican states to particular U.S. metropolitan areas, Caribbean and South Asian networks into the Golden Horseshoe and the Northeast, domestic retirement migration from the Northeast and Great Lakes into Florida and the Sun Corridor. These networks route capital, remittances, and information as well as people, and they bind regions that are not geographically adjacent.
9. Distributed work. The post-2020 normalization of hybrid arrangements has changed the commuting field into something with a longer radius and a lower frequency. A worker who appears at an office twice a month can live 200 miles away. This does not create megaregions where none existed, but it thickens the outer bands of existing ones and makes the commuter-shed method of delineation less reliable than it was.
Growth
Megaregional growth in North America draws on four distinct sources, and they do not operate uniformly. Natural increase remains the dominant driver in Central Mexico and the Bajío. International migration is the dominant driver in the Canadian corridors and the U.S. gateway megaregions. Domestic migration drives the Sun Belt cases — Texas, Florida, Piedmont Atlantic, Sun Corridor — largely at the expense of the Northeast, the Great Lakes, and coastal California. Reclassification and infill — the absorption of previously separate places into the field — adds population to megaregional totals without adding any to national ones, and accounts for a good deal of apparent megaregional growth in mature cases.
IV. A Typology
The eleven-or-so named regions are usually presented as a single list, which implies they are a single kind. They are not. The following typology classifies on two axes: the primary binding agent and the morphology. A third dimension, degree of consolidation, is applied as a qualifier.
Axis One: Binding Agent
Type A — Commuter-Field Megaregions. Integration runs through overlapping daily labor sheds. Adjacent metropolitan areas share workers directly; boundaries between them are administrative rather than functional. Cases: the Northeast; Southern California; Northern California; the Golden Horseshoe within the Windsor–Quebec corridor; the core of Central Mexico. These are the only cases that satisfy the strict Gottmann criterion, and they are the minority.
Type B — Transactional Corridor Megaregions. Integration runs through business travel, headquarters relationships, professional service networks, and firm-level interdependence, with little daily commuting between the major nodes. Nobody commutes from Houston to Dallas, but the two cities share law firms, energy finance, and executive labor. Cases: the Texas Triangle; Piedmont Atlantic; Windsor–Quebec at full extent; Calgary–Edmonton; Cascadia. This type is what most analysts actually mean when they defend the megaregion concept against the commuting-data objection, and it deserves to be named rather than smuggled in.
Type C — Production-Sharing Megaregions. Integration runs through supply chains operating under a shared trade regime. The unit is a manufacturing system, not a settlement system, and its geography is set by tariff rules, logistics costs, and plant siting rather than by contiguity. Cases: the Great Lakes–Bajío automotive complex; the Monterrey–Texas industrial axis; the border maquiladora pairings. Type C is the only kind whose cohesion can be substantially reduced by a single policy decision — which makes it the most fragile and the most politically consequential.
Type D — Extractive-Logistical Megaregions. Integration runs through resource extraction, processing, pipeline networks, and port capacity. Settlement follows the plant and the terminal. Population may be modest relative to economic weight, and the internal structure is a chain of specialized single-industry nodes rather than a hierarchy of general-purpose cities. Cases: the Gulf Coast; the Alberta corridor in part; the Permian-adjacent fringe of the Texas Triangle.
Type E — Amenity-Migration Megaregions. Integration runs through in-migration driven by climate, cost, retirement, and lifestyle, with a settlement structure that is unusually low-density and unusually recent. Employment is disproportionately in construction, health care, hospitality, and the services that in-migration itself generates. Cases: Florida; the Arizona Sun Corridor; the Front Range in part; the Yucatán–Caribbean corridor. Type E regions are the most demographically volatile, since the flows that build them can reverse on cost, insurance availability, or water.
Axis Two: Morphology
M1 — Coalescent Polycentric. Multiple historically independent cities of comparable weight that have grown into contact. No single dominant center. Northeast, Great Lakes.
M2 — Primate-Radial. One overwhelming center with absorbed satellites arranged around it. Growth is mostly a matter of the core’s expanding shadow. Central Mexico, Golden Horseshoe, Guadalajara, Southern California in a qualified sense.
M3 — Linear Corridor. A single axis, usually following a river valley, a coast, or a mountain-front, with limited lateral development. Windsor–Quebec, Cascadia, Front Range, Calgary–Edmonton, Sun Corridor.
M4 — Constellational. Three or more centers arranged with substantial empty space between them, joined by high-volume intercity movement rather than continuous settlement. Texas Triangle, Piedmont Atlantic, Bajío.
M5 — Seam. A binational pairing whose internal structure is organized around a controlled crossing. Density peaks at the border rather than declining toward it — an inversion of ordinary metropolitan form. San Diego–Tijuana, Paso del Norte, Laredo–Nuevo Laredo, Detroit–Windsor.
The Cross-Classification
| Region | Binding Agent | Morphology | Consolidation |
|---|---|---|---|
| Northeast | A | M1 | Consolidated |
| Great Lakes | A / C | M1 | Consolidated, unevenly |
| Southern California | A | M2 | Consolidated |
| Northern California | A | M2/M3 | Consolidating |
| Central Mexico | A | M2 | Consolidated |
| Windsor–Quebec | B (A in the Horseshoe) | M3 | Consolidated |
| Texas Triangle | B | M4 | Consolidating rapidly |
| Piedmont Atlantic | B | M4 | Consolidating |
| Cascadia | B | M3 | Consolidating |
| Calgary–Edmonton | B | M3 | Emergent |
| Bajío | C | M4 | Consolidating |
| Monterrey–Northeast Mexico | C | M4 | Consolidating |
| Gulf Coast | D | M3/M4 | Consolidated, low-density |
| Florida | E | M4 | Consolidated |
| Arizona Sun Corridor | E | M3 | Consolidating |
| Front Range | E | M3 | Emergent |
| Yucatán–Caribbean | E | M3 | Emergent |
| Border seams (four) | C | M5 | Policy-contingent |
Two payoffs follow from this arrangement.
First, it explains why megaregional policy proposals succeed in some places and fail in others. High-speed rail is a Type A and Type B instrument — it addresses either a thick commuter field or a dense business-travel market. Proposing it for a Type D or Type E region misreads what binds the region together. Conversely, port capacity, pipeline permitting, and customs throughput are Type C and Type D instruments, and are largely irrelevant to a Type E region’s actual constraints, which are water, insurance, and construction labor.
Second, it identifies which regions are fragile and in what way. Type C cohesion is policy-contingent and can be reduced quickly. Type E cohesion is resource-contingent and can be reduced slowly but decisively. Type A cohesion is the most durable, which is why the oldest megaregions are Type A and why the newer ones aspire to become Type A without necessarily being able to.
V. The Governance Gap
Every megaregion in North America is a jurisdictional orphan. It is too large for any metropolitan planning organization, too small and too oddly shaped for any state or province, and in four important cases it crosses an international border. There is no megaregional electorate, no megaregional revenue, and no megaregional executive.
The instruments that exist are partial and were built for other purposes:
- Interstate compacts (the Port Authority of New York and New Jersey, the Delaware River Basin Commission, the Great Lakes–St. Lawrence River Basin Compact, WMATA) — powerful within narrow subject matter, and requiring congressional consent.
- Metropolitan planning organization federations and voluntary coalitions — coordination without authority.
- Binational bodies — the International Boundary and Water Commission, the International Joint Commission, the North American Development Bank, and regional groupings such as the Pacific Northwest Economic Region: real, but subject-limited and diplomatically constrained.
- Provincial growth plans — Ontario’s Greater Golden Horseshoe plan is arguably the strongest single megaregional planning instrument on the continent, and it exists only because the megaregion in question happens to sit inside one province.
- Federal-entity coordination in Mexico — the Comisión Ambiental de la Megalópolis is the only body that names the scale outright, and its powers are confined to environmental matters.
The pattern is that megaregional governance succeeds where the region happens to fall inside one general-purpose jurisdiction, and otherwise operates only through single-purpose bodies created around an unavoidable shared physical asset: a river, an airport complex, a harbor, an airshed. Where no such asset forces the issue, coordination does not happen.
This is worth stating precisely because it is the inverse of the more familiar coordination problem at tri-state junctions and other jurisdictional margins. There, the difficulty is that a place is peripheral to every authority that touches it, and none has an interest in developing it. In the megaregion, the difficulty is the opposite: the place is central to every authority that touches it, each has a strong interest in it, and the resulting competition — for headquarters relocations, for stadium and plant siting, for tax base — produces a race that none of them can unilaterally exit. Peripheral orphanage and central rivalry are two different failure modes, and they call for different institutional remedies. Compact bodies with independent revenue and narrow mandates work reasonably well against the second. They are largely irrelevant to the first.
VI. Objections
A responsible treatment should register the case against the concept.
The unit was built to sell a program. The most influential U.S. delineation emerged from an advocacy organization making a case for intercity rail investment. Boundaries drawn to make a corridor look like a system will tend to make corridors look like systems. This does not make the delineation wrong, but it does mean the burden of proof sits with the delineator.
The commuting evidence is weaker than the rhetoric. Algorithmic partitions of actual commuting flows do not reliably reproduce the named megaregions. They tend to produce more, smaller units, with boundaries sensitive to the clustering threshold chosen. On the strict Gottmann criterion, several celebrated megaregions are not megaregions at all. The typology above accommodates this by distinguishing Type B from Type A rather than by pretending the difference away — but a critic could fairly reply that Type B is a rescue operation for a category that the data did not support.
The concept may add nothing to existing statistical geography. Combined Statistical Areas already capture much of what megaregion advocates describe. If the megaregion is merely a CSA with better publicity, the analytical gain is small.
International boundaries are not merely friction. Treating San Diego–Tijuana or Detroit–Windsor as a single unit understates how completely sovereign policy governs the seam. Wage levels, labor law, currency, health systems, and legal regimes differ discontinuously at the line. A region whose two halves cannot share a labor market, a tax base, or a court system is integrated in a much thinner sense than the word ordinarily implies.
Cohesion is reversible. Border thickening after 2001, pandemic closures, and tariff volatility have each demonstrated that Type C and Type M5 cohesion can be reduced substantially within months by decisions taken in national capitals. Analysis that treats megaregional integration as a one-way developmental process mistakes a policy equilibrium for a trend.
The reasonable position is that megaregions are real as functional economic geographies and unreal as political geographies, and that most of the disagreement in the literature comes from arguments in which one side means the first and the other means the second.
VII. A Research Agenda
Five questions seem to hold the most promise, and each is empirically tractable:
- Threshold sensitivity. How much do megaregional boundaries move under alternative delineation methods — commuting, freight tonnage, air passenger volume, firm ownership networks, migration flows? A region whose boundaries are stable across all five is a different object from one that appears only under a single measure.
- Type C durability. How much did integration in the automotive production region actually fall during periods of trade disruption, and how quickly did it recover? This is the best available natural experiment on whether policy-created cohesion is durable.
- Type E resource ceilings. Water allocation in the Colorado Basin and insurance availability in Florida are the two clearest binding constraints on amenity-migration megaregions. At what point do they turn migration flows, and are there observable leading indicators?
- The distributed-work radius. Has the outer band of the commuter field extended permanently, and by how much? This determines whether Type A regions are quietly becoming larger or whether the change was transient.
- Compact effectiveness. Paired comparison of megaregional infrastructure outcomes where a multi-jurisdiction body with independent revenue exists against otherwise similar cases where coordination is voluntary. This is the question with the most direct bearing on institutional design, and it is the one the literature has done least with.
VIII. Conclusion
North America contains roughly a dozen and a half regions large enough and integrated enough to warrant the megaregional label, holding most of the continent’s population on a small share of its land. They are not, however, a single phenomenon. Some are joined by workers who cross their internal boundaries every morning. Some are joined by executives who fly between their nodes twice a month. Some are joined by parts crossing an international border six times before final assembly. Some are joined by pipelines. Some are joined by nothing except a shared reputation for warm winters and low taxes, which is a real binding agent but a different one, with different vulnerabilities.
Treating these as one category has produced a policy discourse in which the same instruments are proposed everywhere and work in a few places. Sorting them by what actually holds them together — and by the shape that binding agent produces — makes it possible to ask the more useful question: not whether a place is a megaregion, but what kind of integration it has, how fragile that integration is, and which of the very limited institutional tools available is matched to it.
The governance problem remains unsolved and is likely to stay that way. Megaregions will continue to be the scale at which the continent’s economy operates and no scale at which anyone votes. The practical response is not to invent a megaregional government, which is not going to happen, but to keep building single-purpose bodies around the shared physical assets that make coordination unavoidable — and to be clear-eyed that where no such asset exists, coordination will not occur, and planning that assumes otherwise is planning for a jurisdiction that does not exist.
