Self-Storage Demand: Per-Capita Saturation and the Three-Mile Logic

Self-storage sells cubic feet to households in transition, and the number the industry quotes for demand, rentable square feet per resident, is the one number in the asset class that the public record cannot verify. No federal series counts rentable square feet. County Business Patterns counts establishments with employees, the Nonemployer Statistics count businesses without them, the American Community Survey counts the households that move, and local permit and assessor records hold the building areas. A demand model built honestly from those sources reaches a different and more useful place than the national rule of thumb: a per-capita figure that is local by construction, because it only exists where a permit or an assessor roll has recorded a building's area, divided by a population that has been apportioned to a drive-time trade area rather than a circle. This article sets out how to build that figure, why three miles is the conventional radius and where the convention breaks, and what the federal record can and cannot say about the supply side. It is the self-storage chapter of the library's demand analysis by asset class pillar, and it pairs with the benchmarks pillar's storage supply metric, which builds the supply side from permits and assessor rolls in detail.

The demand unit: households in transition

The customer of a storage facility is not a resident but a transition: a move, a marriage or a divorce, a death in the family, a downsizing, a child leaving or returning, a business that has outgrown its back room. The Census Bureau measures the largest of these directly. In the 2024 American Community Survey one-year estimates, 11.8% of the population aged one year and older had moved to a different residence in the previous twelve months; 8.9% had moved within the same state and 2.1% had arrived from a different state (U.S. Census Bureau, ACS 1-year migration estimates, 2024). The tables behind those figures, B07001 for mobility by age and B07013 for mobility by tenure, are published for every county in the five-year file released on 8 January 2026 and for areas of 65,000 residents or more in the one-year file released on 11 September 2025. They carry the single most important fact about storage demand: renters move far more often than owners, which is why a trade area's tenure mix matters more to a storage model than its median income.

The stock of renters is measured quarterly. The Housing Vacancies and Homeownership survey for the second quarter of 2026 put the national homeownership rate at 65.0%, with the rate lowest for householders under 35, at 35.2%, and highest for householders aged 65 and older, at 78.6%; the rental vacancy rate was 7.3%, against 7.0% a year earlier, and the median asking rent for vacant units was $1,531 (U.S. Census Bureau, HVS, release CB26-116, 28 July 2026). A trade area that is young, renting and growing produces transitions at a rate the national figure understates; one that is old, owning and stable produces them at a rate it overstates, even when its household count is identical. The trade-area demographics article in the methodology pillar shows how the ACS tables are apportioned to a drawn area and where ring apportionment of block groups breaks; the self-storage model inherits every one of those cautions.

Growth is the second driver, and the Vintage 2025 population estimates measure it with a reference date of 1 July 2025. The national population reached 341,784,857, up 3.1% from the April 2020 base, but the growth was concentrated: the South grew 6.0%, the West 1.9%, the Midwest 1.1% and the Northeast 0.7%, and at the state level the range ran from double-digit gains to outright losses (U.S. Census Bureau, Vintage 2025 Population Estimates, 2026). Each arriving household is a storage transaction waiting to happen, because interstate movers arrive before they have settled housing, and each departing household is one as well. The model therefore reads growth twice: as a level, through population, and as churn, through the mover tables.

The third driver is the apartment pipeline, and it is the one most storage models ignore. Renters in buildings of five or more units hold less space per person than any other tenure group, and the units completing in a trade area are future storage customers in the most literal sense. The Census Bureau and HUD counted 591,700 completions in buildings of five or more units in 2024, the highest annual total in decades, and 471,800 in 2025 on preliminary figures, a fall of 20.3%, while starts of such units rose 18.0% to 396,600 and permits rose 4.3% to 460,400 (U.S. Census Bureau and HUD, New Residential Construction, December 2025, release CB26-28, 2026). The library's articles on tracking multifamily absorption and on permit data as a pipeline indicator show how to read the county-level version of those series; for a storage model the point is simpler. A submarket that completed a large apartment wave in 2024 and 2025 has just added a cohort of small-unit renters, and the storage demand from that cohort arrives over the following two to three years as leases turn and households re-sort.

What the per-capita number is, and what it is not

Square feet per capita is the asset class's favorite statistic, and the trade publishes national and metro versions of it from proprietary inventories. This library does not reproduce those figures, and not only because they are proprietary: the public record cannot check them, and a figure that cannot be checked is not evidence. What the public record can check is the construction of the metric. Rentable square feet come from local building permits and assessor rolls, which record building area and, in most counties, a use code for mini-warehouses; the storage supply metric article walks through that build county by county. Population comes from the Vintage 2025 estimates at the county level and from the ACS at the block group level. The ratio of the two is the per-capita figure, and it carries a property the national number lacks: a vintage and a geography for both numerator and denominator.

The metric fails in predictable ways when it is built carelessly. Dividing a county's storage area by the county's population ignores the fact that facilities near the county line serve the next county. Dividing a three-mile ring's storage area by the ring's population assumes the ring is the trade area, which a river, a freeway or a ridge line can falsify. Counting facilities rather than square feet treats a 20,000-square-foot in-town conversion and a 120,000-square-foot climate-controlled building as equals. And taking the national per-capita figure as a saturation threshold ignores that the figure is an average of markets that differ by a factor of several in both supply and demand. The honest statement is that saturation is a local ratio compared against the same ratio in comparable local markets and against its own history, not a national line.

The three-mile logic

Three miles is the radius storage analysts conventionally draw, and the convention has a logic that survives scrutiny even though the circle itself does not. Storage is a convenience purchase made under time pressure, and the customer chooses among the facilities reachable in a short drive from home or from the route between home and work. In a suburban road network a ten-minute drive covers roughly three miles, which is where the number comes from. The logic is the drive time; the circle is a drawing convenience. In dense urban cores three miles covers more households than the facility could ever serve and the effective trade area shrinks to a mile or less; in rural markets the customer drives fifteen miles to the nearest facility and the three-mile population is a fraction of the true catchment.

The public record supplies what is needed to replace the circle with the catchment. State departments of transportation publish the road network and its traffic counts, which the library's AADT sources article reviews; the Census Bureau's TIGER/Line files carry every road segment with its classification; and LEHD LODES carries the home-to-work flows that define the commuting corridor a storage customer passes along. A drive-time polygon built on those layers, apportioning ACS block groups by the share of their area or their road length inside it, is the trade area the per-capita ratio should use. Where the polygon and the circle disagree, the polygon is right, and the disagreement is largest exactly where a lender is most likely to be looking: at an interchange site on the edge of a metro, where the circle takes in farmland on one side and a dense subdivision on the other.

The three-mile convention has a second use that the catchment preserves. Because the industry has used it for decades, a three-mile ring figure is comparable across a large body of prior analysis, and a model that reports both the ring figure and the drive-time figure lets a reader see how much the geometry changes the answer. When the two diverge by more than a fifth, the site is one where the choice of trade area decides the conclusion, and the credit file should say so rather than report a single number as if geometry were not a choice.

Building the ratio from public records

The build has six steps, and each carries a source with a vintage. One, inventory: every storage facility in the catchment, from County Business Patterns at the ZIP code level for employer establishments, from the Nonemployer Statistics for those without payroll, and from assessor use codes for the properties themselves, with the three sources reconciled by address rather than summed. Two, area: rentable square feet from the assessor's building area, corrected for the share that is rentable (office, drive aisles and loading space are not), with permits supplying the area of anything built since the roll was last updated. Three, pipeline: permits issued but not yet completed, from the local permit system, and the conversions that appear in the permit record as changes of use. Four, population: Vintage 2025 county estimates carried down to block groups with the ACS five-year file and apportioned to the catchment. Five, churn: the mover rates from ACS tables B07001 and B07013 applied to the catchment's households, so that two catchments of equal population but different tenure mixes produce different demand. Six, the ratio: rentable square feet per resident in the catchment, reported with the drive-time and the ring geometry, with the pipeline added as a second figure, and with every source and date listed.

The discipline on the supply side is reconciliation, and it is where most public-data storage models go wrong. An employer establishment in County Business Patterns is a payroll location, not a building; a nonemployer record is a tax filing, not a site; an assessor parcel is a property, which may hold one facility or several phases. Summing the three triple-counts. Matching them by address, and carrying forward only the parcels whose use code and building area are consistent with a storage facility, produces an inventory that can be defended line by line, and the method is the same one the County Business Patterns article describes for counting competitors in any category. The quantitative core of feasibility analysis article in the lender workflow pillar places the resulting ratio in the larger demand, supply and absorption framework that lenders apply across asset classes.

Counting supply: why establishments are not square feet

County Business Patterns is the only series that counts storage operators everywhere on the same basis, and its basis is payroll. The 2023 file, released on 26 June 2025 and coded to NAICS 2017, counted 18,564 establishments in code 531130, lessors of miniwarehouses and self-storage units, with 48,382 employees during the week of 12 March 2023 and $1.96 billion of annual payroll. The size-class columns tell the real story: 16,965 of those establishments, 91.4%, reported fewer than five employees; 1,226 had five to nine; 276 had ten to nineteen; and 97 had twenty or more (U.S. Census Bureau, County Business Patterns 2023, 2025). No other asset class in the library's demand pillar is so concentrated at the bottom of the size distribution. Across all industries 55.5% of establishments have fewer than five employees; for hotels and motels the share is 30.9%, for car washes 52.3%, for general warehousing 31.1%. The average storage establishment employs 2.6 people against 16.7 across all industries and 26.8 for a hotel.

Two consequences follow. The first is that a large share of storage facilities sit outside the employer universe altogether. A facility run by an owner with no payroll, or by a remote manager on another company's payroll, does not appear in County Business Patterns at all; it appears, if anywhere, in the Nonemployer Statistics, the Census Bureau's annual count of businesses with no paid employees and at least $1,000 of receipts, whose 2023 edition was released on 15 May 2025 on the NAICS 2022 basis (U.S. Census Bureau, Nonemployer Statistics 2023, 2025). The second is that the employer count cannot be turned into capacity. A facility with two employees can hold 300 units or 1,200; the file does not know. Dividing 18,564 establishments by the 1 July 2025 population of 341,784,857 gives 5.4 employer establishments per 100,000 residents, a national density that is useful as a screen and useless as a saturation measure, because it is blind both to the non-employer sites and to the size of every site it counts. The state-level version of that density, which the platform computes from the county and ZIP code files, is the right first screen for a market and the wrong last word on it.

The honest supply count therefore moves off the federal series and onto the records the federal series cannot replace: assessor rolls for building area and use code, permits for additions and conversions, and the address-level reconciliation described above. What the federal series contributes is the frame. Every facility with payroll is in County Business Patterns by ZIP code; every facility without payroll and with receipts is in the Nonemployer Statistics by county; a facility that appears in neither is either new, dormant or misclassified, and each of those is worth a phone call before the inventory is closed. The storage supply metric article carries the reconciliation through to rentable square feet; this article stops at the point where the public record hands off to the local one, because that handoff is the methodological event that most published per-capita figures hide.

Where demand is moving: the state record

Storage demand follows people, and the Vintage 2025 estimates show where people went between April 2020 and July 2025. Idaho grew 9.8%, Florida 8.7%, South Carolina and Texas 8.5%, Utah 7.8%, North Carolina 7.2%, Delaware 6.9%, Arizona 6.1%, Tennessee 5.6% and South Dakota 5.3%, against a national 3.1%. At the other end West Virginia lost 1.4%, Hawaii 1.3%, Louisiana 0.7%, Illinois and New York 0.6% and California 0.4%, while Mississippi was flat and Vermont, New Mexico and Pennsylvania grew less than half a percent (U.S. Census Bureau, Vintage 2025 Population Estimates, state totals, 2026). Texas added 2,471,926 residents in five years and Florida 1,871,193, more than the entire population of many states; the arriving households in those two states alone represent a storage transaction volume that no national per-capita average can describe.

The growth figures are levels; the mover figures are churn; and the two together separate markets that look alike on population. A state growing 8% through in-migration generates storage demand at arrival, at first housing and at the second move two years later, when the newcomer trades a rental for a purchase. A state losing population generates it at departure and then loses it, because the departing household's storage need follows the household to its destination. The trade-area demographics article explains how the ACS mobility tables are read at the county and block-group level; the practical rule for storage is to weight the catchment's population by its mover rate and its tenure mix before comparing it with any other catchment's.

Pipeline and vacancy: the apartment feeder

Multifamily completions are the single best leading indicator of storage demand that the public record offers, and 2025 marked a turn. Completions in buildings of five or more units fell from 591,700 in 2024 to 471,800 in 2025 on preliminary figures, a drop of 20.3%, while starts rose 18.0% to 396,600 and permits rose 4.3% to 460,400; in December 2025 the seasonally adjusted annual rate of five-plus permits stood at 515,000 and starts at 402,000 (U.S. Census Bureau and HUD, New Residential Construction, December 2025, release CB26-28, 2026). The apartment cohort that completed in 2024 is the storage customer base of 2026 and 2027; the cohort starting in 2025 will complete into 2026 and 2027 and feed demand through 2029. A storage model that ignores the apartment pipeline in its catchment is ignoring the one demand driver with a published schedule.

Vacancy sets the context for how those cohorts behave. The Housing Vacancies and Homeownership survey for the second quarter of 2026 put the rental vacancy rate at 7.3% nationally, with the South at 9.5%, the Midwest at 6.9%, the Northeast at 5.9% and the West at 5.3%; a year earlier the figures were 7.0%, 9.0%, 6.6%, 5.2% and 5.7%. Vacancy was 8.0% in principal cities, 6.9% in suburbs and 5.8% outside metropolitan areas; the homeowner vacancy rate was 1.2%, highest in the South at 1.5% (U.S. Census Bureau, HVS, release CB26-116, 28 July 2026). High rental vacancy in a growing Southern market means landlords are competing for tenants with concessions, which raises turnover, and turnover is storage demand. Low vacancy in a Western market means tenants stay put, which lowers it. The multifamily absorption article shows the county-level version of those series and the Survey of Market Absorption that tracks how quickly new units lease.

Who rents, and therefore who stores

Tenure by age closes the loop between the demographic base and the transition rate. In the second quarter of 2026 the homeownership rate was 35.2% for householders under 35 and 78.6% for householders aged 65 and older, against 65.0% overall; by region it ran from 69.0% in the Midwest and 66.9% in the South to 61.6% in the Northeast and 60.5% in the West (U.S. Census Bureau, HVS, 2026). Young renters are the mobile population: they move, they live in small units, and they are the tenants of the apartments completing in the pipeline. Older owners are the stable population until the transition that storage operators know well, the downsizing that empties a house into a unit for a season or a decade. A catchment's age structure, read from the Vintage 2025 single-year-of-age tables, therefore predicts not only how much storage demand exists but what kind: short-term and small-unit where the population is young and renting, longer-term and larger where it is old and owning. The senior housing cohort math article works the older cohort in detail for its own asset class, and its arithmetic transfers directly.

What the lender reads

Self-storage occupies an unusual position in lending guidance. The Small Business Administration's 504 program treats it as a multi-use property rather than a special purpose one, which keeps the standard 10% borrower contribution in place, while car washes, hotels, marinas and cold storage carry the special purpose classification and a 15% contribution (504 Capital Corporation, guidance on SOP 50 10 8, 2025). Lenders read that classification as a statement about alternative use, and it is correct as far as it goes: a storage building can become a warehouse. The demand consequence is the opposite of reassuring. Because storage is easy to build and easy to finance, supply responds quickly to any per-capita figure that suggests room, and the markets that looked under-stored on a national average five years ago are the ones where the permit record shows the most new square feet today. The per-capita figure a lender should want is the one that includes the pipeline, the one this article's method produces, and the public SBA loan datasets that MMCG Analytics' SBA layer is built on carry the industry's lending history by NAICS code for anyone who wants to check how the asset class has actually performed, subject to the rule that no rate is shown for a cohort of fewer than ten loans. MMCG Analytics supplies the data and the analysis; the credit decision belongs to the lender.

Method: the five numbers a storage memo should carry

A defensible storage demand statement carries five numbers, each with a source and a date. First, the catchment population, with the geometry stated, from Vintage 2025 and the ACS five-year file. Second, the catchment's mover rate and tenure mix, from ACS tables B07001 and B07013 and the HVS context. Third, the existing rentable square feet, reconciled by address across County Business Patterns, the Nonemployer Statistics and the assessor roll, with the employer-only count reported alongside so the reader can see how much the payroll series misses. Fourth, the pipeline, from permits and from the multifamily completions that feed demand, with the apartment cohort dated. Fifth, the ratio, reported for the drive-time catchment and for the three-mile ring, with the difference between them stated. A memo that carries those five numbers can be checked line by line by anyone with a browser; a memo that carries a single national per-capita figure and a saturation threshold cannot. The difference is the provenance standard that the provenance article sets out for the whole library, and the asset-class pillar applies it to thirty other property types in the same way.

A worked sequence for one site

The method reads more clearly as a sequence than as a list, so here is the order in which a platform built on public data runs it for a single proposed facility at a suburban interchange. The analyst starts from the parcel, pulls the road network around it and builds a ten-minute drive-time polygon; the three-mile ring is drawn beside it for comparison, not instead of it. The polygon is intersected with ACS block groups, and each block group's households, tenure split and one-year mover rate are apportioned by the share of the block group's road length inside the polygon, a better proxy for where people live than the share of its area. The Vintage 2025 county estimate then scales the block-group totals to the current reference date, since the five-year ACS file is centered two to three years earlier, and the scaling factor is written down rather than buried.

On the supply side the analyst lists every self-storage establishment in County Business Patterns for the ZIP codes that touch the polygon, adds the nonemployer filings for the same counties, and reconciles both against assessor parcels whose use code or building description indicates mini-warehouse use. Each surviving parcel contributes its building area from the roll, adjusted to rentable area with a stated factor, and each permit issued since the roll's date adds or converts area. Facilities that sit inside the three-mile ring but outside the drive-time polygon are kept in a separate column, because a lender will ask about them and because their presence or absence is the clearest illustration of why geometry matters. The pipeline column carries permitted but unbuilt area and the apartment units under construction inside the polygon, each dated.

The output is two ratios and a direction. Rentable square feet per resident inside the polygon, with the ring figure beside it; the same ratio with the pipeline added; and the direction of the demand base, from population change since 2020, the mover rate and the tenure mix, and the apartment completions that will feed transitions over the next three years. None of those numbers is the national figure the trade quotes, and none of them needs it. Each can be checked by anyone who opens the Census tables, the county's permit portal and the assessor's roll, which is the whole point of building the model this way: the credit file carries a demand statement whose every component has a public source and a date, and whose geometry is stated rather than assumed. The platform's job is to make that sequence take minutes instead of days; the lender's judgment about what the ratios mean for the loan remains the lender's.

The same sequence scales. Run for every interchange in a county it produces a map of catchments with their ratios, and the map reveals what the national average conceals: the catchments where supply has run ahead of transitions, typically where the apartment wave completed before 2024 and the permit record shows storage area still coming, and the catchments where transitions have run ahead of supply, typically in the fast-growing Southern and Mountain counties where the Vintage 2025 estimates show the population arriving faster than anyone has permitted storage for it. Those are the two kinds of market a lender most needs to tell apart, and the public record, read in this order, tells them apart.

Frequently asked questions

What is the demand driver for self-storage?

Household transitions: moves, marriages, divorces, deaths, downsizing and business growth. The Census Bureau measures the largest of them directly through the American Community Survey's mobility tables, which showed 11.8% of the population moving in 2024, and the Housing Vacancies survey measures the tenure structure that sets the transition rate.

How is square feet per capita calculated from public data?

Rentable square feet come from local assessor rolls and permits, reconciled by address with County Business Patterns and the Nonemployer Statistics; population comes from Vintage 2025 county estimates apportioned to a drive-time catchment with ACS block-group data. The ratio is reported with both the drive-time and the three-mile geometry and with the pipeline added as a second figure.

Why does the three-mile radius matter?

It approximates a ten-minute drive in a suburban road network, which is the convenience radius storage customers actually use. The circle fails in dense cores, where the catchment is smaller, and in rural markets, where it is larger; a drive-time polygon built on the road network replaces it, and reporting both lets a reader see how much geometry changes the answer.

Does County Business Patterns count all storage facilities?

No. It counts establishments with paid employees, and 91.4% of the 18,564 self-storage establishments it counted in 2023 had fewer than five employees; facilities with no payroll appear only in the Nonemployer Statistics. Neither series records square feet.

How does the apartment pipeline affect storage demand?

New apartment residents hold little space per person and move often. Completions in buildings of five or more units reached 591,700 in 2024 and fell 20.3% in 2025 while starts rose 18.0%; the cohorts completing in a catchment are its storage customers over the following two to three years.

Is self-storage a special purpose property for SBA lending?

Lender guidance on SOP 50 10 8 classes self-storage as multi-use, with the standard 10% borrower contribution on a 504 project, rather than as a special purpose property at 15%. The classification reflects alternative use, not demand; the SBA's public loan datasets carry the industry's actual lending history by NAICS code.

Sources

1. U.S. Census Bureau, County Business Patterns 2023, released 26 June 2025; national file cbp23us, NAICS 531130 and comparison codes. https://www.census.gov/newsroom/press-releases/2025/2023-county-business-patterns.html
2. U.S. Census Bureau, Nonemployer Statistics 2023, released 15 May 2025. https://www.census.gov/newsroom/press-releases/2025/2023-nonemployer-statistics.html
3. U.S. Census Bureau, Vintage 2025 Population Estimates, state totals (NST-EST2025-ALLDATA), reference date 1 July 2025, 2026. https://www2.census.gov/programs-surveys/popest/datasets/2020-2025/state/totals/NST-EST2025-ALLDATA.csv
4. U.S. Census Bureau, Vintage 2025 Population Estimates, county characteristics, press release of 25 June 2026. https://www.census.gov/newsroom/press-releases/2026/vintage-2025-pop-estimates.html
5. U.S. Census Bureau, Migration and Geographic Mobility, American Community Survey 1-year estimates guidance (2024 data), 2025. https://www.census.gov/topics/population/migration/guidance/acs-1yr.html
6. U.S. Census Bureau, American Community Survey data releases: 2024 1-year estimates (11 September 2025) and 2020-2024 5-year estimates (8 January 2026). https://www.census.gov/programs-surveys/acs/news/data-releases.html
7. U.S. Census Bureau and U.S. Department of Housing and Urban Development, New Residential Construction, December 2025, release CB26-28 (2026). https://www.census.gov/construction/nrc/pdf/newresconst_202512.pdf
8. U.S. Census Bureau, Quarterly Residential Vacancies and Homeownership, second quarter 2026, release CB26-116, 28 July 2026. https://www.census.gov/housing/hvs/files/currenthvspress.pdf
9. 504 Capital Corporation, SBA 504 Loans for Special Purpose Properties and Real Estate (guidance on SOP 50 10 8 classifications), 2025. https://504capital.com/blog/financing-special-purpose-properties-sba-504-loans/
10. National Association of Government Guaranteed Lenders, SBA Notice Revising SOP 50 10 8 (Procedural Notice 5000-872764), 2025. https://www.naggl.org/sba-notice-revising-sop-50-10-8/
11. MMCG Research, SBA 7(a) Performance Series: MMCG analysis of the public SBA 7(a) loan register, 2026. https://mmcganalytics.com/sba-default-rates/