Shared Parking: How Mixed Uses Cut the Space Count

A shopping mall builds 1,000 spaces that sit largely empty after 8 p.m. The apartment complex next door builds 500 that sit half-empty from nine to five. Between them, the two owners have paid for 1,500 spaces to serve demand that could have been met with roughly 1,125, because the two demands peak at different hours. At $28,000 per structured space, the difference is about $10 million of concrete built to stand vacant on an alternating schedule.

That arithmetic is shared parking: sizing a parking supply to the combined, time-shifted demand of multiple land uses rather than to the sum of each use counted alone. It is the single largest lever a developer has for cutting parking cost before a shovel touches ground. It is also the reason a well-planned mixed-use district can add a restaurant or a daytime conference center to the same 1,125 spaces without pouring another one. The concept is decades old. What has changed is that municipalities are finally writing zoning that permits it, which means the developers and consultants reading this will be asked to calculate it, defend it, and live with the consequences when the calculation is wrong.

What the ordinance assumes, and what it costs

Traditional zoning assigns each land use its own minimum: commonly 3.5 to 6 spaces per 1,000 square feet of gross leasable area (GLA) for retail, around 4 or more per 1,000 for office, one space per studio or one-bedroom unit and up to two per three-bedroom for residential. Each ratio is defensible in isolation. Stacked together on a mixed-use parcel, they mandate a supply sized for a moment that never occurs: every use at peak simultaneously.

The cost of that phantom moment is concrete. Structured parking runs roughly $18,000 to $50,000 per space above grade and $45,000 to $65,000 or more below grade. Constrained urban sites, with tight parcels, high water tables, and small space counts, can exceed $100,000 per space. Run the campus-scale version. A development required to provide 10,000 spaces at $28,000 each carries a $280 million parking obligation. If a shared parking analysis demonstrates that 7,500 spaces serve the same demand, the developer just saved $70 million. That is why owners commission the study.

The correction has its own failure mode, and the owner should watch for it in the ordinance itself. Some municipalities, having recognized that minimums overbuild, have swung to eliminating parking requirements entirely, including in markets with no meaningful transit and a driving culture that is not going to change on the ordinance's schedule. A development built with little or no parking in that market does not eliminate its demand. It exports the demand to its neighbors, pushes rates up, and makes hiring harder for every business inside it. The remedy for a bad fixed ratio is not zero. It is a calculated number.

Why the sharing works: complementary peaks

Shared parking works only where demand curves offset. An office building peaks weekdays, nine to five. A concert venue peaks evenings and weekends. Put them beside each other and the same stall serves both with minimal overlap; the office worker has left before the ticketholder arrives. Hotels peak overnight, restaurants peak at dinner, and a university's spaces empty on weekends exactly when a stadium needs them. The designer's job is to pair uses whose peaks interlock. The consultant's job is to prove the interlock with data rather than assert it.

The prize for getting it right extends past construction savings into operations. An office garage historically fills to 75-85 percent of capacity once a day and empties once. The same garage ringed with restaurants, a gym, and an event venue can turn each space two to four times daily, and each turn is revenue. Shared parking is not only a smaller building. It is a busier one.

Distance disciplines the pairing. Users tolerate different walks for different trips: up to roughly 1,200 feet for office workers, about 600 feet for retail customers, roughly 350 feet for hospital patients, and 1,500 to 2,000 feet for stadium and convention crowds. A stall can only be shared between uses whose acceptable walking ranges both reach it. Manhattan has run a version of this logic for decades through off-site parking arrangements known as parking easements, which let a developer satisfy a parking obligation with spaces on another parcel, secured by legal agreement among the owners and the city.

Where it breaks: the tenant you didn't model

The shared parking count is a photograph of a moment in the development's life, and the development keeps living. The instructive case is a medical office building. A standard MOB demands 4.0 to 4.5 spaces per 1,000 square feet. But suppose the tenant is a high-end specialty practice seeing one patient per 1,000 square feet, so that roughly 70 spaces serve the whole building. Beside it sits a 65-room hotel peaking evenings and weekends. The two share beautifully. The analysis closes, and the garage is built to the shared number.

Then the specialty practice leaves. A conventional MOB tenant moves in, or an HMO packing more services into the same floor area, and the parking demand roughly quadruples overnight. No one poured the difference. The building's floor area never changed; its parking arithmetic did. This is the standing risk of every shared count. Land uses change, tenants change, and business models change inside the same walls. The owner who accepts a shared count below standard ratios should understand that the discount is underwritten by the current tenancy, and price the risk of the next one.

The other break point is governance. Where multiple owners share a facility, the shared parking agreement carries the whole arrangement, and the disputes that kill these deals are rarely about stall counts. They are about who pays for the membrane replacement in year twelve, who manages enforcement, how income splits, and what happens when one party's use changes. An agreement that specifies policies, allocation, and the division of income, operating cost, and capital cost is infrastructure as surely as the deck itself. Some stakeholders will want the savings and none of the shared responsibility. The interviews that surface this cost nothing compared to discovering it after closing.

The recipe: how the number actually gets built

The calculation itself is a stack of project-specific inputs, and the consultant's craft is in the inputs, not the spreadsheet.

Demand data, measured long enough to mean something. A five-to-seven-day count is a weather report. Occupancy patterns shift with seasons, events, and business cycles. A defensible baseline draws on at least a full month of data and preferably years, especially where an existing facility's demand anchors the model.

The denominator, defined. Ratios are only comparable when the floor-area basis is stated. Gross floor area (GFA) measures to the outside of the exterior walls and typically governs single-tenant buildings. GLA subtracts common areas and governs most multi-tenant projects. Net floor area and net rentable area measure inside the walls. A ratio quoted without its denominator is a number without units.

Ratios fitted to the project, not imported. Residential demand runs on unit type, tenure, and location. A small town may require one space per one-bedroom unit while a transit-rich city functions at 0.25 to 0.75 per unit. Restaurant demand runs on riders per vehicle, and the same restaurant can turn 1.5 to 3 riders per car at weekday lunch, when colleagues carpool from the office they will return to, and one rider per car at dinner, when everyone wants to leave on their own schedule. More people per car at lunch means the same seat count generates less parking demand at noon than at 8 p.m. The ratios move with time of day because behavior does.

Employment, counted, not assumed. An event center can run 80 staff in 2,000 square feet. A high-end salon in the same 2,000 square feet runs 30. Same floor area, 50-space difference. Square footage is a proxy for demand only until the actual business model is known, at which point the actual business model governs.

Seasonality, interrogated. A resort's winter surge and a legislature's spring session are genuinely seasonal; serve them with overflow strategies, park-and-ride and shuttles, not permanent spaces. But a "seasonal" peak that arrives a little higher every year is not seasonal. It is growth wearing a costume, and it may justify supply after all. The consultant's question is whether the peak repeats or compounds.

Operations and equipment, designed for the plan. A shared count assumes parkers can be steered: office monthlies here, hotel valet there, event overflow into the deck after six. That steering is performed by the parking access and revenue control system, guidance signage, and the facility's allocation policy. Modern workforce schedules have scrambled the old Monday-to-Friday rhythm, which makes flexible access control a load-bearing element of the count, not an accessory. A shared plan the equipment cannot enforce is a drawing.

A case in siting: the stadium that moved

The proposal that became SoFi Stadium was originally examined for downtown Los Angeles, with an analysis that leaned on existing parking inventory, including spaces 6,000 to 7,000 feet away (1.14 to 1.33 miles) in Chinatown. On paper, downtown held the supply. Vetted, the supply dissolved. Many of the counted spaces carried no guarantee of after-hours public availability. Signage did not establish the public's right to use them during events. The walking distance sat far beyond what event crowds tolerate, with arriving fans set to collide with the departing downtown workforce at peak. The assumptions failed the audit, and the failure was material to relocating the stadium to Inglewood, where transit connections and controllable parking within workable distances could be verified rather than presumed. The lesson generalizes to every shared count: a space is only shared if it is actually available, legally and practically, at the hour the second user needs it.

Source crosswalk -- Module 3