Your apartment supply forecast may be missing 750,000 units.

Apartment investors often treat Census permits and starts as the supply screen. Those series are useful, and they are incomplete.

CRE Daily's September 4 brief highlighted a Center for Public Enterprise analysis, published August 28, 2026, by fellow Paul Williams. The estimate: roughly 750,000 multifamily units sit in an entitled-but-not-permitted stage that no federal agency systematically tracks. At typical development costs, that inventory represents about $225 billion to $265 billion of potential construction. In other words, a large block of future competition can clear zoning and land-use approvals, then stall on financing, without ever appearing in the permit print your team watches every month.

What the shadow pipeline is

Entitlement is the stage where a project receives discretionary land-use approvals such as zoning changes, conditional use permits, planned unit developments, or environmental clearances. It is often the slowest and most political part of development. In San Francisco, Williams notes that approvals can take at least 15 months. A building permit comes later, once detailed construction documents are ready. In normal conditions, about 80 percent of multifamily projects begin construction within two months of receiving a permit, according to Census Survey of Construction data cited in the analysis.

The commonly cited "authorized but not started" figure captures projects that already have permits but have not broken ground. Williams puts that bucket at about 117,000 multifamily units. The shadow pipeline he estimates is more than six times larger. Another roughly 1 million units were under construction in the data he examined, while annual multifamily permits and completions each ran near 500,000 units. The missing stage is the one between land-use approval and the building permit.

It helps to remember why that gap exists. Federal statistics are built to track permits, starts, and completions. They are not designed as a national registry of every local entitlement. The Census Building Permits Survey covers roughly 900 of nearly 20,000 permit-issuing jurisdictions, and it begins after entitlement. CoStar, Yardi Matrix, and RealPage track pipelines with their own categories, and some cities publish local reports, but there is still no standardized national entitled-but-not-permitted database under Williams' definition.

Why federal data starts too late for your rent case

Sellers and brokers often pair a soft national starts print with an optimistic rent path. The logic sounds tidy: deliveries are rolling off, so vacancy should tighten and rents should firm. The shadow pipeline weakens that leap. Projects that already hold entitlements have cleared a major risk. When interest rates, construction lending, or equity terms improve, those deals can move to permit and start faster than projects that still need years of approvals.

Williams' triangulation used metro-level pipeline reports, a Yardi Matrix planned-pipeline ratio, and a timeline-based flow model. The synthesized point estimate is 750,000 units. The ranges differ, and he is explicit that the figure is an order-of-magnitude estimate, not a parcel-level census. Coastal, high-regulation metros with published pipeline data may overstate the national picture if you scale them naively, while private pipeline databases may undercount smaller projects. That uncertainty cuts both ways for an acquisitions team. It does not erase the underwriting point: two metros with similar permit counts can carry very different inventories of approved-but-not-yet-permitted competition.

Local color from the same analysis shows how uneven the stock can be. San Francisco's planning dashboard has reported roughly 52,000 approved but stalled units. Boston reporting pointed to about 23,000 stuck units, plus additional paused affordable projects. New York City's industry pipeline counts have run near 100,000 units in the broader active set. Those are not your submarket, and they are not a national average you should paste into every Sun Belt OM. They are evidence that entitlement inventories can be large where you actually bid.

What to put in the underwriting memo

When an offering memorandum, or OM, leans on "supply is falling" as the growth case, put the missing stage in the memo before you accept the rent path. Ask what the broker is measuring: permits, starts, under construction, or entitled projects waiting on capital. Ask whether the submarket's local planning board, Yardi or CoStar pipeline, and city dashboard agree. Ask how many competing units already have site control and land-use approval within a competitive radius of the asset.

Then connect that screen to the property file. Rebuild net operating income, or NOI, from the T-12, which is the trailing twelve months of actual income and expenses. Compare in-place rents on the rent roll with what is leasing nearby. Check concessions (free rent or move-in discounts), lease-up velocity, and whether the seller's vacancy forecast assumes deliveries stop while entitled competitors sit idle. If financing conditions ease, those competitors may not stay idle.

Run a simple stress. Suppose your model assumes deliveries drop enough to support 3 percent annual rent growth. If 400 entitled units within three miles could permit within a year of better debt terms, the absorption math changes. You do not need to assume every entitled unit gets built. Williams notes that some entitlements expire, redesign, or change hands. You do need a line in the memo that says which local entitled inventory you checked, and which you ignored.

Rebuild from the property, not the national print

National and metro headlines belong in the investment memo as context. They do not belong as the growth tab. The OM is a marketing document. Trailing actuals still beat seller pro forma. A confident wrong supply assumption is still worse than a slower, cited model.

This is where AcquiOS helps without pretending to decide the market. It moves the OM, T-12, rent roll, and supporting exhibits into your team's existing Excel template with source citations, so analysts spend time testing local pipeline and rent assumptions instead of retyping pages. AcquiScore can mark Proceed, Caution, or Pass against your written buy box after the file is built. It does not tell you whether a shadow pipeline in another metro clears your exit cap rate (the rate a buyer will use to value the property when you eventually sell it).

What to do this week

Take one multifamily OM that cites falling starts or a light permit pipeline. Add a one-page entitlement check for the submarket: local planning dashboard if it exists, CoStar or Yardi planned and proposed units, and any entitled projects within a competitive radius. Then rebuild the rent and vacancy case from the T-12 and rent roll without borrowing the national starts headline as proof of pricing power.

If the seller's growth case only works when you ignore approved-but-unpermitted competition, say so in the memo and size the bid accordingly.

Frequently Asked Questions