How a city funds housing creation
Housing production is a flow; thus, ratios and rates are the measures.
Homes are net new homes in completed projects, counted in the year the project was finished. Population: U.S. Census (2000, 2010, 2020) and the Portland State University estimate for 2025. Households: Census 2000 and 2010, and a 2026 private estimate (Esri). A check on the count: the Census reports 28,201 more housing units in Portland in 2010 than in 2000; whereas this dataset finds net new homes completed over the same years. The reason for the discrepancy is unknown; but affirms the scale and directionality implied by this data, which is through .
Funding new homes
New homes are funded by different sources of money. The size of a project is one way to understand this.
Housing production is now mostly subsidized
Institutional-scale projects are funded either by private investors or by public subsidy. Splitting them shows market rate investment forces historically power housing production.
Subsidized means a project receiving state, local bond or federal tax-credit money, or university housing. A market-rate project that includes required inclusionary units is counted as market rate because it is just a tax on higher income renters. Tags for the largest and most recent projects were checked one by one; older and smaller projects rely on matching to public affordable-housing records.
Where new homes are built is always changing
Each dot is a project, sized by its number of homes and colored by its source of capital. Move through the years to see which parts of the city each source favoured, and when it left. (Who builds the 'burbs, in the core, proximate to transit, etc.)
National recessions reset housing production flows
The city's two "gate-keeping" systems - land use & building permit approvals - shrink the total time possible for larger buildings to gain approvals and be funded/built; slower approvals translate to smaller housing production flows.
Cases for 2010 to mid-2020 were transcribed by hand from city records; later cases were read from the written decisions. Cases before 2010 were read by script from the city's short case descriptions and are provisional. A case is linked to a building permit when the decision names the permit or when a permit of similar size was filed on the same site; some real links are missed, so the share lost is an upper estimate. Approvals from the last three years may still proceed.
City generated disruptions, like a new tax on housing, lower total volumes created.
Portland adopted an inclusionary housing requirement that took effect on February 1, 2017. For buildings of 20 or more homes, a fee, or below-market rents on a share of the homes was imposed; this tax being paid by the building's market rate renters. Applications on file before the deadline exempted a project, so vesting held potential option value; however, this appears to not have translated to realized value. Overall, the disruption left housing production materially lower: about 22% below an uninterrupted flow, one that continued the pace seen from 2014 to 2017.
The 20-home line is the rule's own threshold and does not follow the slider above. Land use applications are design and historic reviews, which only projects in design overlay areas need; they are counted in the month the application was filed, whatever the later decision. The red boxes and arrows are interpretation: each arrow points about 13 months ahead, roughly the time a filed project takes to show up in permits (IH is inclusionary housing). An application counts as never reaching a permit when no building permit has been issued for it to date. It is linked to a building permit when the decision names the permit or a permit of similar size was filed on the same site; some real links are missed in every period, so the rates are upper estimates. Construction cost data is not part of this dataset: the cost explanation is an interpretation of the pattern.
Getting to a permit takes about three times as long as in 2000
Bars show the share of that size's projects issued a permit that year; the count is given at the right of each panel, and some years have only a handful of large projects. The last bar holds every wait of 30 months or more. The dashed lines mark the medians for the normal, higher volume housing production years. Records begin in January 2000, so permits issued in 2000 and 2001 cannot show waits longer than one to two years. 2026 is a partial year.
Why the city overstates housing production
The city's permit system carries a "new units" figure on each permit, but adding that column up counts homes that were never added. These are the ways it happens, with the number found in the record for 2000 to 2026.
| What the record shows | What actually happened | Records | Units or lots |
|---|---|---|---|
| Demolish and replace | A home is torn down and new housing built on the same site. Only the difference is new. | 3,693 projects | 4,604 |
| Remodel recorded as a new unit | A kitchen addition or bathroom remodel carries a new-unit figure, though no home is added. | 576 permits | 1,565 |
| Hotel rooms recorded as units | New hotels and motels carry a new-unit figure. | 15 permits | 1,057 |
| Middle housing land divisions | A land use approval splits an already-permitted fourplex or cottage cluster onto separate lots. The lots are not additional homes. | 237 cases since 2020 | 1,026 |
| Permits that expired | Units on permits that were never built remain in the record. | 428 permits | 558 |
| Whole project counted on every permit | Each townhouse permit in a nine-home project shows nine units. | 42 permits | 206 |
| Replacement or relocated homes | Replacement manufactured homes, moved houses and fire rebuilds. | 137 permits | 157 |
About this data
Sources
The city has two gatekeeping functions that bottleneck / control housing production flows - land use approvals and building permit approvals. All counts are pulled from the city's official land use decisions - via City Auditor's Efiles archives. For building permit approvals, the City of Portland's public permit records are used via PortlandMaps. Distinction between market rate versus public/subsidized housing production comes from subsidy records of the Oregon Affordable Housing Inventory, Metro's Regulated Affordable Housing Inventory, the Portland Housing Bureau and Oregon Housing and Community Services - cross referenced to the specific building permit for each project. Every project counted here can be reviewed in the project list.
What counts as a new home
New construction, conversions of offices and hotels into housing, and homes added inside existing buildings such as accessory dwelling units. Remodels, hotels and replacements do not count. Demos that removed a home to build a new home in its place do not count as adding housing.
Projects
Because townhouses and subdivisions are approved as one project, but each home holds its own building permit, a 40 meters proximity and 270 days relationship was used to distinguish a discrete development. For example, a 14-home townhouse project counts once, at its real size (instead of as 14 specific, one unit development efforts.)
Sources of capital
Projects of 1 to 4 homes are typically funded from households with sufficient wealth to do so. Smaller development projects - from 5 homes up to the institutional thresholds - are typically funded by local and regional developers and investor groups. Larger projects at or above the minimum investment scale thresholds, be they rental or for sale, are those funded by institutional investment (or via public capital (i.e. city taxes, regional bond measures, federal low income housing tax credits, etc.))
Projection
The lighter bars for 2027 and 2028 count only projects already filed and awaiting a permit. They assume household-funded building continues at its recent pace, that between a quarter and 60% of pending mid-sized homes actually manifest into permitted new housing production, that most pending subsidized large projects proceed (60% to 90%), and that few pending market-rate large projects do (none to 25%). Projects not yet filed are not included.
Limits
Affordable-housing inventories lag by one to three years. Grouping by distance can occasionally join deals or split phases. Of about 17,000 projects, around 550 were reviewed individually; the rest follow the rules above. The characterization of all flows should be understood to be accurate, with reasonable precision and project-level verifications to ensure reliance.
About this housing data visualization project
Envision Portland Trends began in the summer of 2020 as an applied study of housing production, city governance and statistics / computer science realms. The effort was funded by Old i, LLC via federal COVID PPP stimulus to create socially helpful learning for Williams College students that also can enable better civic decision making.