The Permit Tax: What Review Time Costs Per Apartment in America's Cities

In the fall of 2000, a developer who filed plans for a modest apartment building in an American city had better than two-to-one odds of walking out with a permit inside six months. The odds today are a coin flip, and a slightly unfavorable one. Across 60 cities and roughly 1.1 million permits, three economists at Princeton and Chicago Booth found that a standardized 20-unit building had a 71 percent chance of approval within six months in 2000. By 2025 the figure was 44 percent.

That is the headline of "Where Is Housing Slow to Build, And Is It Getting Slower?", the September working paper by Leonardo D'Amico, Evan Soltas and Audrey Wang that has been passed around housing circles for a month now. The paper measures time, carefully, from the day plans are submitted to the day the building department signs off, and then from sign-off to the certificate of occupancy. What it does not do is put a price on the wait. That is the question developers ask first, and the one city councils almost never answer: when a plan check takes fourteen months instead of four, who pays, and how much?

This piece is an attempt to answer it in dollars per apartment. The method is not exotic. Every month a project sits in review, somebody is carrying the land, the architects' invoices, the option payments and the loan commitment fees. Those costs compound. Construction prices do not wait for the permit either; they have risen about 4.5 percent a year through 2026. Put the months together with the money and you get a number that behaves a lot like a tax, except that no one voted for it, no treasury collects it, and the people who ultimately pay it are tenants who have never heard of a plan check.

The short version: in the slowest cities the carrying cost of review time alone runs to roughly $20,000 to $28,000 per unit† before a shovel is in the ground. In the fastest it is under a thousand. Between those bookends sits most of urban America, and a story about why the gap has widened for a quarter century.

Four clocks, one word

Before ranking anything, it helps to admit that "permit time" means four different things depending on who is counting.

The Census Bureau's Survey of Construction reports that buildings with two or more units took 2.4 months from authorization to start in 2025. That sounds brisk, and it is, because the clock starts when the permit is issued. It measures none of the review.

Phoenix publishes a turnaround report showing major commercial plans get a first review in about 30 days. That measures one round. A project that needs three rounds, which in San Francisco is the median, spends most of its life in the gaps between rounds, when the file is back on the architect's desk.

Seattle's performance dashboard counts "days in city control" from intake to issuance, and for large multifamily projects the 75th percentile sits at 374 days against a goal of 180. The city's own earlier analysis found total calendar time runs about double the city-controlled figure, because applicant time is excluded.

The D'Amico, Soltas and Wang paper uses a fourth clock, submission to issuance, drawn from the raw permit records of 60 cities. It is the only measure that is built the same way everywhere, which is why it can be ranked. It is also conservative. It omits the discretionary entitlement stage that precedes a building permit in many coastal cities, and in a third of the sample it cannot see applications that were never approved. Where it can see them, multifamily applications that died in review had been sitting an average of 2.6 years. The published durations, in other words, are floors.

The practical lesson for anyone reading a city's self-reported numbers is simple. San Francisco's Budget and Legislative Analyst found in March 2026 that housing building permits issued since 2024 had a median processing time of 280 days, against 91 in Austin and 93 in Washington. The same report notes that applications filed after January 2024 show a 114-day median. Both numbers are true. The second one is faster partly because reform is working and partly because the slow applications in that cohort are not finished yet and therefore are not counted. Cohort statistics flatter whoever reports them most recently.

We use the DSW clock for the ranking because it is the only one that lets a developer in Columbus and a developer in Oakland be measured with the same ruler. Where a city publishes a measured figure of its own, we note it. Where a city publishes only a target, we do not treat the target as a fact.

The ranking

The table below is drawn from Table 2 of the D'Amico, Soltas and Wang paper. The figures are unadjusted mean years from submission to permit issuance and from submission to completion, for all new housing permits in each city over the years its data covers. They mix single-family and multifamily permits, so a city that builds mostly subdivisions will look faster here than it does to an apartment developer. The right-hand column shows where each city falls in the authors' separate ranking of a standardized 20-unit building, which covers the 44 cities with enough apartment permits to estimate.

Source: D'Amico, Soltas and Wang (2026), Table 2 and Figure 1(b). Apartment ranks are read from the figure's ordering; the authors have not published the underlying point estimates. Coverage years vary by city.

Two things in this table deserve a second look. The first is the set of cities where the two columns disagree. Charleston, Durham and Baton Rouge issue single-family permits about as fast as anywhere in the country, and rank among the slowest dozen for apartments. A town can be a builder's paradise for cul-de-sacs and a bureaucratic swamp for a four-story walk-up, and the people who write op-eds about "business-friendly" states rarely check which one they are looking at.

The second is who is missing. Chicago is not in the data because, as Soltas told Honolulu Civil Beat, the city lost 25 years of completion dates in a database migration. Dallas, Charlotte, Las Vegas, Minneapolis and Detroit are absent for similar reasons of record-keeping. The largest cities in Texas and the Southeast are measured here mostly through Houston, Austin, Fort Worth, Atlanta and Nashville, which is enough to see the pattern but not enough to settle every local argument.

The pattern itself is not subtle. The coastal metros that already have the most expensive housing in the country also take the longest to say yes to more of it. In San Francisco, New York and Los Angeles, the authors estimate, a 20-unit building takes about 4.6 years from first filing to completion. In Raleigh, Orlando and Phoenix it takes about two. Permitting alone accounts for 30 to 43 percent of the gap between the slowest and fastest quartiles. The rest is construction, and construction in a city where every subcontractor is also waiting on the city tends to run slow too.

Turning months into money

The most honest thing to say about pricing delay is that the published estimates disagree with each other by an order of magnitude, and the disagreement is informative.

At the low end sits the industry's own figure. The National Association of Home Builders and the National Multifamily Housing Council surveyed 49 developers in 2022 and concluded that regulation accounts for 40.6 percent of the cost of a multifamily project. But the line item they label the "pure cost of delay" is only half a percent. That number counts extra interest, at a 7.65 percent rate on the 75 percent of costs that are financed, and nothing else. It books the taxes, the fee escalations and the redesigns under other headings. It is an accounting answer, and a narrow one.

In the middle is RAND. Jason Ward and Luke Schlake compared 144 projects across California, Colorado and Texas in April 2025 and found the average California apartment project took 49 months from predevelopment to completion against 27 in Texas. As California YIMBY summarized the study, the longer timeline raises total development cost by about 8 percent, or $1,284 per unit for every month of delay. That figure folds in interest, property taxes and construction inflation during the wait, and it is the one most often quoted in legislative testimony.

At the top is what the market reveals. D'Amico, Soltas and Wang find that each additional year of multifamily permitting is associated with 39 percent higher median rent across their cities. They are careful to say this is a correlation, not a mechanism, and they note that a pure capital-cost story would predict something closer to 8 percent. The gap between 8 and 39 is the part of the permit tax that is not carrying cost at all. It is the price of scarcity, paid by renters, in cities where slow approval is one tool among several for keeping supply down.

We built our estimates on the middle rung, with the formula in the open so anyone can rerun it with different assumptions.

Permit tax per unit = Excess months × [ (TDC × LTC × r) ÷ 12 + (TDC × (1 − LTC) × r_e) ÷ 12 + TDC × h × (g ÷ 12) + TDC × ℓ × (t ÷ 12) ]

The inputs, and where they come from:

• Excess months. Each city's multifamily duration minus the fastest city's, times the 30 to 43 percent share of the gap the authors attribute to permitting rather than construction. Raleigh, at about 1.5 years total, is the benchmark. Using excess rather than total delay matters: some review time is legitimate, and charging all of it as a tax invites a fair rebuttal.

• Total development cost (TDC). RAND's per-unit figures, in 2019 dollars: $122,000 in Texas, $193,000 in Colorado, $343,000 in California, $485,000 in the Bay Area. Current costs run roughly 20 to 25 percent higher; we have left them in 2019 dollars so the estimates lean conservative. For cities outside those three states we scale by Rider Levett Bucknall's Q3 2026 metro hard-cost ranges.

• Loan-to-cost (LTC) and rate (r). Banks are lending construction money at roughly 60 to 65 percent of cost. The overnight SOFR rate was 3.88 percent on October 14, after the Federal Reserve raised its target to 3.75 to 4.00 percent on September 16, its first increase since 2023. A 250-basis-point spread gives an all-in debt rate of about 6.4 percent (Innowave calculation).

• Equity return (r_e). Developer equity is not free. We assume 10 percent a year on the unfinanced share, which is below most institutional hurdle rates. No brokerage publishes a 2026 multifamily yield-on-cost series openly, so this is an assumption, and the one most worth arguing about.

• Cost escalation (g). RLB's national index rose 4.45 percent year over year in the third quarter of 2026, applied to the roughly 60 percent of cost that is hard construction (h) not yet locked in.

• Land carry (t). Property tax and insurance on the land share (ℓ), assumed at 1.5 percent a year. This is the smallest term and the least sensitive.

The first term, debt carry, is the one we lead with below, because every input in it is verifiable this week. The full formula, with equity carry and escalation added, roughly doubles the figures (Innowave calculation). We treat that as an upper bound, since it assumes all of the development budget is committed during review when in practice land and predevelopment spending dominate before the permit arrives.

A reader who prefers RAND's rule of thumb can skip the algebra. Scaled to local cost, RAND's $1,284 per month works out to about 0.30 percent of total development cost per month of delay†. In Texas that is roughly $450 per unit per month†. In the Bay Area it is about $1,800†. The two methods land within 10 percent of each other for most cities, which is some comfort that neither is wildly off.

What the wait costs, city by city

Here is what the arithmetic produces for ten cities that span the distribution. The multifamily durations are approximate readings of the authors' chart, as published by Andreas Mueller and confirmed in press interviews; we have asked the authors for the underlying estimates and will update the table when they arrive. "Excess months" is review time beyond Raleigh's. The two dollar columns are the debt-carry rung and the RAND-scaled rung, both per apartment, both in 2019 development-cost dollars. The last column is the rent premium the paper's regression associates with that much extra permitting, offered as a ceiling rather than a cost.

California and Texas cities are shown in dollars because RAND published total development cost for those states. Cities elsewhere are shown per $100,000 of development cost; multiply by local cost per unit. A New York apartment at $450,000 of development cost, for instance, carries a debt-only permit tax of roughly $17,000 to $25,000†.

It is worth pausing on what the San Francisco number means in practice. A developer who closes on a lot and files plans is paying interest on a loan, or forgoing returns on cash, for roughly 14 months longer than a developer doing the identical thing in Raleigh. Over a 20-unit building that is about half a million dollars (Innowave calculation) in carry before the first subcontractor is hired. Over a 200-unit building it is $5 million. That money does not disappear. It shows up in the pro forma as a higher rent needed to make the deal pencil, and if the rent the market will bear does not cover it, the building is not built. The paper finds each extra year of permitting is associated with 3.8 fewer homes permitted per 10,000 residents per year, though the authors note that estimate moves around depending on which controls they include.

The Los Angeles case has an unusually clean second source. In a separate paper this year, Soltas and the MIT economist Jonathan Gruber tracked land sales in Los Angeles County and found that a parcel sold with a ready-to-issue permit fetches about 50 percent more than an identical unpermitted parcel, roughly $48 per square foot of land. Buyers, in other words, are paying real money to skip the line, and the size of the premium is about what you would expect if the line costs what our table says it costs. They estimate permitting explains about a third of the gap between what homes sell for in Los Angeles and what they cost to build.

At the other end, the Texas figures are small enough to seem almost beside the point, and that is the point. Fort Worth's excess permitting time is measured in weeks. The carrying cost per unit is less than the cost of a dishwasher. A city that can approve a 20-unit building in a season has, for practical purposes, no permit tax at all, which is one reason the Dallas-Fort Worth metro keeps permitting more apartments than the entire state of California in some years.

Two caveats travel with every number in the table. The durations describe projects filed in 2015 through 2019, because the authors needed completed buildings to measure total time; several cities have reformed since, and some, notably San Francisco and Austin, have measurable improvements to show for it. And the rent column is an association across cities, not a prediction for any one of them. Slow cities are slow in part because they are desirable, and desirable cities command high rents for reasons that have nothing to do with the plan checker.

The case against the number

Anyone who has sat through a planning commission knows that the permit-tax argument has its own opponents, and some of their objections are good.

The first is the one the authors raise themselves. Permitting duration is positively correlated with land prices, which is the opposite of what a pure cost story predicts; if delay simply added cost, it should depress what a developer will pay for dirt. The likelier reading is that attractive cities regulate more because they can. People want to live there anyway, and the political cost of saying no is low. That makes the 39 percent rent association an upper bound on anything a reform could deliver, and the authors say as much: the relationship is, in Soltas's words to Civil Beat, "not necessarily cause and effect."

The second objection is that the industry's own accounting puts the pure cost of delay at half a percent of a project, which is a rounding error next to land, labor and lumber. The NAHB figure is real, and it is small because it is narrow. It counts only incremental interest on the financed share and assigns everything else that happens during a delay to a different category. Our estimates are larger mainly because they count the equity, the escalation and the land carry that the survey books elsewhere. Reasonable people can disagree about which ledger is right. What is not defensible is quoting the half-percent figure as proof that delay does not matter while also citing the same survey's claim that regulation is 40 percent of cost.

The third is the strongest in the current cycle: interest rates and construction costs are killing more projects than plan checkers are. The evidence for this is not hard to find. Multifamily starts fell to a 344,000 annual rate in August, the Fed is raising rather than cutting, and bank loan officers told the Fed in July that construction lending standards sit near the tightest end of their range since 2005. In Arlington Heights, Illinois, a 135-unit project called Mylo sat for two years after winning zoning approval; the developer told the Daily Herald the obstacle was equity and rates, not the village. In Milwaukee, about 700 approved units were on hold in March over costs. The Groundwork Collaborative made the point bluntly in a March paper: "it will not be enough to simply permit more housing to be built."

All true, and none of it contradicts the permit tax. It compounds it. A higher interest rate raises the price of every month of delay; the September hike added roughly 4 percent to the carrying cost of each review cycle (Innowave calculation). The American Enterprise Institute's rebuttal to Groundwork put it this way: permitting delays "can affect project feasibility as much as, and sometimes more than, financing costs." The two camps are describing the same pro forma from different ends.

The fourth objection is about what reform actually does. Yonah Freemark at the Urban Institute has spent years documenting that upzonings produce fewer units, more slowly, than their sponsors promise. An Urban study of 180 reforms across eight metros found a statistically significant but modest 0.8 percent increase in supply over three to nine years; AEI argued the sample included only two reforms large enough to matter. Process reform is easier to pass than zoning reform and, participants at a Pew roundtable in December 2025 agreed, probably less potent. A shot clock does not add a single unit of allowable density.

And the fifth is the risk that a deadline becomes a weapon. Steve Swedberg of the Competitive Enterprise Institute, no friend of slow permitting, warned in a Seattle op-ed that "a well-designed shot clock must also avoid the inverse risk: a denial by deadline." A city facing a 60-day clock it cannot meet can simply reject the application on day 59 and let the developer start over.

The permit tax survives these objections, but smaller and more conditional than its loudest advocates would like. It is a real cost, on the order of tens of thousands of dollars per apartment in the slowest cities, that falls hardest in exactly the places that can least afford fewer homes. It is not the whole story of why rent is high, and in a year when the ten-year Treasury is above 5 percent, it may not even be the biggest chapter. It is simply the one that a city council can fix without asking anyone's permission.

What is being tried

The fixes fall into three families: deadlines, defaults and outsourcing. All three are spreading faster than the evidence on whether they work.

Deadlines. Washington's SB 5290, in force since January 2025, requires a decision within 65, 100 or 170 calendar days of a complete application depending on whether notice and a hearing are required, with partial fee refunds when the clock is blown. The state's first baseline found construction permits running 81 percent over the statutory limits; the second annual report, covering 2025, found median review times improved about 19 percent while most permit types still exceeded the deadlines on average. Georgia's SB 447, effective July 1 of this year, sets a 45-day first review and a 20-day second, and orders a statewide permit dashboard by 2028. Indiana's HB 1005 added a shot clock in 2025 and enforcement teeth in 2026. The vendor Permittable counts 18 states with some statutory deadline; no neutral tracker has confirmed the tally, which tells you something about how new this is.

Defaults. The larger lever is removing the discretionary step altogether. Texas SB 840, effective September 2025, lets apartments and mixed-use buildings go up by right on land zoned for offices, shops and warehouses in the state's 19 largest cities, with density floors of 36 units an acre and no council vote. AEI counted 8,488 units proposed under it in the first nine months. California's AB 130 and SB 131, signed in June 2025, exempt most infill housing from environmental review. Los Angeles went furthest: Mayor Karen Bass's Executive Directive 1 gave fully affordable projects a 60-day approval, and by the time the council made it permanent in December 2025, approvals were averaging 22 days and 490 projects with more than 40,000 units had applied. The by-right finding has older roots. A 2023 UCLA study of about 350 Los Angeles projects found by-right approvals were permitted 28 percent faster than discretionary ones, with a median of 748 days for discretionary projects outside the transit program.

Outsourcing. Texas HB 14, since 2023, lets an applicant hire a licensed third-party reviewer when a city misses its deadline by 15 days. Florida has expanded its private-provider statute three times since 2025, most recently with changes effective this July. Houston's own legislative staff warned that the Texas version could paradoxically slow things down by 14 to 75 days as cities and private reviewers sort out who is responsible for what.

The federal government joined in July. The 21st Century ROAD to Housing Act, which passed the Senate 85 to 5 and became law on July 11 without the President's signature, funds permit-modernization grants, ties a slice of community development block grants to production benchmarks, and directs HUD to write guidelines for single-stair apartment buildings up to six stories. Its zoning provisions, the law firm Brownstein noted, are largely nonbinding. Washington can pay cities to move faster; it cannot make them.

The early returns are mixed in instructive ways. Austin cut its initial site-plan review from about 90 days to 32 and, Pew reports, roughly halved its permitting times, during a period when inflation-adjusted rents in the city fell nearly 20 percent. Correlation again, in a market that was also absorbing a historic supply wave. San Francisco launched PermitSF in February 2025 and can point to the 114-day cohort figure. Honolulu, which takes about 14 months to permit a 20-unit building, began running applications through AI screening software and shaved about 40 days. Milhaus, an Indianapolis developer, told HousingWire it built a 231-unit project in Manatee County, Florida, in about 14 months under the state's Live Local Act, against the 18 to 24 months a normal entitlement would have taken. "Time is money for us," said its executive Brad Vogelsmeier, which is as concise a statement of this article's thesis as anyone has offered.

What none of these reforms has produced yet is a before-and-after study with a control group. The D'Amico paper notes, almost in passing, that permitting durations appear to have shortened since 2023 for reasons the data cannot explain. It may be the reforms. It may be that fewer projects are filing, so the queues are shorter. The honest answer is that we will know in about three years, when the buildings filed in 2024 are finished and can be measured.

The tax nobody levied

There is a reason the permit tax has gone unpriced for so long, and it is not that the math is hard. It is that no one owns it. The plan checker who asks for a fourth round of corrections is not charged for the month it costs. The council member who tables a hearing until after the election does not see the interest accrue. The developer who eats the carry passes what he can to the tenant and absorbs the rest, and the tenant sees only the rent. A cost that appears on no one's budget is a cost that no one has an incentive to cut.

The D'Amico, Soltas and Wang data changes that, a little, by making the time visible and comparable. What this exercise adds is the conversion rate. Fourteen months of excess review in San Francisco is, give or take, $25,000 per apartment† in carrying cost alone, and a good deal more once you count the equity and the escalation. Two months in Fort Worth is the price of a kitchen appliance. Those are the stakes of the next permitting ordinance in each place, stated in the only unit a pro forma understands.

The number will move. Rates will change, and the authors may yet publish the city estimates that would turn our approximations into something firmer. We will update this table when they do. In the meantime, the question for any city that wants more housing is not whether review time costs money. It is whether the city would rather collect that money as a tax it can see, or keep paying it as one it cannot.

† Innowave calculation with the formula and inputs set out above.

A note on method. City durations come from D'Amico, Soltas and Wang (2026), Table 2 and Figure 1(b); multifamily values are approximate readings of the published figure. Excess months are each city's multifamily duration minus Raleigh's, times the 30 to 43 percent permitting share the authors report. Development cost per unit is from RAND (2025), in 2019 dollars. Debt carry assumes 65 percent loan-to-cost at SOFR plus 250 basis points (6.38 percent as of October 14, 2026). The RAND-scaled rung applies $1,284 per unit-month to California cost and scales linearly elsewhere. The full-formula figures cited in the text add 10 percent equity carry on the unfinanced share, 4.45 percent annual escalation on hard costs, and 1.5 percent land carry; they should be read as an upper bound. Spreadsheets with every input are available on request.

Sources

• D'Amico, Soltas and Wang, Where Is Housing Slow to Build, And Is It Getting Slower? (September 2026)

• Gruber and Soltas, How Costly Is Permitting in Housing Development? (2026)

• Ward and Schlake, RAND, The High Cost of Producing Multifamily Housing in California (April 2025)

• NAHB and NMHC, Regulation: 40.6 Percent of the Cost of Multifamily Development (June 2022)

• Manville, Monkkonen, Gray and Phillips, Does Discretion Delay Development?, Journal of the American Planning Association (2023)

• San Francisco Budget and Legislative Analyst, Post-Entitlement Permitting, via San Francisco Chronicle (March 2026)

• Federal Reserve, FOMC statement, September 16, 2026 and July 2026 Senior Loan Officer Survey

• Rider Levett Bucknall, North America Construction Cost Report, Q3 2026

• U.S. Census Bureau, Average Length of Time from Authorization to Start (2025)

• Seattle Department of Construction and Inspections, Construction Permit Performance

• Pew Charitable Trusts, Preapproved Building Plans Help Cities Improve Housing Affordability (May 2026)

• Bipartisan Policy Center, What's in the 21st Century ROAD to Housing Act (2026)

• HousingWire, A case for permit shot clocks builds as approval durations drag out (September 2026)

• Honolulu Civil Beat, While Honolulu Waits for Permits, These Cities Are Already Building (October 2026)

• Urban Institute, How Big Upzonings Affect Housing Supply (April 2026)

• Groundwork Collaborative, Fixing Housing Means Fixing Finance (March 2026); AEI, Housing Supply Is Not Primarily a Financing Problem (2026)

Related reading from Innowave Studio: Your Permit Clock Starts Later Than You Think · The True Cost of Site Plan Revisions · The Site Plan Gap: What Construction Lenders Underwrite · What a Parking Minimum Costs You, Measured in Units