Upward Communities Research
Teacher housing is a retention question before it is a real estate question. The cost of losing teachers is already on the district's books, and in a majority of the largest urban districts a beginning teacher can no longer afford local rent. This page sets out what the evidence supports, what existing programs have delivered, and where the data runs out.
Contents
1. Definitions: what counts as teacher housing
2. Key findings
3. Why is teacher pay no longer enough to cover housing?
4. What does teacher turnover actually cost a district?
5. How do school districts finance teacher housing?
6. What have district housing programs actually delivered?
7. What does a district actually do, step by step?
8. Where is the evidence still thin?
9. Frequently asked questions
10. Sources
Four terms get used interchangeably in this field and they do not mean the same thing. The distinctions matter because they determine which funding sources a project can use and who is allowed to live there.
Housing aimed at households that earn too much to qualify for deeply subsidised affordable housing and too little to afford market rent in their own community. There is no single federal definition. In practice the band is usually described as roughly 60% to 120% of area median income, and different programmes draw the line in different places.
Workforce housing where tenancy is restricted to school district employees. This restriction requires legal authority. California's Teacher Housing Act (SB 1413, Leno, 2016) permits local education agencies to establish housing programmes, to use local or state funds including affordable housing tax credits, and to restrict tenancy to "teachers and school district employees." A follow-up measure, AB 3308 (2020), clarified that housing with those tenancy restrictions can still use public funding sources including affordable housing tax credits.
The same idea applied to a broader group: teachers plus first responders, healthcare workers, and municipal and county staff. Widening eligibility can improve a project's financing options and its politics, at the cost of diluting the recruitment benefit for any single employer.
The midpoint household income for a metropolitan area, published annually by the US Department of Housing and Urban Development and adjusted for household size. Affordability limits in most housing programmes are expressed as a percentage of AMI, which is why the same rent can be classified differently in two neighbouring counties.
Spending more than 30% of gross income on housing. This is the standard federal threshold, and it is the test the National Council on Teacher Quality used in 2025 to classify a district as unaffordable for its own teachers.
Key findings
1. Replacing one teacher costs a district between $11,860 and $24,930, depending on district size. The Learning Policy Institute puts the cost at $11,860 in districts under 10,000 students, $16,450 in districts of 10,000 to 50,000, and $24,930 in districts above 50,000, in January 2024 dollars. Learning Policy Institute, 2024.
2. Roughly 6 of 10 teachers hired each year are replacing colleagues who left before retirement. This is a replacement problem, not a growth problem. Learning Policy Institute, 2017.
3. Between 2019 and 2025, rents rose about 51% and home prices about 47%, while beginning teacher salaries rose about 24%. Inflation over the same period was about 26%, so beginning salaries lost ground against prices generally, and lost badly against housing. National Council on Teacher Quality, 2025.
4. Rent is now unaffordable for a beginning teacher with a bachelor's degree in 39 of 72 large urban districts, up from 18 in 2019. For teachers ten years into the job with a bachelor's degree, 18 of those 72 districts are unaffordable, up from just 2 in 2019. National Council on Teacher Quality, 2025.
5. Public school teachers' weekly wages were 26.9% below those of comparable college graduates in 2024, a record high. Counting benefits, which are better for teachers, the total compensation gap was 17.1%. Economic Policy Institute, 2025.
6. One district has published retention figures either side of opening its housing. Jefferson Union High School District reported certificated staff retention of 85% before its housing opened in 2022 and 90% the following year, in its 2023-24 Local Control and Accountability Plan. This is the clearest published before-and-after figure available, and it is a single district over a single year. UC Berkeley Center for Cities + Schools and cityLAB UCLA, 2025.
7. Nine California educator housing developments took an average of 7.5 years from concept to occupancy. That is about the same as any other multifamily project in California, which is the point: this is a development timeline, not a programme timeline. UC Berkeley Center for Cities + Schools and cityLAB UCLA, 2025.
Because housing costs and teacher salaries have moved at different speeds for six years. Between 2019 and 2025, average rents rose about 51% and home prices about 47%, while beginning teacher salaries rose about 24%, according to the National Council on Teacher Quality. General inflation over that period was about 26%, so beginning pay roughly tracked prices while badly lagging housing.
The consequence shows up as a threshold effect rather than a gradual squeeze. NCTQ examined 72 large urban districts and tested whether local rent exceeded 30% of a teacher's salary, the standard federal definition of cost burden. In 2019, rent was unaffordable on that test for beginning teachers with a bachelor's degree in 18 of the 72 districts. By 2025 it was 39 of 72. For beginning teachers with a master's degree the count rose from 11 to 34.
The more telling movement is further up the salary schedule. For teachers ten years into the profession with a bachelor's degree, the number of unaffordable districts went from 2 in 2019 to 18 in 2025. For the same teachers with a master's degree it went from zero to 8. A decade of experience and a graduate degree used to be a reliable route out of cost burden in a large district. In a quarter of the districts NCTQ studied, it no longer is.
San Diego Unified was the most unaffordable district in the 2025 analysis, where rent consumed 45% of the salary of a beginning teacher holding a master's degree.
This sits on top of a compensation gap that predates the housing run-up. The Economic Policy Institute has tracked what it calls the teacher pay penalty since the 1990s. In 2024 the regression-adjusted weekly wage gap between public school teachers and comparable college graduates reached 26.9%, the highest in the series. Counting benefits, which run better for teachers, the total compensation gap was 17.1%. The Learning Policy Institute frames the same problem by career stage: beginning teachers earn about 20% less than college graduates in other fields, a gap that can widen to 30% for mid-career educators.
The practical implication for a district is that salary alone is a slow instrument against this. A raise is taxed, it applies across the whole schedule, and in the markets where the gap is worst it is competing against rent increases of a similar or larger size. Housing acts on the cost side of the same equation, and it does so only for the employees who choose to use it.
Between $11,860 and $24,930 for each teacher replaced, depending on district size, according to the Learning Policy Institute's 2024 figures. The estimate covers separation, recruitment, hiring and training costs borne by the school and the district. It does not attempt to price the loss of institutional knowledge or the effect on students.
LPI's per-teacher figures, in January 2024 dollars, are $11,860 for districts under 10,000 students, $16,450 for districts of 10,000 to 50,000, and $24,930 for districts above 50,000. The figures are built by averaging reported turnover costs across districts in each size band, drawing on five underlying studies. LPI is explicit about the limit of its own number, noting that the estimates do not capture other consequences of turnover such as academic learning losses for students.
The reason this cost is structural rather than occasional is that most teacher movement is neither retirement nor growth. LPI's 2026 analysis of national data found that 15.1% of US teachers moved schools or left the profession between 2020-21 and 2021-22, split 8.0% who moved and 7.1% who left. Nearly three quarters of that movement, 74%, was voluntary and for reasons other than retirement. An earlier LPI estimate holds that about 90% of annual national demand for teachers is created by teachers leaving, not by enrolment growth.
Turnover is also concentrated where districts can least afford it. LPI found that 20.1% of teachers who were not fully certified left their schools, against 14.7% of those who were fully certified. Districts filling vacancies with underqualified staff are therefore buying a higher probability of filling the same vacancy again.
The national vacancy picture gives the scale. LPI's 2025 scan estimated about 411,500 teaching positions nationwide were either unfilled or filled by teachers not fully certified for their assignment, roughly 1 in 8 positions, affecting more than 6 million students.
A district can put its own number on this without a consultant. Multiply the LPI figure for your size band by the number of teachers you replaced last year. That product is a real recurring line in your budget, and it is the number any housing programme should be measured against.
Through four routes in practice: certificates of participation, general obligation bonds, low income housing tax credits, and partnership with a private or nonprofit developer who brings the capital. Across nine California developments studied by UC Berkeley and UCLA in 2025, most projects combined more than one of these rather than relying on a single source.
The Berkeley and UCLA researchers recorded which mechanism each project used. Certificates of participation financed Casa del Maestro entirely, and contributed to 705 Serramonte, The Alameda, Eastmoor Heights, The Sevens and The Acacia. General obligation bonds contributed to 705 Serramonte, The Alameda, Eastmoor Heights, The Sevens, Shirley Chisholm Village and 1701 San Pablo. Low income housing tax credits were the primary source for Sage Park and contributed to Shirley Chisholm Village and 1701 San Pablo.
The table below sets out how each mechanism works and what it demands of a district. The descriptions are structural. Project-specific costs are not included because the Berkeley and UCLA study did not publish total development cost or cost per unit for any of the nine developments, and we have not found a comparable published figure elsewhere.
| Mechanism | How it works | Typical timeline | Best fit | Tradeoffs |
|---|---|---|---|---|
| Certificates of participation (COPs) | The district finances construction against a lease of its own asset. No voter approval required in most states. | Shorter than a bond cycle because there is no election to wait for. | Districts that own suitable land and want to move without a ballot measure. | Debt sits with the district. Capacity used here is capacity not available for facilities. |
| General obligation bonds | Voter-approved debt repaid from property tax revenue. | Governed by the election calendar, so add the wait for the next ballot. | Districts with community support and available bonding capacity. | Requires winning a public vote on housing rather than classrooms. Slowest to start. |
| Low income housing tax credits (LIHTC) | Federal credits sold to investors fund construction in exchange for long-term affordability restrictions. | Competitive allocation rounds, so timing depends on the state agency cycle. | Projects that can meet income limits and want deep subsidy. | Income limits may exclude mid-career teachers. Compliance is permanent and administratively heavy. |
| Surplus land and ground lease | The district contributes land under a long lease and a partner develops on it. | Depends on the partner, not the district. | Districts that hold underused sites but no capital. | Requires suitable land, and the land use approvals are the slow part. |
| Private or nonprofit partnership | A developer finances, builds and operates. The district sets eligibility and approves terms, without taking on construction debt. | Depends on the partner's capital and the same permitting path. | Districts unwilling or unable to carry debt or operate housing. | Rents are set by project economics, so they are usually less deeply subsidised than tax credit projects. |
One finding cuts across every mechanism. The Berkeley and UCLA researchers reported that after financing, navigating the land use and building approval process was frequently cited by developers and districts as the biggest obstacle to building education workforce housing, and that navigating land use regulation and permitting was an onerous and uncertain process for most of the districts studied. Financing gets the attention. Permitting is what actually sets the calendar.
Buildings, at a documented scale, on a slow timeline. The best available evidence covers nine California developments delivering 50 to 144 units each, which took an average of 7.5 years from concept to occupancy. On retention, the outcome everyone wants to know about, exactly one district has published before-and-after figures.
The most rigorous public study of delivered projects is Housing California Educators: Insights from Nine Education Workforce Housing Developments, published in April 2025 by the Center for Cities + Schools at UC Berkeley and cityLAB at UCLA. Its nine developments and unit counts are The Alameda at 50, Eastmoor Heights at 56, Casa del Maestro at 70, Sage Park at 90, The Acacia at 110, 1701 San Pablo at 110, 705 Serramonte at 122, Shirley Chisholm Village at 135, and The Sevens at 144. Buildings run from two to six storeys at densities of 18 to 141 dwelling units per acre.
On rents, the finding is unambiguous and worth quoting precisely: all of the developments' one-bedroom units are affordable to the respective district's entry-level teachers, at close to or below 30% of their annual income.
Jefferson Union High School District in Daly City opened 705 Serramonte in 2022. In its 2023-24 Local Control and Accountability Plan the district reported certificated staff retention of 85% before the housing opened and 90% the following year. That is a five point improvement, self-reported, for one district, across one year, with no control for other changes in that period.
It is also the strongest published evidence in the country, which tells you how thin this literature is. Beyond it, the Berkeley and UCLA researchers recorded that several districts provided anecdotal evidence in interviews that turnover had declined after their project opened and that housing availability was persuasive with candidates. One survey finding is sharper: every employee for whom housing was part of a hiring offer listed it as a main reason they accepted the position. That is a recruitment finding, not a retention finding, and the two should not be merged.
Three things, per the same study. Most developments had longer construction timelines than the district expected, delayed by COVID-19 cost increases and work slowdowns or by city permitting and inspections. Land use approval, not financing, was most often named as the single biggest obstacle. And affordability erodes: Casa del Maestro, the oldest project in the study, has moved from roughly 50% of market rent at opening to about 80% of market today. A rent that is attainable at delivery is not automatically attainable in year fifteen.
Residents were not uniformly satisfied either. In the tenant survey, 36% were dissatisfied with storage and only 25% were extremely satisfied with safety. Average commute time was cut in half, though 85% of respondents drove alone both before and after moving in, so the gain is drive time rather than a shift to transit.
Precedent in Texas and Florida is at an earlier stage, and that is a finding rather than a gap. What exists is a run of documented board decisions. Hays CISD in Texas voted to proceed with attainable housing for district employees, covered by KXAN, KUT News, Community Impact and Yahoo News between January 2025 and June 2026. Lockhart ISD trustees approved a housing partnership for staff, reported by KXAN in April and May 2026. In Florida, the Marion County school board was reported by the Ocala Gazette in August 2025 as treating the question as not if but when. A district in Taylor, Texas was reported by the Taylor Press in April 2026 to have the question under consideration.
These are approvals and proposals, not delivered units or occupancy figures. For a superintendent in Texas or Tennessee the honest summary is that the playbook is established in California and the decisions are now being taken closer to home, but nobody in these states can yet show you a stabilised building with published retention data.
Seven steps, in this order, with two points where a board should be willing to stop. The sequence matters: districts that run a land inventory before a demand analysis often end up designing for the site they have rather than the workforce they need to house.
Multiply the Learning Policy Institute figure for your size band by the number of teachers you replaced last year. This is the baseline every later decision is measured against, and it takes an afternoon.
Compare local market rent to 30% of your actual starting salary and your salary at ten years. If neither is cost burdened, housing is not your retention lever and you should stop here.
Survey staff on interest, household size, and what rent they would actually pay. Colorado's statewide survey found 58% of more than 3,200 educators interested in district housing, but your own number is the one that sizes the building.
Identify district-owned sites, then assess zoning, utilities and likely community response for each. Permitting is the step most often named as the biggest obstacle, so screen for it early rather than discovering it in year three.
Decide whether the district will carry debt, seek voter approval, pursue tax credits, or contribute land to a partner instead. Each route has a different timeline and a different consequence for your facilities capacity.
Who qualifies, for how long, what happens when someone leaves the district, who is the landlord of record, and who handles maintenance at 11pm. These questions are cheap to answer on paper and expensive to answer after occupancy.
Decide in advance how rents will escalate relative to market and to your salary schedule. Casa del Maestro moved from about 50% of market rent at opening to about 80% today, which is the failure mode to design against.
Expect the whole sequence to take years rather than months. The nine California developments averaged 7.5 years from concept to occupancy, and about five years from the point the district signed a contract with a developer to the point staff moved in.
Candour
In four places, and we would rather name them than write around them. Anyone telling you the retention case is settled is overselling a literature that currently rests on one district's annual report.
1. Retention outcomes are barely measured. One district, Jefferson Union, has published before-and-after retention figures. Everything else in the published record is anecdotal or is a recruitment finding presented as a retention finding. Nobody has run a controlled study.
2. Cost per unit is not public. The Berkeley and UCLA study of nine developments does not publish total development cost or cost per unit for any project. We could not find comparable published figures elsewhere. Any per-unit number you see quoted in this field deserves a source check.
3. Tenant satisfaction data is narrower than it looks. The resident survey in that study covered three of the nine developments, and the report does not state how many people responded. Satisfaction findings should not be generalised to all nine, and we have not done so above.
4. Long-run affordability has one data point and it is a warning. Casa del Maestro is the only project old enough to show a rent trajectory, and it moved from roughly 50% of market to about 80% over its life. One case is not a trend, but it is the only case there is.
We will update this page as better evidence appears, and the last-updated date at the top will change when we do. If you believe a figure here is wrong, tell us and we will correct it and note the change.
Teacher housing is rental housing where tenancy is restricted to school district employees, usually at rents set below the local market. Restricting tenancy by employer requires legal authority. California established that authority through the Teacher Housing Act (SB 1413) in 2016.
The Learning Policy Institute estimates $11,860 in districts under 10,000 students, $16,450 in districts of 10,000 to 50,000, and $24,930 in districts above 50,000, in January 2024 dollars. These cover separation, recruitment, hiring and training. They exclude the effect on student learning, which LPI states plainly.
The published evidence is one district. Jefferson Union High School District reported certificated staff retention of 85% before its housing opened in 2022 and 90% the year after. Other districts have reported improvements anecdotally in interviews. No controlled study exists, so anyone claiming a settled retention effect is going beyond the evidence.
Four main routes: certificates of participation, general obligation bonds, low income housing tax credits, and partnership with a developer who brings the capital. Across nine California developments, most projects combined more than one source rather than relying on a single mechanism.
Nine California developments averaged 7.5 years from concept to occupancy, and about five years from the district signing a developer contract to staff moving in. That is comparable to other multifamily development in California. Land use approval, not financing, was most often named as the biggest obstacle.
No, but owning a suitable site materially changes the economics and the options. Districts without land can contribute other things a developer needs, including a defined tenant pool and a long-term commitment. Districts with land can contribute it under a ground lease rather than spending capital.
Interest is high where housing costs are high. A 2025 Keystone Policy Center survey of more than 3,200 Colorado educators found 58% interested in affordable housing options through their district. Of those who answered a further question about the district acting as landlord, 70% said they would be comfortable with it, though the report notes it received limited responses to that specific question.
Affordable housing generally serves households well below area median income and uses deep subsidy with strict income limits. Workforce housing targets households that earn too much to qualify for those programmes and too little to afford local market rent. Teacher housing is usually workforce housing with an employment restriction added.
Yes, and this is a real design risk. Low income housing tax credits impose income ceilings that can disqualify mid-career teachers, particularly those with a master's degree or a working partner. A project financed primarily through tax credits may end up serving early-career staff only.
Not automatically. Casa del Maestro in Santa Clara Unified, the oldest project with a public record, has moved from roughly 50% of market rent at opening to about 80% of market today. Districts should set an escalation policy relative to their own salary schedule at the outset rather than after occupancy.
The published California evidence covers districts of varying size, with delivered projects ranging from 50 to 144 units. Smaller districts have historically run much smaller programmes. We have not yet verified small-district outcome data to the standard we apply on this page, so we are not making a claim about it here.
There are documented board decisions but not yet stabilised buildings with published outcomes. Hays CISD and Lockhart ISD in Texas have approved staff housing partnerships, and the Marion County board in Florida has publicly moved toward it, all reported by local news outlets between 2025 and 2026. California remains the only state with a body of delivered projects to study.
Sources
We are not affiliated with, endorsed by, or in partnership with any organisation cited below.
Learning Policy Institute. (2017). What's the Cost of Teacher Turnover? Published 13 September 2017. Source for the more than $20,000 per new hire figure in urban districts, the 6 of 10 replacement figure, and the 20% beginning and 30% mid-career pay gap.
Learning Policy Institute. (2024). About 2024 Teacher Turnover Calculations. Published 17 September 2024. Source for $11,860, $16,450 and $24,930 per teacher replaced, in January 2024 dollars, and for the stated exclusion of student learning losses. Companion to 2024 Update: What's the Cost of Teacher Turnover?
Tan, T., Wei, W., Carver-Thomas, D., & García, E. (2026). Teacher Turnover in the United States: Who Moves, Who Leaves, and Why. Learning Policy Institute. Source for 15.1% turnover, the 8.0% and 7.1% split, 74% voluntary and non-retirement, and 20.1% against 14.7% by certification status.
Comai, S., Kemper Patrick, S., & Tan, T. (2025). 2025 Update: Latest National Scan Shows Teacher Shortages Persist. Learning Policy Institute blog, 15 July 2025. Source for 411,500 positions, 1 in 8, and more than 6 million students.
Carver-Thomas, D., & Darling-Hammond, L. (2017). Teacher Turnover: Why It Matters and What We Can Do About It. Learning Policy Institute. Source for the estimate that about 90% of annual national teacher demand is created by teachers leaving. Note this is a 2017 estimate.
Bowser, K. (2025). Priced Out: The Growing Challenge of Teacher Pay and Housing Costs. National Council on Teacher Quality, District Trendline, 8 May 2025. Source for home prices up 47%, rents up 51%, beginning salaries up 24% from 2019 to 2025; for the 72 district sample; for 39 of 72 and 34 of 72 unaffordable for beginning teachers; for 18 of 72 and 8 of 72 at ten years of experience; and for San Diego Unified at 45%.
Allegretto, S. (2025). The Teacher Pay Penalty Reached a Record High in 2024. Economic Policy Institute. Source for the 26.9% regression-adjusted weekly wage penalty and the 17.1% total compensation gap. These are weekly wages, not annual pay.
Hinkley, S., & Proussaloglou, E. (2025). Housing California Educators: Insights from Nine Education Workforce Housing Developments. Center for Cities + Schools, UC Berkeley, and cityLAB, UCLA. April 2025. Source for all nine unit counts, the 2 to 6 storey and 18 to 141 units per acre ranges, one-bedroom affordability at or below 30% of entry-level income, financing mechanism by project, the 7.5 year average concept to occupancy, the five year contract to move-in figure, the Jefferson Union 85% and 90% retention figures from that district's 2023-24 Local Control and Accountability Plan, the Casa del Maestro rent trajectory from about 50% to about 80% of market, the halved commute time, the 36% storage and 25% safety satisfaction figures, and the permitting and construction delay findings.
Gottlieb, A., Lagana, M., Spearman, K., & Schoales, V. (2025). We Can't Live Where We Teach: Teacher Perspectives. Keystone Policy Center, August 2025. Source for the survey of more than 3,200 Colorado educators, 58% interest in district housing, and the 70% comfortable with the district as landlord figure, which the report itself notes drew limited responses.
Texas and Florida board decisions reported by: Community Impact and KUT News on Hays CISD; KXAN on Lockhart ISD; Ocala Gazette on Marion County; Taylor Press on Taylor, Texas.
The first two steps in the process above take an afternoon and need nothing from us. If you would rather talk it through, or you want the affordability test run against your own salary schedule, we will do that.
Get in touch →Upward Communities develops workforce housing for public sector employees. We have a commercial interest in this subject and would rather say so here than have you find out later. Everything on this page is sourced so you can check it without taking our word for anything.
Monthly updates on Upward Communities.
Something went wrong. Please try again, or email info@upwardcommunities.com.
