On November 13, 2026 the 10-year Treasury yielded 5.15%. The San Francisco-Oakland apartment buildings that secure CMBS loans earn 4.9% of their appraised value. That metro ranks first of the 50 largest in the United States on a screen built for an apartment buyer who wants rising rents, tight vacancy and little new supply.
You are on the acquisitions team of an institutional investor that buys apartment buildings, and the first question on any new mandate is where to look. We asked Compound to rank the 50 largest metros and then to test the top five against what buildings there earn, what lenders appraised them at, what public REITs paid for theirs and what debt costs today. It ran on November 14, 2026 and returned one Excel workbook.
What the workbook shows:
- San Francisco-Oakland ranks first. Asking rent rose 11.8% in a year, the fastest of the 50 metros, and vacancy is the lowest at 3.4%.
- The same metro has the highest appraised value per unit of the top five. Its CMBS-financed buildings carry $509,461 of appraised value per unit and earn 4.9% of that value, 23 basis points below the 10-year Treasury.
- Income over appraised value and rent growth run in opposite directions across the top five. Detroit ranks fourth with 4.2% rent growth, and its CMBS buildings earn 7.5% of appraised value, 233 basis points above the Treasury.
- Austin ranks last. Asking rent there is 6.6% lower than three years ago, vacancy is 9.0%, and permits for apartment buildings run at 2.33 units per 1,000 residents.
- Long rates rose while SOFR did not. The 10-year Treasury is up 104 basis points in a year and the 30-year mortgage rate is 7.11%. SOFR is 3.90%, 8 basis points lower.
Each figure in the workbook names its source. Asking rents and home values come from Zillow’s ZORI and ZHVI indexes. Vacancy, concessions and time on market come from Apartment List. Fair Market Rents come from HUD. Jobs come from the Bureau of Labor Statistics. Property income and appraisals come from CMBS loan-level filings on SEC EDGAR, and REIT cost from Schedule III of each REIT’s 10-K. The rates are the Daily Treasury Par Yield Curve, SOFR from the New York Fed and the Freddie Mac mortgage survey. Permits for each metro come from the Census Bureau’s Building Permits Survey.
“I am screening U.S. metro areas for buying apartment buildings (multifamily acquisitions). Use the stored public data wherever it has the figure. For the 50 largest metros, build a screen with: asking rent level and 1-year and 3-year rent growth (Zillow ZORI); apartment vacancy, concession rate and time on market with their change over the past year (Apartment List); HUD 2-bedroom Fair Market Rent next to market rent; typical home value against annual rent as a gauge of how expensive owning is versus renting (Zillow ZHVI); new multifamily supply from building permits for buildings with 5 or more units if the stored data has it; and job growth. Add the financing backdrop: 10-year Treasury yield, SOFR and the 30-year mortgage rate, now and a year ago. Score and rank the metros for a buyer who wants rising rents, tight vacancy and limited new supply. Then, for the top 5 metros, pull actual multifamily loans from CMBS loan-level filings (property name, units, occupancy, net operating income, debt service coverage, appraised value and value per unit) to show what apartment buildings there earn and what they are valued at. Deliver an Excel workbook: a Screen sheet with every metro and the composite score, a sheet of the CMBS multifamily properties for the top 5 metros, a Financing sheet, and a one-page Summary of where to look and why, with what the data cannot tell me. Cite every figure.”
That is the request as we sent it. The REIT sheet came from a second message on the same task, quoted further down.
How a multifamily market screen scores 50 metros
A market screen turns each measure into a distance from the 50-metro average, weights the measures and adds them up. The weights are the buyer’s view of what matters, so they sit in cells you can change.
Compound chose eight measures from the request and wrote the weights on the Screen sheet, in cells D6 to D13.
| Measure | Source | Weight | Better when |
|---|---|---|---|
| 1-year asking rent growth | Zillow ZORI | 20% | Higher |
| 3-year asking rent growth | Zillow ZORI | 10% | Higher |
| Apartment vacancy rate | Apartment List | 15% | Lower |
| 1-year change in vacancy | Apartment List | 5% | Falling |
| Concession rate | Apartment List | 5% | Lower |
| Time on market, days | Apartment List | 5% | Lower |
| Permits for 5+ unit buildings per 1,000 residents | Census Bureau | 25% | Lower |
| 1-year nonfarm job growth | BLS | 15% | Higher |
Rent growth carries 30% of the score, vacancy and leasing 30%, new supply 25% and jobs 15%. Each measure becomes a z-score: the metro’s value less the average of the 50 metros, divided by the standard deviation. A metro with a missing value scores 0 on that measure, which is the average.
The composite score is the weighted sum. A score of 0 is an average metro. San Francisco-Oakland scores 1.64 and Austin scores -1.14.
Composite score: the ten highest and five lowest of 50 metros
Weighted sum of eight z-scores. 0 is the 50-metro average
Compound's Screen tab, column AK. Rents: Zillow ZORI, September 2026. Vacancy, concessions and time on market: Apartment List, October 2026. Permits: Census Building Permits Survey, January to September 2026. Jobs: BLS Current Employment Statistics, September 2026.
Compound said in its answer that it chose the weights and that the ranks update when the weights change. We would treat that as an invitation. A buyer who cares most about supply can raise the permit weight and watch which metros move. A buyer of older, cheaper buildings might drop job growth and add a measure of rent against income.
Multifamily markets ranked: the top 10 and the bottom 5
The top ten are in California, the Midwest and on the East Coast, and eight of them permitted fewer than one apartment unit per 1,000 residents (a count we made from column Y of the Screen sheet). The bottom five are in Texas, Colorado, Virginia and North Carolina, where permit rates are among the highest of the 50 metros.
| B | C | I | J | K | M | N | Q | Y | AB | AK | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 18 | Rank | Metro (CBSA) | 1-yr rent growth | 3-yr rent growth | Vacancy Oct-2026 | Vacancy chg (pp) | Concession rate | Time on market (days) | 5+ permits per 1,000 | Job growth 1-yr | Composite score |
| 19 | 1 | San Francisco-Oakland-Fremont, CA | 11.8% | 19.7% | 3.4% | (1.3%) | 8.7% | 25.2 | 1.10 | 0.5% | 1.64 |
| 20 | 2 | Virginia Beach-Chesapeake-Norfolk, VA-NC | 7.1% | 16.4% | 3.6% | (1.4%) | 4.0% | 15.0 | 0.85 | (0.4%) | 1.19 |
| 21 | 3 | San Jose-Sunnyvale-Santa Clara, CA | 8.3% | 16.2% | 4.1% | (1.2%) | 6.8% | 30.0 | 1.63 | 1.8% | 1.18 |
| 22 | 4 | Detroit-Warren-Dearborn, MI | 4.2% | 11.8% | 4.6% | (0.2%) | 6.9% | 23.7 | 0.40 | (0.1%) | 0.77 |
| 23 | 5 | Milwaukee-Waukesha, WI | 5.1% | 13.5% | 5.1% | (1.6%) | 8.6% | 0.68 | 0.0% | 0.76 | |
| 24 | 6 | Fresno, CA | 3.5% | 9.5% | 0.33 | 2.4% | 0.75 | ||||
| 25 | 7 | Providence-Warwick, RI-MA | 4.2% | 14.5% | 4.7% | (0.1%) | 12.6% | 0.61 | 0.1% | 0.65 | |
| 26 | 8 | Chicago-Naperville-Elgin, IL-IN | 4.7% | 15.9% | 5.9% | 0.6% | 4.9% | 37.3 | 0.39 | (0.0%) | 0.61 |
| 27 | 9 | St. Louis, MO-IL | 4.1% | 13.1% | 6.4% | (0.2%) | 23.7% | 31.7 | 0.69 | 0.9% | 0.56 |
| 28 | 10 | Grand Rapids-Wyoming-Kentwood, MI | 4.2% | 12.1% | 5.9% | (0.1%) | 13.0% | 0.46 | 0.0% | 0.53 | |
| 64 | 46 | Dallas-Fort Worth-Arlington, TX | 0.7% | (0.3%) | 8.7% | (0.0%) | 42.7% | 39.7 | 1.95 | 0.8% | (0.84) |
| 65 | 47 | Denver-Aurora-Centennial, CO | 0.0% | (1.3%) | 7.9% | (0.9%) | 46.7% | 39.7 | 1.80 | 0.1% | (0.86) |
| 66 | 48 | Richmond, VA | 2.5% | 11.3% | 7.1% | 0.5% | 18.0% | 3.00 | (1.3%) | (0.94) | |
| 67 | 49 | Raleigh-Cary, NC | 1.4% | 1.1% | 8.1% | (0.0%) | 44.7% | 41.3 | 3.27 | 2.1% | (0.98) |
| 68 | 50 | Austin-Round Rock-San Marcos, TX | 0.7% | (6.6%) | 9.0% | (0.7%) | 60.7% | 42.0 | 2.33 | 0.9% | (1.14) |
San Francisco-Oakland has the fastest one-year and three-year rent growth and the lowest vacancy of all 50 metros, with concessions and time on market both falling. Its concession rate fell from 16.5% to 8.7% in a year, and a listing now takes 25.2 days to lease, down from 30.8.
Virginia Beach-Norfolk is the surprise at second. Compound’s summary gives the reasons: the third-fastest one-year and second-fastest three-year rent growth, the second-lowest vacancy at 3.6%, the lowest concession rate, and the shortest time on market of the metros Apartment List covers, at 15.0 days. Jobs there fell 0.4% over the year.
San Jose is third on 8.3% rent growth and 1.8% job growth, held back by permits of 1.63 units per 1,000 residents, above the top-five norm. Detroit and Milwaukee are fourth and fifth with rent growth of 4.2% and 5.1%, low vacancy, few permits and flat jobs.
Two rows need a caution. Fresno ranks sixth, and Apartment List does not cover it, so it scores average on vacancy, concessions and time on market. Its rank rests on few permits and 2.4% job growth. Milwaukee, Providence and Grand Rapids have no time-on-market figure and score average on that measure.
The bottom five are Dallas-Fort Worth, Denver, Richmond, Raleigh and Austin. Raleigh permitted 3.27 apartment units per 1,000 residents in nine months, the most of the 50 metros, and Richmond 3.00. Austin’s asking rent is 6.6% below its September 2023 level, the largest three-year fall on the sheet, with vacancy at 9.0% and a 60.7% concession rate.
There is a fair case for the bottom of the list. Raleigh’s jobs grew 2.1% and Austin’s 0.9%, and a metro where rents have already fallen can be a cheaper place to buy. The screen was asked to find rising rents and limited supply, and it ranks on that request. A buyer with a five-year hold and a view that Sun Belt supply is ending would weight it differently.
Asking rent, HUD Fair Market Rent and the cost of owning
Rent growth says where rents are moving. The level says what tenants already pay, and two comparisons on the sheet put the level in context: the HUD Fair Market Rent and the price of a home.
| C | F | G | H | T | U | V | W | |
|---|---|---|---|---|---|---|---|---|
| 18 | Metro (CBSA) | ZORI Sep-2026 ($/mo) | ZORI Sep-2025 ($/mo) | ZORI Sep-2023 ($/mo) | HUD 2BR FMR FY27 ($/mo) | ZORI / FMR | ZHVI Sep-2026 ($) | Home value / annual rent (x) |
| 19 | San Francisco-Oakland-Fremont, CA | $3,445 | $3,082 | $2,878 | $3,697 | 0.93 | $1,120,379 | 27.1x |
| 20 | Virginia Beach-Chesapeake-Norfolk, VA-NC | $1,886 | $1,761 | $1,620 | $1,667 | 1.13 | $371,181 | 16.4x |
| 21 | San Jose-Sunnyvale-Santa Clara, CA | $3,812 | $3,519 | $3,282 | $3,272 | 1.17 | $1,523,615 | 33.3x |
| 22 | Detroit-Warren-Dearborn, MI | $1,514 | $1,453 | $1,354 | $1,459 | 1.04 | $264,316 | 14.6x |
| 23 | Milwaukee-Waukesha, WI | $1,549 | $1,474 | $1,365 | $1,263 | 1.23 | $384,496 | 20.7x |
Zillow’s ZORI puts the typical asking rent in San Francisco-Oakland at $3,445 a month in September 2026, against $3,082 a year earlier and $2,878 in September 2023. San Jose is higher at $3,812. Detroit and Milwaukee are near $1,500, at $1,514 and $1,549.
The HUD Fair Market Rent is the 40th-percentile gross rent for a two-bedroom unit, utilities included, and housing voucher payments are based on it. In San Francisco the asking rent index is 0.93 of the Fair Market Rent of $3,697. In Milwaukee it is 1.23 of a $1,263 Fair Market Rent. Compound’s summary calls the ratio a rough gauge because the two figures measure different units, and we agree. It is still useful to a buyer of workforce housing, who needs to know whether voucher rents sit above or below the market.
The last column divides the typical home value by a year of rent. A typical San Jose home costs 33.3 years of rent and a San Francisco-Oakland home 27.1 years. In Detroit the figure is 14.6 years. Where owning costs that much more than renting, fewer renters leave to buy, which supports occupancy. The same figure warns that land is expensive, and expensive land shows up in the price of the building.
New supply: permits for 5+ unit buildings per 1,000 residents
New supply is the largest single weight in the score at 25%, and it is measured from the Census Bureau’s count of apartment units permitted in each metro.
Compound took permits for each metro from the Census Building Permits Survey year-to-date file on census.gov: units authorized in buildings with five or more units, January to September 2026. It named that file in its answer and on the sheet. It then divided by the July 2025 population estimate so that New York’s 32,968 units and Fresno’s 399 can be compared.
Apartment units permitted per 1,000 residents, January to September 2026
Units in buildings with 5 or more units. The top five metros of the screen, with the five lowest and five highest of the 50. The number before each name is the metro's rank on the screen
Compound's Screen tab, column Y: U.S. Census Bureau Building Permits Survey, year to date through September 2026, over Census population estimates for July 1, 2025.
Detroit has one of the five lowest rates at 0.40 units per 1,000 residents: 1,764 units in a metro of 4.39 million people. Virginia Beach-Norfolk and Milwaukee are below 1.00. San Francisco-Oakland, at 1.10, is close to the middle of the 50 metros, and San Jose at 1.63 is above it. The two Bay Area metros rank high on rents and vacancy, and supply holds them back a little.
Permits lead deliveries by a year or more, which is the main weakness of this input and is covered in the limits below.
What apartment buildings earn: CMBS loan-level evidence
A screen ranks markets on rents and vacancy. Loan filings show what individual buildings in those markets earn and what lenders valued them at, property by property.
Every CMBS deal files a monthly asset data file with the SEC that lists each loan and property: units, occupancy, net operating income, debt service coverage and appraised value. Compound took up to ten of the largest multifamily properties in each of the top five metros, 41 in all, from reports dated July 2026 or later.
| C | D | E | F | G | H | I | |
|---|---|---|---|---|---|---|---|
| 6 | Metro | Properties | Units | Avg occupancy | Value per unit ($, aggregate) | Annualized NOI / value | Avg DSCR (NOI, x) |
| 7 | San Francisco-Oakland-Fremont, CA | 10 | 6,257 | 94.2% | $509,461 | 4.9% | 1.96x |
| 8 | Virginia Beach-Chesapeake-Norfolk, VA-NC | 9 | 2,190 | 94.8% | $136,233 | 6.3% | 1.63x |
| 9 | San Jose-Sunnyvale-Santa Clara, CA | 2 | 838 | 99.0% | $413,365 | 5.7% | 5.52x |
| 10 | Detroit-Warren-Dearborn, MI | 10 | 2,796 | 95.4% | $143,124 | 7.5% | 1.85x |
| 11 | Milwaukee-Waukesha, WI | 10 | 1,595 | 99.3% | $197,085 | 6.5% | 1.26x |
The San Francisco-Oakland sample is ten properties with 6,257 units, 94.2% occupied on average. Their annualized net operating income is 4.9% of appraised value, and the appraised value is $509,461 a unit. Detroit’s ten properties earn 7.5% of an appraised $143,124 a unit. Milwaukee’s ten are 99.3% occupied and earn 6.5%.
CMBS multifamily NOI over appraised value, against today's rates
Aggregate annualized NOI over appraised value of the listed properties in each metro. Appraisals date from 2017 to 2026
Compound's Financing tab, rows 15 to 19, from SEC Form ABS-EE asset data files; Daily Treasury Par Yield Curve; Freddie Mac Primary Mortgage Market Survey.
The property rows explain the aggregates and show their limits.
| D | E | G | H | M | N | O | P | Q | S | T | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 14 | Property | City | Units | Occupancy | Annualized NOI ($) | DSCR (NOI, x) | Appraised value ($) | Value date | Value per unit ($) | NOI / value | Figures basis |
| 15 | Parkmerced | San Francisco | 3,221 | 94.2% | 60,469,764 | 4.06x | 1,430,000,000 | Feb 26, 2026 | 443,962 | 4.2% | At securitization (underwritten) |
| 16 | THE GATEWAY | San Francisco | 1,254 | 97.0% | 36,271,818 | 2.91x | 868,800,000 | Feb 1, 2018 | 692,823 | 4.2% | Servicer most recent |
| 17 | Nema San Francisco | San Francisco | 754 | 97.5% | 36,297,476 | 1.34x | 328,800,000 | Oct 23, 2023 | 436,074 | 11.0% | Servicer most recent |
| 18 | THE COVE AT TIBURON | Tiburon | 283 | 87.0% | 9,668,733 | 1.21x | 228,000,000 | Jun 27, 2025 | 805,654 | 4.2% | Servicer most recent |
| 19 | VENTANA RESIDENCES | San Francisco | 193 | 96.0% | 4,324,188 | 0.94x | 112,000,000 | Oct 28, 2024 | 580,311 | 3.9% | Servicer most recent |
| 25 | Cottage Grove Apartments | Newport News | 552 | 92.0% | 3,078,785 | 1.13x | 66,800,000 | Feb 3, 2025 | 121,014 | 4.6% | Servicer most recent |
| 26 | CHRISTOPHER CROSSING APARTMENTS | Newport News | 531 | 93.0% | 4,717,032 | 1.68x | 58,100,000 | Oct 9, 2024 | 109,416 | 8.1% | Servicer most recent |
| 33 | STERLING OAKS | Norfolk | 160 | 98.0% | 512,634 | 1.71x | 7,500,000 | Jul 12, 2017 | 46,875 | 6.8% | Servicer most recent |
| 34 | HERITAGE PARK APARTMENTS | Sunnyvale | 508 | 99.0% | 12,452,917 | 6.23x | 203,900,000 | Aug 10, 2020 | 401,378 | 6.1% | Servicer most recent |
| 35 | WILLOWBEND APARTMENTS | Sunnyvale | 330 | 99.0% | 7,143,380 | 4.80x | 142,500,000 | Sep 6, 2018 | 431,818 | 5.0% | Servicer most recent |
| 36 | Edge at Novi | Novi | 264 | 96.2% | 3,677,486 | 1.24x | 64,400,000 | Mar 16, 2026 | 243,939 | 5.7% | At securitization (underwritten) |
| 37 | INFINITY PARK APARTMENTS | Detroit | 480 | 99.0% | 3,591,995 | 1.66x | 53,800,000 | Jul 9, 2025 | 112,083 | 6.7% | At securitization (underwritten) |
| 39 | Diamond Forest Apartments | Farmington Hills | 264 | 94.3% | 6,538,988 | 3.45x | 38,800,000 | Feb 21, 2018 | 146,970 | 16.9% | Servicer most recent |
| 40 | The Kensley | Howell | 136 | 94.9% | 2,151,904 | 1.27x | 37,575,000 | Mar 24, 2026 | 276,287 | 5.7% | At securitization (underwritten) |
| 46 | Village 43 | Brown Deer | 156 | 95.5% | 2,746,765 | 1.30x | 46,600,000 | Jul 30, 2026 | 298,718 | 5.9% | At securitization (underwritten) |
| 47 | Wisconsin Avenue Portfolio | Milwaukee | 244 | 100.0% | 2,155,606 | 1.17x | 36,050,000 | Nov 28, 2023 | 147,746 | 6.0% | Servicer most recent |
| 48 | The Emree | West Bend | 105 | 97.1% | 1,979,106 | 1.36x | 32,400,000 | Jun 23, 2026 | 308,571 | 6.1% | At securitization (underwritten) |
| 55 | ATKINSON PORTFOLIO | Milwaukee | 143 | 100.0% | 1,271,311 | 1.13x | 24,900,000 | Jul 8, 2024 | 174,126 | 5.1% | Servicer most recent |
Parkmerced is 3,221 of San Francisco’s 6,257 units. Its $60.5 million of net operating income is the figure underwritten at securitization, and its $1.43 billion value is dated February 2026. The Gateway’s 1,254 units carry an $868.8 million appraisal from February 2018, or $692,823 a unit, against income annualized from the first half of 2026. Two buildings drive the San Francisco aggregate, and their income and value dates are years apart.
Coverage tells a second story. Ventana Residences in San Francisco covers its debt service 0.94 times, which means income is below the loan payments. Milwaukee’s buildings are full and cover their debt between 1.13 and 1.36 times. Cottage Grove in Newport News covers 1.13 times. The two Sunnyvale properties in the San Jose sample cover 6.23 and 4.80 times. The San Jose sample has only those two properties, so its 5.7% rests on 838 units.
Detroit’s 7.5% needs the most care. Diamond Forest in Farmington Hills shows income of 16.9% of a 2018 appraisal. The three Detroit properties appraised in 2026, Edge at Novi, Chestnut Woods and Chestnut Crossing, and The Kensley, earn 5.7% to 6.0%. We would read the newer appraisals as closer to what a buyer pays today, and they put Detroit between 55 and 90 basis points above the Treasury (our arithmetic), well under the 233 the aggregate suggests.
The financing backdrop: Treasury, SOFR and mortgage rates
The benchmark under fixed-rate debt, the 10-year Treasury yield, is about one percentage point higher than a year ago. SOFR, the benchmark under floating-rate debt, is slightly lower.
| B | C | D | E | F | G | |
|---|---|---|---|---|---|---|
| 5 | RATES | Latest date | Latest (%) | Year-ago date | Year ago (%) | Change (bp) |
| 6 | 10-year U.S. Treasury par yield | Nov 13, 2026 | 5.15 | Nov 13, 2025 | 4.11 | 104 |
| 7 | SOFR (Secured Overnight Financing Rate, overnight fixing) | Nov 12, 2026 | 3.90 | Nov 12, 2025 | 3.98 | (8) |
| 8 | 30-year fixed mortgage rate (U.S. average) | Nov 12, 2026 | 7.11 | Nov 6, 2025 | 6.22 | 89 |
| 10 | SPREADS | Latest (bp) | Year ago (bp) | Change (bp) | ||
| 11 | 30-yr mortgage less 10-yr Treasury | 196 | 211 | (15) | ||
| 12 | 10-yr Treasury less SOFR (curve slope) | 125 | 13 | 112 | ||
| 14 | CMBS MULTIFAMILY YIELDS VS. 10-YR TREASURY (top 5 metros) | NOI / value | Spread to 10-yr (bp) | |||
| 15 | San Francisco-Oakland-Fremont, CA | 4.9% | (23) | |||
| 16 | Virginia Beach-Chesapeake-Norfolk, VA-NC | 6.3% | 110 | |||
| 17 | San Jose-Sunnyvale-Santa Clara, CA | 5.7% | 51 | |||
| 18 | Detroit-Warren-Dearborn, MI | 7.5% | 233 | |||
| 19 | Milwaukee-Waukesha, WI | 6.5% | 137 |
The 10-year Treasury par yield was 5.15% on November 13, 2026, up from 4.11% a year earlier. Freddie Mac’s 30-year fixed mortgage rate was 7.11%, up from 6.22%. SOFR was 3.90%, against 3.98%. The gap between the 10-year yield and SOFR widened from 13 to 125 basis points.
For a buyer, the comparison that matters is the yield on the purchase price against the cost of the loan. Apartment loans from the agencies and from CMBS price at a spread over the Treasury, so the Treasury yield is the floor on fixed-rate debt and the true cost is higher. San Francisco-Oakland’s CMBS-financed buildings earn 4.9% of appraised value, 23 basis points below that floor. A buyer who paid those appraised values and borrowed at a fixed rate would start with negative leverage: the debt would cost more than the building earns. The 4.9% divides reported income by appraisals, some of them years old, so it is not the yield on a purchase made today. The actual spread depends on the purchase price and the loan terms. On the same measure, Virginia Beach-Norfolk sits 110 basis points above the Treasury, Milwaukee 137 and San Jose 51.
The request asked for benchmark rates. Loan spreads, proceeds and terms come from a lender’s quote, and a floating-rate loan would be priced off the SOFR forward curve with a cap.
REIT cost basis against CMBS appraised value
Public apartment REITs list every property they own in Schedule III of the 10-K, with what it cost. In the two Bay Area metros, the REITs’ historical cost per unit is 15% to 20% lower than the per-unit values CMBS lenders appraised.
We asked for this as a follow-up on the same task.
| C | D | E | F | G | H | I | J | |
|---|---|---|---|---|---|---|---|---|
| 6 | Metro | REIT properties | With units given | Units | Gross cost of rows with units ($) | REIT gross cost per unit ($) | CMBS appraised value per unit ($) | REIT cost / CMBS value |
| 7 | San Francisco-Oakland-Fremont, CA | 105 | 89 | 25,572 | 10,366,268,384 | 405,376 | 509,461 | 0.80x |
| 8 | Virginia Beach-Chesapeake-Norfolk, VA-NC | 3 | 0 | 0 | 0 | n/a | 136,233 | n/a |
| 9 | San Jose-Sunnyvale-Santa Clara, CA | 58 | 58 | 17,756 | 6,230,889,045 | 350,917 | 413,365 | 0.85x |
| 10 | Detroit-Warren-Dearborn, MI | 0 | 0 | 0 | 0 | n/a | 143,124 | n/a |
| 11 | Milwaukee-Waukesha, WI | 0 | 0 | 0 | 0 | n/a | 197,085 | n/a |
In San Francisco-Oakland, five REITs report 105 apartment properties. Unit counts are given for 89 of them: 25,572 units at $405,376 of gross cost each, against the $509,461 a unit that CMBS lenders appraised. In San Jose, three REITs report 58 properties with 17,756 units at $350,917 each, against $413,365.
Per-unit value: CMBS appraisals against REIT gross cost
Dollars per apartment unit. REIT gross cost is historical cost from fiscal 2025 10-Ks
Compound's REIT Holdings tab, rows 7 and 9: Schedule III of each REIT's fiscal 2025 10-K, and SEC Form ABS-EE asset data files.
The sheet’s rows, summed by owner, show how much the figure depends on when each REIT bought or built. These sums are our arithmetic on the sheet’s rows.
| Metro | Owner | Properties | With units | Units | Gross cost per unit |
|---|---|---|---|---|---|
| San Francisco-Oakland-Fremont | Apartment Investment & Management Co (Aimco) | 2 | 0 | no units given | |
| San Francisco-Oakland-Fremont | AvalonBay Communities | 30 | 30 | 8,640 | $345,587 |
| San Francisco-Oakland-Fremont | Equity Residential | 25 | 24 | 6,940 | $458,179 |
| San Francisco-Oakland-Fremont | Essex Property Trust | 35 | 35 | 9,992 | $420,400 |
| San Francisco-Oakland-Fremont | UDR | 13 | 0 | no units given | |
| Virginia Beach-Chesapeake-Norfolk | Mid-America Apartment Communities (MAA) | 3 | 0 | no units given | |
| San Jose-Sunnyvale-Santa Clara | AvalonBay Communities | 12 | 12 | 4,727 | $271,967 |
| San Jose-Sunnyvale-Santa Clara | Equity Residential | 10 | 10 | 3,053 | $323,452 |
| San Jose-Sunnyvale-Santa Clara | Essex Property Trust | 36 | 36 | 9,976 | $396,732 |
Equity Residential’s 855 Brannan in San Francisco has 449 units at $326.5 million of gross cost, about $727,217 a unit. Essex’s Palm Valley in San Jose has 1,100 units at $490.7 million, about $446,088 a unit.
The other three metros read differently. In Virginia Beach-Norfolk, MAA owns three properties, in Chesapeake, Hampton and Newport News, and its Schedule III gives no unit counts, so there is no per-unit figure. In Detroit and Milwaukee, none of the twelve public apartment REITs Compound checked reports a property. An institutional buyer there competes with private owners, and the public comparables that exist for the Bay Area do not exist for those two metros.
Gross cost is what the REIT paid or spent to build, plus capital spent since. It is historical cost, and it can sit above or below what a building is worth today. Many of these buildings were bought or built ten or more years ago. The CMBS figures are appraisals, mostly from loan origination. Compound’s summary calls the gap a rough gauge, and we read it the same way: the Schedule III figures of about $350,000 to $405,000 a unit record what long-term owners spent on Bay Area apartments, before depreciation.
What the public data cannot tell you
The screen says where to look. Pricing a building takes figures the public sources do not publish, and Compound listed them on the Summary sheet at our request.
- Rents. ZORI is a typical asking rent on new leases across all rental homes, single-family houses included. It does not show in-place rents, renewals or loss-to-lease at a building, and a metro average hides differences between submarkets and building classes.
- Vacancy, concessions and time on market. Apartment List covers only the properties listed on its site. Fresno has none of the three measures, and nine metros have no time-on-market figure. A missing figure scores as average.
- Supply. Permits count units authorized from January to September 2026. They do not count units under construction or delivered. Most completions in the next two years come from permits issued in 2023 to 2025, which the screen does not count. Supply is scaled per resident, and a scale per existing apartment would rank some metros differently.
- Values. The public sources used here publish no transaction prices and no market cap rates. CMBS values are lender appraisals dated from 2017 to 2026. Net operating income is the servicer’s latest figure or the underwritten one, and a partial-year figure is annualized. The sample leans to larger buildings that could be financed in a CMBS deal. Three large properties are left out of the sample because their names did not match in the filing text.
- REIT cost. Schedule III is historical cost. Properties a REIT did not tag one by one in its filing data are missing, among them 18 Essex lines, and UDR and Aimco give no unit counts.
- Costs and rules. Property taxes, insurance, operating costs and local rent regulation are absent, and they can change returns more than rent growth does.
- Jobs and comparability. BLS metro job figures are preliminary and get revised. The HUD Fair Market Rent measures a different unit from ZORI. Every figure describes the past.
- Debt. The rates are benchmarks, without the loan spreads, proceeds and terms of a specific deal.
Where we would look, and what would change it
We would take the screen’s five metros as two different trades and send an analyst into each with a different question.
The Bay Area trade is growth at a low income yield on appraised value. Rents are rising fastest there, vacancy is lowest, and the lenders’ appraisals put San Francisco-Oakland buildings at $509,461 a unit. A buyer who paid those appraised values would earn less than the Treasury yield and would need rent growth to continue, so the question for San Francisco-Oakland and San Jose is how long 8% to 12% rent growth can last. San Francisco’s asking rent is 19.7% above September 2023.
The Detroit, Milwaukee and Virginia Beach trade is yield with slower growth. CMBS-financed buildings earn, on appraised value, roughly half a point to a point and a half over the Treasury, few apartments are being permitted, and jobs are flat. The question there is whether income can grow without job growth, and in Milwaukee whether coverage of 1.1 to 1.4 times leaves room for a rate reset at refinancing.
The other side deserves a hearing. The screen’s inputs are metro averages of asking rents, its appraisals are old, and its bottom-ranked metros are where prices may have fallen most. A buyer who believes Sun Belt deliveries peak in 2026 could reasonably start at the bottom of the list.
Three things would change our view. Broker cap rates and recent sales for each metro would replace the appraisal-based yields. Counts of units under construction would replace permits. Submarket rents by building class would show whether the metro average describes the buildings we would buy. From there the work moves to one property: a real estate market analysis of its trade area and submarket, the rent roll and the pro forma. To triage the deals a market turns up, real estate deal pipeline screening ranks a few dozen on shared loan terms, and a value-add multifamily screen tests one renovation deal on a single sheet.
U.S. metro multifamily screen (Excel)Compound’s workbook from this run: a one-page Summary, the Screen of 50 metros with every input and the score weights, 41 CMBS multifamily properties in the top five metros, the Financing rates, and the apartment properties public REITs report in those metros.Download · FreeFAQ
What measures belong in a multifamily market screen?
Rent growth, vacancy, new supply and job growth are the core four, each from a source that publishes every metro on the same definition. This screen adds concessions, time on market, the HUD Fair Market Rent and home value against rent. Rent against household income and population growth are common additions.
Is CMBS net operating income over appraised value a cap rate?
It is a yield on an appraisal, and the appraisal can be years old. A market cap rate comes from a sale price and the income at the time of sale. In this workbook the appraisals date from 2017 to 2026, so the ratio overstates the yield wherever values have risen since.
Why does an asking rent index differ from the rents in a rent roll?
The Zillow Observed Rent Index (ZORI) tracks asking rents on units listed for new leases across a whole metro, houses and apartments together. A rent roll shows what tenants in one building pay under leases signed over the past year or more. The index shows the direction and pace of market asking rents, and it does not measure a building’s loss-to-lease. Loss-to-lease is the market rent for comparable units less the in-place rent on the rent roll, so a buyer underwrites it from rent comparables for the same unit types.
Why measure permits per 1,000 residents?
A raw permit count ranks metros by size: New York permitted 32,968 apartment units in nine months and Fresno 399. Dividing by population puts them on one scale, 1.64 against 0.33 per 1,000 residents. Dividing by the existing apartment stock is a sharper measure where that count is available.
What is negative leverage in an apartment acquisition?
Negative leverage means the interest rate on the loan is higher than the yield the building earns on its price. Each borrowed dollar then lowers the cash return, and the purchase depends on income growth or a later sale at a higher price. San Francisco-Oakland’s CMBS-financed buildings earn 4.9% of appraised value against a 5.15% Treasury yield, so a buyer paying those appraised values would start with negative leverage before any loan spread. The actual spread depends on the purchase price and the loan terms.
How often should a market screen be rerun?
Monthly is enough for the market inputs. Zillow, Apartment List, the Census permits survey and BLS each publish monthly, and Treasury yields and SOFR publish every business day. CMBS servicer reports are also monthly, and REIT Schedule III changes once a year with the 10-K.





