Real estate market analysis example: the model's rent is too high

You are the associate on the buy side of Cedar Fork Commerce Park, the 248,400 square foot industrial and flex park in Lewisville, Texas from our real estate pro forma example. The broker asks $48.0M. Our model bids $46.4M for an 8.0% unlevered return, and it leases every suite that rolls at the appraisal’s market rent: $14.50 a square foot for small bays and $12.75 for large bays in 2027, triple net. It holds general vacancy and credit loss at 1.0%, on top of six months of downtime when a tenant leaves.

Before the bid goes to the investment committee, someone has to ask whether the market supports those inputs. That is the job of a market analysis. We gave Compound the underwriting model and asked it to build the trade area from Census data, read the submarket from the brokers’ Q3 2026 reports, and tell us whether the model’s rent and vacancy sit inside what the market shows. It ran once, on November 3, 2026.

The answer: the model’s rent sits above the submarket and its vacancy sits below it. Compound would raise general vacancy to 3.0% and run a downside at lower rents.

Cedar Fork Commerce Park, its tenants and its underwriting model are illustrative: a made-up property built to read like a real deal. The Census, BLS and broker figures are real public data. The workbook and the one-page slide on this page come from one real Compound run, quoted unedited.

How to do a real estate market analysis

A real estate market analysis tests a property’s income against the market it competes in.

It works outward from the site in five steps, and Compound’s run followed them in order.

1. Fix the site and draw the trade area. The usual rings are 1, 3 and 5 miles, with population, households, income, the labor force, education, commuting and housing inside each.
2. Compare the trade area with its city, county, metro, state and the US, so each figure reads as high or low.
3. Read growth and the labor market from series with history: population, jobs and wages.
4. Read the submarket for the property type: vacancy, asking rent, net absorption and space under construction, from each broker that publishes it.
5. Tie the result back to the underwriting model. Do the model’s market rent and vacancy sit inside the market’s range, and if not, what changes?

The weight on each step depends on the property. An industrial park leans on labor and the construction pipeline, a multifamily deal on household growth and rent against income, and a retail center on spending in the trade area.

The address on Cedar Fork’s OM is invented, so it cannot be geocoded. Compound anchored every ring on a real point instead, I-35E and Corporate Drive in south Lewisville, and wrote that on the first sheet of the workbook and on the slide.

Population by radius: 1, 3 and 5-mile demographics around the site

The first sheet is the ring table, the part of a market analysis most readers look for first.

Within 3 miles of the site live 96,722 people in 38,546 households, with an approximate median household income of $101,091 and 4.1% unemployment.

Compound took the 2026 TIGER/Line block group boundaries around the site and intersected them with circles of 1, 3 and 5 miles. Before it took any shares, it cut out water with the Census hydrography layer: Lewisville Lake, Grapevine Lake and smaller water bodies. A block group then counts by the share of its land area inside the ring. If 40% of a block group’s land is inside the 3-mile circle, 40% of its people and households are counted.

The other common method counts a block group whole if its center point falls inside the ring and not at all if it falls outside. Compound kept that centroid count as a check in rows 43 to 45. At 3 and 5 miles the two methods agree within 390 people. At 1 mile, where only 23 block groups touch the circle, the centroid method gives 16,380 people, 2,282 fewer than the area share. A small ring is where the method matters most.

Some rows are built differently, and the method column says which. Population, households, aggregate income, the labor force and unemployment come straight from block groups. A median cannot be added up across areas, so the ring’s median income is an approximation: tract medians weighted by the ring’s households. Education, commuting and housing are read for ZIP code areas (ZCTAs) and allocated by the share of each ZCTA’s population that lives in the ring’s block groups. Median home value and median rent are medians of medians and are labeled approximations. The mean household income, aggregate income over households, is built from block group totals, which makes it the better figure to compare across rings.

For a flex park, the rings describe the people who could work for its tenants and the small businesses that lease its bays. Cedar Fork’s HVAC, fire protection, cabinetry and dental lab tenants hire locally and serve the households around them. The 3-mile ring holds a civilian labor force of 59,117, with a 25.7-minute average commute.

The 1-mile ring looks different from the other two. Its households are small (2.08 people each, against 2.51 at 3 miles), its approximate median income is $69,462, and the ZIP-allocated housing figures put more renter units than owner units inside it (4,622 against 3,023). It reads as a rental corridor along the highway, with a younger, lower-income population than the suburbs further out.

Block group vs ZIP code: the ZIP shortcut missed 30% of the 1-mile ring

An earlier quick ring check on this site, dated October 28, split ZIP code totals by the share of each ZIP’s area inside the ring.

Compound set those figures beside the block group figures. At 3 and 5 miles they agree within 3%. At 1 mile they do not.

The ZIP split put 12,980 people and 5,320 households inside 1 mile. The block groups put 18,662 people and 8,971 households there, 44% and 69% more. The cause is in Compound’s note on the sheet. ZIP 75067 holds 66,913 people, and the ZIP split spread them evenly over its whole area, water included. The block groups near I-35E and Corporate Drive are denser than the ZIP’s average, so an even spread puts too few people near the site.

Households miss by more than people because the people near the site live in smaller households. The ZIP split implies 2.44 people per household (our arithmetic), and the ring’s block groups have 2.08. Across 3 and 5 miles, the dense and sparse parts of each ZIP average out, which is why the two methods converge there.

The income rows move 9% to 12% in both directions, and Compound marked that comparison as not like for like. The ZIP split read median income off the ACS income brackets for each ZIP. The block group method weights tract medians by the ring’s households. The comparison that holds is population and households, and there the ZIP split came in 30% short on people and 41% short on households.

For a buyer, the point is which ring the market story rests on. A 1-mile population of 13,000 and one of 18,700 describe different neighborhoods, and a broker’s OM that quotes ZIP-based rings can carry the smaller one without saying so.

Trade area analysis: the 3-mile ring against Lewisville, Denton County and DFW

A trade area figure means little alone.

The third sheet sets the 3-mile ring beside the city of Lewisville, Denton County, the Dallas-Fort Worth-Arlington metro, Texas and the US, all from the same ACS 2020-2024 release.

The headline comparison needs care. The ring’s approximate median income of $101,091 is 12.0% above the metro’s $90,275. Its mean income, the figure built from block group totals, is $123,028, 1.1% below the metro’s $124,387. We read the trade area as level with the metro on income. The median runs high because tract medians weighted to the ring are not the same statistic as a published median.

The other rows point the same way. The 3-mile ring is better educated than the metro (43.0% with a bachelor’s degree against 39.4%), more of it works from home (20.5% against 17.8%), and its commutes are 4.6 minutes shorter. Unemployment is 4.1% against the metro’s 4.6%. Home values sit 13.2% above the metro, and rent 3.5% above it. Against Denton County, the ring is less affluent: the county’s mean household income is $145,847 and 49.4% of its adults hold a degree. Cedar Fork sits in the older, denser south end of a fast-growing county.

Growth and the industrial labor market for a flex tenant

The fourth sheet keeps to series with published history and projects nothing.

Denton County grew from 914,517 people in July 2020 to 1,069,346 in July 2025, 3.2% a year, against 2.0% for the metro and 0.6% for the US. Covered employment in the county grew 3.9% a year from March 2014 to March 2026, and the metro’s 2.2%.

The latest year is softer than the long run. Denton County’s population grew 2.2% in the year to July 2025, its slowest in the series, and county employment in March 2026 was 0.4% below March 2025. Per capita personal income grew 4.7% a year from 2010 to 2024, to $80,108 (BEA). A flex park does not need the fastest-growing county in Texas. It needs enough small businesses to fill bays when leases roll, and the county is still adding them, more slowly.

The fifth sheet narrows to the jobs a flex tenant fills: construction, manufacturing, and transportation and warehousing. In Denton County those three sectors held 60,290 jobs in March 2026, 19.0% of covered employment and 3.7% more than a year earlier. Construction did most of that, up 11.4%. In the metro the same three sectors are 19.7% of jobs.

One gap in the public data shaped the sheet. BLS publishes the metro’s sector employment from QCEW with a lag, and the quarters after Q2 2025 show as 0 in the series, meaning not disclosed. Compound took June 2025 as the latest usable metro sector quarter and said so on the sheet. To bring the metro up to date, it added September 2026 from the monthly CES survey, which groups construction with mining and logging, and transportation with utilities. On that basis the metro’s industrial jobs grew 2.8% (construction group), 0.0% (manufacturing) and 3.0% (transportation group) over the year.

Unemployment in September 2026 was 3.9% in Denton County and 4.1% in the metro, each 0.1 points lower than a year earlier (BLS LAUS, not seasonally adjusted). Wages are the other half of a tenant’s labor cost. A county manufacturing job paid $1,870 a week in Q1 2026, 36% above the all-industry average. A transportation and warehousing job paid $1,171, 15% below it. Cedar Fork’s contractor tenants hire in the sector whose jobs grew fastest, construction, and pay its above-average wages.

Industrial submarket analysis: Q3 2026 vacancy, rent and absorption

The ring and the labor market describe demand.

The submarket shows what landlords are getting for space now. Compound read the two public Q3 2026 industrial reports that break out a submarket containing the site. It placed the site on each broker’s submarket map: CBRE calls it Lewisville, in its Northwest Dallas district, and Cushman & Wakefield calls it Far North/I-35. It also read the metro-wide flex segment from each, the closest product match to a small-bay park. Newmark’s 3Q26 report keeps its submarket table behind a form, so Compound did not use it.

The two brokers agree on vacancy and disagree on almost everything else. CBRE’s Lewisville submarket is 11.0% vacant with an average direct asking rent of $11.29 a square foot, triple net. Cushman’s Far North/I-35 is 10.2% vacant at $9.97. Both submarkets hold about 37.3 million square feet, but they are not the same 37.3 million. CBRE’s Lewisville lost 106,000 square feet of occupancy in the quarter. Cushman’s Far North/I-35 gained 2,055,003. A buyer who quoted only one report would tell the committee either that the submarket is softening or that it is absorbing space fast.

Compound’s sheet says why the figures differ. The boundaries differ, so inventory and absorption differ. CBRE quotes average direct asking rent, triple net, for buildings of 10,000 square feet and over. Cushman quotes an overall weighted average net rent. Neither is wrong. Compound used the two submarkets as the ends of a range, 10.2% to 11.0% vacancy and $9.97 to $11.29 rent, and the two flex segments as a second range, 6.4% to 6.6% vacancy and $12.83 to $12.98.

The submarket against its metro tells the rest. Both brokers put the submarket’s vacancy above the metro’s (7.8% at CBRE, 7.3% at Cushman). Flex vacancy, 6.4% to 6.6%, sits below it. Small-bay space is tighter than the market as a whole. The reports do not split the submarket’s pipeline by building size, so how much of CBRE’s 1.89 million square feet under construction competes with small bays is a question for the brokers.

Market rent and vacancy vs the underwriting model

The last analysis sheet reads the model’s inputs from the pro forma workbook, without changing them, and tests each one against the two ranges.

Neither model rent sits inside the submarket range. Small bay at $14.50 is $3.21 above the top of the submarket’s $9.97 to $11.29 and $1.52 above the flex segment’s $12.83 to $12.98. Large bay at $12.75 is $1.46 above the submarket and just under the flex range.

There is a fair case for the model. Its rents come from the appraisal’s market rent conclusion and from the park’s own lease comps, which average $14.70 for small bays and $12.75 for large bays. Broker averages blend every building type and age in a submarket, and a 20,000 square foot flex suite rents for more than a 300,000 square foot warehouse. So the gap does not prove the model wrong. It does mean the rent rests on one appraisal and a handful of park-level leases, and no public market figure confirms it.

Vacancy is the clearer miss. General vacancy and credit loss of 1.0% sits on top of modeled downtime, and Compound worked out what the two add up to. Over the 2027 to 2036 hold, downtime and general vacancy remove 3.3% of potential rent and recoveries. Submarket vacancy is 10.2% to 11.0%, and flex vacancy 6.4% to 6.6%. Physical vacancy and economic vacancy are not the same measure, and the model’s 3.3% leaves out free rent. Counted with free rent, the model’s figure is 5.2%. Either way, the model sits below both ranges.

Compound’s recommendations, from the sheet:

The run stops at the inputs. It does not rerun the pro forma at them, so the IRR and bid at 3.0% vacancy and downside rents are the next step, in the real estate pro forma workbook. Both changes push the same way, toward a bid below $46.4M.

The market analysis report: one page for the investment committee

A real estate market analysis report for a committee is one page, answer first.

Compound ended the run with that page, in the house slide template, shown here with one label corrected.

Where the data comes from: real estate market analysis tools and sources

Every figure in the workbook comes from a public source, cited on the sheet that uses it.

Compound pulled the Census and BLS series itself and saved the two broker reports beside the workbook.

The ring step is GIS work, and Compound kept its block group shares on the BG Data sheet, so every ring figure traces to a block group.

FAQ