The Market Is Flat. Your Next Deal May Not Be.
Across 20 sampled metros, advertised occupancy in older apartment buildings barely moved on average. Underneath them, ZIP codes, local pockets and individual properties told different stories.
By Tal Berger · lizo Insights
National research: March 29–November 2, 2026. Dallas follow-up: March 29–November 8, 2026.
What this research shows
1. The question the average can’t answer
A property is losing ground. Is it the building, or is the neighborhood struggling with it?
For an investor, those are two different problems. One may prompt a closer look at the property's pricing, condition or operations. The other calls for understanding the local market before assuming the building can recover on its own.
Neither answer is visible in a metro average.
In our analysis of older apartment stock across 20 sampled metros, advertised occupancy barely moved at the metro level. Yet in every metro, some ZIP codes gained ground while others lost it. A closer look at Dallas revealed another layer: properties and local pockets moving in different directions within the same ZIP code. [1, 2, 3]
We followed that trail from the metro to the property to see how “the market is flat” can become a misleading starting point for an investment decision.
Advertised occupancy is a listing-derived indicator calculated from recorded vacant units and the units in each property record. It is not verified rent-roll occupancy. Changes can reflect listing coverage and advertising practices as well as leasing.
2. First, separate older and newer stock
Before asking where performance diverged, separate what you’re comparing. In our fixed four-date sample, advertised occupancy in buildings recorded as built through 2021 finished close to where it began: 96.64% to 96.61%. Buildings recorded as built in 2022–2025 rose from 90.00% to 93.37%. The combined sample improved, but that improvement did not describe the older group. [1]
That is why we look at older stock separately in the metro comparison below. Did flat metro results mean flat local markets? This next comparison uses a different panel and the rent screen described in the methods. Construction years alone do not verify stabilization or lease-up. [1]
3. The illusion of a flat market
In the rent-screened comparison, between March 29 and November 2, advertised occupancy in older stock changed by less than half a percentage point in either direction in each of our 20 sampled metros. Viewed only at that level, the markets looked largely unchanged. [1]
Inside them, the picture was different. Every metro contained ZIP codes gaining and losing advertised occupancy. Across the 1,205 ZIP codes meeting the sample requirements, the middle 90% of changes ranged from approximately −1.8 to +1.8 percentage points. These results describe the covered sample, not every apartment in those markets. [1]
Even without the rent filter, all 20 metros still had ZIP codes moving in both directions. [1]
What’s happening inside your market?
Explore ZIP-code trends in your market. Choose a metro, then hover over a ZIP code to see its change directly on the map. On mobile, tap a ZIP code; enlarge the map for smaller areas. The maps use the same screened older-stock comparison as the national chart: March 29–November 2, with at least 1,000 screened units per displayed ZIP code. They describe covered records, not every apartment or a verified comp set. Each map identifies any missing geometry.
The differences don’t stop at the ZIP code. Let’s zoom into one Dallas ZIP.
Dallas–Fort Worth shows why the distinction matters to an investor. Using the same older-stock and rent-data screen as the metro chart, its metro result was just +0.04 percentage points. ZIP code 75228 gained 3.1 points, while 76116 lost 2.4. “Dallas is flat” would be a poor description of either local trend. [1]
The map marks both ZIP codes and a third, 75219, where we take the next step. We use 75219 as an illustrative follow-up because its tracked records let us compare local groups and one asset inside the same ZIP code. It is not presented as a typical Dallas ZIP or a verified comp set. The follow-up uses a separate sample and period, explained below.
4. A ZIP code is not one market
The Dallas 75219 follow-up is a separate sample. It includes newer buildings as well as older ones and runs from March 29 through November 8, rather than ending November 2 like the metro research. [2]
Of the 56 non-test property records we tracked, 31 had usable observations at both dates. Together, they represented 7,570 units. Recorded vacant units fell from 519 to 314, equivalent to a 2.7-percentage-point rise in advertised occupancy. The other 25 records lacked a usable comparison and are not counted as unchanged. [2]
Newer buildings contributed, but they did not account for the whole change. Properties recorded as built in 2022 or later supplied 38% of the net decline in vacant units; older buildings supplied the rest. Construction years alone do not establish opening dates or lease-up status. [2]
To see how that improvement was distributed, we grouped the tracked properties by census tract, an existing geographic boundary. We use “local pocket” for these groups, which cover only sampled properties inside 75219, not every building in a tract. These are geographic groups, not verified comp sets. [3]
Of the seven pockets containing at least two properties with observations at both dates, advertised occupancy rose in five, fell in one and was unchanged in one. Three other pockets had only one property with observations at both dates; advertised occupancy fell in each. [3]
The ZIP code's overall gain therefore contained both shared local weakness and individual properties moving against the wider trend. Those patterns lead to different investigations.
5. Two kinds of weakness, two different investigations
The distinction becomes clearer when the two patterns are set side by side. In one pocket, all five tracked records recorded more vacant units. Elsewhere, Eclipse Oak Lawn weakened while the wider ZIP-code sample improved. These are different starting points for an investigation—not diagnoses. [2, 3, 4]
The three other tracked properties in Eclipse's assigned tract recorded a combined decline in vacant units, from 27 to 12. One property accounted for 13 of the 15 fewer vacancies; another was unchanged. The buildings differ in recorded age and class, so this is geographic context, not broad-based strength or a verified comp set. It narrows the question without answering it: why did Eclipse finish below its March level while that combined count fell? [4]
These are two observed patterns, not two proven causes. The first calls for examining shared local pressures. The second calls for testing the property's position against relevant alternatives. Neither proves that a renovation, a pricing change or new management would solve the problem.
Eclipse was selected after the outcomes were examined. It illustrates a question to investigate; it does not establish how common the pattern is. [4]
Fewer recorded vacancies. More apartments advertised.
From October 10 to November 8, Eclipse Oak Lawn’s recorded vacant units fell from 23 to 14. Yet the share of apartments advertised for rent rose from 31.17% to 33.77%. Advertised availability includes apartments offered for a future move-in, not just those vacant now. The lower vacancy count alone therefore does not establish a leasing recovery. [4]
6. What an investor should do next
Before underwriting a turnaround, establish what the property is competing with. A geographic group helps locate the question. A relevant comp set should test which alternatives a renter might actually choose.
Where vacant units are rising across several nearby buildings, investigate competing supply, pricing and concessions, and the fit between the available apartments and local renter demand. A property plan that assumes the surrounding market will recover needs evidence for that assumption.
Where the property moves against nearby trends, compare its asking rents and concessions with the relevant alternatives, then examine condition, unit mix, amenities, renter fit and operations. The question is whether there is a specific, supportable explanation for the gap, rather than a generic expectation that a renovation will close it.
Eclipse adds another discipline: read the signals together. Vacancies falling while availability rises should prompt a check of what is being offered, what is leased and when units become available. Until those records are reconciled, a better-looking vacancy count is a question to resolve, not a recovery to underwrite.