Office-to-residential conversion does not pencil in Austin CBD
I built a public-data screener of 25 Class B/C office buildings in Austin's CBD to see which ones would convert to residential. The honest finding: zero of them pencil at current rents and 350 per sf hard cost. The breakeven rent is 6.52 per sf per month against a current comp set near 3.00.
This is the short version of a paper I wrote to send to Austin CRE shops as the lead artifact in a cold-outreach round.
The question
Every CRE conference for the last three years has had a panel on office-to-residential conversion. The story writes itself: hybrid work emptied Class B inventory, Austin still needs housing, downtown wants the foot traffic back. The thing nobody at those panels seems to commit to is the actual arithmetic. If you put real construction cost against real residential rent, what does the spread look like?
I wanted to know the answer for Austin specifically, before I cold-emailed any CRE shop pretending to have an opinion about conversion deals.
The setup
I curated 25 Class B and C office buildings in the Austin CBD from publicly known inventory. For each one I have year built, stories, building square footage, parcel size, last assessed value, and office class. That gets me the screening inputs. It does not get me anything useful about the actual interior, which is the limit I keep coming back to.
The feasibility score is a weighted blend of plate depth, vintage, assessed price per square foot, office class, and size band. Plate depth is the big one because residential layouts need a window within roughly 32 feet of the unit interior. Vintage is the second one because pre-1990 buildings tend to have smaller plates and accessible plumbing risers. The other three are acquisition-side proxies.
The underwriting is one set of assumptions: 350 per sf hard construction cost, 1.25x soft cost multiplier, 85 percent sellable area, 750 sf average unit, 2.80 per sf per month stabilized rent, 60 percent NOI margin, 5.25 percent exit cap. Acquisition basis is the assessed value.
The headline
Zero of the 25 buildings produce a positive yield-to-cap spread. Median stabilized yield on cost is 2.25 percent against a 5.25 percent exit cap. That is a 300 basis point gap. The median all-in cost per door is 670,000 dollars and the median stabilized value per door is 245,000 dollars. Every building underwrites to a roughly 400,000 dollar per door loss before any tenant cleanout or zoning friction.
The cleanest way to state the gap is in rent terms. To break even, the model needs 6.52 per sf per month in residential rent. The Austin CBD luxury new-construction comp set is closer to 3.00. v1 conversion needs rents 2.2x today’s market.
What the ranking is actually for
If the answer is “nothing pencils,” then why bother with a ranking at all. Because the ranking still does work: it tells you which 8 of the 25 buildings are worth a walking tour and which 17 are not. The top of the list is exactly where you would expect: the pre-1990 small-plate buildings. Norwood Tower (1929), Capital National Bank (1955), Stokes Building (1962), and Trinity Building (1965). The Norwood Tower already has a partial residential conversion completed, which is a small confirmation that the score is finding real signal rather than random noise.
The other thing the ranking does is set up the second conversation: what would have to be true for any of these to pencil? Hard cost below 200 per sf, exit caps below 4 percent, rent recovery to 3.75 by 2028, a federal conversion tax credit, an Austin-specific TIF. Each one of those is its own essay. The screener is the thing that lets you skip the eight buildings where even the best subsidy stack will not close the gap.
What public data cannot do
Five things this screener has no answer for. Interior column grid and structural slab capacity. Plumbing stack count and routing, which is the single biggest cost variable. Window operability. Parking ratios. Tenant cleanout cost on remaining leases. Public data has none of this and you cannot pencil a real deal without it.
That is exactly why the v2 path is short and concrete. Walk the top 5 buildings. Photograph facades, count visible stacks from lobbies, sketch real plates. Pull adaptive-reuse zoning incentives and TIF availability for each parcel. Add a partial-conversion case where the lower floors stay office and the upper floors convert, because that is how the Norwood Tower deal actually worked and probably the more honest base case for everything else on the list.
Why this is worth showing to CRE shops
The useful version of this paper is not “I found the deal of the decade.” The useful version is “the deal of the decade does not currently exist at v1 cost assumptions in this submarket, and here is what would have to change for it to exist.” Endeavor, Stream, Aquila, Treaty Oak, Rocky Point: they all know the conversion narrative is overstated. The paper that lands with them is the one that names that overstatement and quantifies it instead of repeating the talking points.
I would rather show up with a model that says no than a deck that says yes when the math says no.
PDF is here. If you work at any of those CRE shops and want to argue with the assumptions, the email is at the bottom of this site.