The market pays for a guest house, not for its size

What the market pays for

Casita, guest house, mother-in-law suite, garage apartment, ADU. Different words, one question: if I build it, what is it worth when I sell?

The MLS carries an accessory-unit flag that listing agents answer on about a quarter of listings, and a size field on top of it. Across the closings where the flag was answered, 2,002 carried an accessory unit. Comparing each of those against non-unit sales in the same subdivision — holding living area, lot size, year built and sale date constant — a unit adds 7.2% (CI +5.5 to +9.0).

That is a real premium and a substantial one. It is also an average across two quite different things, and separating them is the first thing worth doing.

Attached or detached: the field that does not exist

A freestanding casita in the back yard and a guest suite folded into the main roofline are recorded identically. Both tick the accessory-unit box. Neither the flag nor the type field says which one you are looking at, and in practice agents do not specify in the remarks whether the unit is attached or detached either.

Using the presence of a detached structure elsewhere in the record — a guest house, separate entry quarters, a second residence — as a proxy for a freestanding unit, and comparing each group against no-unit houses on the same streets:

The gap holds at every price level. Banded by the price of the subdivision it runs +7.8% against +4.4% below $400,000, +12.9% against +6.7% from $400,000 to $700,000, and +11.2% against +6.0% above $700,000. So it is not simply that freestanding units cluster in expensive neighbourhoods.

There is a second reason for caution. Properties with a freestanding unit are not otherwise similar to properties with an attached one:

The model holds living area and lot size constant, but it cannot hold constant the kind of property that tends to come with an acre, a pool and a separate building. Some part of that 11.3% is the unit and some part is the property it sits on, and this data cannot cleanly separate the two.

So the defensible statement is narrower than the headline figure. Freestanding units are associated with a larger premium, that association survives controlling for the price level of the neighbourhood, and it is consistent with what a buyer would plausibly pay more for — a building someone can occupy without routing them through the house. It is not established that the form itself is worth double.

For an appraiser this is the whole assignment in miniature. The record establishes that something exists. What it is, whether it is permitted, whether it has its own entrance, kitchen, meter and address, and whether someone can occupy it without walking through the main house — none of that is in the file, and all of it is what the market is pricing.

Whatever the true difference between the two forms, the record does not let anyone measure it, because the field does not exist.

The finding that matters: size does not pay

Put two terms in the same model — one for whether there is a unit, one for how many square feet it has — and they separate cleanly.

The size term cannot be told apart from zero. On 753 units with a recorded size, the data give no evidence that a larger accessory unit is worth more than a smaller one.

And it is not an artefact of mixing forms. Run the same two terms inside the detached group alone and the result holds: having one is worth +8.2%, each additional 100 sf +0.15% with an interval from −0.70% to +1.01% (p = 0.73). Whatever the market is paying for, it is not floor area.

Put the other way round: the median accessory unit in this market is 600 square feet. Doubling that to 1,200 buys you, as far as 753 sales can tell, nothing.

What that does to the economics

Published build costs for a detached accessory unit run $150 to $300 per square foot. That puts the median detached unit here, 659 sf, at $99,000 to $198,000 — and a 1,200 sf one at $180,000 to $360,000.

Against that, an accessory unit contributes about 7% of the house — roughly $25,000 in a mid-priced neighbourhood and $103,000 at the top of the market. Units with a freestanding structure of record contribute more, around $87,000 on the median house that has one, though how much of that is the building rather than the property it sits on is not settled above.

So recovery runs from about a fifth to most of the build cost, depending on the neighbourhood and on how large the unit is built.

The attached case is weaker on the market side, at about $32,000, but it is also far cheaper to build: shared walls, shared foundation, shared roof, shared systems, and no separate utility runs. I have no defensible per-foot figure for attached construction in this market, so I will not invent one — but the two halves of that comparison move in the same direction, and the honest statement is that attached space costs less and returns less.

For comparison, a swimming pool in this market returns roughly a third to three quarters of its cost, and owned solar about 40%. A detached guest house sits in the same range as a pool. What does not pay is building a large one.

Which points at one clear decision. Build it detached, and build it small. A freestanding 450 sf casita costs roughly $68,000 to $135,000 and carries the same measured premium as a freestanding 1,200 sf one costing $180,000 to $360,000. If the point is housing a parent, four hundred square feet houses a parent.

And if the point is rent, the arithmetic is different again and it is not a resale question. A unit that rents is an income stream; whether it pays depends on the rent, the vacancy, the financing and how long you hold it, none of which appear in a sale price.

Where it pays and where it does not

Banded by the price level of the subdivision, the premium runs:

All four are statistically distinguishable from zero. The pattern is the familiar one: the same feature is worth more where the houses are worth more, and the dollars diverge by roughly nine to one across the range while the percentages move by about two to one.

It is also worth knowing how common they are. Among listings where the flag was answered, an accessory unit appears on 5.1% of sales in the $300,000 to $500,000 band and 24.6% above $1.2 million. At the top of the market a guest house is close to expected; in the middle it is unusual, and unusual features are the ones that price unpredictably.

Compared with everything else in the yard

The MLS also records other structures on the lot, on far more listings than the accessory-unit flag. Measured identically:

The workshop line is the interesting one, because a workshop is cheap. A metal building on a slab with power is a fraction of what habitable space costs to put up, and it returns 5.7% against a guest house's 9.7%. Per dollar spent, the shop is the better buy by a wide margin.

And unlike the guest house, the shop scales with land:

A shop on a tenth of an acre is storage. A shop on five acres is why somebody bought five acres. The guest house does not behave that way — it is worth about the same on a suburban lot as on acreage, because what it does for the buyer is the same either way.

They also sell more slowly

Within the same subdivisions, houses with an accessory unit went under contract in a median 32 days against 27, and closed at 96.3% of original asking against 97.1%. Split by form, the detached ones are the slow half: a median 41 days against 32 for attached units and 32 for houses with no accessory unit at all.

It runs the opposite direction from a pool, which sells eight days faster. A pool is a common feature in the neighbourhoods that have them. A detached guest house is a specific answer to a specific need, and the buyer who has that need is a smaller slice of the market — it is worth more, to fewer people, and it takes longer to find them.

A note on how these were banded

This is technical, and it is here because it changed several of the numbers above by a wide margin before they were published.

When you want to know whether a feature is worth more in expensive neighbourhoods than cheap ones, the obvious move is to split the sales into price bands and run the model inside each. It is also wrong, and wrong in a direction that matters.

Sale price is the thing being explained. Splitting on it conditions on the outcome. Inside a band of, say, $250,000 to $300,000, a house that has the feature and still landed in that band must be weaker in some other respect than a house without it that landed in the same band — otherwise the feature would have carried it into a higher band. The comparison is rigged against the feature, and the rigging is worst in the bottom band where the floor truncates hardest.

The fix is to band on something that is not the outcome. Here that is the median price of the subdivision — a description of the neighbourhood, fixed before any individual house sells. Every banded figure above uses that.

Handling one in a report

Establish the form before anything else. Detached or attached is worth roughly two to one in this data, and no field records it. Answer it from the plat, the sketch, the photographs and the inspection, not from the flag.

Do not cost it. A cost approach on an accessory unit produces a number three to five times what the market is paying. Contributory value here is a market-extracted figure or it is nothing.

Do not adjust on square footage. There is no support in this data for a per-square-foot adjustment on accessory space, and a per-foot adjustment on a 1,200 sf unit produces an indefensible number. Bracket on presence, not on area.

Establish what it legally is. Permitted second dwelling, converted garage, unpermitted addition, or a shop somebody put a bathroom in. These are different properties with different marketability, and the MLS flag does not distinguish them. Utilities, kitchen, separate entry, separate meter and the permit history all matter more than the size field.

Check whether it is occupied. A unit with a tenant in place on a lease is an encumbrance as well as an improvement, and the assignment conditions need to say which one is being valued.

And separate the unit from the land it sits on. An accessory unit is frequently on a larger-than-typical lot, which is why the naive comparison overstates it: controlling only for city, the same flag reads +12.6%. Most of that gap is lot.

What this does not say

The flag is optional and thinly answered. About a quarter of listings answer it. Everything above compares houses where an agent said yes against houses where an agent said no — not against the market at large. Units that exist and were never flagged are sitting in the comparison group, which biases the premium down.

The size result rests on 753 sales. It establishes that any size effect is small, not that it is exactly zero. A finding of no effect on a modest sample is weaker evidence than a finding of an effect.

Quality and legality are invisible, and so is form. A permitted 700 sf detached casita with its own kitchen and meter records identically to a converted garage with a hot plate, and to a guest suite off the back hall.

The attached/detached split rests on a proxy, not on a field. It infers a freestanding unit from a detached structure appearing elsewhere in the record, which will miss detached units nobody flagged. That biases the two groups toward each other, so the measured two-to-one gap is a floor rather than a ceiling.

The label made a difference and I cannot fully explain it. Listings tagged “Guest Quarters” came in at +11.4%; those tagged “Accessory Dwelling Unit (ADU)” at +2.8% and not distinguishable from zero. That may be language, or it may be that the two words describe genuinely different things — guest space attached to a nice house versus a rental unit in a yard. On 128 ADU-tagged sales I would not build an adjustment on it, but it is worth knowing before you choose a word.

And this is Dallas–Fort Worth, 2023 to 2026, a market where land is comparatively cheap and detached construction is comparatively easy. It should not be carried to a city where an accessory unit is the only way to add a second household to a lot.

Terrence Bilodeau is a Certified Residential Appraiser (TX-1360232) and a Texas real estate broker. Clearfork Appraisals provides private-client appraisals in Fort Worth and across Tarrant, Parker, Johnson and Hood counties. This article is a general market study. It is not an appraisal, not tax advice, and not a valuation opinion about any specific property.