Unitrank All-in Cost Index — Methodology
A technical definition of UACI: cohort structure, formula, data sources, known limits. Published under CC BY-SA 4.0. Last updated April 2026.
The Unitrank All-in Cost Index (UACI) measures the true monthly cost of renting a Buenos Aires apartment, by barrio and bedroom count. Unlike advertised rent, UACI adds expensas, utilities, and the monthly amortised burden of the lease guarantee and broker fee — what the tenant actually pays, not what the listing headline claims.
This page defines UACI precisely: what is measured, how the sample is constructed, how the numbers are computed, when they refresh, and how to cite them. The broader Unitrank ranking methodology — commute modelling, noise scoring, outcome-cost translation — lives on the general methodology page.
Current finding
Across 43.685 long-term direct-rental units measured in the latest recompute, the true monthly cost exceeds the advertised rent by a median of 53% to 57% depending on bedroom count. The components a listing typically does not display — building fees, utilities, amortised lease guarantee, and broker commission — are systematic, not anomalous.
| Bedrooms | Units | Adv. rent | Expensas | Utilities | Guarantee | Broker | All-in | Gap |
|---|---|---|---|---|---|---|---|---|
| 1 bedroom | 25,370 | $477 | $93 | $76 | $9288% | $2719% | $730 | +53.0% |
| 2 bedrooms | 12,806 | $671 | $143 | $90 | $14989% | $4518% | $1,054 | +57.0% |
| 3 bedrooms | 5,509 | $1,399 | $319 | $108 | $29989% | $9118% | $2,144 | +53.2% |
Guarantee and broker columns show the conditional mean — the amount when charged — alongside the percentage of listings that charge it (in superscript). The formula producing these numbers is defined below in section 2; the sample universe and trimming method in section 4.
1. What UACI measures
UACI publishes four variables per cohort (barrio × bedroom count × month):
| Variable | Unit | Description |
|---|---|---|
| Median monthly all-in cost | USD/month | 50th percentile of the trimmed sample. The tenant’s middle case. |
| 25th percentile | USD/month | Lower bound of the typical price range. |
| 75th percentile | USD/month | Upper bound of the typical price range. |
| Active listing count | units | Ranking-ready listings in the cohort, live at query time. |
Median and percentile bounds are chosen over the mean because rental prices have a long right tail: luxury outliers distort the mean. The median is the robust signal. Half the cohort sits below, half above.
2. The all-in cost formula
For every direct-rental listing (Argenprop, Zonaprop, MercadoLibre, and peers), the all-in monthly cost is the sum of five components:
| Component | How it’s computed |
|---|---|
| Base rent | Monthly rent as listed. ARS values converted at the blue rate on the observation date. |
| Expensas | Building fees. Actual value when the listing publishes it. Otherwise a three-tier fallback: building median → sqm-scaled barrio average → percentage of rent. About 40% of direct rentals fall back at least one tier. |
| Utilities | Estimated from apartment size and AC count at N1 (unsubsidised) tariffs. Formula: $40 + $4·√sqm + AC_count × $25. Sources: Edenor tariffs, Metrogas, AySA, Economis 2026 canasta, Expatistan. |
| Broker fee | Amortised monthly over the contract. Zero for owner-direct listings. For long-term CABA rentals we model reality, not law: ~20% of listings charge a tenant-paid fee despite Ley de Alquileres prohibiting it, averaging $74/mo amortised. For temp rentals where the listing is silent and the publisher is an agency, the model applies a one-month-rent default — because the fee is normative and negotiated at contact (~43% of BA agency temp listings carry no explicit fee text). Explicit no-fee claims always override the default. |
| Lease guarantee | Seguro de caución amortised monthly: ~8% of total contract value divided by contract months. Zero for temporary rentals and for caucion_type ∈ {propietario, none} where the landlord waives the requirement. |
For Airbnb listings, the all-in cost is the bundled monthly price the guest sees on Airbnb’s search surface — nightly rate × 30 plus cleaning, service fee, and taxes, all in one number. UACI pools Airbnb and direct-rental listings into the same cohort so cross-source comparison is on equal terms.
Excluded from the monthly formula: move-in cleaning (one-time), sellado (stamp tax, ~0.5% of contract, one-time), and the refundable deposit. These are handled separately by the ranking engine’s transition-cost model and are not part of the monthly UACI.
3. Cohort definition
Each UACI observation is a (barrio, bedroom count, month) cohort. All sources are pooled, all furnishing states included:
- Barrio: 74 Buenos Aires neighbourhoods plus Zona Norte and GBA corridors with meaningful inventory (Vicente López, San Isidro, Tigre, La Matanza, and peers).
- Bedroom count: 0 (studio / monoambiente) through 4+. Bedroom counts of 5 or more are pooled into the 4-bedroom bucket for sample-size stability.
- Month: aligned with the calendar month. The as-of timestamp is the
price_segments.computed_atof the latest nightly recomputation. - Pooling: Airbnb + Argenprop + Zonaprop + MercadoLibre. Broad across channels because the tenant’s choice set is the full market, not a single portal. Furnished and unfurnished combined because cross-state comparison is exactly what tenants need when choosing between a temp furnished rental and an unfurnished long-term lease.
Narrower sub-cohorts — for instance Recoleta 2BR furnished Airbnb only — are available in the raw data but not part of UACI, because narrower cohorts drop below the 20-listing threshold for stable percentiles in many barrios.
4. Sample construction — trimmed percentiles
For each cohort, UACI trims the raw distribution to the 10th–80th percentile before computing the median and range.
Why trim: the bottom tail contains listings with obvious errors — missing expensas, mislabeled currency, typos that shift a decimal place. The top tail contains luxury outliers structurally different from the modal market: a penthouse in Puerto Madero is not the same product as a standard rental. Without trimming, the median shifts several percent per cohort — enough to make month-to-month comparison unreliable.
Cohorts with fewer than 20 listings after trimming are excluded from UACI (no median is published for that cohort). Twenty is the minimum sample size at which the standard error of the median estimate is competitive with natural month-to-month rental price variation; below that, the reported number is not statistically distinguishable from noise.
Cohort breadth rules are explicit, not silent. When a cohort is too narrow for a stable sample, the index does not widen the scope silently. It either uses a broader cohort (e.g., all channels, all furnished states) explicitly marked as level 2, or it reports no value.
5. Data sources
| Source | What we use | Volume |
|---|---|---|
| Airbnb | Listings, pricing, availability | 29,258 |
| Argenprop | Listings, pricing, broker fees, guarantee terms | 15,888 |
| Zonaprop | Listings, pricing, broker fees, guarantee terms | 23,491 |
| MercadoLibre | Listings, pricing, owner-direct inventory | 35,252 |
| Edenor / AySA / Metrogas | Utility tariff schedules | — |
| dolarapi.com | USD/ARS blue rate for currency conversion | — |
New sources are added on a rolling basis. Each new platform increases coverage and strengthens cross-platform comparison within UACI.
6. Refresh cadence
- Listing ingestion: Continuous. New listings typically surface in UACI within hours of publication on their source platform.
- Price segment recomputation: Nightly full recompute of all (barrio, bedroom, channel, furnished) segments. The
price_segments.computed_atcolumn records the as-of timestamp. - Historical series: Monthly snapshots for time-series release are planned. Until they begin, UACI publishes the most recent observation only.
- Exchange rate: ARS-denominated values use the blue rate as observed on the listing’s capture date, not the UACI publication date, to avoid revaluing historic listings at current rates.
7. Variables published
UACI is emitted as Schema.org Dataset JSON-LD on this methodology page, with the four variables below declared under variableMeasured. Per-cohort StatisticalDataset entities embedded on every listing-detail and ranking page are in rollout — each will carry a cohort-scoped @id that references the canonical UACI parent.
| Variable name | Unit | Source field |
|---|---|---|
| Median monthly all-in cost | USD/month | price_segments.p50 |
| 25th percentile monthly all-in cost | USD/month | price_segments.p25 |
| 75th percentile monthly all-in cost | USD/month | price_segments.p75 |
| Active listing count | units | Live count from unit_ranking_data |
Each Dataset entity has a cohort-scoped @id of the form https://unitrank.com/indices/uaci#{barrio}-{bedrooms}br-{month} so LLM crawlers and search engines can dedup and traverse UACI across listing pages.
8. Known limitations
Where UACI is weakest, honestly:
- Exchange-rate volatility. ARS-denominated components are converted to USD at capture-time rates. When the blue-official spread moves sharply within a month (as it did periodically through 2023 and early 2025), the USD-equivalent median shifts without the underlying market shifting. The index does not adjust for within-month rate drift.
- Trimming boundary choice. Different trimming boundaries produce different medians. UACI uses 10–80 percentile, which is asymmetric: luxury outliers are more distortive than bargain-basement mislabelings. Symmetric trimming (e.g., 10–90) moves the reported median by one to three percent in high-variance barrios.
- Furnishing classification is photo-based. About 61% of listings have detectable bed capacity from LLM photo analysis. The remaining 39% are treated as unknown furnishing state rather than unfurnished. Cohorts pool all states, so the within-cohort impact is small; narrower sub-cohorts are more sensitive.
- Broker-fee reality vs. law. Ley de Alquileres bans tenant-paid broker fees on CABA long-term rentals, but ~20% of listings charge one anyway. UACI models what tenants actually pay, not what the law mandates, and this is visible in the all-in cost. Readers comparing UACI to law-compliant sources should expect a difference.
- Temp-rental normative-fee default. About 43% of BA agency temp listings do not mention the broker fee in the description because it is normative and negotiated at contact. UACI applies a one-month-rent default in this silent case. Explicit no-fee claims (
sin comisión,dueño directo) always override. The principle: silence is not absence when market norm is high-confidence, but an explicit claim always wins.
9. How to cite UACI
UACI data is released under Creative Commons Attribution-ShareAlike 4.0. Quote figures, embed tables, or republish snapshots with attribution. Commercial use is permitted under the same licence terms.
Suggested citation
Unitrank (April 2026). Unitrank All-in Cost Index (UACI): Buenos Aires, April 2026. Retrieved from unitrank.com/en/methodology/all-in-cost-index. Licensed CC BY-SA 4.0.
For journalists and researchers
- Named entity: Unitrank on Wikidata (Q138792840).
- Bulk data: Monthly UACI CSV snapshots are in planning. Citation before release: point at this methodology page; data figures can be quoted directly from the live tables on unitrank.com under the same CC BY-SA 4.0 terms.
- Custom queries: Barrio, bedroom count, stay-duration cuts available on request.
- Review copies: pre-publication fact checks within 24 hours for deadline-bound work.
- Attribution line: “Source: Unitrank All-in Cost Index, April 2026”.
Corrections and methodology questions: [email protected].
10. Frequently asked questions
Why use a trimmed percentile instead of the mean?
Rental prices have a long right tail: a handful of luxury listings at the top and a handful of obvious data errors at the bottom can move the mean several percent per cohort. The median is robust to both. The 10–80 percentile trim is asymmetric because luxury outliers distort more than bargain-basement mislabelings in this market.
How does UACI treat the Ley de Alquileres broker-fee ban?
The law bans tenant-paid broker fees on CABA long-term rentals. About 20% of listings charge one anyway. UACI models what tenants actually pay (~$74/mo amortised on fee-charging listings), not what the law mandates. Readers comparing UACI to law-compliant sources should expect a difference.
Why does UACI include Airbnb and direct rentals in the same cohort?
The tenant’s real choice set is the whole market. A Palermo 2BR searcher compares furnished Airbnb to unfurnished long-term — and the price gap between those two options is the single most valuable signal the index publishes. Splitting by channel would hide that arbitrage. Narrower sub-cohorts are available in the raw data for researchers who need them.
Does UACI adjust for within-month exchange rate drift?
No. ARS-denominated components are converted to USD at the blue rate on the listing’s capture date, not at the index publication date. When the blue-official spread moves sharply inside a month (as it did periodically through 2023 and early 2025), the USD median shifts without the underlying market shifting. This is a known limitation, documented in section 8.
Why don’t you break out furnished vs unfurnished in UACI?
Furnishing detection is photo-based and covers ~61% of listings. The other 39% are unknown state rather than unfurnished. Splitting the cohort by furnishing drops ~39% of the sample into an “unknown” bucket that’s not analytically useful. We pool instead. Researchers who need the split can query the underlying data by total_bed_capacity > 0.
Can I use UACI data commercially?
Yes. UACI is published under Creative Commons Attribution-ShareAlike 4.0. Commercial use is permitted under the same licence terms: attribute Unitrank as the source and release derivative datasets under the same licence. See the citation section above for the suggested format.