Gross rental yield
Script: scripts/modules/yields.py · Public data: data/out/yields/ · Site data: site/data/yields.json
All figures are statistical estimates built only from Dubai Land Department (DLD) open data published on data.dubai under the Dubai Open Data Licence. They are not valuations and say nothing about any individual property.
What gross yield means here
Gross rental yield = annual rent ÷ purchase price, for comparable homes. It ignores service charges, maintenance, vacancy, agency and registration fees, mortgage costs and tax, so a net yield is always lower.
Because DLD does not publish a rent and a price for the same home, we compare typical rents and typical prices of the same kind of home in the same place and period: the same community, the same segment and the same number of bedrooms, over the same 12 months.
Inputs
Rents (Ejari, staged table rents). The residential rent universe is every contract with
usage = 'Residential', property type Flat, Studio, Villa or Complex Villas, exactly one property in the contract
(no_of_prop = 1) and all quality flags true (valid date, annual rent ≥ AED 1,000, valid area). This excludes labour
camps, staff accommodation, rooms and multi-unit bulk leases. Each contract counts once: a second Ejari line of a
one-property contract repeats line 1, so only the lowest line_number of each contract_id is kept, as in
rents_market.md (about 390 such lines in the 12 months to October 2026). Both new contracts and renewals are included, because the
yield is meant to describe the rent that the existing stock actually earns. The contract month is the contract start
month; contracts starting after October 2026 are ignored.
Prices (DLD transactions, enriched residential basis of sales_market). Residential sales (Apartment, Villa,
Townhouse) with all quality flags true, plus the houses that sales_market adds to its enriched basis: villa sales with a
Commercial, Multi-Use or Other usage label (common in Palm Jumeirah, Jumeirah Park, Emirates Living and Jumeirah
Islands; the price index includes them too) and houses that DLD registers on their plot (type Land; bulk plot
registrations, which are not house sales, are left out as in sales_market.md). Restricted to
individual sale procedures (Sell, Delayed Sell, Sell - Pre registration, Sale On Payment Plan;
development registrations and other bulk procedures are excluded. Unlike the
price index, the "Development" sale procedures, used for unit sales in projects on development-title land such as Dubai
Silicon Oasis or Dubai Investments Park, are also left out here; they are about 3% of ready sales) and to price per
sq ft between AED 100 and 50,000. Only ready sales are used for the yield, because rented homes are completed
homes; off-plan prices reflect a different product (future delivery, payment plans) and are excluded. Ready means the
market of sales_market: title-deed sales, except houses registered on their plot in a project that DLD does not list
as finished by the sale date, which are off-plan.
Segment. Flat and Studio contracts and apartment sales = Apartment. Villa and Complex Villas contracts and villa or townhouse sales = Villa (DLD does not separate townhouses from villas).
Bedrooms. Parsed from the Ejari property sub type (Studio = 0, "N bed rooms+hall" = N) and from the DLD rooms field for
sales. For villa sales without a DLD bedroom count, the count comes from the DLD buildings register through the plot match
of sales_market (the register house with the same plot size in the same project, or master project): about 91% exact
on the imputed sales. The weaker project-mode and master-project-mode fallbacks (about 70% and 58% exact) are not used.
The share of villa sales with a register bedroom count is published per cell (share_sales_bedrooms_imputed_pct).
Before 7 October 2026 only DLD bedroom counts were used; those cover 45–72% of ready villa sales and are cheaper
than the rest (about 8% lower in the same community and bedroom group), so villa yields were too high. Groups: Studio,
1, 2, 3, 4, 5+ bedrooms. Contracts or sales without a bedroom count only enter the all-bedroom columns, never a
yield.
Community. Rents and sales are given a community name by the same function used by the price index module
(resolve_community in scripts/modules/price_index.py), so the names match the price index and community_map.csv.
The function applies, in order, hand-written rules on project names, aliases for DLD master project names, aliases for
DLD area names, and finally the DLD area name itself. Sales use area, master project, project and building name; Ejari
contracts carry no building name, so rents use area, master project and project name. The rent-side mapping is published in
rent_community_map.csv (DLD area and master project → community, with contract counts; project names are not published).
If the price index module is ever unavailable the script falls back to "master project if present, else DLD area" and says
so in its output.
Method
For every month M from January 2018 to October 2026, using the 12 months ending in M:
- Bedroom cell. For each community × segment × bedroom group: median annual rent of the contracts and median price of the ready sales. A cell needs at least 10 rent contracts and at least 10 ready sales; otherwise it has no yield.
- Primary yield (bedroom-matched). Cell yield = median annual rent ÷ median sale price × 100.
- Plausibility guards.
- Same stock. A community's rents and ready sales must come from the same DLD areas: the overlap of their area
mixes (the sum over DLD areas of the smaller of the two shares; published as
rent_sale_area_overlap_pct) must be at least 50%. Otherwise the community has no yield (quality_flag = stock_mismatch_suppressed). Example: in Nad Al Sheba the rents are mostly older villas in Nad Al Shiba Third while the sales were the Meydan gated villa projects in Nad Al Shiba First. - Range. A cell yield below 1.5% or above 12% is treated as a mismatch between the homes that were rented and the
homes that were sold (for example a cluster of unusual sales in one building) and is suppressed
(
quality_flag = implausible_suppressed). Cell yields above 9% are published withquality_flag = high_yield. Until 7 October 2026 the upper bound was 15%, which never applied.
- Same stock. A community's rents and ready sales must come from the same DLD areas: the overlap of their area
mixes (the sum over DLD areas of the smaller of the two shares; published as
- Community × segment yield (
bedrooms = All,gross_yield_pct) = average of that community's bedroom-cell yields, weighted by the number of rent contracts in each cell. This is mix-adjusted: a community where small flats are rented and large flats are sold is not pushed up artificially. - Dubai-wide yield (
dubai_monthly) = the same rent-weighted average over all published bedroom cells in Dubai (segment All), or over apartment or villa cells only.
Why bedroom matching is the primary method: it works for both apartments and villas, and it does not depend on floor area. Ejari areas and DLD sale areas are recorded differently, and for villas the recorded area is often the plot rather than the built-up area, so per-sq-ft rents and prices of villas are not comparable.
Cross-checks kept as columns
gross_yield_psf_pct(apartments only): median rent AED/sq ft ÷ median ready-sale price AED/sq ft. Rent per sq ft uses contracts with an Ejari area of 150–15,000 sq ft and rent of AED 5–1,000 per sq ft. It is computed per bedroom cell and averaged in the same rent-weighted way. Not computed for villas.yield_unadjusted_pct(bedrooms = All): median rent of all contracts ÷ median price of all ready sales in the community and segment, ignoring bedrooms. Differences fromgross_yield_pctshow how much the bedroom mix matters.
Changes (dubai_monthly)
mom_pct,qoq_pct,yoy_pct: relative change of the yield level versus 1, 3 and 12 months earlier (for example 6.0% → 5.7% is −5%).*_ppcolumns give the same in percentage points.*_same_cells_pct: the same changes computed only over bedroom cells published in both months, so that cells entering or leaving the sample do not move the figure.psf_*_pct: changes of the apartment per-sq-ft cross-check. Because each point already averages 12 months, month-on-month changes are small and lag turning points by several months.
Quality flags (quality_flag)
thin_sales: the yield rests on fewer than 30 ready sales in 12 months. Typical for villa areas where most homes are rented and few are sold (older villa districts); the few sales are often newer or larger homes, which lowers the yield.methods_disagree(apartments): the bedroom-matched and per-sq-ft yields differ by more than 30%.implausible_suppressed,stock_mismatch_suppressed,high_yield: see step 3;bedroom_cells_suppressedon an All row means at least one of its bedroom cells was suppressed and the community figure uses the remaining cells.
Publication and privacy rules
- Only community-level and Dubai-level aggregates are published; no row relates to a unit, contract or sale.
- Counts below 5 are blank (
null; shown as "<5" on the site). Medians are only shown where that side has at least 10 records (formedian_rent_aed_sqft: at least 10 contracts with a usable area); yields need at least 10 rents and 10 sales. - No company names are published. Some community names contain a brand because that is the name of the place, as in the price index.
Files
| File | Content |
|---|---|
dubai_monthly |
month, segment (All / Apartment / Villa), gross_yield_pct, gross_yield_psf_pct (apartments and All), counts of cells, communities, contracts and ready sales used, changes |
community_monthly |
month × community × segment × bedrooms (All row = mix-adjusted community yield); counts, medians, yields, flags |
community_latest |
the October 2026 slice of community_monthly |
coverage_latest |
how many communities have rents, ready sales, both, and a published yield; share of rent contracts covered |
rent_community_map |
DLD area × master project × segment → community for rent contracts (Feb 2017–Oct 2026) |
Coverage (October 2026, 12-month window)
Rent contracts map to about 230 community names, ready and off-plan sales to about 190. In the latest window, 178 communities
have apartment rents and 84 have ready apartment sales (all 84 also have rents); 145 have villa rents and 105 ready villa sales
(102 in both). A yield is published for 70 apartment and 45 villa communities, 325 bedroom cells in total. The published cells
contain about 56% of apartment contracts and 58% of villa contracts in the rent universe. The same-stock guard removed two
villa communities (Al Barsha, Nad Al Sheba), the range guard no bedroom cell, and 8 published cells carry high_yield.
Dubai gross yield, October 2026: all homes 6.57%, apartments 6.78%, villas 4.78%. (Before the decisions of 7 October
2026 the September 2026 villa yield was 5.19% on 27 communities: the price side used only villa sales with a DLD bedroom count, which
are cheaper, and left out villas with a commercial usage label and houses registered on their plot.)
Sales and rents without a bedroom count form their own unmatched group and never enter a bedroom cell. (Until
7 October 2026 they were counted in the 5+ cell, because least(NULL, 5) returns 5 in DuckDB: 1,442 ready villa sales
without bedrooms in the 12 months to September 2026, against 105 genuine 5-bedroom sales, which made 34 villa 5+ cells
publishable. After that fix 4 villa 5+ cells remained and the Dubai villa yield for September 2026 was 5.19% instead of
5.03%. With register bedroom counts for villa sales (7 October 2026) 21 villa 5+ cells are published, each resting on
sales whose bedroom count is known. The module checks on every run that the 5+ cell holds only rows with 5 or more
bedrooms.)
The unmatched rents are mostly in older parts of Dubai (for example Deira, Bur Dubai, Karama, Al Nahda, Al Qusais) where homes are rarely or never sold to individuals, so there is no price to compare with. The Dubai-wide yield therefore describes the freehold, actively traded part of the market, and its sample grows over time (about 50 communities in 2018, 80 to 95 in 2024–2026).
Caveats
- Rents and prices of "the same kind of home" are medians of different homes. Within a community and bedroom group, rented and sold homes can still differ in age, size, view and finish. New buildings often enter the sales data years before they enter the rent data, which can lower yields in fast-growing communities.
- Renewals are often below current market rent (rent-increase caps), so yields on new lettings are higher than these. Ready-sale prices are also lower than off-plan prices, so including off-plan sales would give lower yields.
- Ejari registration can lag the contract start; the latest month may have fewer contracts than it will eventually have.
- The plausibility range (1.5–12%), the 9% flag, the 50% area overlap and the minimum counts are judgement calls; they are constants at the top of the script.
- 2018 values rest on far fewer communities than recent values; compare levels across years with care.
- Dubai-wide changes mix true yield movements with changes in which cells are published; the
*_same_cells_pctcolumns remove that effect.
Net yield calculator (site only)
The site's yield section has a calculator that runs in the browser; nothing is sent or stored. It starts from the
community's 12-month medians in communities_latest (price and annual rent of the bedroom group with the most rent
contracts, or of another group the user picks), the Dubai median size of registered homes with that bedroom count (DLD
units register, supply_stock), and
the Dubai median service charge of residential building budgets (service_charges, latest budget year), or the charge of
the building chosen in the service-charge section. The user can change every field. With all bedrooms chosen, the
price and rent are the community's medians for all homes and the size starts empty, so the net yield leaves out the
service charge until a size is entered (the report says so), and the report explains why that gross yield differs from
the published bedroom-weighted yield; sizes and service charges follow the site's area unit (sq ft or sq m).
- Gross yield = annual rent / price.
- Net yield = (annual rent − service charge per sq ft × size − vacancy and management (% of rent) − other annual costs) / (price × (1 + purchase costs %)).
- Default purchase costs are 6%: the DLD transfer fee of 4% of the price plus an assumed 2% agent commission. Registration and trustee fees, mortgage costs and VAT are not included. The default 5% for vacancy and management is an assumption, not a statistic.