A Decade of Dubai’s Indicative Gross Yields

Dubai apartment and villa gross rent-to-price ratios have not moved in lockstep. Our area-rate proxy for ready apartments fell from 7.39% in 2016–17 to 4.94% in 2021–22, then recovered to 6.21% in 2025–26. The equivalent villa proxy ended at 5.08%.

The raw data layer

This history measures a gross rent-to-price rate proxy, not the yield earned by an owner. For each of ten complete September–August windows, from September 2016–August 2017 through September 2025–August 2026, the calculation divides the observation-count-weighted mean annual rent per square metre on New residential physical-unit leases by the observation-count-weighted mean price per square metre on Existing Properties residential sales. Apartments and villas are calculated separately. The rental and sale observations are independent: a leased home is not paired with its purchase transaction. The published samples apply the source’s positive-area, positive-amount and broad rate/value filters. They count qualifying registrations, not homes available to let or the housing stock. The underlying source is the Dubai Land Department’s real-estate data; the annual aggregates used here are frozen in data/evidence.json:yields.

The apartment proxy was 7.39% in the first window, fell to 4.94% in the window ending August 2022, and stood at 6.21% in the latest window. The villa proxy moved from 6.31% to 5.08% between the first and latest windows. These percentages answer a narrow question: how did the recorded rate paid by new tenants compare with the recorded rate paid by ready-property buyers in each period? They do not show the rent received by buyers in the sale sample, or what an owner bought years earlier and later sold.

The sample sizes matter to that interpretation. The earliest apartment calculation draws on 107,433 new leases and 12,848 ready sales; the latest draws on 208,295 leases and 36,712 sales. For villas, the corresponding counts are 14,509 leases and 4,359 sales at the start, and 26,546 leases and 5,484 sales at the end. A larger sample describes more registrations under the filters. It cannot make two independently selected groups equivalent. In particular, a change in the communities, sizes or specifications represented among sales can move the denominator without an equivalent change among new lettings.

“Gross” also has a precise boundary. The numerator is annual contractual rent expressed per unit of measured area; it is not cash collected after vacancy, service charges or repairs. The denominator is a recorded sale-price rate, not an owner’s entire acquisition and disposal cost. Financing, management, insurance, taxes where applicable, capital expenditure and changes in resale value are outside the calculation. Nor is the proxy the average of property-level yields: dividing a mean rent rate by a mean sale rate is a different operation from averaging rent divided by price for matched properties. No such matching is supplied.

The choice of cohorts is defensible but consequential. An off-plan purchase may precede the possibility of letting the home, so the ready-sale denominator is closer to an immediately lettable asset. A New lease is closer to a current letting decision than a renewal under an existing tenancy. Neither choice ensures that the typical ready property sold has the characteristics of the typical property newly leased. Annual averages also give more influence to months with more qualifying registrations; they are not a fixed-property index. Those distinctions should accompany any use of the historical percentages, rather than disappear behind the word “yield”.

The deeper analysis layer

The apartment series is not a continuous recovery story. It reads 7.39%, 7.25% and 7.20% in the first three windows, then 6.54%, 5.55% and 4.94% in the next three. It subsequently rises through 5.40%, 5.82% and 6.24%, before edging down to 6.21%. Using the displayed annual ratios, the fall from the first reading to the August 2022 low is 2.45 percentage points; the rise from that low to the window ending August 2025 is 1.30 points. The latest reading remains 1.18 points below the first. Thus, selecting the trough as a starting point supports a recovery description, while selecting the full supplied span supports a lower-ending-ratio description. Both are true; neither is an account of an individual investment.

The companion unrounded rate series identifies both sides of the apparent downturn. Between the windows ending August 2021 and 2022, the New-apartment rent rate rose from AED 53.60 to AED 59.50/sqft/year, an increase of 11.0% using the first rate as denominator. The ready-apartment sale rate rose faster, from AED 966.12 to AED 1,204.89/sqft, or 24.7% on its own earlier rate. Their quotient consequently fell from 5.55% to 4.94%. This directly rules out describing that year’s lower proxy as a fall in observed New-lease rents. These are different registered populations; faster growth in one group’s mean does not demonstrate that comparable homes followed the same path.

From the windows ending August 2022 to 2025, the numerator rose from AED 59.50 to AED 94.58/sqft/year, about 58.9%, while the ready-sale denominator rose from AED 1,204.89 to AED 1,515.40/sqft, about 25.8%. The rent rate grew faster than the sale rate, lifting the gross proxy from 4.94% to 6.24%. Both inputs increased; neither a sale-price collapse nor a static rent denominator explains this period. The following window offers a counterexample to an uninterrupted recovery: the quotient moved slightly down to 6.21%. Observed cohort means identify this arithmetic, not why either market moved or what a matched owner earned.

Villas complicate a single-market narrative. In the window ending August 2022, their proxy was 5.19%, above the apartment proxy of 4.94% by 0.25 percentage points. In the next window, the villa reading of 5.71% again exceeded the apartment reading of 5.40%, by 0.31 points. Yet in the latest window the ordering had reversed: 6.21% for apartments against 5.08% for villas, a 1.13-point apartment lead. At the beginning, apartments had led by 1.08 points. A buyer cannot treat the asset-class ordering as permanent, or infer from a citywide crossover that a particular villa was once a better rental investment than a particular flat.

The villa path also differs from the apartment recovery. After 5.19% in the window ending August 2022, the villa proxy reached 5.71% in the next window, then 5.67%, 5.38% and 5.08%. Meanwhile the apartment proxy rose through the windows ending August 2025. Independent shifts in what sold and what was newly let could contribute to either pattern. Villa sale prices may reflect land and other attributes differently from a lease rate per measured area; the aggregates cannot separate those attributes or establish that tenants’ willingness to pay weakened.

Observation counts provide a second, distinct check on interpretation. Between the first and latest windows, qualifying apartment ready-sale registrations rose from 12,848 to 36,712, about 2.86 times the initial count. New apartment lease registrations rose from 107,433 to 208,295, about 1.94 times their initial count. For villas, ready-sale counts went from 4,359 to 5,484, about 1.26 times, while new-lease counts went from 14,509 to 26,546, about 1.83 times. These comparisons use the first-window count as each denominator. They show that the two sides of each quotient did not acquire observations at the same pace. They do not measure changes in Dubai’s housing stock, prove an expanding set of comparable homes, or indicate the direction of composition bias.

A particularly tempting mistake is to treat transaction depth as validation of an owner’s yield. The villa sale sample reaches its supplied minimum of 2,394 in the window ending August 2019; even that is a substantial number of recorded sales for a citywide description. But count alone says nothing about whether those sales resemble the villa leases in location, plot share, size or condition. The appropriate test would compare consistently defined property cohorts—or, for realised investor returns, link a property’s purchase, actual tenancy receipts, costs and eventual sale. The frozen yield table does neither. Its strength is a transparent comparison of two market flows across consistent annual windows, not a reconstruction of household or landlord accounts.

Impact on renters, sellers and investors

For an investor considering a ready apartment, the latest 6.21% is best understood as an initial screening signal: the annual new-lease rate was that fraction of the ready-sale rate after the specified aggregation, not that a purchasable unit will earn it. To make the financial distinction concrete, apply 6.21% to an assumed, rather than observed, AED 1,000,000 purchase. That produces AED 62,100 in hypothetical annual contracted rent. If the unit earns no rent for an assumed one month out of twelve, receipts fall by AED 5,175 to AED 56,925, before any other deduction. If assumed annual operating expenses are AED 10,000, the remainder is AED 46,925, or about 4.69% of that assumed purchase price, before acquisition costs, finance and resale outcomes. None of those assumed cash flows is measured in the DLD yield series. In a jointly owned apartment, Dubai Law No. (6) of 2019, Articles (16), (25) and (27) makes service/usage charges an owner-liability issue (subject to lease wording), allocates approved common-parts costs by unit area, and requires RERA approval. The hypothetical AED 10,000 expense is not an observed or legally prescribed service charge; obtain the actual unit's approved charge budget, lease and any arrears. The example shows how quickly an apparently precise gross rate ceases to describe money retained by an owner.

There is a further practical obstacle before applying even the gross number to a deal. An investor needs a sale-price comparison and an achievable lease for sufficiently similar ready homes, over compatible dates and area definitions. If the lease observation is from one set of communities and the sale observation from another, their citywide quotient can be arithmetically correct but financially irrelevant to the proposed purchase. A signed lease on the target property, or a defensible range from closely comparable lettings, answers a different question from the historical market-flow proxy. The series supplies no vacancy frequency, expense distribution or matched rent-price records with which to put an observed uncertainty range around that property’s return. An occupied property's future renewal requires a different input from the New-lease numerator: DLD's Rental Index calculates the renewal context from the existing rent and property/expiry details. Dubai Decree No. (43) of 2013, Articles (1) and (3) sets maximum renewal-increase bands against its RERA-approved similar-unit index, not against this article's citywide New-contract rate. No such property-specific calculation is in the frozen yield series.

For a seller, relevance is indirect. The ratio cannot determine an individual asking or achieved sale price. It can, however, expose why an income-focused buyer may resist a price increase that is not accompanied by comparable achievable rent: a higher purchase-price denominator reduces the gross entry quotient if the rent numerator stays fixed. Conversely, a higher proxy need not indicate stronger tenant demand; a lower sale-price rate can also raise it. The correct negotiation evidence remains transactions and leases for comparable property, not a citywide percentage applied as a valuation multiple.

For renters, relevance is also indirect. A rising new-lease rent rate can affect the price of entering a tenancy, but this calculation does not measure the rent change faced by an existing tenant renewing the same home. Its sales denominator adds no information about what a particular renter owes. New-lease observations and sale observations may also occupy different parts of the market. The decade-long ratio should therefore not be presented as a renter affordability measure: a lower quotient could coexist with a higher annual rent rate if sale-price rates rose faster. The useful renter question is what happened to comparable new-lease fees and renewal fees, separately, not whether landlords’ synthetic gross quotient rose.

Across all three audiences, the timing matters. A September–August point combines registrations throughout a complete annual window; it does not locate the month of a turning point. An owner who bought before a price rise experiences that rise differently from a new buyer facing the later denominator. A renter entering a new contract experiences a different flow from someone renewing. Calling these outcomes “the market yield” compresses distinct decisions into a single division.

Data in context with international markets

An official international series offers context for the rental numerator, but not a foreign yield with which to rank Dubai. Eurostat’s EU-27 Harmonised Index of Consumer Prices for actual rentals for housing records 117.41 in August 2024 and 121.10 in August 2025, an increase of approximately 3.14% between those two monthly index observations. The source is Eurostat’s prc_hicp_midx API, COICOP CP041, index unit I15, whose supplied snapshot was updated 6 February 2026.

This is an appropriately identified rent-price measure, rather than a mortgage rate or a sale-price index mislabeled as a rental return. It is nevertheless not matched to the Dubai numerator. Eurostat measures an index of actual housing rentals across the EU-27; the Dubai proxy uses the mean area-normalised rate on registrations classified as New leases. The Eurostat comparison is August-to-August, whereas each Dubai observation pools a September–August window. Its supplied observations also end before the latest Dubai window ends. Neither its index level nor its percentage change can be divided by Dubai sale prices to create a yield.

A defensible international gross-yield comparison would require foreign rent and ready-sale rates for compatible property types, areas, transaction periods and aggregation rules, with their underlying samples disclosed. The frozen Eurostat rent index does not provide that sale denominator. Eurostat’s separate house-price index measures price change, not the price per measured area paid for the homes generating the rent observations. Consequently, this article gives no numerical international yield rank. The international evidence illustrates a measurement distinction: even a well-defined official rent series cannot turn Dubai’s independently sampled rent-to-price quotient into a cross-country investor-return league table.

Summary and clear conclusion

The testable result is narrow but informative. Under the frozen definition—New residential lease rent rate divided by Existing Properties residential sale-price rate, separately for apartments and villas, across complete September–August windows—the apartment proxy finishes at 6.21% in the window ending August 2026, below its 7.39% starting value despite recovering from 4.94% in the window ending August 2022. The villa proxy finishes at 5.08%, below its 6.31% starting value. Another extraction using the same cohorts, filters, weighting and windows could falsify those statements directly.

What the history rules out is an uncomplicated “rents rose, therefore entry yields rose” explanation. The apartment quotient fell into August 2022 and later recovered because its rent-rate numerator changed relative to its ready-sale-price denominator. The companion annual rate aggregates show that both rates rose across the 2021–22 decline and the 2022–25 recovery, but at different relative speeds. The changing apartment-versus-villa ordering and the different growth in sale and lease sample counts further caution against interpreting a citywide line as the experience of an unchanged property cohort. These are observed relationships, not evidence that one side of the market caused the other to move.

The practical conclusion is equally bounded: the series can identify periods worth investigating in the companion rent-rate and sale-price-rate data, and it can check whether a proposed account of a turning point is arithmetically possible. It cannot establish net income, total return, a typical matched-property yield or an international ranking. Any claim that an investor buying a particular Dubai home will realise the latest displayed percentage would require property-paired prices and leases, vacancy and expense records, and a defined holding period—evidence absent from this frozen history.