Can Dubai property outgrow comparable global cities?

Part I · The conclusion in plain language

What the models say

According to our statistical models, Dubai is not clearly undervalued today. However, the models become materially more explanatory and more reliable outside the estimation sample when a GCC regional factor is included. After rent, income, population, safety and other observable characteristics are considered, that factor is associated with lower property prices in Gulf cities than in comparable cities elsewhere. The residual discount may intuitively reflect the market's relative youth, limited awareness among some international investors, restricted access to naturalisation, regional geopolitics, high construction capacity and other unmeasured differences. Regression cannot determine how much belongs to each explanation. Importantly, Dubai's own unexplained gap appears to have narrowed over time, which is consistent with the country's development and the market's growing maturity. Continued fundamental growth and institutional progress could therefore create additional upside if they lead investors to apply a smaller regional discount.

To test this proposition, we compare Dubai with 30 global cities, examine comparable model estimates for 2012, 2020 and 2025, and compare Dubai's recent price path with six European markets. The detailed variables and statistical checks are documented in Part II.

Three main findings

The answer is more nuanced than “cheap” or “expensive”

Finding 1 · Today

The models do not show clear undervaluation

Dubai's observed price in our standardised 2025 dataset is approximately USD 7,602 per square metre. The preferred models, which account for rent or income and a range of city characteristics, place Dubai at roughly USD 6,700–7,800 per square metre under the market's present structure.

In other words, Dubai is close to the level the models expect today. Strong rental yields, safety, infrastructure and a favourable tax regime are real advantages, but the market already prices in a substantial part of them. This finding does not mean that every building or unit is fairly priced. It applies only to the city-level average.

The 31-city comparison

City-centre residential prices in 2025

Dubai ranks 23rd in the sample at USD 7,602/m²—below Madrid, Dublin and Los Angeles, but above Lisbon, Melbourne and Montreal. The table is ranked from the highest to the lowest average price.

Source: Deutsche Bank, Mapping the World's Prices 2025, based on city-centre market inputs. These are broad city averages, not matched-property valuations. Property mix, sample size and data collection can differ between cities.

Finding 2 · The unexplained discount

Dubai still carries a regional residual

Even after the measurable advantages are included, several models assign Dubai and other GCC cities a lower price than comparable cities outside the Gulf. We refer to the remaining difference as the GCC residual or discount.

This is not a special penalty applied only to Dubai, and the model does not tell us exactly why it exists. It may combine investor caution about the region, limited familiarity with a relatively young international market, geopolitical perceptions, differences in legal permanence and citizenship, high construction capacity, liquidity, transparency and imperfectly comparable data.

The discount should therefore be treated as a bundle of factors, not as proof of prejudice and not as a number guaranteed to disappear.

Finding 3 · Change over time

Dubai's own gap appears to be narrowing

When we repeat the same leave-Dubai-out test for 2012, 2020 and 2025—including the GCC regional factor—the residual gap narrows in the point estimates. Dubai's observed price was approximately 31%, 26% and 2% below the corresponding conditional model estimates. The 2025 figure does not mean that Dubai was only 2% cheaper than peer cities. It means that the observed price of USD 7,602/m² was 2.3% below the model estimate of USD 7,780/m² after the existing GCC discount and the other controls had already been applied.

This is directional evidence of convergence, not a precise measure of market undervaluation. The statistical range is wide, so it is not conclusive proof. Nevertheless, the direction is consistent with what has happened on the ground: stronger transaction activity, more international buyers, better data, improving transparency, a more developed regulatory framework and longer residence options.

A separate annual comparison also identifies a clear relative repricing after 2022. Golden Visa reform may have contributed, but it coincided with reopening, migration and geopolitical capital flows, so no single cause can be isolated.

Why further convergence is possible

Country development can change how investors price risk

Dubai's future price advantage does not depend only on buildings. It depends on whether the UAE continues attracting residents, companies and capital while improving the institutions that make a market predictable and investable.

The UAE has demonstrated strategic continuity, execution capacity and an ability to adapt through difficult regional and global conditions. We discuss this separately in our article, Why Leadership Is an Investment Factor. Effective leadership does not guarantee property returns, but it raises the probability that infrastructure, economic diversification and long-term population growth remain supportive.

If these fundamentals remain stronger than in mature peer cities and investor confidence continues to improve, Dubai can plausibly achieve faster long-term price growth through two channels: better economic performance and partial reduction of the residual discount.

What the scenarios mean

Partial—not complete—convergence is the credible case

The table illustrates what would happen in one stable model if different portions of the regional residual disappeared. It is a sensitivity exercise, not a price forecast.

A 25% reduction would imply a value around USD 9,036/m²; a 50% reduction, around USD 11,714/m². Spread over ten years, the convergence component would add approximately 1.7 or 4.4 percentage points a year respectively. Actual growth could be lower or higher because rents, supply, financing and the required return also change.

Full removal produces an extreme figure of USD 19,688/m² and is not a reasonable base case. With rent unchanged, gross yield would fall from roughly 7–8% to 2.7–3.1%. This reality check is one reason we focus on partial convergence.

What could invalidate the thesis

Faster growth is possible, not automatic

The thesis weakens if new supply grows faster than population and rents, if the UAE economy or migration disappoints, if financing or geopolitical risk rises, or if the legal and residency differences between the Gulf and Europe remain permanently important to buyers.

It also weakens if today's prices have already captured most of the expected improvement. And even a correct city-level thesis can produce a poor investment in an overpriced unit with high service charges, heavy competing supply or weak resale liquidity.

Reading the evidence

The thesis remains conditional

The main section above explains the investment conclusion. The technical appendix below documents the samples, variables, tests and limitations behind it.

Part II · Technical appendix

How the testing was performed

This appendix documents the sample, variables, time horizons and statistical checks behind the conclusions above. It is included so that the result can be challenged and reproduced rather than treated as a marketing opinion.

Data universe

Samples and horizons

Model A · Current relative value

Variables and results

The valuation models use rent or net salary as the economic anchor. Additional variables cover population, safety, infrastructure, investment residence, access to naturalisation, personal income tax, expected growth and regional indicators. The GCC indicator absorbs common Gulf characteristics that remain after the measurable inputs are included.

Preferred specifications produce Dubai values of approximately USD 6,700–7,800/m² against an observed USD 7,602/m². Stronger specifications have in-sample R² of about 0.57–0.68, adjusted R² of 0.51–0.63 and leave-one-city-out R² of about 0.38–0.54. Rent is the most stable positive anchor; safety is generally positive, while several policy variables are sensitive to specification.

Model B · Convergence over time

Has Dubai's conditional gap changed?

Comparable leave-Dubai-out models are estimated for 2012, 2020 and 2025. The actual-to-predicted ratio rises from 0.695 to 0.738 and 0.977, equivalent to gaps of −30.5%, −26.2% and −2.3%.

A paired bootstrap across the 30 peer cities shows convergence in 79.5% of resamples, but the 95% interval still crosses zero. This is directional evidence, not a statistically final proof. Transparency improved from 3.10 in 2012 to 2.38 in 2024, while Dubai Land Department reported 226,000 transactions in 2024; these indicators are supportive but not a causal market-maturity regression.

Model C · Annual panel and events

Did Dubai reprice relative to European peers?

The panel covers Dubai, Berlin, Dublin, Madrid, Paris, Amsterdam and Lisbon. Explanatory variables are national real GDP growth, city population growth, built-area growth and real-estate transparency, with market and year fixed effects.

The normalised Dubai × time coefficient is +0.270 with HC3 standard error 0.077 and p below 0.001. It remains between +0.218 and +0.279 when each peer is removed in turn. The post-2022 effect is +0.220; the post-2019 effect is statistically insignificant. Model R² ranges from 0.53 to 0.68.

With one treated city and only two post-2022 observations, the test cannot identify Golden Visa reform as the cause. Reopening, migration, capital flows, financing and supply changed at the same time.

Statistical controls

How robustness was assessed

City boundaries, property mixes, data years and index methodologies are not perfectly identical. Regression measures conditional association, not causation. Scenario prices are model outputs, not property valuations or guaranteed forecasts.