DXB REX · Dubai Real Estate Index

Methodology

How DXB REX, the Dubai Real Estate Index, is built: the data, the model, and how the series are updated and revised.

1. Data

The indices are built from Dubai Land Department (DLD) open data:

  • Sales: every registered residential sale since 2015, more than 700,000 transactions, ready and off-plan. Each record has the price, size, property type, project, area and registration type.
  • Rents: every registered Ejari tenancy contract since 2015, more than 3 million residential contracts, plus commercial leases. Each record has the annual rent, size, property type, area, contract dates and whether it is a new lease or a renewal.

The data is enriched with project, developer and location information, such as distances to malls, beaches, main roads and the airport, and cleaned before estimation:

  • outliers in price or rent per square metre and in size are trimmed;
  • registrations below a minimum value, such as gifts and nominal transfers, are removed;
  • categories with too few observations, such as small projects, are grouped together.

2. The model: hedonic regression

A simple average price per square metre changes whenever the mix of what is sold changes: more villas, a new prime community or a wave of off-plan launches can move it without any home becoming more expensive. The indices remove this effect with a hedonic regression, the standard method for property price indices.

The model explains the log price, or log rent, of each transaction from its characteristics, plus a dummy for each quarter:

IndexCharacteristics held constant
Residential salesSize, bedrooms, project or area, developer, off-plan or ready, freehold, amenities (parking, pool, shops), distances to malls, beaches, main roads and the airport
Residential rentsSize, location, new lease or renewal, freehold, number of units in the contract, tenant type
Commercial rentsSize, location, new lease or renewal, contract length, tenant type, freehold, and for shops the unit type

The quarter dummies measure how much prices or rents changed between quarters for property with the same characteristics. That quality-adjusted change is what the index tracks.

International standard. This is the hedonic time-dummy method set out in the Handbook on Residential Property Prices Indices (RPPIs), published by Eurostat in 2013 together with the ILO, IMF, OECD, UNECE and the World Bank. The Handbook is the international reference for compiling house price indices and is used by national statistical offices and central banks. For the commercial rent indices, the method follows the same principles as Eurostat's Commercial Property Price Indicators: Sources, Methods and Issues (2017).

3. Rolling windows and chain-linking

Rather than one model over the full history, the model is re-estimated for each quarter on the latest six quarters of data. Each window gives the quality-adjusted change from the previous quarter to the latest one. These quarterly changes are chain-linked into the index, which is set to 100 in the base quarter.

This lets the value of characteristics, such as the premium for a waterfront location or for off-plan, change over time, so the index follows the market through booms and corrections.

4. Segments and aggregation

  • Residential sales: villas and six apartment market clusters (Waterfront, CBD, Old & East, and Inland low, mid and high) are estimated separately. Apartments and All residential combine them each quarter, weighted by transaction value.
  • Residential rents: apartments and villas are estimated separately and combined into All residential.
  • Commercial rents: offices, shops, hotel apartments and warehouses each have their own model. The series follow new leases of single units. Office business-centre desk packages, which flooded Ejari from 2020, are excluded because they are not comparable with regular office leases.
IndexBaseFrequency
Residential sales prices2015 Q1 = 100Quarterly
Residential rents2015 Q1 = 100Quarterly
Commercial rents2019 Q1 = 100Quarterly

5. Updates and revisions

The indices are updated every quarter once the quarter's registrations are in. Each update re-estimates the full series on the latest data.

Data revisions: Dubai Land Department (DLD) and Ejari keep updating their open data. Transactions and tenancy contracts can be registered late, reclassified or corrected after they are first published. Each release is re-estimated on the latest data, so values for earlier periods, and the most recent quarter in particular, can change from one release to the next.

Tenancy contracts in particular are often registered weeks or months after they start, so the latest quarter of the rent indices is the most likely to change.

6. Reading the indices

  • The indices measure change, not price levels. A residential sales index of 146 means quality-adjusted prices are 46% above 2015 Q1. Two series at the same value do not have the same price per square metre.
  • Turning points show with a short delay. Because each quarter is estimated on six quarters of data, the series are smoother than raw averages.
  • Standards. The method is in line with the international Handbook on Residential Property Prices Indices (Eurostat and others, 2013; see section 2), and the model's accuracy is checked out of sample every quarter.
  • Not an official index. The indices are produced by Spek Advisory FZE from DLD open data as published, and are not a DLD index or publication.

Area, project and unit-type indices, rental yields and occupancy estimates are built with the same models and are available on request.

Contact us

For area- and project-level data, custom indices, yield and occupancy studies, or market reports for investors, developers, lenders and advisers, please get in touch.

Email Spek Advisory