1.   An Overview of Various Price Indexing Methodologies

Price indices are used worldwide to give observers straightforward visibility on the performance and attractiveness of specific real estate markets.

Individual markets – whether at the urban, nationwide, or regional levels of granularity – are often dictated by interdependent demographic, financial, and socio-economic conditions, changes in which inevitably cascade to impact the real estate sector in terms of demand, materialized supply, transactional volumes, and transactional price points. For this reason, once a market has sufficiently matured, it is often useful to compile observed transactional movements into a single number that conveys a unified representation of market trends: the price index.

While there are many different price-indexing methodologies, selecting the appropriate one is typically dependent on the specific characteristics of the studied real estate market, including, among other factors, its intensity in terms of transactional volume and the distribution of price points at the global and granular levels.

Notwithstanding, three indexing approaches have gained traction since the inception of the indexing practice, all of which are recognized by the World Bank:

  1. The repeat sales method, which is particularly useful in high-intensity property markets, measures price movements for the same property by grouping its transactions into sequential sale pairs.
  2. The sale price / appraisal ratio method, which is useful for lower-intensity markets, boosts transactional data with government-approved valuations of real estate, and tracks the temporal changes on the basis of a combination of both.
  3. The hedonic method – which is particularly useful in even lower-intensity markets where characteristic data is nonetheless abundant – uses multi-variable regression to estimate market prices based on known, recorded property characteristics such as location, size, and layout, among others.


2.  Abu Dhabi Real Estate Market Context

Abu Dhabi’s real estate market has witnessed considerable growth since the emergence of freehold areas and prominent property developers, factors which have effectively dictated Abu Dhabi’s real estate market trends in the previous decade.

Notwithstanding, special-case transactions remain a hallmark of the Abu Dhabi property landscape; these can refer to purchases of large, contiguous land plots for development purposes, the bulk purchase of multiple residential units by corporate investors and clients, and other ad hoc real estate transactions. Such special-case transactions can often skew performance analysis by assigning heavy weightages to otherwise dormant (in the transactional sense) geographies. It is for this reason that the first iteration of Abu Dhabi’s real estate price indices focuses on several Abu Dhabi development and investment areas, where more or less organized, consistent, and considerable transactional volumes are witnessed.

With the above in mind, sale price indices have been compiled for the following categorical and geographical taxonomies:


Abu Dhabi City

  • Al Raha Beach
  • Al Reef
  • Al Reem Island
  • Al Saadiyat Island
  • Yas Island


Abu Dhabi City

  • Al Raha Gardens
  • Al Reef
  • Al Saadiyat Island
  • Yas Island


Abu Dhabi City


3.  Utilized Data

In terms of data availability, both transactional and listing data have been used to create a model that combines the price movements of both transacted and listed units into a unified real estate market indicator.

To successfully implement the hedonic price index, the key consideration lies in the simultaneous availability of large datasets and a coherent, consistent data structure.

The primary source of real estate sale records in the Emirate of Abu Dhabi is the Emirate’s Department of Municipalities and Transport (DMT), which collects a multitude of data points pertaining to transactional activity and granular, physical property characteristics.

Despite the abundance of property-specific data in the DMT datasets, the relatively modest intensity of transactional volumes in the Emirate of Abu Dhabi precludes the synthesis of price indices on regular temporal basis.

To address this shortage, property listing data from a number of reliable, online sources were merged with the set of actually-materialized transactions. The pertinent data points were then matched to ensure a consistent data flow and, upon successful implementation of the former process, the materialized sale and property listings datasets were consolidated for further processing.

Moreover, in the context of the Abu Dhabi real estate market, data cleansing was applied prior to executing the calculation process. To do this, observations were clustered on the basis of observation period, geographical location, property typology, and bedroom layout. Further classification methods were then applied to optimize the typological categories.

Finally, the pre-processed dataset was subjected to rigorous outlier detection and filtration logic. This was done by implementing “hard-stops” for unrepresentative price points, after which three statistical metrics – the z-score (based on standard deviations from the mean), mad-score (median absolute deviation), and interquartile range computations were applied. Data points that pass the outlier treatment stage then advance to be used in the calculation models.


4.  The Hedonic Price Index Methodology

The hedonic model, as the name might suggest, assigns value to properties by taking into account physical characteristics such as unit type, unit size, unit location, etc.

The model also uses time – in the form of a dummy variable representing various quarters of the year – as an imputable characteristic of the transaction itself.

The model then calculates the contribution of the aforementioned factors, i.e., their weights, in determining the estimated value of a property transaction.

The following mathematical equation represents the hedonic price indexing model:

As inferred above, the equation can effectively be broken down into three individual components, with each component serving a specific role to ensure calculation integrity:

  1. Laspeyre’s formula compares the aggregate value of transactions at time  to the aggregate value of  transactions at base period
  2. Exponentiation diminishes the effects of heteroskedasticity, i.e., increasing errors with the broadening of observed price ranges
  3. The hedonic linear equation represents the estimated impact of variables at time  vis-a-vis their impact at base period , with the impact weight represented by the corresponding coefficients