House prices - first semester 2022
Statbel, the Belgian statistical office, publishes today the house prices for the first semester of 2022 in Belgium, based on the deeds of sale registered by the FPS Finances.
The main results are the following:
- The Belgian median price for an attached or semi-detached house amounted to 252,000 euros.
A buyer had to pay 357,000 euros for a detached house.
- The Walloon Region is the cheapest with a median price of 170,000 euros for an attached or semi-detached house, and 285,000 euros for a detached house.
- The Brussels-Capital Region is the most expensive region: attached and semi-detached houses cost there 486,000 euros. The price for a detached house amounted to 1,150,000 euros.
- The Flemish Region is in between. Attached and semi-detached houses cost there 280,000 euros, and detached houses cost 390,000 euros.
- Flemish Brabant and Walloon Brabant are the most expensive provinces. The cheapest Flemish province is Limburg, the cheapest Walloon province is Hainaut.
- The most expensive municipalities are Knokke-Heist and Ixelles. The cheapest houses can be found in Hastière.
The statistic on house prices is based on the deeds of sale registered by the FPS Finances, and more specifically the General Administration of the Patrimonial Documentation (GAPD), better known as the land registry. Until 2004, the necessary data were transmitted to Statbel on paper. From 2005 onwards, the data were transmitted in digital format, based on the registration system CADNET/LOCO. From 2016 onwards, the STIPAD system replaced the CADNET/LOCO system within the GAPD. This document describes the methodology used to calculate house prices.
Direct sales versus public sales
In the case of a direct sale of real estate, the buyer and the seller agree on a selling price.
In the case of a public sale, the property is sold according to the highest bid system. The sale is usually conducted under the direction of a notary in a public space (e.g. a café) and is announced in a local newspaper, for example.
Real estate prices as calculated by Statbel are based on both direct sales and public sales.
Preliminary sales agreement versus deed of sale
In the case of a direct sale of real estate, a preliminary sales agreement is first signed between the seller and the buyer. The signature of the notarial deed of sale takes place with a notary within 4 months of the signature of the preliminary sales agreement.
All preliminary sales agreements do not result in an actual sale. It is possible that the buyer does not find the necessary funding after the signature of the preliminary sales agreement because the bank refused the mortgage credit, for example.
Real estate prices as calculated by Statbel are based on the notarial deeds, i.e. on effective sales, and not on preliminary sales agreements.
In addition to the agreed sales price, the buyer still has to pay additional costs, such as registration fees (or VAT in the case of a new building) and the notary's fees for drawing up the deed. If you take out a mortgage loan, the mortgage registration fees and any bank processing fees are added. For the calculation of real estate prices, we only take the agreed sales prices into account, excluding additional fees.
Type of property
Residential real estate
The data set of the land registry, to which Statbel had access from 2005 to 2017 for the calculation of real estate prices, included the following categories of properties:
- regular houses
These categories are based on the nature mentioned on the cadastral plan. This nature is determined once the new building is completed. This nature is not always updated if the purpose of the building changes.
The breakdown into ‘regular house’ versus ‘villa’ is based on the (subjective) judgement of the evaluator of the land registry.
In the rest of the document, real estate prices calculated according to the breakdown above are mentioned as real estate prices according to the ‘former methodology’. From 2017 onwards, a more comprehensive data set, including a detailed description of the property, was made available to Statbel. It provides two additional variables: the nature according to the deed of sale and the construction code.
This additional information makes it possible to further optimise the methodology:
- In order to determine the type of building, we no longer use the nature mentioned in the cadastral plan, but the nature mentioned in the deed of sale. This nature is more up to date because it is identified (by the notary) at the time of each sale.
- Houses are now objectively broken down according to the number of facades. The former method made the distinction between regular houses and villas, but this qualification is subjective. In the modified methodology, we now distinguish attached and semi-detached houses (2 or 3 facades) and detached houses (4 facades or more).
In the rest of the document, real estate prices calculated based on the nature in the deed of sale are mentioned as real estate prices according to the ‘new methodology’.
The additional variables do not make it possible to calculate reference prices for building plots. The reason is that a large part of actual building plots was assigned to a residential category, in which we can find anything: agricultural lands, industrial lands, building plots, etc. This concerns more than 50 % of all building plots.
Other types of properties
In the same way, for non-residential properties (agriculture, commercial, etc.), the database of the land registry does not allow to calculate results. Properties in this category are very often sold in lots, with other plots with different purposes for a total price, so that it is not possible to calculate a price per individual property and therefore the median prices per category.
In the former methodology, all transactions that were in the summary data set of the land registry were included in the calculation. It also included part of the new constructions.
Thanks to the more detailed data set available to Statbel since 2016 and after a joint analysis with the land registry, it appeared that not all actual new constructions were included in the data set and that the data on new constructions that were included were not always accurate. The land registry’s database does not allow to build up a correct picture of the total sales of new constructions. For this reason, the new optimised methodology only takes into account the secondary real estate market (resale) and all transactions relating to new constructions are eliminated.
Cleansing of data sets
The data sets provided by the land registry are cleaned in two ways.
On the one hand, all transactions that do not belong to the statistical universe are eliminated. These transactions are not taken into account in the calculation of the reference price or of the number of transactions. This concerns for example transfers due to inheritances. In the optimised methodology, new constructions are also eliminated.
In the same way, transactions for which essential data are unknown - e.g. the date of sale or the type of property - are also eliminated.
On the other hand, filters are used to identify missing data. In this case, the missing data make it possible to include transactions in the calculation of the number of transactions, but they are not taken into account in the calculation of the reference price. This concerns for example transactions for which only the price is missing. Transactions relating to the sale of several properties in one lot for an overall price are also included in this category.
In the former methodology, the results calculated include average prices, median prices, percentile prices, the number of transactions, total prices and total surface areas.
In the new optimised methodology, the results calculated include median prices, percentile prices and the number of transactions.
The results are grouped per year, per semester or quarter, per type of dwelling and per location (municipalities, districts, provinces, regions and country).
Reference price: average price versus median price
For statistical breakdowns characterised by observations with extreme values (e.g. the sales price of real estate), it should be taken into account that the average, as the central measuring tool, is strongly influenced by these extreme values: an extreme value will steer the average towards it to a large extent.
The median price is a more appropriate measuring tool. This is the price against which 50% of transactions are cheaper and 50% more expensive. The influence of these extreme values on this measuring tool is minimal or non-existent: the median does not change, or hardly changes, if an extreme value is added to or removed from the calculation (unlike the average). So, the median price is a more stable measuring tool than the average price.
In Statbel's publications, the median price is used as reference price.
Percentile prices are additional measuring tools that, with the median price, give a better picture of the statistical breakdown of prices. The 25th percentile (also called the first quartile: Q25) is the price against which 25 % of transactions are as expensive or cheaper, while 75 % of transactions are as expensive or more expensive.
The 75th percentile (also called the third quartile: Q75) is the price against which 75 % of transactions are as expensive or cheaper, while 25 % of transactions are as expensive or more expensive.
Median and percentile prices are published only for 16 transactions with valid prices per aggregate, in order to achieve a certain degree of representativeness and to guarantee the confidentiality of individual data. The land registry’s database sometimes contains transactions with no valid price. These transactions are included in the total number of transactions, but not in the price calculation. If, for an aggregate, there is one or more transactions whose price is not valid, the number of transactions for which prices are shown will be greater than the limit of 16 transactions referred to above, in accordance with the number of transactions whose price is not valid.
Given the relatively high number of transactions for which the price is missing (especially since STIPAD), we no longer publish the total price in the new optimised methodology.