Census 2021 round (cens_21)

1.1. Contact organisation

Statbel (Directorate-general Statistics – Statistics Belgium)

1.2. Contact organisation unit

Databases Citizens

1.3. Contact name

Confidential because of GDPR

1.4. Contact person function

Confidential because of GDPR

1.5. Contact mail address

Boulevard du Roi Albert II, 16 - 1000 Brussels

1.6. Contact email address

Confidential because of GDPR

1.7. Contact phone number

Confidential because of GDPR

1.8. Contact fax number

Confidential because of GDPR

2.1. Metadata last certified

23 December 2022

2.2. Metadata last posted

6 November 2024

2.3. Metadata last update

6 November 2024

3.1. Data description

The data present the results of the 2021 EU census on population and housing, following Regulation (EC) 763/2008; Regulation (EU) 2017/543; Regulation (EU) 2017/712 and Regulation (EU) 2017/881.

3.1.1. Impact of the COVID-19 pandemic on census methodology

Belgium conducted a register based census. The COVID-19 pandemic had very little or no impact on the availability of administrative data. Therefore the impact of the pandemic on the Belgian census is negligible.

3.2. Classification system

The following classification systems should be used:

1. Nomenclature of Territorial Units for Statistics (NUTS 2021)

2. International Standard Classification of Education (ISCED-2011)

3. International standard classification of occupations (ISCO-08)

4. Statistical Classification of Economic Activities in the European Community (NACE Rev 2)

3.3. Coverage - sector

Not applicable.

3.4. Statistical concepts and definitions

The information is given separately for each census topic. See the sub-concepts 3.4.1 - 3.4.37.

3.4.1. Statistical concepts and definitions - Usual residence

‘Usual residence’ is the place where a person normally spends the daily period of rest, regardless of temporary absences for purposes of recreation, holidays, visits to friends and relatives, business, medical treatment or religious pilgrimage.

The following persons alone shall be considered to be usual residents of the geographical area in question:

(i)                 those who have lived in their place of usual residence for a continuous period of at least 12 months before the reference date

(ii)               those who arrived in their place of usual residence during the 12 months before the reference date with the intention of staying there for at least one year

Where the circumstances described in point (i) or (ii) cannot be established, ‘usual residence’ shall mean the place of legal or registered residence.

In the Belgian 2021 census, Statbel used the concept of usual residence, not the concept of registered residence.

3.4.2. Statistical concepts and definitions - Sex

‘Sex’ refers to the biological and physiological characteristics that define men and women.

This variable identifies males and females as indicated in the population register.

3.4.3. Statistical concepts and definitions - Age

The age reached in completed years at the reference date.

The definition used by Statistics Belgium is in line with the definition in the Eurostat Regulation.

3.4.4. Statistical concepts and definitions - Marital status

Marital status is defined as the (legal) conjugal status of an individual in relation to the marriage laws (or customs) of the country (that is the de jure status).

A person shall be classified according to his/her most recently acquired legal marital status at the reference date.

In Member States where the legislation includes provisions for married partners or partners in registered partnership to be ‘legally separated’, such ‘legally separated’ persons are classified under ‘Married or in registered partnership’.

Persons registered in the population register as being "séparées de corps et de biens" (legally separated) are included in the section "married" for the census.

Belgian legislation recognises the concept of legal cohabitation, a concept which corresponds only partly to that of registered partnership. For instance, there may be legal cohabitation between a brother and sister, or between a father and his son. In order to provide the most accurate assessment possible of the term registered partnership, individuals who are related to each other have been removed from the population of legally recognised cohabitors.

Note that marriages with persons outside the household are possible. The marital status is the legal situation. This is a big difference with the notion of a married couple used in a family nucleus.

3.4.5. Statistical concepts and definitions - Family status

The family nucleus is defined in a narrow sense; that is as two or more persons who belong to the same household and who are related as husband and wife, as partners in a registered partnership, as partners in a consensual union, or as parent and child. Thus, a family comprises a couple without children or a couple with one or more children, or a lone parent with one or more children. This family concept limits relationships between children and adults to direct (first-degree) relationships, that is, between parents and children.

'Skip-generation households' (households consisting of a grandparent or grandparents and one or more grandchildren, but no parent of those grandchildren) are not included in the definition of a family nucleus.

Consensual union is by definition impossible to measure directly via administrative registers. The population living as husband and wife without a contract within the same household has been evaluated either by the presence in the household of a common child, or by using an algorithm based on potential partnerships. In each household, all individuals aged 18 years or more of opposite sex, without a current spouse (in the same household) or registered partner have been identified; if only two persons meet these criteria and are not related to each other, they then form a consensual union couple. If a number of potential partners are present in the household, the couple with the smallest age gap is then declared as living in consensual union, provided that the age difference with the other potential partner(s) is 15 years or more. Otherwise, no consensual union is counted.

Couples married to each other can't be in consensual union. However, if a person is married with someone outside the household, he/she can be in consensual union because he/she isn't considered as part of a married couple inside the family nucleus.

3.4.6. Statistical concepts and definitions - Household status

Private households are identified using the ‘housekeeping concept’ or, if not possible, using the ‘household- dwelling concept’.

1) Housekeeping concept

According to the housekeeping concept, a private household is either: (a) a one-person household, that is a person who lives alone in a separate housing unit or who occupies, as a lodger, a separate room (or rooms) of a housing unit but does not join with any of the other occupants of the housing unit to form part of a multiperson household as defined below; or (b) a multiperson household, that is a group of two or more persons who combine to occupy the whole or part of a housing unit and to provide themselves with food and possibly other essentials for living. Members of the group may pool their incomes to a greater or lesser extent.

2) Household-dwelling concept

The household-dwelling concept considers all persons living in a housing unit to be members of the same household, such that there is one household per occupied housing unit.

 

The housekeeping concept has been retained. A number of households can therefore occupy the same dwelling.

The Population Register cannot be used to identify the homeless. While they form part of the population legally residing in Belgium, they are included in the total figure for the census population and are entered in the population register with an "administrative address". This concept of administrative address does not refer to the individual's actual place of residence but, for example, to a social assistance centre.

3.4.7. Statistical concepts and definitions - Current activity status

‘Current activity status’ is the current relationship of a person to economic activity, based on a reference period of one week, which may be either a specified, recent, fixed, calendar week, or the last complete calendar week, or the last seven days prior to enumeration.

'Employed' persons comprise all persons aged 15 years or over who during the reference week:
(a) performed at least one hour of work for pay or profit, in cash or in kind, or
(b) were temporarily absent from a job in which they had already worked and to which they maintained a formal attachment, or from a self-employment activity.
The 'unemployed' comprise all persons aged 15 years or over who were:
(a) 'without work', that is, were not in wage employment or self-employment during the reference week; and
(b) 'currently available for work', that is, were available for wage employment or self-employment during the reference week and for two weeks after that; and
(c) 'seeking work', that is, had taken specific steps to seek wage employment or self-employment within four weeks ending with the reference week.

For CAS, LPW, SIE and IND, the main data source is de social security register. The data concerning workers are quarterly, and it is possible to know whether the persons in question worked on the last day of the quarter, in our case 31 December 2020. The information on retired persons and the unemployed refers to the same date. Furthermore, for persons with more than one job, their main job is determined according to the time spent on it and on the income generated (for persons working abroad, only their taxable income is used).

With regard to the variable "Current activity status", the definitions used by Statistics Belgium correspond largely to those requested by Eurostat. Nevertheless, a number of adjustments had to be made before these definitions could be applied. As regards persons with a job, the data can easily be used to identify those working in Belgium. However, in this group, those aged 60 years or over are over-estimated. A number of corrections had to be made, taking account of earned income, information available through fiscal data. For persons working abroad (who appear as inactive in the data of the social security register), tax declarations also had to be used. However, since these data are based on the total for the year, only people who had an continuous income for both 2020 and 2021 are considered (very likely) as employed at the census reference moment.

The definition of unemployed used by Statistics Belgium meets the criteria a) and b) above. However, with the available databases, it is impossible to evaluate whether the persons are actively looking for a job, since no information on steps taken by jobseekers exists. Criterion c) therefore has not been met.

A census was taken of the student population using the education databases of the various communities which record those registered in the various teaching institutions. This information was supplemented by data on family benefits, because these benefits are received on the condition that the child is in education. This operation made it possible to recover specific sub-populations, such as pupils in home schooling.

The order of priorities (identification first of the working population, then of unemployed persons, and lastly of inactive persons) is respected.

3.4.8. Statistical concepts and definitions - Occupation

Occupation refers to the type of work done in a job. ‘Type of work’ is described by the main tasks and duties of the work.

Persons are classified according to the occupation they had during their most recent employment. The categories included in the breakdown 'occupation' correspond to the major groups of the ISCO-08 (COM) classification.

Persons under the age of 15 years, as well as persons aged 15 or over that were:

— not economically active during the reference week, or

— unemployed, never worked before (i.e. they have never been employed in their lives)

are classified under 'not applicable'.

The allocation of a person within the breakdowns of the topics 'Occupation', 'Industry' and 'Status in employment' is based on the same job.

3.4.9. Statistical concepts and definitions - Industry

Industry (branch of economic activity) refers to the kind of production or activity of the establishment or similar unit in which the job of an employed person is located.

Persons doing more than one job shall be allocated an industry (branch of economic activity) based on their main job, which is to be identified according to the time spent on the job or (if not available) the income received.
The breakdown by industry is available for persons aged 15 or over that were:
— employed during the reference week, or
— unemployed during the reference week, but have ever been in employment
Persons are classified according to the industry they worked in during their most recent employment. The categories included in the breakdown 'industry' list the 21 sections of the NACE Rev. 2 classification and appropriate aggregates

3.4.10. Statistical concepts and definitions - Status in employment

An ‘employee’ is a person who works in a ‘paid employment’ job, that is a job where the explicit or implicit contract of employment gives the incumbent a basic remuneration, which is independent of the revenue of the unit for which he/ she works. An ‘employer’ is a person who, working on his or her own account or with a small number of partners, holds a ‘self- employment’ job and, in this capacity, on a continuous basis (including the reference week) has engaged one or more persons to work for him/her as ‘employees’.

The definitions used by Statistics Belgium are consistent with those of Eurostat.

3.4.11. Statistical concepts and definitions - Place of work

The location of the place of work is the geographical area in which a currently employed person does his/her job.

The place of work of those mostly working at home is the same as their usual residence. The term ‘working’ refers to work done as an ‘employed person’ as defined under the topic ‘Current activity status’. ‘Mostly’ working at home means that the person spends all or most of the time working at home, and less, or no, time in a place of work other than at home.

Information on persons who do not have a fixed place of work but who report to a fixed address at the beginning of their work period (for example bus drivers, airline crew, operators of street market stalls that are not removed at the end of the workday) refer to that address. This group may also include individuals who travel to work, on a regular basis, across the border to a neighbouring country. Breakdown ‘No fixed place of work (inside or outside the Member State)’ includes all persons without fixed place of work but will also refer to persons such as sailors, fishermen and offshore workers for whom it may not be possible to allocate the place of work.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

For employed workers, the information on the place of work is available in the social security register.

For self-employed workers, the information on the place of work is not available in the social security register. It had to be searched for in the company database:

- If the person is the authorised representative of a business which has only one establishment, the address of the business shall be considered as his place of work.

- If the business has several establishments, the profession code (specific code of business register) of the person is matched with the activity code of each of the establishments in order to allocate the person to the most relevant establishment.

Furthermore, the database of businesses is incomplete. If the person cannot be found on this database, his professional address shall be that of his private address providing that a company registered on that address.

3.4.12. Statistical concepts and definitions - Educational attainment

Educational attainment refers to the highest level successfully completed in the educational system of the country where the education was received. All education which is relevant to the completion of a level is taken into account even if this was provided outside schools and universities.

Persons aged 15 years or over are classified under only one of the categories according to their educational attainment (highest completed level). Persons under the age of 15 years are classified under 'not applicable'.

The definition used by Statistics Belgium is identical to that recommended by Eurostat.

3.4.13. Statistical concepts and definitions - Size of the locality

A locality is defined as a distinct population cluster that is an area defined by population living in neighboring or contiguous buildings.

Such buildings may be either:
(a) form a continuous built-up area with a clearly recognizable street formation; or
(b) though not part of such a built-up area, comprise a group of buildings to which a locally recognized place name is uniquely attached; or
(c) though not meeting either of the above two criteria, constitute a group of buildings, none of which is separated from its nearest neighbor by more than 200 meters.

 

Statistics Belgium applies criterion c.

Almost each building has geographical coordinates. Based on that coordinates, (c) can be applied and the localites can be computed. In a few cases, some households cannot be coupled to a cadastre parcel. Is geographical coordinates of the household are known, the locality is the locality of the nearest building where coordinates are known. In the other very rare cases which rest, values were imputed in cases where households could not be coupled to a dwelling. If the statistical sector of the household's residence comprises only one locality, it is attributed to the household. However, if a number of localities make up the statistical sector, non-coupled households are divided between them in proportion to the distribution observed for coupled households.

3.4.14. Statistical concepts and definitions - Place of birth

Information on the ‘Place of birth’ is collected according to the place of usual residence of the mother at the time of the birth, or, if not available, the place in which the birth took place, on the basis of international boundaries existing on 1 January 2021.

For reporting countries that are EU Member States, the sub-category under the category 'other EU Member State' that refers to their Member State does not apply. For reporting countries that are not EU Member States, the category 'Other EU Member State' shall be read as 'EU Member State'.
The category 'Information not classifiable according to current borders' covers those persons whose country of birth existed at the time of the birth, but no longer exists at the time of the census, and who cannot be allocated uniquely to one country existing at the time of the census, i.e. according to current borders.
The category 'outside any country' covers persons for whom the usual residence of the mother at the time of the birth is not known and who where born outside the borders of any country, for example at sea or in the air.

The place of birth is always defined as the place where birth actually took place; the place of residence of the mother at the time of birth is entered in the population register but isn't always known (especially in the case of migrants where the mother never stayed in the country).

3.4.15. Statistical concepts and definitions - Country of citizenship

Citizenship is defined as the particular legal bond between an individual and his/her State, acquired by birth or naturalization, whether by declaration, option, marriage or other means according to the national legislation. A person with two or more citizenships is allocated to only one country of citizenship, to be determined in the following order of precedence:

1. reporting country

2. if the person does not have the citizenship of the reporting country: other EU Member State

3. if the person does not have the citizenship of another EU Member State: other country outside the European Union.

'EU Member State' means a country that is a member of the European Union on 1 January 2021. The list of countries in the breakdown 'Country of citizenship' shall only apply for statistical purposes.
For reporting countries that are EU Member States, the sub-category of the category 'citizenship not of reporting country, but other EU Member State' that refers to their Member State does not apply. For reporting countries that are not EU Member States, the category 'citizenship not of reporting country, but other EU Member State' shall be changed to 'citizenship of an EU Member State'.
Persons who are neither citizens of any country nor stateless and who have some but not all of the rights and duties associated with citizenship shall be classified under 'recognised non-citizens'.

The definition used by Statistics Belgium is identical to that requested by Eurostat. Belgium does not take into consideration the concept of "recognised non-citizens".

3.4.16. Statistical concepts and definitions - Year of arrival in the country

The year of arrival is the calendar year in which a person most recently established usual residence in the country.

The data for 2021 refer to the time span between 1 January 2021 and the reference date.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.4.17. Statistical concepts and definitions - Residence one year before

The relationship between the current place of usual residence and the place of usual residence one year prior to the census.

For all persons that have changed their usual residence more than once within the year prior to the reference date, the previous place of usual residence is the last usual residence from which they moved to their current place of usual residence.

The definition used by Statistics Belgium is identical to that requested by Eurostat. The population register is a system for the ongoing collection of information, and all the successive residences of inhabitants are recorded in it. Identifying the place of residence of the population one year before the census therefore does not pose a major problem.

3.4.18. Statistical concepts and definitions - Housing arrangements

The topic ‘Housing arrangements’ covers the whole population and refers to the type of housing in which a person usually resides at the time of the census. This covers all persons who are usual residents in different types of living quarters, or who do not have a usual residence and stay temporarily in some type of living quarters, or who are roofless, sleeping rough or in emergency shelters, when the census is taken.

'Other housing units' are huts, cabins, shacks, shanties, caravans, houseboats, barns, mills, caves or any other shelter used for human habitation at the time of the census, irrespective if it was designed for human habitation.
'Collective living quarters' are premises which are designed for habitation by large groups of individuals or several households and which are used as the usual residence of at least one person at the time of the census.
'Occupied conventional dwellings', 'other housing units' and 'collective living quarters' together represent ‘living quarters'. Any 'living quarter' must be the usual residence of at least one person.

The sum of occupied conventional dwellings and other housing units represents 'housing units'.
The homeless (persons who are not usual residents in any living quarter category) can be persons living in the streets without a shelter that would fall within the scope of living quarters (primary homelessness) or persons moving frequently between temporary accommodation (secondary homelessness).

The information contained in the register of the building and dwelling register can be used to identify the occupants of standard housing units and of other housing units, with the exception of caravans, boats and huts.

The occupants of a collective dwelling are identified in the the population register.

However, the population register cannot be used to identify the homeless. While they form part of the population legally residing in Belgium, they are included in the total figure for the census population and are entered in the population register with an "administrative address". This concept of administrative address does not refer to the individual's actual place of residence but, for example, to a social welfare centre or a body responsible for handling asylum requests.

3.4.19. Statistical concepts and definitions - Type of family nucleus

The family nucleus is defined in a narrow sense; that is as two or more persons who belong to the same household and who are related as husband and wife, as partners in a registered partnership, as partners in a consensual union, or as parent and child. Thus a family comprises a couple without children or a couple with one or more children, or a lone parent with one or more children. This family concept limits relationships between children and adults to direct (first-degree) relationships, that is, between parents and children.

Child (son/daughter) refers to a blood, step- or adopted son or daughter (regardless of age or marital status) who has usual residence in the household of at least one of the parents, and who has no partner or own children in the same household. Foster children are not included. A son or daughter who lives with a spouse, with a registered partner, with a partner in a consensual union, or with one or more own children, is not considered to be a child. A child who alternates between two households (for instance if his or her parents are divorced) shall consider the one where he or she spends the majority of the time as his or her household. Where an equal amount of time is spent with both parents the household shall be the one where the child is found at the time on census night.
The term couple shall include married couples, couples in registered partnerships, and couples who live in a consensual union.
Registered partnerships are reported only by countries that have a legal framework regulating partnerships that:
(a) lead to legal conjugal obligations between two persons;
(b) are not marriages;
(c) prevent persons to commit themselves into multiple partnerships with different partners.
Two persons are considered to be partners in a 'consensual union' when they
— belong to the same household, and
— have a marriage-like relationship with each other, and
— are not married to or in a registered partnership with each other.
'Skip-generation households' (households consisting of a grandparent or grandparents and one or more grandchildren, but no parent of those grandchildren) are not included in the definition of a family.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

Married couples are identified using the population register, thanks to a code which specifies the relationship between the members of the household and the information on marital stutus. For couples in a registered partnership or in a consensual union, see points 3.4.4 and 3.4.5, respectively.

Parent-child links are identified using the population register by means of the relationship code and information on filiation.

3.4.20. Statistical concepts and definitions - Size of family nucleus

The family nucleus is defined in a narrow sense; that is as two or more persons who belong to the same household and who are related as husband and wife, as partners in a registered partnership, as partners in a consensual union, or as parent and child. Thus a family comprises a couple without children or a couple with one or more children, or a lone parent with one or more children. This family concept limits relationships between children and adults to direct (first-degree) relationships, that is, between parents and children.

3.4.21. Statistical concepts and definitions - Type of private household

Private households are identified using the ‘housekeeping concept’ or, if not possible, using the ‘household- dwelling concept’.

1) Housekeeping concept

According to the housekeeping concept, a private household is either: (a) a one-person household, that is a person who lives alone in a separate housing unit or who occupies, as a lodger, a separate room (or rooms) of a housing unit but does not join with any of the other occupants of the housing unit to form part of a multiperson household as defined below; or (b) a multiperson household, that is a group of two or more persons who combine to occupy the whole or part of a housing unit and to provide themselves with food and possibly other essentials for living. Members of the group may pool their incomes to a greater or lesser extent.

2) Household-dwelling concept

The household-dwelling concept considers all persons living in a housing unit to be members of the same household, such that there is one household per occupied housing unit.

Statistics Belgium applied  applied the housekeeping concept.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.4.22. Statistical concepts and definitions - Size of private household

Private households are identified using the ‘housekeeping concept’ or, if not possible, using the ‘household- dwelling concept’.

1) Housekeeping concept

According to the housekeeping concept, a private household is either: (a) a one-person household, that is a person who lives alone in a separate housing unit or who occupies, as a lodger, a separate room (or rooms) of a housing unit but does not join with any of the other occupants of the housing unit to form part of a multiperson household as defined below; or (b) a multiperson household, that is a group of two or more persons who combine to occupy the whole or part of a housing unit and to provide themselves with food and possibly other essentials for living. Members of the group may pool their incomes to a greater or lesser extent.

2) Household-dwelling concept

The household-dwelling concept considers all persons living in a housing unit to be members of the same household, such that there is one household per occupied housing unit.

 

'Primary homeless persons' are persons living in the streets without a shelter that would fall within the scope of living quarters.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.4.23. Statistical concepts and definitions - Tenure status of households

The topic ‘Tenure status of households’ refers to the arrangements under which a private household occupies all or part of a housing unit.

The population register has been coupled with the register of the Administration générale de la Documentation patrimoniale (the "cadastre") to determine the tenure status of the household.

The data taken from the dwelling register can be used to identify individuals with a title deed to the housing in which they live. Households in which none of the members own at least part of the housing unit are classed in the category "Households in which at least one of the members is the tenant of all or part of the housing unit". The available sources of information do not make it possible to identify "Households occupying all or part of a housing unit under some other form of tenure".

3.4.24. Statistical concepts and definitions - Type of living quarter

A living quarter is housing which is the usual residence of one or more persons.

'Conventional dwellings' are structurally separate and independent premises at fixed locations which are designed for permanent human habitation and are, at the reference date, either used as a residence, or vacant, or reserved for seasonal or secondary use.
'Separate' means surrounded by walls and covered by a roof or ceiling so that one or more persons can isolate themselves. 'Independent' means having direct access from a street or a staircase, passage, gallery or grounds.
'Other housing units' are huts, cabins, shacks, shanties, caravans, houseboats, barns, mills, caves or any other shelter used for human habitation at the time of the census, irrespective if it was designed for human habitation.
'Collective living quarters' are premises which are designed for habitation by large groups of individuals or several households and which are used as the usual residence of at least one person at the time of the census.
'Occupied conventional dwellings', 'other housing units' and 'collective living quarters' together represent‘living quarters'. Any 'living quarter' must be the usual residence of at least one person.

 

The information in the register of the dwelling register makes it possible to identify "other housing units", with the exception of caravans, boats and huts.

Collective dwellings are identified in the population register.

3.4.25. Statistical concepts and definitions - Occupancy status

‘Occupied conventional dwellings’ are conventional dwellings which are the usual residence of one or more persons at the time of the census. ‘Unoccupied conventional dwellings’ are conventional dwellings which are not the usual residence of any person at the time of the census.

The definition used by Statistics Belgium is identical to that requested by Eurostat. The population register is linked with the dwelling register. Occupancy status can be derived from the matching of those 2 registers.

3.4.26. Statistical concepts and definitions - Type of ownership

‘Type of ownership’ refers to the ownership of the dwelling and not to that of the land on which the dwelling stands. It shows the tenure arrangements under which the dwelling is occupied.

'Owner-occupied dwellings' are those where at least one occupant of the dwelling owns parts or the whole of the dwelling.
'Rented dwellings' are those where at least one occupant pays a rent for the occupation of the dwelling, and where no occupant owns parts or the whole of the dwelling.

Dwellings are either occupied by their owner or considered as rented; no information is available on the category: "dwellings in other types of ownership" (see point 3.4.23.) Where several households occupy the same dwelling and at least one person in one of the households is the owner, the dwelling is considered to be owner-occupied.

3.4.27. Statistical concepts and definitions - Number of occupants

The number of occupants of a housing unit is the number of people for whom the housing unit is the usual residence.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.4.28. Statistical concepts and definitions - Useful floor space

Useful floor space is defined as the floor space measured inside the outer walls excluding non-habitable cellars and attics and, in multi-dwelling buildings, all common spaces; or the total floor space of rooms falling under the concept of 'room'.

The number of rooms is reported instead of the useful floor space.

3.4.29. Statistical concepts and definitions - Number of rooms

A ‘room’ is defined as a space in a housing unit enclosed by walls reaching from the floor to the ceiling or roof, of a size large enough to hold a bed for an adult (4 square meters at least) and at least 2 meters high over the major area of the ceiling.

 The definition of the dwelling register of the uses the 4m² criterion, but no reference is made to ceiling height.

3.4.30. Statistical concepts and definitions - Density standard (floor space)

The topic ‘Density standard (floor space)’ relates the useful floor space in square meters to the number of occupants, as specified under the topic ‘Number of occupants’.

The denisty standard (number of rooms) is reported instead.

3.4.31. Statistical concepts and definitions - Density standard (number of rooms)

The topic ‘Density standard (number of rooms)’ relates the number of rooms to the number of occupants, as specified under the topic ‘Number of occupants’.

This indicator refers to the number of rooms per occupant.

3.4.32. Statistical concepts and definitions - Water supply system

Whether the conventional dwelling is equipped with piped water.

Based on sample surveys in the past, we know the percentage of dwellings with running water is almost 100%: a decision was therefore made to count all dwellings as being connected to running water.

3.4.33. Statistical concepts and definitions - Toilet facilities

Whether the conventional dwelling is equipped with toilet facilities.

Based on sample surveys in the past, we know the percentage of dwellings with toilet facilities is almost 100%: a decision was therefore made to count all dwellings as having toilet facilities.

3.4.34. Statistical concepts and definitions - Bathing facilities

Whether the conventional dwelling is equipped with bathing facilities.

The data from the dwelling register make it possible to calculate this variable directly for the vast majority of dwellings.

However, it has been noted that the dwelling register underestimates the number of bathrooms and renovations of dwellings, which are incorrectly recorded. Therefore, information of both the dwelling register and the 2011 census are used to derive the variable "bathing facilities". The algorithm used therefore stipulates that any dwelling built after 2001 has a bathroom. (last traditional census) With regard to older dwellings:

- Where the dwelling register or the 2011 census mentions a bathroom, the dwelling is considered to have a bathroom.

- Where both sources explicitly mention the absence of a bathroom, the dwelling is classed as being without a bathroom.

- Where the situation is unknown in the 2011 census, it is still considered to be unknown in 2021.

3.4.35. Statistical concepts and definitions - Type of heating

Conventional dwelling is considered as centrally heated if heating is provided either from a community heating centre or from an installation built in the building or in the conventional dwelling, established for heating purposes, without regard to the source of energy.

Because of an under registration of central heating in the dwelling register, the 2011 census is combined with the dwelling register to derive the information of type of heating.

The algorithm for all dwellings is as follows:

- Where one of the two sources (dwelling register or 2011 census) mentions the existence of central heating, the dwelling is equipped with central heating.

- Where both sources specify the absence of central heating, this information is used.

- Where the situation is unknown in the 2011 census, it is still considered to be unknown in 2021.

3.4.36. Statistical concepts and definitions - Type of building

The topic ‘Dwellings by type of building’ refers to the number of dwellings in the building in which the dwelling is placed.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.4.37. Statistical concepts and definitions - Period of construction

The topic ‘Dwellings by period of construction’ refers to the year when the building in which the dwelling is placed was completed.

The definition used by Statistics Belgium is identical to that requested by Eurostat.

3.5. Statistical unit

The EU programme for the 2021 population and housing censuses includes data on persons, private households, family nuclei, conventional dwellings and living quarters.

3.6. Statistical population

The persons enumerated in the 2021 census are those who were usually resident in the territory of the reporting country at the census reference date.

3.7. Reference area

Data are available at different levels of geographical detail in EU countries: national, NUTS2/NUTS3 regions and local administrative units (LAU), grids.

3.8. Coverage - Time

Data refer to the situation in the reporting country at the census reference date.

For Belgium is this 1st January 2021.

3.9. Base period

Not applicable.

Counts of statistical units should be expressed in numbers and where is needed rate per inhabitants enumerated in the country.

Information is provided in the sub-concepts 5.1 - 5.3.

5.1. EU census reference date

1 January 2021

5.2. National census reference date

1 January 2021

5.3. Differences between reference dates of national and EU census publications

No differences.

6.1. Institutional Mandate - legal acts and other agreements

National legal background for the population and housing censuses, as requested by Reg. 2017/881, Annex point 1.1.

The census in Belgium is regulated at the European level.

Regulation (EC) 763/2008; Regulation (EU) 2017/543; Regulation (EU) 2017/712 and Regulation (EU) 2017/881.

There isn't a specific law at national level for the 2021 census. Of course the Belgian statistical law of 4 July 1962 applies here.

Law of 4 July 1962 on official statistics (ammended by the Act of 1 August 1985 on tax and other measures, the Act of 21 December 1994 on social and other provisions, the Law of 2 January 2001 on social provisions, Budget and miscellaneous and by the program law of 22 December 2008).

6.1.1. Bodies responsible

Direction générale Statistique – Statistics Belgium

6.2. Institutional Mandate - data sharing

Not applicable.

7.1. Confidentiality - policy

Confidentiality - Regulation (EU) 2017/712 Art 4 and 5

7.2. Confidentiality - data treatment

Confidentiality - Regulation (EU) 2017/712 Art 4 and 5

by Regulation 2017/881, Annex point 3.4.

There was some research in order to decide which SDC-method to use.

First the microdata were aggregated without SDC. These results were checked for sensitive information. The first conclusion was that for the hypercubes as well for the GRID-data, there are not a lot of sensible frequency counts. So, a very light SDC-method should suffice.

Different SDC-methods were considered:

• Cell key method was never considered as a serious candidate because of the lack of additivity. Our dissemination team found it too difficult to explain to the general public that additivity is not preserved.
• We performed record swapping in 2011, but due to some inconsistency problems in 2011 for the household data, there was very little support to use this method again. There was also a lot of criticism from census users regarding the record swapping method.
• Also, cell suppression was considered as a bad choice due to reasons Eurostat mentioned in several meetings.

Statbel came up with a new method we could call "geographical perturbation".  This method only works for a very light SDC approach when only a very small percentage of the frequency counts is considered as confidentially problematic. This method is now applied for the grid data. Briefly explained, this method works as follows:

It is a pretabulated method, which means some of the modifications is done in the microdata before aggregation.

1. Based on some confidentiality rules, we determine which grid cells have to be protected. These cells are determined on the basis of the aggregated data before a SDC method is applied.
2. Then we add some perturbations to the microdata. Households in a grid cell to protect are virtually moved to a point with different geographical coordinates. This is a point in another grid cell that should not be protected. This point is always located in the same municipality (same LAU2). We try to choose to the greatest extent possible a point in the same neighbourhood. In Belgium we have some statistical geographical areas that are very detailed. We call them the statistical sectors and these are more detailed than the LAU2 level. Statistical sectors can be considered as the neighbourhoods and are very important for our national dissemination. We choose, if possible, the nearest point (=inhabited building) in the same statistical sector. But (in some rare cases) if this point is too far away, we choose a point closer to the original point in another statistical sector, but always in the same LAU2 area.
3. Grid cells are recalculated based on the perturbated points. After that, the dataset is aggregated again.

Of course, the result of this method depends on the confidentiality rules used in step (1). We have tested this method where we started from different confidentiality rules, used in step 1 and we've compared the results with each other. These rules in step 1 are criteria based on frequency counts in grid cells such as number of people, number of households and the other frequency counts by grid cell we have to deliver to Eurostat in 2024. We compared relatively mild criteria with more severe criteria. After analysing the results, the conclusion was that one of the mildest criteria was sufficient for the protection of confidential data. Keeping the detailed rules secret is an important aspect to prevent anyone from being able to undo the protection.

There were no additional confidentiality rules applied for the other datasets.

8.1. Release calendar

The schedule of census release dates.

By the end of 2022, population data on grid level will be available.

The national publication of the census is planned very soon after the delivery to Eurostat (2nd half of April 2024).

A workshop on Census is scheduled for October 2024.

8.2. Release calendar access

Access to the release calendar information at national level.

The national publication of the census will be very soon after the delivery to Eurostat (2nd half of April 2024).

A workshop on Census is scheduled for October 2024.

8.3. Release policy - user access

In line with the Community legal framework and the European Statistics Code of Practice Eurostat disseminates European statistics on its website respecting professional independence and in an objective, professional and transparent manner in which all users are treated equitably.

Decennial.

Information is provided in the sub-concepts 10.1 - 10.7.

10.1. Dissemination format - News release

Press release(s) concerning the 2021 census, as part of the requirements of Regulation 2017/881, Annex point 3.4.

April 2024.

10.2. Dissemination format - Publications

Publications concerning the 2021 census, as part of the requirements of Regulation 2017/881, Annex point 3.4.

April 2024.

10.3. Dissemination format - online database

On-line database on the 2021 census, as part of the requirements of Reg. 2017/881, Annex point 3.4.

There will be an online database available. But too early to communicate a link.

10.4. Dissemination format - microdata access

Census data should be accessible as microdata, as part of the requirements of Regulation 2017/881, Annex point 3.4.

According to the Belgian statistical law of 4 July 1962, researchers have access to microdata. It concerns pseudonymised data. Researchers can submit an application to the DPO committee of statbel. Based on the principles of finality and proportionality (described in Belgian law), an assessment is made for researcher's application.

Finality = clear description of the purpose for which the data will be used. The researcher can only use the data for the purposes discribed in his application.

Proportionality = The researcher can only request the data that are really necessary for his research. (e.g. if the researcher only needs data on NUTS 3 level, he can't obtain data on LAU 2 level)

Annexes:
access to microdata for research

10.5. Dissemination format - other

Some open data sets will be available on Statbel's website.

10.6. Documentation on methodology

There are different methodological notes available:

• On the labour related variables => see in the annexes
• data on education => not yet translated to English
• We participated also for a grant on education data. You can consult the grant report.
• For demography and dwelling data, there exist detailed powerpoint presentations in dutch.

10.7. Quality management - documentation

See point 11.

Annexes:
Information on quality

11.1. Quality assurance

Inside Statistics Belgium, a service which is independent of the service responsible for producing figures, is responsible for the final quality management of the data. Each year there is a survey that has to be filled in for all statistics that evaluates the different quality dimensions of the statistic.

See also the attached link.

 

Annexes:
Information on quality

11.2. Quality management - assessment

• Additivity in the tables are checked.
• Consistency between tables is checked.
• Validation rules in tables should reveal any suspicious values.
• Values are compared to other statistics.
• Values are compared longitudinally with previous ones to see if differences are plausible or can be explained.

e.g. cross tabulations of the variables with age are very important to detect suspicious values, such as very young people that are married, number of very old people still working, etc.

For the grid data:

• Additivity of population figures and surface data were checked.
• Comparison was performed with the demographic statistics already published. The numbers match exactly. The total area is comparable with cadaster results.
• For each grid cell, the values of 2021 were compared with the grid in 2016. The differences are very acceptable. 

Annexes:
Quality check grid data 2021

11.2.1. Coverage assessment

The population register does not completely cover the de facto population due primarily to the under-declaration of arrivals in Belgium. For instance, there is of course no information on the population in an illegal situation. Furthermore, some foreigners do not register even though they have a residence permit; in particular Europeans who do not wish to settle permanently in the country. Foreigners whose duration of stay does not exceed three months are not required to register. However, non-registration of the latter category of foreigners does not pose a problem for the 2021 census because only individuals residing in the country for more than one year or who intend to stay there for longer than a year must be taken into account.

Departures abroad also seem to suffer from under-registration, for both the Belgian and the foreign population. Although some persons do not declare even their definitive departure from Belgium, this may be on account of insufficient knowledge of the law, to avoid an administrative procedure deemed to be tedious or to keep benefits which depend on residence.

Nevertheless, such under-recording should be kept in perspective: on the one hand, registering in a municipality is of obvious interest for permanent residents and, on the other hand, controls are run on the accuracy of declarations from individuals. These primarily involve identifying individuals who have left the territory of the municipality permanently without having made a prior declaration. For example, if no response to administrative letters is received, or when a new occupant in a dwelling makes a registration request, the municipality can conduct a procedure to verify the address. Moreover, foreigners whose residence permit expires are systematically deleted from the register after eight days. The municipalities also look into undeclared arrivals and may conduct inquiries based, for example, on the opening of electricity meters or telephone lines.

Following undeclared departures, the municipalities carry out automatic removals from the population register. If automatically removed individuals are re-registered before 1 March of the following year, Statistics Belgium considers these removals to be incorrect. These removals are therefore cancelled and the information on the address of re-registration is used on the date of the incorrect removal. These removals considered to be incorrect therefore result in the cancellation of around half of automatic removals.

The population register is used to build the population database, which is used to create all demographic variables, including those concerning the nuclear family and the household. The housing database is created on the basis of the register of the Administration Générale de la Documentation Patrimoniale (land register). The main objective of this administrative source is the collection of taxes on real estate. It is possible that some dwellings are not included in this register, particularly where dwellings are subdivided.

11.2.2. Post-enumeration survey(s)

Statbel didn't conduct any post-enumeration surveys for the census 2021.

12.1. Relevance - User Needs

There are different working groups and task forces with different kind of users.

Each year the different users inside the IIS (Interfederal Institute for Statistics) can submit projects to improve statistics, develop new statistics or can ask to add new information in existing statistics.

There are also multiple task forces with researchers to discuss this.

In the high council of statistics, the most important stakeholders come together to follow up the statistical production. Specific working groups can be established if some new needs are expressed or if a reform of a particular statistic is necessary.

12.2. Relevance - User Satisfaction

See attached link.

Annexes:
satisfaction survey

12.3. Completeness

All the variables in the EU census regulation are available.

For national purposes, we have some additional variables: e.g. study domein, more detailed classification of some variables, etc.

At a regular basis there are task forces with users to know their needs for data and define priorities for new data.

13.1. Accuracy - overall

Information is provided in the sub-concepts 13.1.1 - 13.1.35.

13.1.1. Overall accuracy - Usual residence

Highly accurate.

Statbel uses the concept of usual residence and tries to deduce this from the data in the population register. The population register disposes of longitudinal information. Therefore, we have information of moving citizens. Based on that information, it's possible to see if condition (i) of the definition of usual resident population is satisfied:

"(i) those who have lived in their place of usual residence for a continuous period of at least 12 months before the reference date"

Indeed, we can check whether someone lives already for 12 months in Belgium.

It's more difficult to see if condition (ii) is satisfied:

"(ii) those who arrived in their place of usual residence during the 12 months before the reference date with the intention of staying there for at least one year"

In the population register, we have no information on the intention to stay. But, based on the records with information of moving citizens in the months after the reference date, it is possible to estimate this. 

Remarks

• This differs from the figures statbel publishes on its own website, where the registered population is used at midnight 31 December 2020 (=1st January 2021). Moreover, asylum seekers are included in the figures we submit to Eurostat which isn't the case in our national publications.
• Both census statistics and demography use the same reference moment and data source (population register). As a consequence, the demographic part of the census is consistent with the demographic statistics published earlier.

Once the usual resident population is determined, we can precise with more geographical detail the usual residence. Municipality codes are available in the population register. Therefore, all levels NUTS1,2,3 and LAU2 can be derived from that information. To see in which km² grid cell the citizen has his/her usual residence, we use the address from the population register.

There is a geocoding of these addresses in what we call the CSAB file established by Statbel.

In a first step we require that members of the same household are attributed to the same geographical point. Therefore, it is important to determine the household id's before a grid cell id can be allocated. The composition of the households is also derived from information in the population register. Here, we distinguish between private households and collective living quarters. Private households are mainly determined based on information of the reference person in the household, whereas collective living quarters are determined by the information in the address.

To establish the CSAB file (geocode the addresses), the information in the population register has to be compared with geographical data coming form databases from the 3 regions (each NUTS 1 region has his own database). There is an automated step first. To compare addresses between the population register and the geographical data, intelligent algorithms are used as e.g. based on TF-IDF methods. Sometimes the exact address isn't available as such in one of the geographical registers. e.g. for a given street we have nr. 20 in the population register, but in the geographical databases we have nr. 18 and nr. 22. In those cases geographical interpolation can be used to estimate the geographical coordinates. In some other cases extrapolation can be used.

After that there is a manual step to geocode the rest of the addresses whereby geographical coordinates where not found in the automated step. A possible example of a manual treatment is an address on the corner of 2 streets where the population register allocated one street, but the geographical registers the other street. The geographical coordinates can be found by consulting a map. 

The CSAB file is updated each year with all new addresses. In the Belgian geographical databases, Lambert 72 (EPSG:31370) projection system is used. To attribute a grid ID, this is converted to the ETRS89 / LAEA Europe (EPSG:3035) system.

It is difficult to mention the exact accuracy of the geographical points, since this can change from address to address. Interpolated and extrapolated points are far less precise than addresses available in the original geographical databases.  Addresses in the original geographical databases are precise to the metre and sometimes even more detailed. Only 0,32% of the addresses is inter- or extrapolated. Exceptionally, it can be less precise for some extrapolated points. Mostly, only for points very near to the border of 2 different grid cells there is a very small risk the address is allocated to the wrong grid cell.

For the km² grid, not only population figures were supplied, but also land surface data. In most cases this is exactly 1 km². For border and coastal areas, the area was computed based on a GIS intersection of the GRID layer with the layer of the official Belgian borders both converted to the ETRS89 projection. Moreover, areas of water bodies such as lakes were subtracted from the initial surface. This information is estimated based on Corine Land Cover 2018 data.

13.1.2. Overall accuracy - Sex

Highly accurate.

13.1.3. Overall accuracy - Age

Highly accurate.

There are no particular reasons for data unreliability for this topic

13.1.4. Overall accuracy - Marital status

Registered partnership is not directly measured, but derived from legal cohabition. In order to provide the most accurate assessment possible of the term registered partnership, individuals who are related to each other have been removed from the population of legally recognised cohabitors.

13.1.5. Overall accuracy - Family status

Consensual union is not measured directly but is evaluated using an algorithm which may be the source of bias. See 3.4.5. for more details.

13.1.6. Overall accuracy - Household status

Some children are registered alone in a household. It rarely occurs (approximately 5000 cases younger than 15 years old), but it could happen. There are several reasons why this could happen:

• Wrong registrations in the population registers.
• Administrative delay
• Temporary situation (in combination with an administrative delay)

An example where this can happen is where a parent died just before the census moment (in a single parent household). It can be possible there’s an administrative delay in the address change in the population register. So the child is registered temporary as a one-person household at the census date in this case.

13.1.7. Overall accuracy - Current activity status

With regard specifically to the variable “Current activity status (CAS)”, the modalities and definitions used by Statbel correspond to those of Eurostat, and the order of priorities in the situation on the labour market (first identifying the employed population, then the unemployed and finally the inactive) is respected. Nevertheless, the implementation required some assumptions and adaptations in line with the Belgian legislation.

 

Generally speaking, the administrative data make it easy to identify the employed persons in Belgium. For people working abroad, Statbel assumes that those defined as employed persons are those who are neither unemployed nor in retirement and who are either registered with a Belgian mutual insurance as a person working abroad or who declare having received an income during the years before and after the Census. In this way, they are more likely to be working during the reference week.

Moreover, not all workers in the European and international institutions are detected. Indeed, only the workers in the European Parliament, the European Commission, the European Council and the European Economic and Social Committee as well as people who have declared in their tax return that they work in a European or international institution are considered as employed.

We also requested data from the NATO, but didn't receive them yet. We've got some aggregated data for (almost) all European institutions in Belgium to get an idea of the coverage of the employed population in the census. So, there's a slight underestimation of the number of employed people in our data. They are included in the categorie "not stated". Approximately 500.000 citizens are in this categorie "not stated", so only a minority of them are employed, the others can be considered as "others not economically inactive".

The Eurostat definition stipulates that the main job is determined based on the time devoted to it and, failing that, the income received. For several categories of workers, notably own-account workers, employers and workers abroad, it is difficult, using administrative databases, to discern their working time. Only the income received as recorded in tax databases are therefore used for these types of workers.

 

Unemployed people identified by Statbel follow Eurostat’s definition. However, administrative databases do not make it possible to assess the steps taken to find a job. Although in Belgium there are controls on the efforts made to find a job and a time limit on registration with one of the regional employment agencies, it is difficult to state with certainty that an unemployed person has taken sufficient steps to be considered to be actively looking for work.

 

Although Eurostat does not explicitly state this in the definitions, a correction is made in relation to the legal age of retirement in Belgium, which is currently set at 65. This correction takes into account the work income as declared in the tax data.

 

Capital income recipients are only defined based on capital and property income declared in the tax data for the year preceding the Census and the following year. This way, Statbel captures the continuity of income.

 

Students are identified using databases relating to enrolment in a school in Belgium. However, some students could not be identified using the community and university data, in particular those following a particular type of education or an education abroad. In order to avoid any loss of information in school enrolments, this information has been supplemented by data relating to family allowances, as these are received on condition that the child is enrolled in school. To take into account the compulsory schooling in Belgium, a correction has also been made for people under the age of 18.

 

Unfortunately, the administrative databases do not provide information on the labour market situation for all the people included in the Belgian population. Indeed, come categories cannot be identified:

• job seekers not registered in an unemployment or placement agency;
• helpers non-administratively registered;
• new self-employed workers who are not yet administratively registered;
• housewives/househusbands;
• voluntary workers;
• undeclared employment (“black economy” and illegal activities).

13.1.8. Overall accuracy - Occupation

The information used to identify the occupation of each worker comes from 89 administrative sources. Administrative sources are received for several years. A longitudinal study is done to see if the information on occupation is steady or fluctuates a lot. For some ISCO classes it's clear that the occupation is rather steady. If for those classes, the ISCO code of previous and next years is also retained, 72% of the employed population can be covered.

Under 18.1.1.1., you'll find a more complete list of data sources we received. For the occupation, it consists on mainly data from professional orders, occupations at various government agencies, occupations from the health sector, data coming from some large companies, data on self-employed workers, etc.

The broad ISCO groups that were exhaustively covered are

• 0: Militaries
• 6: agriculture.

All broad ISCO-groups are partially covered by some administrative sources, however not every ISCO group is covered to the same extent. Data from sample surveys (LFS, SES, SILC, AES) were used to estimate the ISCO distribution for the rest of the employed population. In the mathematical model, we put several years together to increase the sample size. However carefully examination is necessary because the distribution in the part of the population not covered by administrative sources isn't the same as the ISCO frequency distribution predicted from administrative sources. We developed an imputation model for 28% of employed people to avoid bias. 

• Officials in European and international institutions

Given the scarcity of information available in the administrative databases received on officials in the European and international institutions making it possible to assign an ISCO code to these workers with a specific profile, Statbel has taken the information from the different types of jobs performed within the European Commission in order to draw up a table of ISCO probabilities according to gender. The final allocation of an ISCO code for all officials within the European and international institutions is therefore based on the assumption that they follow the same distribution as the members of the European Commission.

• Imputation model - learning basis

The imputation of occupation for the rest of the employed population takes into account the probabilities, determined based on a multinomial logistic regression model, of being in an ISCO category according to various individual characteristics.

In order to approximate the economic reality, Statbel considers that the population follows an overall occupational distribution similar to that determined by the Labour Force Survey (LFS). The learning basis for the logistic regression is therefore made up of individuals who described their occupation in the LFS survey between 2015 and 2022 and who do not have an occupation determined based on purely administrative data. Given that administrative sources do not cover homogeneously the entire employed population, the profile of individuals with an unknown occupation differs from the characteristics of the population as a whole. It is therefore essential to remove the profiles of these individuals and their associated weights from the learning basis. In addition, individuals included in the ISCO 0 (military) and ISCO 6 (farmers) categories are also excluded from the learning basis, given that these occupational categories are considered exhaustive in the administrative databases.

In order to counterbalance the coverage bias of the available administrative sources, the survey weight of the individuals was taken into account in the logistic regression for the individuals included in the learning basis. The aim of the imputation model is to extrapolate the profiles observed in the learning basis to all the profiles for which the occupation is undetermined.

• Construction of the imputation model and choice of variables

Once the learning basis has been defined, it is randomly divided into two parts. In order to check that the random nature has only a minor impact on the prediction parameters, the learning basis was split into these two parts several times. One of the parts, covering 2/3 of the learning basis, is used for training the model, i.e. calculating the parameters corresponding to the impact of each modality of the explanatory variables on the probability of being in an ISCO category. The other part is treated as if the ISCO variable were missing and serves as a verification test.

These tests are carried out at individual level in order to ensure that the probability of predicting an ISCO category at individual level is correct, but also within a group of individuals in order to compare the distribution of ISCO categories defined in the learning basis and the distribution of those predicted in the test part. This test at the level of a group's ISCO distribution can be carried out for various breakdowns of interest and ensures that prediction errors partially make up for each other within a breakdown.

Several modelling runs were produced in parallel to assess, among other things, the robustness of the modelling and its accuracy at several detail levels. In addition, prior to modelling, a number of variables were analysed to determine which would be used as explanatory variables in the model. The choice of variables was guided by Cramér's V indicator, which measures the intensity of dependence on the predictor variable (ISCO). In order to reduce the number of dimensions entering the model without losing predictive information, certain modalities were grouped, taking into account the thematic aspect and the effectiveness in relation to individual prediction. The explanatory variables selected are mainly constructed as part of the Census, thus making it possible to obtain individual information for the majority of cases to be imputed.

In order to capture the effect on probabilities of combinations of several variables, these are entered differently in the imputation model, on the one hand directly as explanatory variables and on the other hand as stratification variables. By adding stratification variables, the learning basis is dissociated so that several distinct models run independently in parallel. This method makes it possible to identify sub-populations with very different distributions and a very different link with the occupation.  In addition, the effects of crossing several variables can be captured without revealing problems of codependency and collinearity between the stratification variables and the explanatory variables. The latter play a purely direct role in the prediction of the variable of interest.

 

The stratification variables selected for the imputation model are:

• sex
• status of employment
• sector of economic activity
• educational attainment (3 broad groups)

As explanatory variables for the logistic regression, Statbel selected:

• age groups
• working time
• professional income
• nationality (broad groups)
• country of birth (broad groups)
• province of the place of work is similar to that of the place of residence
• detailed sector of economic activity
• field of study (study domain)

13.1.9. Overall accuracy - Industry

The implementation of the branch of economic activity by Statbel is consistent with the Eurostat definitions and provides greater detail. The sector of economic activity identified is that of the establishment for which the person is working. The establishment is registered by the employer in the case of employees. Temporary agency workers and consultants are attached by default to the temporary employment agency or consultancy company for which they work. The sector of economic activity is therefore that of the enterprise directly employing them, and does not correspond to economic reality. In some cases, a correction is made based on the data obtained directly from temporary employment agencies. For the self-employed, the establishment taken into account is the one for which the person is appointed as a representative and for which the sector of economic activity is the closest to the occupation identified. In the absence of an establishment number, if the enterprise number is known, Statbel uses the percentage distribution of each enterprise's staff across their various establishments and the geographical proximity of the workplace to the place of residence to determine the establishment to which the worker is attached.

 

For workers for whom no establishment could be linked, a displacement model was drawn up and imputations were made according to a probabilistic distribution.

The results were validated. Comparisons with LFS and the national accounts were made.

13.1.10. Overall accuracy - Status in employment

The definitions used by Statbel are consistent with those of Eurostat and provide some additional detail in the modalities. The distinction between own-account workers and employers is made based on the number of persons employed in the enterprise of the self-employed person. The identification of “aid workers” listed as “other employed persons” is based on the legal form of the enterprise to which they are attached.

13.1.11. Overall accuracy - Place of work

The implementation of the place of work by Statbel is consistent with the Eurostat definitions and provides details on the exact place of employment. The place of work is based on the number of the establishment for which the person is working. The establishment is registered by the employer in the case of employees. Temporary agency workers and consultants are attached by default to the temporary employment agency or consultancy company for which they work. The place of work therefore corresponds to the head office of the enterprise directly employing them, and does not correspond to economic reality. In some cases, a correction is made based on the data obtained directly from temporary employment agencies. For the self-employed, the establishment taken into account is the one for which the person is appointed as a representative and for which the sector of economic activity is the closest to the occupation identified. In the absence of an establishment number, if the enterprise number is known, Statbel uses the percentage distribution of each enterprise's staff across their various local units and the geographical proximity to the place of residence to determine the local unit to which the worker is attached.

 

For workers for whom no establishment could be linked and for whom no place of work in Belgium could be identified, a displacement model was drawn up and imputations were made according to a probabilistic distribution.

 

In general, Census statistics are based on administrative data that are relatively homogeneous across the country. Nevertheless, given that not all workers in European and international institutions are detected, an under-representation of workers in the Brussels-Capital Region should be considered.

 

It is also important to note that during the COVID period, most workers carried out their work from home. However, no correction was applied in the Census statistics. The actual place of work remains similar to the place of the establishment for which the person is working.

13.1.12. Overall accuracy - Educational attainment

The Census 2011 served as basis to build a database on the education level of the population on 01 January 2021.

This register was improved and updated by the delivery of new databases on education. These databases come from the Communities and the Regions. Indeed, in Belgium, education is not a federal responsibility but a community or regional competence depending on the type of education.

Data collected for the Census 2021 also come from other public administrations, databases received for the topic “Labour Market” and surveys.

 

In order to guarantee the constitutional right to education, compulsory education has been introduced for all children aged 6 to 18 years old residing in Belgium. Compulsory education however does not equal the duty to attend school. Parents may choose home schooling for their children. Besides, a pupil must attend fulltime compulsory education until the age of 15. From 15 onwards students may engage in part-time schooling and opt for a structured learning path, which combines part-time vocational education in an educational institution with part-time employment. This if often called apprentice training or alternated learning.

 

The Communities mainly organize education in Belgium. This means that administrative data regarding education (registrations, study domains, diplomas and certificates…) are collected by the different Communities in Belgium, are differently structured and, for a same source, can also change from one year to the other. A source that is available one year may no longer be available the following year, then come back a few years later (e.g.: FWB juries).

 

However, the Belgian educational system is more complex. In fact, there are within each Community different educational networks organising education and there are different types of education in Belgium. Some of these types are organised by the Regions and not by the Communities.

 

Education is also a sector that is constantly evolving. Therefore, we must pay regular attention to the various reforms implemented in the relevant administrations. To this end, the mapping sent annually by the competent authorities to Eurostat is very useful and constitutes our basis to determine an ISCED level.

 

Statbel received administrative databases on diplomas (12). 4 of them are new compared to the Census 2011.

 

Concerning diplomas obtained abroad, only the holders who asked for an equivalence in Flanders or in compulsory education in the FWB which was accepted are included in the database.

 

We have not received the diplomas:

• of social advancement education in the FWB which delivers diplomas of secondary and higher education
• of equivalences of higher education diplomas in the FWB 
• of the first and second degrees of secondary education in the FWB
• of juries in primary and higher education in the FWB
• of primary education in the Flemish Community
• of higher education in the German-speaking Community from 2018 onwards
• of education of apprentices and business leaders in the Brussels-Capital Region that delivers upper secondary education and higher education diplomas
• of apprentices and business leaders in the German-speaking Community from 2017/2018 onwards
• of private schools in primary, secondary and higher education
• of European schools in primary and secondary education
• of private schools, including European schools, that are not governed by the Communities or the Regions.

Therefore, the delivered diplomas are not included in the databases.

 

Statbel made a request to the Secretary-General of the European Schools (5 schools including 4 in Brussels-Capital Region). It has been rejected (February 2019). The European School claims that it has no legal obligation to provide the data. We are therefore afraid that we will never be able to measure education in Brussels correctly.

 

Besides, Statbel has also received information on registrations in schools. These registrations helped to deduce diplomas.

 

For cases where the diploma has not been provided in the sources mentioned above, Statbel also uses data from declarative sources. Namely:

• Census 2001: Highest level of education successfully completed by the residents in Belgium on 01 January 2001
• Crossroads Bank for Social Security: Data on job seekers (FOREM, ADG , VDAB & ACTIRIS)
• Bulletins of births: Education level of the father / mother (from eBirth)
• LFS: Highest education levels collected in the Labour Force Survey.
• Labour market database: Data collected among enterprises on their workers and survey data (AES and SILC)

 

With the data received, Statbel could cover the entire territory. However, some data are still missing (EDU=UNK). In 2021, the percentage of missing data in Belgium amounted to 5.14%. It is better than in 2011, when it amounted to 7.9%.

13.1.13. Overall accuracy - Size of the locality

Locality is imputed in 0.27% of cases. Variable is of very good quality. Definition of 200 m criterium is used.

Based on the geographical coordinates, distances between buildings are calculated. Pairs of buildings closer than 200 m from each other form the edges of a graph. The localities are determined by computing the connected components of the graph.

13.1.14. Overall accuracy - Place of birth

There are no particular reasons for data unreliability for this topic. Not the usual residence of the mother is used, but the country where the birth took place.

13.1.15. Overall accuracy - Country of citizenship

There are no particular reasons for data unreliability for this topic

13.1.16. Overall accuracy - Year of arrival in the country

There are no particular reasons for data unreliability for this topic.

The population register was used to allocate the year of arrival. When a person migrates twice (or more) two the country, the last migration was taken into account.

13.1.17. Overall accuracy - Residence one year before

There are no particular reasons for data unreliability for this topic.

The population register was used. In the population register, we have information of moving citizens (longitudinal database of addresses). Data from the last year was used to allocate this variable. When someone moved several times, the last move was taken into account.

13.1.18. Overall accuracy - Housing arrangements

There are no particular reasons for data unreliability for this topic.

13.1.19. Overall accuracy - Type of family nucleus

There are no particular reasons for data unreliability for this topic.

The population register was used.

13.1.20. Overall accuracy - Size of family nucleus

There are no particular reasons for data unreliability for this topic.

The population register was used.

13.1.21. Overall accuracy - Type of private household

There are no particular reasons for data unreliability for this topic.

The population register was used.

13.1.22. Overall accuracy - Size of private household

There are no particular reasons for data unreliability for this topic.

The population register was used.

13.1.23. Overall accuracy - Tenure status of households

There are no particular reasons for data unreliability for this topic.

The population register was coupled with the dwelling register to determine the tenure status of households. Coupling made it possible to find information in almost all cases. Only in 0,27% of the cases, an artificial (probabilistic) linkage was effectuated.

13.1.24. Overall accuracy - Type of living quarter

There are no particular reasons for data unreliability for this topic.

13.1.25. Overall accuracy - Occupancy status

The population register was coupled with the dwelling register to determine the occupancy status of the dwelling. The official residence was used here. This can raise problems in some cases if people don't stay where they're officially registered. Dwellings of a secondary use such as holiday houses can't be distinguished with this method and are considered as vacant dwellings.

13.1.26. Overall accuracy - Type of ownership

There are no particular reasons for data unreliability for this topic.

The population register was coupled with the dwelling register to determine the tenure status of households. Coupling made it possible to find information in almost all cases. Only in 0,27% of the cases, an artificial (probabilistic) linkage was effectuated.

Note that the houskeeping concept was used here. Multiple households can be allocated to the same dwelling. A dwelling is considered as owner-occupied if at least one of the household members of one of the households is owner. Therefore, there's a difference between type of ownership (dwelling related variable) and "tenure status of household" (household related variable).

13.1.27. Overall accuracy - Number of occupants

There are no particular reasons for data unreliability for this topic.

The variable is based on the official residence. This can raise problems in some cases if people don't stay where they're officially registered.

13.1.28. Overall accuracy - Useful floor space

Number of rooms was reported instead of useful floor space.

13.1.29. Overall accuracy - Number of rooms

There are no particular reasons for data unreliability for this topic.

The definition in the dwelling register does not refer to ceiling height; however, the very concept of a room with a ceiling lower than 2m is unrealistic, which mitigates this limitation.

13.1.30. Overall accuracy - Density standard (floor space)

Instead, the density standard (number of rooms) was reported.

13.1.31. Overall accuracy - Density standard (number of rooms)

There are no particular reasons for data unreliability for this topic.

13.1.32. Overall accuracy - Water supply system

Is reported as 100% of the occupied conventional dwellings has a water supply system.

13.1.33. Overall accuracy - Toilet facilities

Is reported as 100% of the occupied conventional dwellings has toilet facilities.

13.1.34. Overall accuracy - Bathing facilities

There is an undercoverage of the number of bathrooms in the dwelling register. Therefore a correction algorithm was applied using the previous census and the dwelling register:

- Where the dwelling register or the 2011 census mentions a bathroom, the dwelling is considered to have a bathroom.

- Where both sources explicitly mention the absence of a bathroom, the dwelling is classed as being without a bathroom.

- Where the situation is unknown in the 2011 census, it is still considered to be unknown in 2021.

13.1.35. Impact of the COVID pandemic on data accuracy

No impact, the covid crisis had minimal impact on the use of administrative data.

13.2. Sampling error

Not applicable for register-based and traditional censuses.

13.3. Non-sampling error

Not applicable.

14.1. Timeliness

For the grid and usual residence:

The reference date of the Belgian 2021 census was 1/01/2021. Data for the grid on km² level was ready on 1st December 2022 for

• total population
• number of males
• nomber of females
• population under 15 years old
• population between 15 and 64 years old
• population 65 years or older.

This means a process time of 23 months. Publication in the census hub is foreseen during the month of December 2022. Reasons why it takes 23 months at this moment:

• Each year Statbel receives an important update of the population register in the beginning of March. Statbel receives weekly updates of the population register, but every year there is a major reset to avoid differences between the copy at statbel and the register saved at the Ministry of Internal Affaires. There can be administrative delays on some data. e.g. for a person moved today, it is possible it takes some days before it's updated in the population register. The census reference date is 1st January. By taking a reset in the beginning of March, Statbel is sure that most updates are available and so few data are still missing due to administrative delays.
• It takes approximately a month to load the yearly update in the database.
• Then it takes some weeks to process the data. Algorithms must be run to determine, for example, the official population. 
• The official population figure is ready by the summer according to the Belgian population definition. This definition differs from the Eurostat definition based on the concept of usual residence.
• By the end of the year it is possible to produce all demographic tables according the Eurostat population definition.
• To produce the grid data, addresses in the population register have to be geocoded. This work can only start after the yearly big reload of the population register which is ready just before summer (approximately 6 months after reference date).
• To geocode addresses, the information in the population register has to be compared with geographical data coming form databases from the 3 regions (each NUTS 1 region has his own database). There is an automated step first.
To get geographical coordinates for everyone in the population: we need
• the population according to the Eurosat definition
• geocoding
• => this is only possible after approximately 12 months.
• After that there is a manual step to geocode the rest of the addresses whereby geographical coordinates where not found in the automated step. This phase takes several months.
• Once the geographical coordinates are available, Statbel can compute the grid ID.
• At this point, There are still some hassles to solve. It is important to have everybody of the same household in only one grid cell for consistency reasons. The problem here is the Eurostat population definition. According to the definition, some citizens have to be included in the population but are temporary absent at the moment of the census. This makes it very complicated to allocate a household ID for citizens that don't stay in the country at the time of the census. Some algorithms were developed to solve this issue, but it took some time.
• After this step, there are some quality checks and validation steps. This took some weeks.
• For the 2021 census there was some research about SDC methods. This was the first time for Statbel. So SDC research and discussion took several weeks. This is one of the main reasons why it took up to 23 months before grid data was available. Next time, the SDC method will be clear and the time to publish grid data will be reduced with some months.

 For the other variables:

• Demographic data are available witin 12 months.
• Data from the social security register is available after 24 months. Afterwards, data processing is still needed. Consequently (in the moste positive case), data on lablour are available after 27 months.
• Data on education: 12 months after reference moment.
• Data on housing: 12-18 months after reference moment.

14.2. Punctuality

Grid data for total population was delivered on time before the deadline of 31 December 2022.

All other data and metadata was delivered on time before the deadline of 31 March 2024.

15.1. Comparability - geographical

See 15.1.1

15.1.1. Geographic information - data quality

In all regions of Belgium, the data from the geographical databases are very precise. In most cases more precisely than one metre. See also 13.1.1

15.2. Comparability - over time

Not applicable.

15.3. Coherence - cross domain

Not applicable.

15.4. Coherence - internal

There is exact coherence with the already published population statistics.

There is coherecne between the datasets and inside the datasets (additivity).

Cost associated with the census data collection and production of a census product and burden on respondents.

Statbel estimated the cost of the production of the census for Eurostat: this is approximately 3.3 million EURO.

The burden for the population is very low because, Belgium uses administrative sources. Only some enterprises had to respond concerning the occupation (ISCO).

17.1. Data revision - policy

See attached link.

Annexes:
revision policy

17.2. Data revision - practice

See 17.1.

The grid data will be each year one shot. Census data and population statistics have the same reference date and uses the same administrative sources for the demographic part. As a consequence, no revisions are needed.

For the occupation (ISCO) data, a revision was sent in October 2024.

18.1. Source data

Information is provided in the sub-concepts 18.1.1 - 18.1.4.

18.1.1. List of data sources

See subcategories

18.1.1.1. List of data sources - Data on persons

Population register

Data on education:

• Census 2011 (important to update with data between 2011 and 2021, e.g. old diploma's came from the previous census)
• Flemish data on education
• LED (data on diplomas and study certificates  excluding primary school diplomas). Equivalents of diplomas obtained abroad (previously NARIC) are now included in LED.
• AHOVOKS data on school-age population on tertiary education
• Agodi data on school-age population on compulsary education
• NARIC: Equivalences between 2014 and 2018 for secondary and higher education diplomas. Included in the LED after 2018
• Data on education from the French speaking community
  • Compulsory education (registrations and compulsory education diplomas excluding the first and second degrees of secondary education diplomas)
  • data on distance learning (registrations)
  • centre jury (diplomas)
  • equivalent diplomas obtained abroad (! Compulsory education only)
  • CREF: data from universities  (registrations and diplomas)
  • SATURN: data from tertiary non-university education (registrations and diplomas)
• Data on education from German speaking community (registrations and diplomas)
• IFAPME: education data from Walloon region (registrations and diplomas)
• Royal Military Academy: diplomas and registrations
• Other declarative sources
• Population and housing census of 2001 (diplomas)
• Crossroads Bank for Social Security: Data on job seekers (FOREM, ADG , VDAB & ACTIRIS) (diplomas)
• Labour market database: Data collected among enterprises on their workers (diplomas)
• Bulletins of births: Education level of the father / mother (from eBirth) (diplomas)
• Labour force survey (LFS): Highest education levels collected in the Labour Force Survey (diplomas)

Data on labour and social security:

• National Register of Natural Persons (NR): database centralising information identifying the population on 1 January of the reference year and their demographic characteristics. Every municipality is obliged to keep a population register. All these registers are grouped into one “National Register of Natural Persons” managed by the Federal Public Service Home Affairs (law of 8 August 1983).
• Calculated personal income tax (IPCAL): database of taxable income in Belgium from the Federal Public Service Finances.
• Database on school enrolments: data centralised in Statbel coming from several administrative sources: (FWB (counting, juries, equivalences), Cref, Saturn, DGOV (German-speaking community), IFAPME, ERM, AGODI, LED, DHO, Home schooling, Distance learning).
• Data warehouse Labour Market and Social Protection of the Crossroads Bank for Social Security (DWH LM&SP - CBSS): register aggregating socio-economic data from Belgian social security institutions and other public bodies. This database is used in particular to distinguish the working population from the inactive population and to determine the status of the employed population.
• Crossroads Bank for Enterprises (CBE): database created within the FPS Economy, SMEs, Self-employed and Energy containing information on the identification of enterprises and their establishment units.
• National Institute for the Social Security of the Self-employed (NISSE): database centralising information supplied by the social insurance funds on all taxable persons who were affiliated for at least one quarter during the year in question.
• DBRIS: business register compiled by Statbel based on the CBE’s data and completed by other data sources (data from NISSE, NSSO, surveys, etc.). The information contained in this database is needed to identify own-account workers and employers based on the functions performed within enterprises, and to determine the branch of economic activity in which they are employed, as well as their place of work.
• National Office for Social Security (NSSO) – DMFA: database containing all data on the work performances of all workers declared to the NSSO on a quarterly basis by employers.
• National Office for Social Security (NSSO) – WABRO: database containing information on secondments from Belgium and simultaneous employment in several countries for employees subject to Belgian social security on a mission abroad (“Working ABROad”).
• National Office for Social Security (NSSO) – DIMONA: immediate daily declaration by the employer of all entries into service and exits of their workers. This database is mainly useful for determining student employment during the reference period.
• National Employment Office (RVA-ONEM): database containing information on the status and reasons for temporary unemployment of benefit recipients (fully unemployed receiving benefits or others) charged to the RVA-ONEM budget.
• Regional employment public services (VDAB - FOREM - ACTIRIS - ADG): database containing the characteristics of job seekers registered with a regional employment office (VDAB, FOREM, ACTIRIS, ADG) and making it possible to identify job seekers who are unemployed, available for work and looking for work who are not receiving any benefits.
• Federal Pension Service (FPS): database containing information on the retired.
• Occupational Diseases Fund (FMP-FBZ): database containing information on the occupation of employees recognised as suffering from an occupational disease.
• National Institute for Health and Disability Insurance (INAMI-RIZIV) - GRI-CMI: database containing information of the professional situation of people recognised as disabled by the Invalidity Medical Council (GRI-CMI), particularly frontier workers.
• National Intermutualist Board (CIN): database containing all the persons affiliated to the Alliance nationale des mutualités chrétiennes, the Union nationale des mutualités libres, the Union nationale des mutualités socialistes, the Union nationale des mutualités neutres, the Union nationale des mutualités libérales and the Caisse auxiliaire d'assurance maladie-invalidité, who were in primary incapacity for work under the scheme for employees or under the scheme for the self-employed and who exceeded the period of guaranteed pay (the first 30 days) of the period of primary incapacity for work. This information makes it possible to identify the professional situation of frontier workers affiliated to a Belgian mutual insurance.
• Birth bulletins (E-birth): data from the regions, centralised within Statbel, relating to information on the professional situation and in particular the occupation of parents at the birth of their child.
• Death bulletins (DM1A and DP1A): data from the regions, centralised within Statbel, relating to information on the professional situation and in particular the occupation of deceased persons.
• Brussels Regional Employment Office (ACTIRIS) – professional experience: database containing information on the previous professional situations and in particular the previous occupations of job seekers registered with the public employment service in the Brussels-Capital region.
• Brussels Regional Employment Office (ACTIRIS) – hoped for occupation: database containing information on the professional aspirations of job seekers registered with the employment public service in the Brussels-Capital region. This information is used to feed the occupation imputation model in order to define a future occupation.
• Arbeitsamtder Deutschsprachigen Gemeinschaft Belgiens (ADG) – professional experience: database containing information on the previous professional situations and in particular the previous occupations of job seekers registered with the public employment service in the German-speaking community.
• Arbeitsamtder Deutschsprachigen Gemeinschaft Belgiens (ADG) – hoped for occupation: database containing information on the professional aspirations of job seekers registered with the employment public service in the German-speaking community. This information is used to feed the occupation imputation model in order to define a future occupation.
• Office wallon de la formation professionnelle et de l’emploi (FOREM) – professional experience: database containing information on the previous professional situations and in particular the previous occupations of job seekers registered with the public employment service in Wallonia.
• Office wallon de la formation professionnelle et de l’emploi (FOREM) – hoped for occupation: database containing information on the professional aspirations of job seekers registered with the employment public service in Wallonia. This information is used to feed the occupation imputation model in order to define a future occupation.
• Vlaamse Dienst voor Arbeidsbemiddeling en Beroepsopleiding (VDAB) – professional experience: database containing information on the previous professional situations and in particular the previous occupations of job seekers registered with the public employment service in Flanders.
• Vlaamse Dienst voor Arbeidsbemiddeling en Beroepsopleiding (VDAB) – hoped for occupation: database containing information on the professional aspirations of job seekers registered with the employment public service in Flanders. This information is used to feed the occupation imputation model in order to define a future occupation.
• Office wallon de la formation professionnelle et de l’emploi (FOREM) – service vouchers: database containing information on workers in the service voucher system in Wallonia.
• VLAAMSE GEMEENSCHAP - Departement Werk en Sociale Economie (WSE) – service vouchers: database containing information on workers in the service voucher system in Flanders.
• Brussels Regional Public Service - Brussels Economy and Employment (BEE) – service vouchers: database containing information on workers in the service voucher system in the Brussels-Capital Region.
• DEFENCE: database containing information on the professional situation, in particular the occupation, of defence members (including the military).
• Federal Police: database containing information on the professional situation, in particular the occupation, of staff members of the Federal Police (policemen and CALog).
• FPS Policy and Support (BOSA): database containing information on the professional situation, in particular the occupation, of (statutory and contractual) federal civil servants.
• AGENTSCHAP OVERHEIDSPERSONNEL (AGO): database containing information on the professional situation, in particular the occupation, of civil servants in Flanders.
• Brussels Regional Public Service (BRPS): database containing information on the professional situation, in particular the occupation, of civil servants within the BRPS.
• Public Service of Wallonia (PSW): database containing information on the professional situation, in particular the occupation, of civil servants in the PSW.
• Federation Wallonia-Brussels (FWB) – staff: database containing information on the professional situation, in particular the occupation, of civil servants in the Federation Wallonia-Brussels.
• Birth and Children Office (ONE) – staff: database containing information on the professional situation, in particular the occupation, of the staff working in the ONE.
• Federation Wallonia-Brussels (FWB) – education: database containing information on the professional situation, in particular the occupation, of the teaching staff in the French and German-speaking communities.
• Agentschap voor Onderwijsdiensten (AGODI): database containing information on the professional situation, in particular the occupation, of the teaching staff in the Flemish Community.
• National Institute for Health and Disability Insurance (INAMI-RIZIV): database containing information on the professional situation, in particular the occupation, of healthcare workers.
• FPS Health, Food Chain Safety and Environment (HEALTH): database containing information on the issue of the visa required to work in the healthcare sector (work certificate).
• Brussels Regional Public Service - Brussels Economy and Employment (BEE) – Work permits: database containing information on work permits in Brussels as an employee.
• Brussels Regional Public Service - Brussels Economy and Employment (BEE) – business cards: database containing information on work permits in Brussels as a self-employed.
• Federation Wallonia-Brussels (FWB) – non-profit employment register: database containing information on workers in the non-profit sector in the Federation Wallonia-Brussels.

 

Statbel also uses data directly from certain enterprises, relating to their staff members:

• European Commission: database containing information on civil servants in the European Commission. This information is necessary to identify civil servants in the European institutions.
• European Council: database containing information on civil servants in the European Council. This information is necessary to identify civil servants in the European institutions.
• European Economic and Social Committee (EESC): database containing information on civil servants in the EESC. This information is necessary to identify civil servants in the European institutions.
• European Parliament: database containing information on civil servants in the European Parliament. This information is necessary to identify civil servants in the European institutions.
• Accent for People: database containing information on the professional situation, in particular the occupation, of employees and temporary agency workers of Accent for People.
• BELFIUS: database containing information on the professional situation, in particular the occupation, of Belfius employees.
• BPOST: database containing information on the professional situation, in particular the occupation, of Bpost employees.
• DELHAIZE: database containing information on the professional situation, in particular the occupation, of Delhaize employees.
• FORTIS: database containing information on the professional situation, in particular the occupation, of Fortis employees.
• HR-rail: database containing information on the professional situation, in particular the occupation, of employees in HR-rail, Infrabel and SNCB-NMBS.
• Ordina: database containing information on the professional situation, in particular the occupation, of Ordina employees.
• Start People: database containing information on the professional situation, in particular the occupation, of employees and temporary agency workers of Start People.
• STIB-MIVB: database containing information on the professional situation, in particular the occupation, of STIB-MIVB employees.
• Synergie: database containing information on the professional situation, in particular the occupation, of permanent and temporary employees working for Synergie.
• University of Hasselt: database containing information on the professional situation, in particular the occupation, of the staff members (professors and others) of the University of Hasselt.

 

In order to identify certain categories of occupations, Statbel collaborates with certain orders to obtain data:

• Professional Institute of the Real Estate Agents (BIV-IPI): database containing information on the professional situation of real estate agents.
• Commission of Psychologists (COMPSY): database containing information on the professional situation of psychologists.
• Order of expert surveyors: database containing information on the professional situation of expert surveyors.
• Institute of Company Lawyers (IBJ-IJE): database containing information on the professional situation of company lawyers.
• Institute of Auditors (IBR-IRE): database containing information on the professional situation of auditors.
• Chamber of Notaries: database containing information on the professional situation of notaries.
• Architects' Association: database containing information on the professional situation of architects.
• French- and German-speaking Bars Association: database containing information on the professional situation of French- and German-speaking lawyers.

 

Finally, within Statbel, certain surveys are used to improve the quality of the administrative data collected, in particular for the allocation of an occupation code:

• Structure of Earnings Survey (SES): information on the professional situation, in particular the occupation, based on the survey among a sample of enterprises.
• Labour Force Survey (LFS): information on the professional situation, in particular the occupation, based on the survey among citizens.
• European Union – Statistics on Income and Living Conditions (EU-SILC): information on the professional situation, in particular the occupation, based on the survey among a sample of households.
• Adult Education Survey (AES): information on the professional situation, in particular the occupation, based on the survey among citizens.

18.1.1.2. List of data sources - Data on households

Population register

18.1.1.3. List of data sources - Data on family nuclei

Population register

18.1.1.4. List of data sources - Data on living quarters

Land register (cadastre) => Dwelling and building register is derived from land register.

Address registers with geographical information: CRAB, URBIS, ICAR

18.1.1.5. List of data sources - Data on conventional dwellings

Land register (cadastre) => Dwelling and building register is derived from land register.

18.1.2. Classification of data sources

Classification of the data sources as requested by Reg. 2017/881, Annex point 2.1.

18.1.2.1. Classification of data sources - Data on persons

04.Combination of register-based censuses and sample surveys

18.1.2.2. Classification of data sources - Data on households

02.Register-based censuses

18.1.2.3. Classification of data sources - Data on family nuclei

02.Register-based censuses

18.1.2.4. Classification of data sources - Data on living quarters

02.Register-based censuses

18.1.2.5. Classification of data sources - Data on conventional dwellings

02.Register-based censuses

18.1.3. List of data sources per topic

See Excel file attached: worksheet 3. Administrative data sources

18.1.4. Adequacy of data sources

See subcategories

18.1.4.1. Adequacy of data sources - Individual enumeration

The characteristics of each statistical unit are recorded separately, so that each characteristic can be cross-classified with others.

This is the case for all data sources.

18.1.4.2. Adequacy of data sources - Simultaneity

All information refers to the same point in time (reference date).

This is the case for all data soucres except for LFS and SES.

18.1.4.3. Adequacy of data sources - Universality within the defined territory

Data are provided for all statistical units in a defined territory (for persons in particular, data are provided for all usual residents in a defined territory).

This is the case for all data sources except for LFS and SES.

18.1.4.4. Adequacy of data sources - Availability of small-area data

Data are available for small geographical areas and for small subgroups of statistical units.

This is the case. The only variable where this is more difficult is occupation.

18.1.4.5. Adequacy of data sources - Defined periodicity

decennial

18.2. Frequency of data collection

Decennial.

18.3. Data collection

See subcategories

18.3.1. Data collection - Questionnaire based data

not applicable

18.3.2. Data collection - Register based data

The 2021 census in Belgium is a registered based census.

The 2011 census was the first census for which the population is not directly approached. It draws solely on databases. All the files used are primarily administrative in purpose, with the exception of the databases on education which were created in 2001 especially for the 2011 census, with a view to updating the information collected during the previous census. Since education is a community matter in Belgium, the data are collected separately by the various communities.

The methodology used in 2021 is very similar compared to the 2011 census. Now the 2011 census is used as a starting point (e.g. the information on education). This information from 2011 has been updated with administrative data between 2011 and 2021. Some new databases such as new data sources on education and occupation should lead to an improvement compared to 2011.

The population register forms the core of the census, and makes it possible to identify the entire Belgian population. Each municipality is obliged to keep a register of the population. All these registers grouped into a "Registre National des Personnes Physiques" are managed by the Service Public Fédéral Intérieur (Home Affairs Federal Public Service) (Law of 8 August 1983). Belgians residing in a municipality of the country or entered in a consular register, and foreigners authorised to stay in the country for a period longer than three months must register there. A unique identification number is then allocated to them. Subsequently, any change in the situation of individuals (arrival, departure, internal migration, change in civil status, etc.) must be reported within the deadline of eight working days. There are very few late declarations because the population on 1 January 2011 is calculated using the data from the Registre National as available on 1 March 2021.

This administrative source does not completely cover the Belgian population, mainly due to the under-declaration of arrivals (population without the right of residence, European population not wishing to settle permanently in Belgium). Departures abroad also seem to suffer from under-registration, for both the Belgian and the foreign population. Nevertheless, under-registration or failure to de-register should be kept in perspective: the municipalities regularly organise local procedures to check the accuracy of the content of their population registers.

These demographic data from the Registre National are coupled with a series of other databases in order to establish the variables listed in the various themes of the census. See 18.1 for a list of data sources. These different connections are made using the unique identifier of each individual as key.

For socio-economic data, the social security register, business register and tax register was used. The social security register contains both id's: national register number (person id) and enterprise number. Due to the social security register, it's possible to link the databases on persons with the business register. Also the tax register contains the national register number.

For education: mainly databases from the communities were used. Most (but not all) of then contains the national register number which makes it easy to link with the other databases. For those files that don't contain the national register number, a matching was done based on other characteristics (name, first name, birith date, etc.).

For dwellings the land register was used. A derived framework register was created, namely the dwelling and building register. Linkage between demographic data and the dwelling register is based on address information. The dwelling register contains geographic coordinates. In the database there's owner information for each dwelling. This can be an enterprise id or national register number depending whether the owner is a physical person or an enterprise. This information is used to create variables such as OWS and TSH.

18.3.3. Data collection - Sample survey based data

not applicable

18.3.4. Data collection - Data from combined methods

This is only the case for occupation.

The information used to identify the occupation of each worker comes from 89 administrative sources.

Under 18.1.1.1., you'll find a more complete list of data sources we received. For the occupation, it consists on mainly data from professional orders, occupations at various government agencies, occupations from the health sector, data coming from some large companies, data on self-employed workers, etc.

18.4. Data validation

Inside Statistics Belgium, a service which is independent of the service responsible for producing figures, is responsible for the final validation of the data. Internal validation procedures have been set up in order to ensure consistency between all hypercubes and grid data. An external evaluation was also carried out. The aim is, on the one hand, to ensure that the statistics produced are part of a time trend or, if interruptions are found, that these are the consequence of specific, cyclical socio-economic factors. Furthermore, the figures set out in cubes have been compared with information from other sources (LFS, SILC, national accounts for example).

So, inside Statistics Belgium, there's a double validation check: first by the census team and afterwards by the validation team.

18.5. Data compilation

Records from the various registers are usually coupled to each other using the national number for individuals as a key (the single identification number allocated by the population register). However, other link variables are used in the following cases:

- In order to identify in which dwellings individuals live, the demographic data are coupled with those of the land register (cadastre). The address is then used to facilitate a link between the various registrations.

- The first name, family name, gender and date of birth are used to couple the demographic data with files where the national number is missing. This is the case for some files coming from the French-speaking community (tertiary education), although, the situation is improved compared to 2011. The national number is also missing in the data for education from the German-speaking community and in some of the databases on occupation. In other cases, i.e. in the French-speaking community for compulsory education and in the Dutch-speaking community, the coupling key is still the national number.

The population register is the source of primary data used to generate households and family nuclei. This database contains information on household composition: the members of the households as declared to the communal services and the reference person, defined as the member of the household who is the most often in contact with the municipality. Moreover, the relationship of each household member with the reference person is known. To create the seven variables for the families/households area, the operational introduction of two concepts called for particular care.

First of all, it was necessary to assess the population living in a registered partnership. The Belgian concept of legal cohabitation differs from the definition favoured by Eurostat, since it does not imply a marital relationship between the two contracting parties. To obtain an assessment of the population in registered partnerships, individuals who are related to each other have been excluded from the population of legal cohabitants. The family relationships between individuals have been identified by the relationship with the reference person using data on family relationships.

The algorithm for family nuclei in 2021 is improved compared to the 2011 census. Now we have more detailed information on parentage, even if the parents don't live in the same household. This ensures that family ties can be mapped much more accurately.

It is impossible to measure directly consensual union. The population living as husband and wife without a contract within the same household has been evaluated either by the existence of a common child, or by using an algorithm based on potential partnerships. In each household, all individuals aged 18 or over and without a family link have been identified. The couple with the smallest age gap is then declared as living in consensual union.

- The business number is also used to couple certain records from the land register with the business register in order to determine whether certain individuals are owners of the dwelling in which they live via their business. The link between the two databases is then established using the business number.

 

Data on education

The variable EDU was created in several steps:

1. Upload of each database separately
2. Cleansing of databases:
3. Deletion of duplicates
4. Research of unique identifier, national register number, if not provided
5. Creation and harmonisation of variables useful for the processing
6. Creation of specific tables for diplomas and registrations for each source
7. Pooling of all imported tables (diplomas + registrations)
8. New cleansing and corrections
9. Deduction of diplomas based on the registrations if the diplomas are not known
10. Introduction of declarative databases
11. Matching with population
12. Adding the highest level of diploma obtained based on the priority rule in the origin of the source

 Data on dwellings. Dwelling variables were created according the following steps:

1. Cleaning of the database
2. Linking population register with land register.
3. Creation of "virtual dwellings"
4. Creation of register of dwellings and buildings
5. Probabilistic linking of the remaining 0.27% households to a dwelling
6. Creating of the final dwelling variables.

18.6. Adjustment

Not applicable.

No comments.

Further qualitative metadata on the 2021 Population and Housing Census
Methodological note on the economic variables
Methodological note on the education variables

The data present the results of the 2021 EU census on population and housing, following Regulation (EC) 763/2008; Regulation (EU) 2017/543; Regulation (EU) 2017/712 and Regulation (EU) 2017/881.

6 November 2024

The information is given separately for each census topic. See the sub-concepts 3.4.1 - 3.4.37.

The EU programme for the 2021 population and housing censuses includes data on persons, private households, family nuclei, conventional dwellings and living quarters.

The persons enumerated in the 2021 census are those who were usually resident in the territory of the reporting country at the census reference date.

Data are available at different levels of geographical detail in EU countries: national, NUTS2/NUTS3 regions and local administrative units (LAU), grids.

Information is provided in the sub-concepts 5.1 - 5.3.

Information is provided in the sub-concepts 13.1.1 - 13.1.35.

Counts of statistical units should be expressed in numbers and where is needed rate per inhabitants enumerated in the country.

Records from the various registers are usually coupled to each other using the national number for individuals as a key (the single identification number allocated by the population register). However, other link variables are used in the following cases:

- In order to identify in which dwellings individuals live, the demographic data are coupled with those of the land register (cadastre). The address is then used to facilitate a link between the various registrations.

- The first name, family name, gender and date of birth are used to couple the demographic data with files where the national number is missing. This is the case for some files coming from the French-speaking community (tertiary education), although, the situation is improved compared to 2011. The national number is also missing in the data for education from the German-speaking community and in some of the databases on occupation. In other cases, i.e. in the French-speaking community for compulsory education and in the Dutch-speaking community, the coupling key is still the national number.

The population register is the source of primary data used to generate households and family nuclei. This database contains information on household composition: the members of the households as declared to the communal services and the reference person, defined as the member of the household who is the most often in contact with the municipality. Moreover, the relationship of each household member with the reference person is known. To create the seven variables for the families/households area, the operational introduction of two concepts called for particular care.

First of all, it was necessary to assess the population living in a registered partnership. The Belgian concept of legal cohabitation differs from the definition favoured by Eurostat, since it does not imply a marital relationship between the two contracting parties. To obtain an assessment of the population in registered partnerships, individuals who are related to each other have been excluded from the population of legal cohabitants. The family relationships between individuals have been identified by the relationship with the reference person using data on family relationships.

The algorithm for family nuclei in 2021 is improved compared to the 2011 census. Now we have more detailed information on parentage, even if the parents don't live in the same household. This ensures that family ties can be mapped much more accurately.

It is impossible to measure directly consensual union. The population living as husband and wife without a contract within the same household has been evaluated either by the existence of a common child, or by using an algorithm based on potential partnerships. In each household, all individuals aged 18 or over and without a family link have been identified. The couple with the smallest age gap is then declared as living in consensual union.

- The business number is also used to couple certain records from the land register with the business register in order to determine whether certain individuals are owners of the dwelling in which they live via their business. The link between the two databases is then established using the business number.

 

Data on education

The variable EDU was created in several steps:

1. Upload of each database separately
2. Cleansing of databases:
3. Deletion of duplicates
4. Research of unique identifier, national register number, if not provided
5. Creation and harmonisation of variables useful for the processing
6. Creation of specific tables for diplomas and registrations for each source
7. Pooling of all imported tables (diplomas + registrations)
8. New cleansing and corrections
9. Deduction of diplomas based on the registrations if the diplomas are not known
10. Introduction of declarative databases
11. Matching with population
12. Adding the highest level of diploma obtained based on the priority rule in the origin of the source

 Data on dwellings. Dwelling variables were created according the following steps:

1. Cleaning of the database
2. Linking population register with land register.
3. Creation of "virtual dwellings"
4. Creation of register of dwellings and buildings
5. Probabilistic linking of the remaining 0.27% households to a dwelling
6. Creating of the final dwelling variables.

Information is provided in the sub-concepts 18.1.1 - 18.1.4.

Decennial.

For the grid and usual residence:

The reference date of the Belgian 2021 census was 1/01/2021. Data for the grid on km² level was ready on 1st December 2022 for

• total population
• number of males
• nomber of females
• population under 15 years old
• population between 15 and 64 years old
• population 65 years or older.

This means a process time of 23 months. Publication in the census hub is foreseen during the month of December 2022. Reasons why it takes 23 months at this moment:

• Each year Statbel receives an important update of the population register in the beginning of March. Statbel receives weekly updates of the population register, but every year there is a major reset to avoid differences between the copy at statbel and the register saved at the Ministry of Internal Affaires. There can be administrative delays on some data. e.g. for a person moved today, it is possible it takes some days before it's updated in the population register. The census reference date is 1st January. By taking a reset in the beginning of March, Statbel is sure that most updates are available and so few data are still missing due to administrative delays.
• It takes approximately a month to load the yearly update in the database.
• Then it takes some weeks to process the data. Algorithms must be run to determine, for example, the official population. 
• The official population figure is ready by the summer according to the Belgian population definition. This definition differs from the Eurostat definition based on the concept of usual residence.
• By the end of the year it is possible to produce all demographic tables according the Eurostat population definition.
• To produce the grid data, addresses in the population register have to be geocoded. This work can only start after the yearly big reload of the population register which is ready just before summer (approximately 6 months after reference date).
• To geocode addresses, the information in the population register has to be compared with geographical data coming form databases from the 3 regions (each NUTS 1 region has his own database). There is an automated step first.
To get geographical coordinates for everyone in the population: we need
• the population according to the Eurosat definition
• geocoding
• => this is only possible after approximately 12 months.
• After that there is a manual step to geocode the rest of the addresses whereby geographical coordinates where not found in the automated step. This phase takes several months.
• Once the geographical coordinates are available, Statbel can compute the grid ID.
• At this point, There are still some hassles to solve. It is important to have everybody of the same household in only one grid cell for consistency reasons. The problem here is the Eurostat population definition. According to the definition, some citizens have to be included in the population but are temporary absent at the moment of the census. This makes it very complicated to allocate a household ID for citizens that don't stay in the country at the time of the census. Some algorithms were developed to solve this issue, but it took some time.
• After this step, there are some quality checks and validation steps. This took some weeks.
• For the 2021 census there was some research about SDC methods. This was the first time for Statbel. So SDC research and discussion took several weeks. This is one of the main reasons why it took up to 23 months before grid data was available. Next time, the SDC method will be clear and the time to publish grid data will be reduced with some months.

 For the other variables:

• Demographic data are available witin 12 months.
• Data from the social security register is available after 24 months. Afterwards, data processing is still needed. Consequently (in the moste positive case), data on lablour are available after 27 months.
• Data on education: 12 months after reference moment.
• Data on housing: 12-18 months after reference moment.

See 15.1.1

Not applicable.