1. From census to Census
Belgium has a long tradition of organising general population and housing censuses. The first Belgian census was organised in 1846 by Adolphe Quetelet. More or less 10 years separated the different censuses: 1846, 1856, 1866, 1876, 1880, 1890, 1900, 1910, 1920, 1930, 1947, 1961, 1970, 1981, 1991 and 2001. Not only was the census a population count, but it also had a socio-economic and scientific value.
In the early stages, however, the administrative value of a general census was particularly high: it was used to establish the official population figures for each municipality. The population figures of a municipality are for instance still important for the allocation of subsidies, the salaries of local government officials and for establishing the number of municipal and provincial councillors, to name but a few examples.
The 1980s saw the introduction of the national register of natural persons. This population register tremendously facilitated the population count, allowing much quicker publication of population figures. The general census consequently lost its administrative nature.
In 1991, part of the variables in the population and housing census came from the national register. This eventually led to a name change in 2001, because it was no longer the census but the national register that provided the population count. Hence the new name “General socio-economic survey 2001”, abbreviated to SEE 2001. The SEE 2001 is in fact a collection of important socio-economic data and consists of two main parts: population characteristics and the housing stock. From 2011 onward, the survey is referred to as ‘Census’.
Why a census if it no longer counts the population?
As mentioned above, the censuses did not only have an administrative purpose, but also socio-economic and important scientific value. The name may have changed, but the collected data are still invaluable.
On policy level: the statistical information can be used to take certain political decisions.
For the scientific world: universities among others regularly use data from previous censuses and from the SEE 2001. Also in 2011 there is a large demand for the Census. It is a source of very detailed information on small geographical areas and smaller sections of the population. It can produce interesting studies, for example in the field of sociological research. Smaller subpopulations cannot be studied with a small sample.
For international institutions: e.g. Eurostat, OECD, United Nations. This allows comparisons between different countries.
For public social welfare centres and other municipal public services: they regularly carry out studies based on data from the censuses.
For private individuals and enterprises: statistical information can help decide what part of the city is most suitable to start an enterprise. The neighbourhood may depend on the type of enterprise.
2. A new name, a new way of working
The approach to the 2011 Census will be completely different from the previous traditional censuses, which consisted of exhaustive surveys each time. In the past we could only obtain considerable parts of the necessary data by asking the person in question. Now a vast amount of information can be recovered from administrative databases. It is therefore more suitable to use these administrative data than to burden the citizens with a survey. The following administrative databases are vital for the Census 2011:
The national register of natural persons
DBRIS2: statistical database derived from the Crossroads Bank for Enterprises (CBE)
The data warehouse Labour Market and Social Protection of the Crossroads Bank for Social Security (CBSS)
The General Administration of the Patrimonial Documentation (former land register)
Educational data from the Communities
See following websites
The Central Identification Address Database (CIAD)
Fiscal data from the FPS Finances
Many administrative data on education are available from 2001 onward. The level of education of people who graduated before 2001 is available through the SEE 2001.
A number of existing sample surveys will be used to collect part of the lacking variables in administrative databases. Estimation procedures similar to the model the Netherlands has been using for years (the virtual Census) will be used to integrate the survey data with the exhaustive data.
A schematic overview of the working method can be found below:
The different arrows indicate which databases need to be linked. Two arrows are thicker because they refer to several types of links between the databases.
The new working method was tested by means of a pilot project. In 2002 the Higher Council of Statistics appointed an academic research team to map every potential information source for the administrative census. Three years of scientific research were concluded with a report that contains these potential data sources. Subsequently the pilot project Microcensus 2006 was launched to already collect data and test some databases. This project also took three years and was also supported by an academic research team. A sample of 20% for the situation on 1 January 2006 (halfway between both censuses) revealed that the data from administrative sources were sufficient to organise an administrative census for the first time in 2011.
3. Your data are in good hands
The new method unites different information sources. Data confidentiality of course cannot be compromised. Legal and technical measures ensure citizen privacy.
Statistics Belgium employees are subject to statistical confidentiality. They may never release identifiable information. Also non-identifiable individual information requires approval from the privacy commission. Only anonymous and global statistics are disseminated directly.
The administrative data used by Statistics Belgium are requested through the privacy commission or "Vlaamse Toezichtcommissie". They will only approve cases if it can be proven for each requested item that it is vital for the project (the requested data must be in proportion to the project needs) and that data protection can be guaranteed sufficiently.
Public services such as the land register deliver data to Statistics Belgium, but confidential information never flows back to these services. It can therefore be seen as one-way traffic. When Statistics Belgium finds a mistake in administrative data, they cannot give individual feedback to the public service. This should prevent negative consequences for citizens. When linking administrative data to survey data reveals an undeclared conversion or renovation for example, this will not have fiscal consequences for the citizen.
Of course no more people than strictly necessary have access to the census data. Only those people from the census team and definitely not everyone in Statistics Belgium have access to some of the data. The technical security of the database system and of the IT network is very strict.
There is also a general procedure for granting access. This procedure is also repeated regularly for people who already have access to certain data to ensure nobody can access tables they do not strictly require for their work. A logging system also exists, that saves the actions performed when consulting data. The data warehouse system shows the data used by each civil servant.
Statistics Belgium has already proven that statistical data are always treated with the highest regard for privacy.
4. Benefits of the new method
The way the Census is organised in 2011 (based on administrative databases) has a number of benefits compared to a classic survey. A few examples are listed below:
The cost of an administrative census is much lower than that of a traditional census. It does no longer require poll-takers and processing of paper surveys. Using the national register and a survey by post already reduced the cost by half in 2001. Now that almost all data originate from administrative sources, the cost for the 2011 Census will be even lower. It will become even much lower in the future because creating quality databases still requires much time.
It is clear that the administrative census also requires less paper.
The burden on citizens disappears. Why ask it to people if the data already exist in databases? As part of administrative simplification it became increasingly difficult to support a traditional census. In 2000 the government therefore decided to organise an exhaustive survey for the last time in 2001 and to look for alternatives to reduce the burden on citizens.
Contrary to the General socio-economic survey 2001 (SEE 2001), which was a recording of a given moment in time (except for the national register), the Census 2011 has a number of source files with historical data.
CBE / DBRIS
It will be possible to study the evolution of certain data. In the long run a retrospective approach of the data will also be possible.
There are different approaches to linking data. The patrimonial documentation can for example be linked to DBRIS (thicker arrow on figure). This can be done by means of the location of the property and allows us to deduct its purpose. The two files can also be linked by means of the owner of the property, if the owner is an enterprise. This link has another meaning: the owner of a property can be determined by means of his/her enterprise. These different approaches allow a wide range of possibilities.
Possibility of a potential increased frequency of the Census. Once the new system is up and running, results will be available much more quickly.
Availability of geographical coordinates:
It will be easier to visualise data on maps. Calculations with adapted software (e.g. route planners) are possible. Previous subjective questions will be replaced by objective criteria.
There will no longer be a risk of misunderstanding questions or of answering questions incorrectly. There will also no longer be confusion between certain terms as there was in the 2001 questionnaire. Respondents are no longer obliged to choose one specific answer (see below for examples).
No problems with coding of data (e.g. NACE codes).
No more confusion of the terms households, houses and buildings. The SEE 2001 for example equated a household with a house. However, one of the questions asked if respondents shared some of their rooms with other households, which is impossible according to the above principle.
The number of floors of the building respondents live in will be more correct. During the SEE 2001 some people included the ground floor, others did not. This was especially noticeable since some households who lived on the same address gave different answers.
The purpose of the property will be much more detailed based on databases. Not only will we be able to see if a property has other purposes besides housing, but we will also be able to indicate in great detail if respondents for example live above a café, a bakery, and so on. DBRIS (or CBE) data can be linked to the housing stock.
Data on the date of construction and possible conversions or renovations will be much more correct in the housing stock than in the SEE 2001. Especially tenants do not always know exactly when ‘their’ home was built. Additionally the exact year will be known, instead of categories of years usually.
Homeowners: since the Patrimonial Documentation’s mission is to tax owners of real estate, data on property rights (from the Patrimonial Documentation) are by definition the official and correct data. They are much more objective than the answers to the 2001 census questions. The data are also more detailed and precise. This allows many options that were previously impossible.
All the different kinds of rights of disposal and enjoyment (e.g. usufruct) can be mapped.
We can see how large someone’s share in the ownership of a property is, how many proprietors there are, and so on.
We can see the difference between direct owners and owners who have the property through their enterprise.
Not only occupied properties but also vacant houses and houses for secondary use will appear in the housing stock. This was not the case in 2001.
The possibility exists to create age-specific fertility figures without delay, compared to the sometimes years of delay for figures based on birth certificates.
Not only the highest completed level of education will be available. The highest completed level of education does not always equal the most recent completed level of education. Longitudinal studies will be possible.
The Flemish Region has very detailed information available through the variable administrative group.
Much smaller risk of errors. In 2001 people for example responded that they had a high level of education when they were still studying (they confused their current ongoing education with their level of education). This is suggested from comparisons of the data from databases with the SEE 2001 responses.
Labour market and social security
Much more objective than in 2001. Back then, people were forced to select a specific answer (e.g. student, employed, retired, and so on) whereas in reality it is possible to be retired and still working. CBSS data resolve this issue and allow mapping of the different possible combinations.
5. Results and use
The Census results will be saved in a data warehouse, facilitating the dissemination of the results. Also modern dissemination methods have been adopted, such as web services and dynamic tables on our website (be.STAT).
Data will of course become available for Eurostat, but additional tables will also be available for all census users. Politicians, public institutions, enterprises, universities, students, private individuals and so on will all be able to use the census data, which will be available as global and anonymous tables. Additionally, authorised institutions will be able to order individual coded data for scientific research by filing a request to the Statistical Supervisory Committee.
This is the first time Belgium has organised the Census based on administrative data. The start-up of this project requires additional work, since the administrative data need to be prepared for statistical purposes for the first time. When the new system is fully up and running, we will be able to publish statistics much more quickly (no longer comparable to a traditional census) and we will be able to organise the Census more frequently.
6. And Europe?
2011 is the first census year bound by a European regulation. European regulation 763/2008 of 9 July 2008 obliges us to organise a census at least on a ten-year basis, compared to the non-binding gentleman’s agreement in the past. Belgium has chosen 1/01/2011 as a reference date for this census.
Next to the European regulation there are a number of commission rules that give more details on Europe's needs. On the one hand they establish the definitions of the variables. These definitions are based on the recommendations of the United Nations. On the other hand they set the minimum requirements of the level of detail. They determine which contingency tables are definitely needed and they also ask for a quality report. The European regulation is completely output-oriented, which means that it establishes which data should definitely be available. The regulation makes the census internationally comparable, which is a good thing. However, it does not impose a certain method to organise the census.
It is clear that the information from the Belgian census will be much more elaborate than what Europe prescribes. We have conducted a study into the needs of census users. Based on the results we have made a selection of the data that will be included in the Census.