Bundesverfassungsgericht

You are here:

Provisions regarding the 2011 Census are constitutional

Press Release No. 74/2018 of 19 September 2018

Judgment of 19 September 2018
2 BvF 1/15, 2 BvF 2/15

The challenged provisions governing the 9 May 2011 census of the population, buildings and housing (2011 Census) are compatible with the Constitution. They do not violate the obligation to realistically determine the population of the Länder; in particular, they do not conflict with the essential matters doctrine, the requirement of specificity or the right to informational self-determination. The provisions also do not violate the principle of equal treatment in the federal order (Gebot föderativer Gleichbehandlung), since the unequal treatment of municipalities with less than 10,000 inhabitants is justified given that it was based on factual reasons and promised sufficiently comparable results. This was decided by the Second Senate of the Federal Constitutional Court in a judgment pronounced today, following the applications of the governments of the Land Berlin and the Free and Hanseatic City of Hamburg in abstract judicial review proceedings (abstrakte Normenkontrolle). In its reasoning, the Court held that the legislature is afforded a margin of prognosis, appreciation and manoeuvre when setting out the procedure for collecting data.

Facts of the case:

From 2001 to 2003, a “test census” was carried out to test the register-based census method and to develop it further. The statistical offices of the Federation and the Länder derived recommendations for carrying out a future census from this test. The 2011 Census marked a change in methodology from traditional censuses, which are carried out by means of complete data collection, to data collection primarily based on existing register data. The constitutional objections raised by the applicants focus on this change in methodology. They claim that § 7(1) to (3), § 8(3), § 15(2) and (3) and § 19 of the Act on a Register-Based Census in 2011 (Census Act 2011, Zensusgesetz 2011 – ZensG 2011) of 8 July 2009 and § 15 of the Act on the Preparation of a Register-Based Census in 2011 including Buildings and Housing (Census Preparation Act 2011, Zensusvorbereitungsgesetz 2011 – ZensVorbG 2011) of 8 December 2007 as well as § 2(2) and (3) and § 3(2) of the Ordinance on the Procedures and Extent of Sample Household Surveys accompanying the Census Act 2011 (Sample Ordinance Census Act 2011, Stichprobenverordnung Zensusgesetz 2011 – StichprobenV) are unconstitutional – in particular in respect of the principle of equal treatment in the federal order and among municipalities.

Key considerations of the Senate:

I. Formally, the challenged legal provisions are constitutional. In particular, the Federation has legislative competence with regard to statistics for federal purposes (Art. 73(1) no. 11 of the Basic Law, Grundgesetz – GG). A census organised by the state and carried out by analysing existing registers and additionally surveying households is included in the concept of statistics under constitutional law and also serves federal purposes.

II. Substantively, the statutory provisions are compatible with the Basic Law.

1. § 7(1) to (3) ZensG 2011 violates neither the requirements of the essential matters doctrine arising from Art. 20(1) to (3) GG nor Art. 80(1) second sentence GG. The constitutional law review of § 7(1) to (3) ZensG 2011, which comprises the legal basis for taking household samples, takes into account the requirements of a census and in particular the interest in realistically determining the population. On this basis, the provision is not objectionable with regard to the degree of regulation (Regelungsdichte) and specificity required, given that it contains the essential framework for household samples, for the programme of the Sample Ordinance and for the administrative procedure.

2. The procedures laid down in § 7(1) to (3) and § 15(2) and (3) first alternative ZensG 2011 for correcting incorrect civil register data in the context of determining the official population of large municipalities satisfy the constitutional requirements.

a) The Basic Law attaches specific significance to the population of the Länder for determining the voting powers of the respective Länder in the Bundesrat, for their status in the financial redistribution scheme between the Federation and the Länder (Bund-Länder-Finanzausgleich) and for the requirements of a potential revision of the Länder boundaries (Art. 29 GG). To this extent, the legislature must ensure that the data is determined in a sufficiently realistic manner. The procedure must be designed in such a way that it satisfies the requirements regarding a “valid” prognosis in particular.

b) The legislature relies upon the accuracy of population numbers that can be expected from a complete census. It can be assumed that the constitutional legislature’s conception of the accuracy of the population numbers as required under the Constitution was based on a complete census, which is the traditional method of data collection. Consequently, a new statistical method cannot be required to be more accurate than a complete census.

c) Determining the population with sufficient accuracy in respect of a certain reference date according to a procedure provided for by law is a complex legislative task; it also conflicts with the legally protected interests of other parties. Therefore, such a provision requires the balancing of different interests, with a margin of appreciation and manoeuvre for the legislature. Both with regard to the selection of the procedure and its specific set-up, different aspects such as accuracy, necessity of interferences with fundamental rights and resources required must be considered. The legislature must also monitor the development of statistical methods with a view to potentially finding procedures that interfere less with fundamental rights.

According to the current state of statistical research, it cannot be established that a complete census is clearly superior to a register-based census. Experience has shown that complete censuses are associated with inaccuracies in primary statistical surveys, with complications that are inevitable in mass procedures and with difficulties in guaranteeing a uniform procedure and training a large number of data collection officials. Compared to a complete census, the register-based census chosen here is linked to significantly lower burdens placed on the census subjects. This paves the way for greater acceptance in society and reduces the risk of incorrect or incomplete responses or subjects refusing to respond, and thus also results in more precise data. Furthermore, European Union law expressly allows Member States to choose either of the two procedures as well as combined methods, given the equivalence of complete censuses and register-based censuses from a scientific point of view. On this basis, many other Member States have also opted for a register-based procedure.

d) The decision in favour of a register-based census provided for in the 2011 Census Act is also based on a valid prognosis. Following the fundamental political decisions in 1997/98, the test census constituted an empirical assessment of such a register-based approach, which tested different elements of the procedure and the suitability of the registers for this type of data collection. Several approaches were compared on this basis, and a data collection method was recommended that essentially corresponds to the method that has become law. In addition, the legislative procedure was supported by a census commission consisting of independent scientists and by two expert hearings.

To the extent that the legislature assumed that the approach examined in the test census could be used to obtain data of the required quality, this is not objectionable. All elements of the register-based census – apart from the final procedures to correct false register entries – had already been tested in the test census and had been found to be, in principle, suitable. Therefore, the legislature was not obliged to re-examine the entire procedure by means of a further test, which the administrative authorities would have been barred from using as a basis for future decisions.

e) Deviations from the legislature’s prognosis that were only recognisable after the fact do not call into question its validity. In fact, the official evaluation shows that, to a large extent, the objectives were achieved.

f) However, the legislature must take the experience gained in the course of the 2011 Census, which is the subject of these proceedings, into account for future censuses and review whether changes are necessary.

3. a) The procedures for determining the official population which vary depending on whether a municipality has more or fewer than 10,000 inhabitants may have different effects in the respective Länder due to their respective population structures. These differences must be measured against the principle of equal treatment in the federal order (Art. 20(1) GG). Where different census methods are applied, this constitutes unequal treatment requiring justification, unless it appears likely that the different methods only have minor effects on the comparability of results. This also applies to the challenged provisions, since the share of the population living in large and small municipalities differs considerably across the Länder, and, at the very least, the two procedures do not lead to identical results.

Yet the unequal treatment is justified, because it is based on factual reasons and was only minor from the ex ante perspective required here.

b) Based on the knowledge available to the legislature, it appears justifiable that in order to correct for over-coverage and under-coverage in the civil registers different procedures were used pursuant to § 7(1) second sentence no. 1 in one case and § 16 ZensG 2011 in the other. It is evident that these differences were based on the notion that the procedures were suitable for determining the population of the Länder, at least with comparable accuracy, despite the different structuring of their municipalities. In addition to lowering the administrative burden, limiting household samples to municipalities with at least 10,000 inhabitants made it possible to prevent additional interferences with fundamental rights due to the lower number of census subjects. It was also appropriate for the legislature to expect that the reduced volume of data collection resulting from the limited sample procedure would lead to higher quality primary statistical data. It is apparent that the legislature followed assessments that were based on the experience of official statisticians.

c) Similar considerations apply to the different provisions regarding the review of data pertaining to persons with more than one register entry (Mehrfachfallprüfung) pursuant to § 15(2) and (3) ZensG 2011. The legislature argued that different methods were used to correct data in cases with more than one register entry because it was sufficient that the data of persons with more than one primary residence that were registered in municipalities with at least 10,000 inhabitants was reviewed in the context of the household samples, thus essentially drawing on the differentiation of § 7(1) second sentence no. 1 and § 16 ZensG 2011. The legislature clearly assumed that this would not lead to significant distortions of the results and that it would therefore be possible to forgo the additional surveys that would have been required if the procedure pursuant to § 15(3) ZensG 2011 had been extended to larger municipalities. This is not objectionable. Based on the results of the test census, the legislature assumed that municipalities below the threshold of 10,000 inhabitants could essentially use the procedure tested there, while a comprehensive correction of errors on the basis of samples would have to be carried out for larger municipalities. In particular, the test census had also suggested that the vast majority of ambiguous cases that had come up in the review of data of persons with more than one register entry could be clarified without further inquiries. The evaluation of the 2011 Census has confirmed this assessment.

d) The threshold for using the respective methods is also based on factual considerations. The test census established the error rates in the registers for municipalities with fewer than 10,000 inhabitants, for municipalities with 10,000 to 50,000 inhabitants, for municipalities with 50,000 to 800,000 inhabitants and for municipalities with more than 800,000 inhabitants. On this basis, the statistical offices of the Federation and the Länder recommended that the register-based census be accompanied by samples in municipalities with more than 10,000 inhabitants.

This was based on the correlation, established in the test census, between the number of inhabitants of a municipality and the (unadjusted) error rates in the registers as well as on the consideration that, in order to deliver sufficiently accurate results, collecting samples would increasingly come close to a complete data collection the lower the number of inhabitants in a municipality. Furthermore, the method of individual surveying was deemed unsuitable for municipalities with more than 10,000 inhabitants.

4. There are no indications of structural deficits in enforcing legislation. In this regard, it must also be taken into account that the Länder were directly involved in the design and implementation of the 2011 Census via their statistical offices. From the outset, the Länder, which were mainly responsible for implementation, had sufficient opportunities for influence and participation to secure their interests in adequate implementation. In particular, they controlled all required surveys and the collection of primary statistical data. Gaps in control lying outside the sphere of responsibility of the Länder, are, at most, only conceivable in respect of centralised processes; yet it is not apparent that the statistical offices of the Länder expressed doubts as to the transparency of the procedural steps, or that such queries could not have been clarified between the official statisticians involved.

5. Moreover, the challenged provisions do not violate the right to informational self-determination (Art. 2(1) in conjunction with Art. 1(1) GG). While the 2011 Census Act does authorise interferences with the right to informational self-determination, these interferences were suitable, necessary and reasonable (zumutbar) for the census subjects. In this respect, it must be taken into account that a high level of accuracy is required to ensure that official statistics have informative value. The sample collection procedure chosen here guaranteed the required degree of accuracy, while at the same time limiting interferences with the fundamental right to informational self-determination to as low a level as possible.

6. The challenged provisions also do not violate any constitutionally guaranteed legal protection interests of the Länder or municipalities.

a) To the extent that the Basic Law sets out an obligation for the Federation to realistically determine the population, the principle of federalism under Art. 20(1) GG at least gives rise to the Länder’s entitlement to being treated equally by the federal legislature. Yet no general claim to enforcement can be derived from Art. 20(1) GG.

Therefore, errors in implementing the 2011 Census are not suited, in principle, to infringing the right of the Länder to equal treatment in the federal order. The challenged deletion provisions pursuant to § 8(3), § 19(1) and (2) ZensG 2011 and § 15 ZensVorbG 2011 only concern the implementation and monitoring of the census.

More extensive possibilities for legal protection are not required under constitutional law. It is not objectionable under constitutional law that the Länder do not have recourse to the regular courts against the determination of their official population. The population was determined by authorities of the Länder in the first place. Given that they were competent for this matter, it belonged to their responsibilities to ensure that the data was collected in a lawful and uniform manner.

b) The challenged deletion provisions do not violate Art. 28(2) GG.

Insofar as so-called city states are concerned – as in this case, where the applicants are the Land Berlin and the Free and Hanseatic City of Hamburg –, they are not covered by the guarantee of municipal self-government. Their status as municipalities is completely set aside because their status as Länder under constitutional law is their only significant legal relationship to the Federation in this case.

The legal status of other municipalities as protected by Art. 28(2) GG is not affected. The 2011 Census Act does not govern the legal relationships between municipalities and the Federation nor between municipalities and the Länder. In this respect, the deletion provisions challenged here cannot have any effect on the guarantee of protected subjective rights conferred on the municipalities, which they can invoke against the state.

III. § 2(2) and (3) and § 3(2) of the Sample Ordinance satisfy the constitutional requirements applying to them. The provisions of the Sample Ordinance comply with the requirements laid down in its enabling statute. They do not impermissibly delegate tasks to the administration or to private individuals. § 2(2) of the Sample Ordinance also does not violate the principle of clarity and consistency of the legal order derived from the principles of democracy and the rule of law (Art. 20(1) to (3) GG).