Page 19 - Invited Paper Session (IPS) - Volume 1

IPS30 Stefan D. et al.
size of the operations used for data collection in traditional, but also many
combined censuses goes along with rather long production times so that
results can only be produced with some delay. As the experiences of countries
with register-based census illustrate, production time can be shortened by
using a register-based approach. At the same time, a frequency of ten years is
considered too long by many users. Providing annual results for a limited
subset of census variables could help to remedy this concern. Finally, the
potential of the census to inform local and regional decisions regarding
population and housing can only be tapped if data are provided in grid-based
form allowing for variable tabulations also for non-administrative areas.
Despite the fact that the combined census model in Germany already led
to important cost reductions, its components related to traditional data
collection (in particular the complete enumeration of buildings and dwellings
and the relatively large supplementary household sample survey) still go along
with relatively high cost. Experiences from countries with purely register-based
censuses indicate that important cost reductions can be achieved (UNECE
2014).
With further efforts to make progress in the digitisation of public
administration, the register infrastructure in Germany is under constant
development. This development may lead to further harmonisation and
linking of registers, which could open new opportunities for the use in official
statistics and in particular the census. With new technologies applied for data
protection and encryption, some of the legal challenges in connection with
the constitutional courts judgement of 1983 could possibly be solved.
The introduction of a fully register-based census presupposes that a
number of preconditions are being met regarding register access, data quality
and the possibility to link the records from several registers:
• Data access: To move to a register-based census, the necessary data in the
fiields of demography, buildings and dwellings, household and family types,
as well as labour market and education must exist and be accessible to official
statistics. While a significant part of the information of interest is already
available in registers, some still need to be created. For instance, no
nationwide registers on buildings and dwellings as well as on the educational
attainment of the population are currently existing.
• Data quality: To be used for statistical purposes, registers need be of sufficient
quality and approaches need to be developed in order to correct the data for
errors. This is of particular importance in the case of administrative registers,
which are usually not kept for statistical purposes and therefore often need
data processing. As shown above, the census 2011 has shown that the
population registers are subject to significant over- and undercoverage. We
are currently exploring suitable ways to correct for these errors by linking the

8 | I S I W S C 2 0 1 9