The results of migration statistics and, consequently, of current development from intercensal population updates are as from reference year 2016 only to a limited extent comparable with previous year’s values. This is due to methodological changes, technical developments regarding data transmission from registration offices to statistical offices, and the changeover to a new statistical processing procedure.
Difficulties regarding the registration of asylum seekers, problems caused by technical changes to the data transmission system and inconsistencies detected in the statistical processing result in a limited accuracy of the results for 2016 and the monthly results for 2017. These problems have largely been eliminated in the final annual results for 2017.
Methodical changes and impact on comparability:
- Up to 2015, the results of migration statistics included the arrivals and departures recorded by the registration offices until the end of the relevant month. As from January 2016, migration statistics changed over to a new data transmission and processing procedure. Because of this, for a transition period, all arrivals and departures with an arrival or departure date in 2015 or 2016 that were reported to the statistical offices between January 2016 and May 2017 were processed for the reference month corresponding to the date of the event (that is, date of arrival, departure or change in housing status). As from June 2017, migration moves have been processed for the reference month in which the registration office reported the event to the statistical office if the event was in the same month – otherwise the migration move has been processed for the month preceding data transmission from the registration office. The modified procedure for 2016 and until May 2017 produced an anticipatory effect, which means that part of the migration moves were included in the statistics earlier than usual. For example, a departure to another country that happened in March 2016 and was reported to the statistical office in February 2017 was included for reference month March 2016 rather than for January 2017, which would be in line with the current principle of processing.
- As from 2016, the results of migration statistics have included only those arrivals and departures whose date (of arrival or departure) was in the reference year or the preceding year. Previously, earlier arrivals and departures were included too, provided that they took place after the 2011 Census reference date. Such earlier arrivals and departures are only relevant for adjustments of population figures through intercensal population updates, not for the migration statistics of the current year. This is why, as from 2016, they have been included only in the intercensal population updates. As a result, the migration figures are slightly lower than in previous years.
- As from 2016, arrivals and departures of Germans “with new/previous residence unknown / no data provided" have includes in the external migration figures, whereas in the past they were not included at all. To determine the population figures, people who had been deregistered “with new residence unknown” and register again are now only processed as arrivals “with previous residence unknown” if the earlier deregistration “with new residence unknown” has been included in the statistics (i.e. if it happened in 2016 or later). As, however, all deregistrations of Germans “with new residence unknown” have been included, the number of registrations “with previous residence unknown” shown is too low when compared with the number of deregistrations “with new residence unknown”, so that the net emigration of Germans shown is too high. This effect is inevitable for methodological reasons and affects especially the results for 2016 and, with a downward trend, the results for subsequent years. The external migration figures of Germans by country of origin/destination are not affected by that methodological change.
Other limitations to the accuracy of the results:
- Inaccuracies have been detected in the registration of asylum seekers. These inaccuracies are mostly attributed to problems in the feedback procedure between registration offices and in the data transmission from the registration offices to the statistical offices. Because of constraints in the legal framework, the statistical offices cannot do much to eliminate these inaccuracies. This means that the accuracy of flow and stock data on asylum seekers is limited for 2016 and 2017.
- The immigration figures for 2016 include delayed registrations of asylum seekers who had entered Germany in 2015 and were registered only in 2016. The number of delayed registrations in Germany is estimated at roughly 90,000 arrivals. The estimate is based on evaluations of citizenships with a high proportion of asylum seekers (Syria, Afghanistan, Iraq, Iran, Pakistan, Eritrea, Somalia, Nigeria, and uncertain citizenship). Adjustments of the population registers regarding asylum seekers whose registered place of residence was an initial reception center and who have moved to another municipality are included in the results for 2016 if the adjustments took place in May 2017 or earlier.
- In the context of the technical changes to the data transmission system and the statistical processing procedure, a number of subsequent problems have been identified. They have led not only to the above methodological effects but also to an overestimation of the number of departures “with new residence unknown” as shown in net external migration and an underestimation of the number of arrivals “with previous residence unknown” as from 2016. This has led to an underestimation of the population of Germany. The quantitative impact on the annual results for 2016 is estimated as follows: underestimation of arrivals “with previous residence unknown” of 9,000 to 10,000 cases, overestimation of departures “with new residence unknown” of 6,000 to 7,000 cases, and underestimation of net migration and of the population of Germany of roughly 16,000 people. Retroactive correction of migration statistics and intercensal population updates released for 2016 is not possible. The detected inaccuracies were largely eliminated from statistics and intercensal population updates until the end of reference year 2017. Corrections of migration moves recorded in the months of 2017 are included in the annual result for 2017 of migration statistics and in the results of intercensal population updates for December 2017. Corrections of migration moves that occurred before 2017 are included only in intercensal population updates.