Migration Data Transformation
Migration Data Transformation
Migration Data Transformation removed the statistical dependency on the data provided by departure cards and transforming external migration data with the sustainable implementation of a new measure of migration.
Methodological changes were made to enable production of the official measures of migration, tourism, and estimates of population to continue with the removal of nearly 7 million paper-based departure cards completed by people leaving New Zealand annually. This involves looking at how to make more use of existing information, such as border crossing information from Customs (passport data), arrival cards, and integrated administrative data.
Impact on variables As a result of the Migration Data Transformation and the removal of the departure cards in November 2018, there were changes to international travel and migration variables (see Impact on variables - from November 2018).
The new measure of migration: Outcomes-based measure
The previous migration measure (permanent and long-term (PLT) migration) was estimated from travellers’ statements on arrival or departure cards – based on how long they intended to stay in New Zealand (or be away). This intentions-based measure was timely; however, traveller behaviour was not always consistent with the stated intentions at their border crossing. This may be due to: circumstances changing, misunderstanding the questions on the traveller cards and incorrectly reporting their intentions, or deciding to extend their visa or stay/absence.
The outcomes-based measure is estimated from the actual travel histories of people travelling in and out of New Zealand. This provides a more accurate measure of migration – a traveller is classified as a migrant based on their actual movements.
The outcomes arrival and departure estimates are consistently higher than intentions PLT estimates.
The outcomes net migration estimates are sometimes higher, and sometimes lower, than intentions PLT estimates. In recent years, net migration using outcomes was lower than intentions.
Why migration estimates change The outcomes-based measure of migration with provisional and final estimates is now the official way we measure migration in New Zealand. The results build on the outcomes-based measure of migration that was released from May 2017 and enabled the removal of the departure card.
The new approach uses passport data to link arrivals and departures and accurately measure how long people spend in, or out of, New Zealand after their initial border crossing. To classify a border crossing as a migrant movement, we need to observe up to 16 months of travel history.
With this new approach it takes 17 months before final migration estimates are available. To produce timely results, we use a statistical model to produce provisional migration estimates. Statistics produced using these provisional estimates have uncertainty for 16 months; after this time we can finalise the classification of all border crossings (according to the 12/16-month rule).
As new data becomes available, the provisional migration model has more information about the border crossings it is trying to estimate. So, with an extra month of data available, this causes shifts in the estimated number of migrant arrivals and migrant departures, and thus changes in the net migration estimates.
For example, the extra data will indicate travellers who have now departed New Zealand, or travellers who were away that have since returned to New Zealand.
Compared with total border crossings, the number of migrants is very small. Of every 50 people crossing our border, typically 49 are short-term movements and only 1 is a migrant arriving or departing.
The migration estimates become more certain after each subsequent month. For a typical month, 1 in 4 arrivals are classified with certainty for the first estimate after six weeks (after the end of the reference month). This increases to 9 in 10 after four months. The monthly revisions can therefore be expected to become relatively small after about five months, as we can calculate the duration of stay/absence more definitively.
Of the 14 million border crossings in the December 2018 year, 81 percent of border crossings were classified with certainty at the time of the first estimate in mid-February 2019. The remaining 19 percent represent 2.6 million border crossings, so a small change in classification can affect the migration estimates.
Customers therefore have a choice of using the most timely migration estimates which have more uncertainty, or waiting a few months until the migration estimates become more certain.