Data Collection
International travel and migration processing system changes in August 2016
From August 2016, international travel and migration data started being processed using a new, upgraded processing system. The new system uses improved methodology, which takes greater account of travellers' history in addition to intentions stated on the arrival and departure cards. It also makes greater use of automation in the processing and classification of passenger types.
The first provisional data available from the new processing system are the Provisional International Travel Statistics released on 19 August 2016, and the first finalised data is the International Travel and Migration: August 2016 monthly release on 21 September 2016. There will be no revisions to the historical data with the new system and methodology changes.
Why have we changed
The previous international travel and migration processing system had come to the end of its life, as it was built from technology that was no longer supported. We have developed the new system to take advantage of technology advancements, future proof the system for anticipated increases in the volume of international travel, and taken the opportunity to make methodology improvements.
en-NZMethodology
New methodology and changes related to the new International Travel and Migration processing system
New methodology and system processes include:
- more automated passenger type classification rules
- greater use of passenger travel history
- greater use of previous passenger type classifications
- imputation of 'country of residence' for overseas visitor arrivals (from citizenship and port)
- move from sampled to full coverage of some variables for visitor departures and resident arrivals, where they were matched to previous movements
- introduction of some new variables for visitor departures and resident arrivals where they were matched to previous movements.
Note: Changes in rules used to determine passenger type have been minimised as much as possible to maintain the comparability in the total passenger movements over time. There have been no changes in the variables captured for permanent and long-term travellers.
Impacts of the methodology and system changes
The following impacts are seen from the changes in the methodology and processing system:
- less manual processing of records
- improved accuracy of automatic passenger type classification
- less 'not stated' 'country of residence' for visitor arrivals
- better coverage available for visitor departures 'country of residence' using matching to previous journeys
- new variables ('travel purpose' and 'length of stay') available for visitor departures using matching to previous journeys
- better coverage available for resident arrivals 'length of absence' using matching to previous journeys
- new variables ('country of main destination' and 'territorial authority') available for resident arrivals using matching to previous journeys.
Changes seen in the data because of methodology and system changes
The following changes in the ITM data are related to changes in methodology:
- Decrease in ‘not stated’ responses for visitor arrivals by country of residence – due to the introduction of new imputation methodology using citizenship and port information.
- New series – due to the use of matching to previous journeys (eg. travel purpose for visitor departures, territorial authority area for resident arrivals).
- Increased accuracy of ‘length of stay’ for visitor departures – as departures can be matched to the visitor’s arrival and calculated, a useful comparison to reported intended length of stay on the arrival card.
- Increased accuracy of ‘length of absence’ for resident arrivals – as arrivals can be matched to the resident’s departure and calculated, a useful comparison to reported intended length of absence on the departure card.
- Increased coverage for some variables – due to the use of matching to previous journeys we can get a more comprehensive set of responses, rather than relying on sampling (eg. previously 1 in 26 visitor departures ‘country of residence’ were sampled, now all visitor departures ‘country of residence’ are captured when matched to their previous arrival).
- Increased sample volatility for some variables – due to different procedures for dealing with manual sample changes compared to the previous system (eg. when passenger type is changed on review, from its automated class).
- Increased ‘not stated’ responses for sampled resident departure variables (eg. ‘country of main destination’) – due to the use of travel history to automate passenger type assignment. Although a departing international traveller may have answered the departure card as if they were a visitor to NZ (eg. not provided information on purpose and country of destination), our automated rules may look at their travel history and determine that for ITM purposes we would class them as a resident of NZ.
Please note that some of the changes may create discrepancies in time series available (particularly with capturing variables from matches to previous journeys and moving from sampled to full coverage). For example Infoshare series on visitor departures by 'country of residence' used sample counts prior to August 2016 and actual counts from matched visitor arrivals from August 2016.
en-NZ
