International Travel

Where the data comes from

International Travel and Migration statistics are based on electronic arrival and departure records for each passenger supplied to Statistics NZ by the New Zealand Customs Service. These electronic records include flight and passport details, such as date of travel, date of birth, sex, and country of citizenship.

The New Zealand Customs Service also supplies Statistics NZ with arrival and departure cards from every passenger. Additional detail such as country of residence and travel purpose is captured from these cards and added to the information from the electronic record.

Immigration New Zealand owns the arrival and departure cards, which are used to administer numerous New Zealand laws, in addition to producing statistics. Statistics NZ only captures information from the cards for statistical purposes, and passes the cards to Immigration New Zealand after it has captured the data.

How the data is collected

Statistics NZ scans every arrival and departure card, and uses image recognition technology to automatically recognise and code responses from each. Processing staff then manually capture any required information which was not captured automatically - as not every response can be clearly recognised by automated software.

The imaging system automatically determines all the information required for almost 9 in every 10 cards (meaning about one million still require manual completion each year).

Some response fields are required from each card. These include the passport number and birth date, so each card can be matched to the electronic record supplied by Customs containing passport and flight details. Also, responses to length of stay/absence and country last/next lived in questions are required from every card, to determine the person's passenger type (overseas visitor, NZ-resident traveller, or permanent and long-term migrant).

Most other response fields are only required for a sample of passengers, and required fields vary by passenger type and direction.

Responses to some questions are not captured at all, including the first and last names of the passenger. Responses to questions on the back of the arrival card are not captured, nor are responses to the seat number, email address, and phone number questions on the front of the arrival card.

Passenger counts

International Travel and Migration statistics relate to the number of passenger movements to and from New Zealand, rather than to the number of people – that is, the multiple movements of individual people during a given reference period are each counted separately. For example, a New Zealand resident making five business trips overseas within a year would be counted as five arrivals and five departures.

Determination of passenger type

Passenger type is determined from responses to questions on the arrival and departure cards about how long the person is in or away from New Zealand, and where they are living for 12 months or more. A person's travel history (their other arrivals and departures) will also be used to determine their passenger type.

Changes in intentions

A person may change their intentions after their arrival or departure, which may mean the recorded passenger type becomes incorrect. For instance, a person who indicated they were arriving for a short-term stay may actually stay permanently. In this case, they would be recorded as a short-term overseas visitor, even though they became a permanent and long-term migrant. Statistics NZ does not revise published statistics to adjust for such changes.

Seasonally adjusted and trend series

Statistics NZ also produces seasonally adjusted and trend series for each of the passenger types and directions (eg total visitor arrivals). These series attempt to remove regular seasonal variation so each month (or quarter) can be compared.

Seasonally adjusted figures are only estimates, and are revised every month as more data becomes available. Trend series are a smoothed version of the seasonally adjusted series.

Statistics NZ can provide assistance on the interpretation of seasonally adjusted and trend series. If quoting seasonally adjusted and/or trend series, this should be clearly stated.

Confidentiality

Statistics NZ releases statistics derived from arriving and departing passenger records. It does not release the records of individual passengers.

Data is continually collected and processed.

International travel and migration data is released at pre-specified times to ensure all users have equal access to the data. These release dates are published well in advance on the release calendar on Statistics NZ's website.

Weekly provisional statistics

Provisional international travel statistics are published on a weekly basis, usually at 2pm on a Friday. These provide provisional figures on visitor arrivals and short-term departures by New Zealand residents, including statistics for ten major source and destination countries. No further detail is available until the release of monthly data.

Release of monthly data

Monthly international travel and migration data is released at 10.45am, usually 15 working days after the end of the month. The release of December and January data takes longer after the end of these months, due to the Christmas-New Year holidays and high passenger volumes.

Additional embargo of port data

New Zealand port and overseas port data is released two working days after the release of monthly international travel and migration statistics. This is part of an agreement with airlines and airports which allows Statistics NZ to release this otherwise confidential data.

Actual and sampled counts

Some fields, such as citizenship, date of birth, and sex, are held for all arrivals and departures. However, some fields are only collected for a sample of passengers.

Only short-term (overseas visitors and New Zealand-resident travellers) are sampled. Required fields are collected for every permanent and long-term migrant.

Data from the sample is used to estimate the characteristics of the full population.

Published counts

The only actual (non-sampled) counts regularly published by Statistics NZ are totals by passenger type and direction (ie total visitor arrivals, total New Zealand resident departures, etc), and statistics for permanent and long-term migrants.

All other statistics published for short-term travellers are derived from a sample of records. This is true even when a field is available for all passengers (for example country of residence for visitor arrivals). This avoids confusion by publishing actual counts for a field in some instances, but sampled counts for the same field if it is cross-tabulated with another field only available for a sample of passengers. The exception is visitor departures by country of residence. With a methodology change in August 2016 it was determined that the actual (non-sampled) counts for visitor departures by country of residence were more suitable for release than sampled counts. Series of visitor departures by country of residence are available on Infoshare using sampled counts prior to August 2016 and actual counts from August 2016 on.

Sampling ratio

Within each passenger type and direction, one in every k passengers is selected for the sample, where k is the sample ratio. Until June 2011, the sample ratios were changed every month to adjust for changes in the number of passengers, so were higher in summer than in winter. From July 2011 to July 2016, the sample ratios were kept the same every month:

• Resident arrivals: 1 in 26
• Visitor arrivals: 1 in 16
• Resident departures: 1 in 20
• Visitor departures: 1 in 26

From August 2016:

• Resident arrivals: sample weights can be 0, 1, or 20. Resident arrivals are matched to their previous departure to gather information. Depending on the sample status and passenger type of the previous journey, as well as the type of variable, the sample ratio will vary.
• Visitor arrivals: sample ratio 1 in 16
• Resident departures: sample ratio 1 in 20
• Visitor departures: sample weights can be 0, 1, or 16. Visitor departures are matched to their previous arrival to gather information. Depending on the sample status and passenger type of the previous journey, as well as the type of variable, the sample ratio will vary.

Resident departure records are selected for the sample simply by choosing every kth record processed for each passenger type and direction.

For visitor arrivals, records selected for the sample are the kth record processed for each country of residence. This ensures that the country totals derived from a sample will be close to the actual totals.

Counts from the sample of records are rated up by the sample ratio to estimate actual counts.

Sample error

Sample error is the difference between the sample estimate and the actual figure.

When the actual figure is small, the error from sampling can often be large as a percentage of this figure. When the actual figure is large, the error from sampling will usually be small as a percentage of the actual figure.

Statistics NZ does not highlight totals or differences under 500 because of the potentially large percentage sample error.

Absolute sampling error can be calculated using the following equation:

Relative sampling error gives a percentage measure of the magnitude of the error. It can be calculated using the following equation:

Sample error example

Suppose that in a particular month, there were 1,000 visitors from Brazil (population) of whom 100 (cell size) visited New Zealand to visit their friends or relatives. In this situation, the absolute sampling error is 76, which means that the true number of people arriving from Brazil for this particular purpose, at the 95 percent confidence level, could fall between 24 and 176.

International travel and migration records are collected from all New Zealand airports and seaports handling international movements.

New Zealand port is available as a data variable, as is territorial authority of residence (captured from address responses on arrival and departure cards) for some passenger types.