Stats NZ has a new website.

For new releases go to

www.stats.govt.nz

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
6 Privacy analysis of proposed data integration

This chapter determines whether the proposed data integration complies with Statistics NZ's Data Integration Policy.

The policy exists to ensure that impacts on privacy are minimised when we consider proposals to create statistical and research outputs using data integration.

In addition, compliance with this policy ensures that the proposed data integration is in keeping with important legislation, including the Privacy Act 1993, the Statistics Act 1975, the Public Records Act 2005, and any other legislation relevant to the source datasets.

Stakeholders consulted

 The following people have been consulted in relation to compiling this business case:

  • Legal Counsel, Statistics NZ
  • Legal Team, MfE
  • Manager, Evidence Reporting, MfE
  • Manager, Customer Policy & Research, Statistics NZ
  • Environmental Reporting Team, Statistics NZ
  • Manager, Security, Statistics NZ
  • Privacy Officer, Statistics NZ
  • Risk Officer, Statistics NZ
  • Senior Advisor, Data Lab, Statistics NZ
  • Integrated Data Infrastructure, Statistics NZ
  • Manager, Business Indicators and Agriculture Production, Statistics NZ
  • Statistical Methods, Statistics NZ
  • Office of the Government Statistician, Statistics NZ
  • Manager, Strategic Communications, Statistics NZ.

How the proposed integration complies with our data integration principles

Statistics NZ will consider integrating data to produce official statistics and related research only when all four of the following principles are met.

Principle 1: The public benefits of integration outweigh both privacy concerns about the use of data and risks to the integrity of the Official Statistics System, the original source data collections, and/or other government activities

a) How will the proposed data integration produce or improve official statistics and/or statistical research?

  1. This data integration will bring together a number of data sources and provide new insights into agricultural land use and freshwater quality changes in New Zealand. It will produce better information on which to base decisions relating to agricultural land use and fresh water.
  2. The new statistics developed in this data integration will help Statistics NZ and the Ministry for the Environment (MfE) meet the reporting needs of the Environmental Reporting Bill. The outputs from this data integration will be published, at national and regional level, in the first Freshwater Domain Report due in 2016, or sooner if results are ready well ahead of the report. The report will be jointly produced by Statistics NZ and MfE.
  3. This data integration supports Statistics NZ’s strategic vision of Unleashing the power of data to change lives and its purpose of Empowering decisions by adding value to New Zealand’s most important data.

b) What are the expected public benefits of the proposed data integration?

  1. Published data will show changing farm outputs and changing water quality. Over the past 10 years, some water quality attributes have shown improvements while others have declined. This data integration will explore if there are statistically robust relationships, such as patterns of change or likely impacts between agricultural land use and freshwater quality changes in New Zealand.
  2. It will also enable relationships to be derived between agricultural land use and freshwater quality.
  3. As a result of this data integration, government agencies, local authorities, Māori, farmers, farmer organisations, media, politicians, and environment advocacy organisations will have better information on which to base decisions relating to agricultural land use and fresh water. Water quality monitoring agencies and policy makers will also be able to make more informed decisions on agricultural land use and water quality policy and practices. Interest may extend beyond New Zealand and have an impact on this country’s export and tourism sectors.

c) What potential privacy concerns, both real and perceived, are associated with the data integration?

  1. The different stages of this project present different issues with respect to privacy. However, all have the potential to derail not only the work this project will produce (and potentially prevent it being repeated in future) but also to damage the collection and production of the databases on which this project relies, as well as the other first intended purposes of the existing agricultural production and land records databases.
  2. Individuals may have concerns that:
    • information may be released that identifies them or their farm
    • information may be used in a manner that is detrimental to them or their farming interest
    • unrelated information may be collected about them or their farm in an ever-growing database for non-specific purpose
    • negative publicity will link farm locations or practices with poor water quality in nearby lakes, rivers, and stream
    • information may lead to punishing legislation for farms or farming practices.
  3. Privacy, security and confidentiality of respondent’s information is protected by the provisions of the Statistics Act 1975, and the policies and procedures Statistics NZ have in place to meet the requirements of that Act. Data is not released in a form where an individual could be identified.
  4. Published outputs will be in the form of aggregated statistics and any tabulated cells that contain small numbers (for example at some regional levels) will be confidentialised using standard methods. The use of the aggregated data for purposes not already defined in this PIA will require a new PIA to be completed.
  5. This data integration will explore if there are statistically robust relationships, such as patterns of change or likely impacts between agricultural land use and freshwater quality changes in New Zealand. Published messaging will also explain that changes in agricultural land use and freshwater quality are, each, standalone indicators and may not necessarily be directly linked in a cause and effect relationship.
  6. When published as a series, this data will inform New Zealanders of changes over time in agricultural land use and freshwater quality so that better-informed decisions about agricultural land use and fresh water can be made.

d) What are the possible risks and potential benefits to the integrity of the Official Statistics System, the source data collection, and other government activities?

  1. A privacy breach has the potential to impact individuals and the ongoing trust and confidence farmers have in Statistics NZ. This could also lead to a sharp deterioration in responses to future Agricultural production surveys and, potentially, all survey work undertaken by Statistics NZ.
  2. The farming sector is an important element of New Zealand’s history and economy. The rural voice is well represented through farmer organisations and specialist boards through to strong representation in Parliament. A negative event associated with this data matching would likely be well-aired in the public arena.
  3. Integrating farm and water quality data as proposed in this PIA has the potential to impact the privacy of individuals involved, as it involves combining their information from different sources.
  4. However, the potential risks for a breach in privacy are easily identified, and mitigation plans are in place, as required by the legislation and policies that apply to this data integration. This includes having staff seconded to Statistics NZ and researchers working as contractors to Statistics NZ, in the confines of secure (Data Lab) environments.

e) Will the data integration be reviewed? If so, what will the review period be?

  1. The process followed to integrate the data will be fully documented and independently reviewed prior to the information being published in the Freshwater Domain Report due in 2016. Two post-doctorate researchers from Landcare Research will review the process and methodology used in this data matching. They are contracted to MfE to undertake the review. They will be seconded to Statistics NZ, during stage 1, working under the provisions of the Statistics Act 1975, and as researchers under s37 of the Statistics Act 1975 during stages 2 and 3.

f) What information is required for linking the datasets together?

  1. The Ministry for Primary Industries (MPI) and the Ministry for the Environment (MfE) have combined information from three databases, AgriBase®, FarmsOnLine and Land Information New Zealand data (see table1). This integrated database contains names and addresses and spatial boundaries for each farm.
  2. The combined dataset will be imported into Statistics NZ and matched to Statistics NZ’s Agricultural production survey (APS) using the names and addresses within each dataset to produce an integrated dataset of farm location, size and production data. At this point names and addresses will be removed and replaced with a farm identifier for modelling and mapping purposes. The end product, of this stage, will be a spatial map of farms in the APS for all available years from 1996 to 2014.
  3. The modelling stage uses production data (eg livestock numbers and fertiliser applications) from the APS to estimate various outputs on each farm including nutrient loss and waste to pasture. The farm identifier (from stage 1) allows this data to be matched to a location.
  4. The mapping stage will locate a farm and its outputs onto a map alongside river, lake, and groundwater quality information. At this point, the individual farm identifier is removed. The final output will show changing farm outputs and changing water quality.
  5. The information will be published in the first Freshwater Domain Report due in 2016. The published results will not divulge information about individual farms, or be presented in a way that can be used to infer information about an individual farm.

g) What policies and practices will be in place to protect the data from unauthorised access?

  1. Stage 1, where the name and address matching occurs, will be undertaken at Statistics NZ by seconded staff. This provides the same legal and physical protections and securities given to all personal information held by Statistics NZ. Signed commitments to the Statistics Act 1975 are lifetime obligations in terms of that Act. No names and addresses will be on the dataset at the end of stage 1.
  2. Stage 2 and stage 3 will be undertaken in Statistics NZ Data Lab conditions. These provide a secure environment and researchers accessing the data will sign documentation relevant to s37C of the Statistics Act 1975. This ensures that only bona fide researchers will access the datasets for bona fide research purposes. The data accessed in these stages will contain no names and addresses and at the end of stage 2, the database to be used as input to stage 3 will have all individual location identifiers removed. It will contain land parcel identifiers (all farms will have been placed into these land parcels) in preparation for aggregation.
  3. Following completion of the modelling and mapping stages, the farm’s identifier will be removed (for use in stage 3 – however, a copy with the identifier intact will be retained by Statistics NZ for potential data quality assurance) and Statistics NZ’s provisions for confidentiality applied so that no farm is recognisable in the published results. The published results will not divulge information about individual farms, or be presented in a way that can be used to infer information about an individual farm.
  4. In addition, data used in matching (stage 1) and outputs used for modelling (stage 2) are stored on limited access drives to ensure only authorised personnel are able to gain access to information while it contains names and addresses and personal identification numbers.

h) Do the outcomes of the data integration justify the impact on individuals' privacy?

  1. The value of this information is very high and will better inform decision-making on agricultural land use and freshwater quality relationships and any potential measures or controls on agricultural land use and freshwater quality.
  2. Environmental reporting of this nature is highly sought after and in strong public demand.
  3. There are manageable concerns and risks to individual privacy, confidentiality and to the security of the personal information used in this data integration. There are also potential impacts that the information itself may inspire. These could lead to impacts on survey response rates. As a result there are five areas of concern that this PIA deals with: privacy, confidentiality, security, survey response, and public response.
  4. The outputs from this data matching will be published in the 2016 Freshwater Domain Report and help the Government Statistician and the Secretary for the Environment meet their obligations under the Environmental Reporting Bill.

For these reasons, the outcomes of this data matching (and the potential associated risks) outweigh and justify those potential risks.

Principle 2: Integrated data will only be used for statistical or research purposes

a) What are the statistical and related research purposes of the proposed data integration?

  1. All use of integrated data must comply with the requirements of the Statistics Act 1975. It and the Privacy Act 1993 require that Statistics NZ only collect and use information that is necessary for statistical purposes. This means that data may not be used for other purposes (eg audits by the data source agencies or regulatory or non-regulatory monitoring). This is an important demonstration of Statistics NZ’s commitment to respondents' privacy.
  2. Appendixes 1 and 2 set out information for this data matching of:
    • variables that are likely to have a statistical or research purpose
    • variables that are unnecessary and will be removed.

b) What is the expected long-term value of the data that will result from the integration?

  1. These data integration outputs are likely to be of interest to most New Zealanders. In particular, government agencies, local authorities, Māori, farmers, farmer organisations, media, politicians, and environment advocacy organisations. They will have better information on which to base decisions relating to fresh water. Water quality monitoring agencies and policy makers will also be able to make more informed decisions on agricultural land use and freshwater quality policy and practices. Interest may extend beyond New Zealand and have an impact on this country’s export and tourism sectors.
  2. It also enables the Secretary for the Environment and the Government Statistician to meet their reporting requirements set out in the Environmental Reporting Bill.
  3. This database matching is being undertaken to produce agricultural land use and freshwater quality information for use in the 2016 Freshwater Domain Report. The report will also be updated for publishing in 2019. It is likely that this database matching and integration will need to be updated (from sources available at that time).
  4. It aligns with Statistics NZ’s strategic vision Unleashing the power of data to change lives and purpose Empowering decisions by adding value to New Zealand’s most important data.
  5. The matched database will not be used for any purpose other than the intentions outlined in this privacy impact assessment (PIA) unless a further PIA is prepared and approved.

c) What policies and practices are in place to ensure the data is only used for appropriate research?

  1. The data matching is being led by the Ministry for the Environment, however Statistics NZ maintains the lead on all statistical matters relating to the data matching and outputs. Seconded staff (stage 1) ensures the same legal and physical protections and securities for the data as that given by Statistics NZ.
  2. Only bona fide researchers undertaking bona fide statistical research under s37C of the Statistics Act 1975 have access to the data sets produced in this data matching.
  3. This ensures that only bona fide researchers can utilise the datasets for bona fide research purposes. The data accessed in these stages will contain no names and addresses, and at the end of stage 2 all individual location identifiers will be removed before stage 3 commences.

d) What confidentiality practices will be in place to ensure that data can only be used for statistical purposes?

  1. Data published from this data matching will be signed off by Statistics NZ to ensure the confidentiality practices have been met. The information will be published in the first Freshwater Domain Report due in 2016 or sooner if results are available well ahead of the report’s expected release. The report will be jointly produced by Statistics NZ and the Ministry for the Environment.
  2. The confidentialisation of statistical outputs refers to the methods in place to ensure that no information is published that could identify an individual. Under the Statistics Act 1975, Statistics NZ does not publish any information that could lead to an individual’s identity.
  3. Statistics NZ will be the final custodian of the Agricultural production survey spatial maps and any outputs from modelling Agricultural production survey information.

Principle 3: Data integration will be conducted in an open and transparent manner

a) How will information about the proposed data integration be made publicly available?

  1. Statistics NZ will publish this privacy impact assessment on its website. The Ministry for the Environment (MfE) and the Ministry for Primary Industries (MPI) will also be invited to publish or link to this document on their websites.
  2. The information will be published in the first Freshwater Domain Report due in 2016 or sooner if results are available well ahead of the report’s expected release. The report will be jointly produced by Statistics NZ and MfE.
  3. Results of this data integration will be released in accordance with the policy for the release of official statistics – Principles and Protocols for Producers of Tier 1 Statistics – that underlie the release of all official statistics.

b) How will information about the statistical results or research outcomes be made publicly available?

  1. The information will be published in the first Freshwater Domain Report due in 2016 or sooner if results are available well ahead of the report’s expected release. The report will be jointly produced by Statistics NZ and the Ministry for the Environment.
  2. The published results will not divulge information about individual farms, or be presented in a way that can be used to infer information about an individual farm.

Principle 4: Data will not be integrated when an explicit commitment has been made to respondents that prevents such action

a) What commitments, both actual and implied, have been given to respondents about how their personal information will be used?

Here are the messages we give to respondents about how their personal information will be used in relation to the following databases:

  • Statistics NZ’s Agricultural production survey
    • Only people authorised by the Statistics Act 1975 will access individual information.
    • Information will only be used for statistical and research purposes.
    • Individual information will be combined with similar information to prepare summary statistics.
    • Individual information will be released to the Ministry for Primary Industries (MPI) but continues to be protected by the Statistics Act 1975 (s37) and must only be used for statistical purposes and must not be released in any way which identifies individual information.
  • MPI’s farm database (FarmsOnLine)
    • The Director-General of MPI may authorise any person to access the public domain information in FarmsOnLine for any lawful purpose. Access for commercial purposes would not generally be permitted.
    • Only information sourced from the public domain are used in this case.
    • Individuals can check and update their own information online.
    • Individuals can request their information be withheld from the FarmsOnLine at any time.
    • The Director-General of MPI has no power to disclose information that has not been sourced from the public domain, unless:
      • the requested information falls within a biosecurity purpose under the Act, or
      • the Director-General has received permission from the person (or their agent) to whom the information relates, to release the information, or
      • legislation (other than the Official Information Act 1982) requires access to be provided.
  • AgriBase®
    This database is prepared by AsureQuality (licensed to the Ministry for the Environment (MfE) for the specific purpose(s) of modelling and reporting on water quality flows and the pressures on water quality and flows throughout New Zealand). AgriBase® data (as stated in their licence agreement with MfE) may not be used for correspondence with, or for directly contacting, landowners or occupiers for purposes relating to compliance, legal action, or the enforcement of customer regulations. Any products or written papers using or referring to the data will not divulge information about individual farms, or be presented in a way that may be used to infer information about an individual farm.

    On their property registration form, where they collect the information stored in AgriBase®, AssureQuality state that:
    • AsureQuality may collate, deal with and use the information in such manner as AsureQuality reasonably thinks fit, and includes to:
      • enhance New Zealand’s agricultural productivity and ability to trade
      • help respond to and manage rural emergencies, diseases, pests, environmental quality issues and product quality issues
      • help produce agricultural statistics
      • assist fundraising by rescue services.
  • Land Information New Zealand
    The land owner data provided by LINZ may be used for any purpose, even commercially, as long as:
    • the provisions of the Privacy Act 1993, including the information privacy principles, are adhered to as for an agency under that Act
    • the data is not misused and is not provided to any individual or entity that has not agreed to the LINZ Licence for Personal Data
    • nothing is done, or anything is not done that would cause LINZ to breach its obligations under the Privacy Act 1993
    • the Data is amended or deleted within five business days if requested by LINZ.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Top
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+