• Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Adding Accommodation Survey data to the Longitudinal Business Frame – Privacy impact assessment

Purpose

The purpose of this business case is to seek approval to add Accommodation Survey data collected by Stats NZ to the Longitudinal Business Database (LBD).

Approval is sought to:

  • extend the LBD by integrating Stats NZ’s Accommodation Survey data – to enable research and to better inform policy decision-making
  • provide access to Stats NZ’s Accommodation Survey data to approved Stats NZ staff and to external researchers approved by the Government Statistician through the microdata access service.

Background

About the LBD

The LBD is the creation of the Improved Business Understanding via Longitudinal Database Development (IBULDD) project. This was a two-year project to test the feasibility of creating a longitudinal firm-level database covering the period 1999–2005/06. The project was funded by Cross Departmental Research Pool money and finished in December 2007 with the successful creation of the prototype LBD.

Since 2007 the LBD has expanded to include additional business data held by Stats NZ and other government agencies.

Reasons to add accommodation data to LBD

New Zealand’s tourism industry has grown significantly over recent years. Over the year to March 2016 expenditure by international visitors grew 25 percent, to reach $10.2 billion. The Ministry of Business, Innovation and Employment (MBIE) forecasts this growth will continue, with international visitor expenditure expected to reach $16 billion by 2022.

This growth has increased demand for MBIE to improve its understanding of the industry, and deliver insights to government ministers and the tourism industry. In response, MBIE’s senior leadership team commissioned a series of Tourism Insights papers to focus on six key issues facing the tourism sector. This series will use new data sources and analysis to help inform decisions by government and industry. Producing these insights papers is a high priority for MBIE.

The sixth paper in this series will focus on the productivity of firms in the tourism sector, which was highlighted in the 2011 Tourism Data Domain Plan as a key topic area of interest. This work would make substantial use of Accommodation Survey data if it were within the LBD. Doing this would allow productivity to be measured at a very granular level, but also to derive insights that wouldn't’t be possible using aggregate data.

Stats NZ’s Accommodation Survey

The Accommodation Survey has collected data on short-term commercial accommodation providers for more than 20 years. The survey covers a range of accommodation providers including hotels and resorts, motels (includes motor inns and serviced apartments), backpacker accommodation (includes short-stay hostels), and holiday parks (includes caravan parks and camping grounds).

Because the survey’s sample frame is based on the Stats NZ Business Frame, the data already contains the unique business identifiers used by the LBD. Names and addresses provided in the datasets will not be required for linking purposes and will not be available in the linked dataset available to researchers.

Stats NZ is seeking to integrate comprehensive datasets that cover all available years of Accommodation Survey data and will update this data annually.

Extending the LBD

The privacy and confidentiality risks related to integration are assessed in the risk and mitigation table. (See Complete risk assessment for Accommodation Survey data in the LBD in ‘Available files’.) This assessment assures that the linking can be carried out in a way that adequately addresses privacy concerns. It also identifies privacy risks and the process by which they will be managed.

Data collected is subject to the requirements of Part 4 of the Statistics Act 1975 for de-identification and confidentialisation. Specifically, sections 37–37C of the Act set out the requirements around security and disclosure of information.

Benefits of adding accommodation data

A key feature of the LBD is its use in analysing business performance and productivity. Integrating Accommodation Survey data would help with research projects on tourism productivity; specifically, investigating the effect of seasonal demand on business productivity (a key topic in the Tourism Data Domain Plan).

Commercial accommodation providers are an essential part of the tourism sector. The Accommodation Survey provides very detailed information on firms in this industry that is not found in other data sources. This information includes: number of guest nights supplied, number of stay units (eg bedrooms) available, a breakdown for international and domestic visitors, and a breakdown of providers by accommodation type (hotels, motels, backpackers, and holiday parks).

Crucially, this information is available monthly, giving insight into how demand varies across the year (seasonal demand). Having the Accommodation Survey linked to the LBD would allow a more detailed analysis of accommodation providers.

Table 1 outlines potential research, and how each relates to MBIE’s business needs.

Table 1

 How potential research relates to MBIE’s business need
 Policy context  Business need  Information from Accommodation Survey in LBD  Potential analysis
Visitor arrivals have grown faster than accommodation supply, leading to an undersupply of hotel rooms in key tourism destinations. To understand how accommodation supply and investment responds to demand. Information on demand for accommodation (number of guest nights) at business and regional level. How changes in demand, and in seasonal variation, affect business profitability and investment.
Large seasonal variation in visitor numbers (seasonality) identified by MBIE and tourism industry as a key challenge facing the sector. To understand how seasonality affects productivity, and what strategies are effective at mitigating this. A measure of seasonality faced by each accommodation provider. How seasonality is related to business investment, employment practices, and productivity.

An economic production function that considers the spare capacity a firm needs to invest in to satisfy peak demand, and how that affects productivity.
Government policies (eg international advertising and funding for major events) to target more tourists outside peak seasons. To quantify the value provided by additional tourists at different times of the year. A measure of seasonality faced by each accommodation provider. A model of the cost of production for accommodation providers, and the marginal cost of additional tourists at different times of the year.
International visitors are forecast to increase considerably over the five years to 2016, changing the tourist mix faced by the tourism industry. To understand how firms can best take advantage of more international visitors. A breakdown of demand into domestic and international tourism. Analysis of how accommodation providers serving international tourists differ from those serving domestic tourism (eg productivity, business practices, ICT use).

In addition, integrating Accommodation Survey data in the LBD would allow regular monitoring and reporting, to give MBIE and the tourism industry greater insight into the operation of the industry. The integrated data could be used in additional research projects, by MBIE or others researching the tourism industry.

Risks of adding accommodation data

This integration proposal relates to businesses. Businesses are generally not covered by the Privacy Act 1993, although some businesses (eg sole traders) can have privacy concerns that may be covered by that Act. However, Stats NZ is still required to protect the confidentiality of individuals and businesses under the Statistics Act.

Although there are risks associated with any use of the LBD integrating Accommodation Survey data does not change the consequence or likelihood of any of these risks and does not introduce any new risks. Consequently no additional mitigations are required.

The risks of integrating Accommodation Survey data were assessed and given an ‘unmitigated risk’ rating, based on the consequence and the likelihood of the risk occurring, using these tables:

  • risk consequence table
  • risk likelihood table
  • risk rating table.

Each risk is also given a residual risk rating, which considers the mitigation or controls that are in place to manage the risk. In assessing the Accommodation Survey data, no new risks were identified. Table 2 outlines the risk to the LBD in general for any risks with a residual risk rating of medium or higher, and the current controls in place to mitigate these risks.

Risks rated as medium or higher

In assessing adding the Accommodation Survey data to the LBD, we identified two risks with a residual risk rating of medium or higher.

Table 2

 Residual risks rated medium or higher, Accommodation Survey data in LBD
Description of risk Existing controls that help manage risk Assessment of residual current risk
 Principle 11: Disclosure of information
Businesses can be re-identified by researchers accessing the data – they recognise them in the information.

All researchers must be approved by the Government Statistician and meet the requirements of the Statistics Act 1975 Section 37C (1). They have the necessary research experience, knowledge, and skills to access and use the information. Processes exist to assess a potential researcher’s integrity and experience.

Once researchers are approved they sign a legally binding Declaration of Secrecy before accessing the data, which requires them to keep any identifying information secret.

Regular audits ensure only approved researchers can access a de-identified dataset.

Consequence = Major
Likelihood = Unlikely
Risk rating = Medium
Businesses can be identified in the research or statistics published.

Researchers and staff are trained to apply confidentiality methods to research and statistics before publication – to ensure they contain no information that could identify an individual or business.

All output is independently checked; where information is released from the secure IDI environment (Data Lab or remote access facility), trained Stats NZ staff perform strict confidentiality checks before release.

Consequence = Major
Likelihood = Unlikely
Risk rating = Medium

Complete risk assessment

See Complete risk assessment for Accommodation Survey data in the LBD in ‘Available files’ for the complete risk assessment table.

Rating consequences and likelihood risks

Tables 3–5 show the tools we use to assess the ratings for the consequence and likelihood of a risk occurring.

Table 3
Risk consequences

 Rating Consequence
 Severe Would require extensive senior management attention and diversion of resources to recover from the risk event.
 Major Significant senior management attention would be required to recover from the risk event.
 Moderate Management effort would be required to prevent the situation from intensifying. Changes to operating procedures would be required.
 Minor Management oversight would be required to ensure effectiveness and efficiency is maintained. Changes to operating procedures may need to be considered.
 Insignificant Management oversight might be required to ensure day-to-day, routine operations are not disrupted.

Table 4
Risk likelihood

 Rating Likelihood of occurrence
 Almost certain The risk consequence will occur in most circumstances.
80–100% expectation in the next 12 months
 Likely The risk consequence will probably occur.
50–80% expectation in the next 12 months
 Possible The risk consequence is liable to occur.
30–50 % expectation in the next 12 months
 Unlikely The risk consequence may occur at some time.
5–30% expectation in the next 12 months
 Rare The risk consequence will only be realised in exceptional circumstances.

Table 5
Risk rating matrix

 Likelihood

 Consequence

 Insignificant  Minor  Moderate  Major  Severe
 Almost certain  Medium  High  High  Very high  Very high
 Likely  Low  Medium  High  High  Very high
 Possible  Low  Low  Medium  High  High
 Unlikely  Very low  Low  Low  Medium  High
 Rare  Very low  Very low  Low  Low  Medium

Planned outputs with accommodation data

Releases from the LBD include:

  • published research reports

Planned outputs using the integrated Accommodation Survey data include:

  • a public report on productivity in the tourism sector, including the analytical methodology
  • code and methodology for other researchers to reuse.

Privacy and confidentiality

The Statistics Act 1975 protects the confidentiality of all information provided to Stats NZ. This ensures that no other government department has the power to obtain information that could identify an individual and no research or statistical results can identify an individual’s details.

Information is retained to ensure that, as new or improved methodology is developed, and as new data is supplied on a regular basis, the infrastructure can be updated to contain the most up-to-date data and methodology.

The LBD retains names, addresses, and encrypted unique identifiers in a dedicated environment. This is separate from the research dataset, and has access limited to Stats NZ employees who are working on integrating data into the LBD.

Information available for research is de-identified and does not contain names or addresses. All unique identifiers are removed and new Stats NZ-assigned identifiers are created. All outputs are checked for confidentiality before they are approved for release or publication.

Access

All data is subject to Stats NZ’s protocols for access to data, to ensure security of the data. Access to the LBD is strictly controlled and will be:

  • used only for approved statistical or research purposes
  • restricted to only the datasets required for each purpose
  • managed in accordance with the Statistics Act 1975 and the Tax Administration Act 1994.

Data will be available for approved researchers through the Stats NZ microdata access service, including the Stats NZ Data Lab and the remote access to microdata service (where this meets security and confidentiality requirements – including those of the Tax Administration Act 1994, the Memorandum of Understanding between Stats NZ and Inland Revenue, and the Deeds of Variation and Memorandum of Understanding between Stats NZ and the Ministry of Social Development).

Any employee of Stats NZ (including any researchers contracted or on secondment to Stats NZ), and any researcher accessing the LBD through the Stats NZ microdata access service who has access to any information provided by Inland Revenue in the LBD, must sign a Declaration of Secrecy, as well as an IR820 certificate which certifies that they have been shown, have read and understood section 87 of the Tax Administration Act 1994. Stats NZ will keep these signed certificates as a permanent record.

Stats NZ works with agencies to support their use of the LBD for statistical and research purposes.

LBD stakeholders

Agencies with data held in the LBD:

  • Stats NZ
  • Inland Revenue
  • Ministry of Business, Innovation and Employment
  • Intellectual Property Office of New Zealand.

Other government stakeholders:

  • Office of the Privacy Commissioner
  • The Treasury
  • Productivity Commission.

Appendix: Variables and linking to the LBD

Variables used for linking

Businesses in the Accommodation Survey are selected from Stats NZ’s register of businesses with each business is given a unique identifier, which is only meaningful within the secure data systems of Stats NZ. This identifier is used to link the Accommodation Survey to the LBD.

Variables that are removed

Accommodation Survey variables that are removed from research database:

  • Business name

Other potentially identifying variables (eg address) are not Accommodation Survey data.

Citation

Stats NZ (2017). Adding Accommodation Survey data to the Longitudinal Business Database. Retrieved from www.stats.govt.nz.
ISBN 978-1-98-852800-7 (online)
Published 25 May 2017

  • 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+