The 12th meeting of the International Roundtable on Business Survey Frames Helsinki 28.9. - 2.10.1998 Session No. 1 Paper No. 11 Country: New Zealand Progress Report Ian Ewing Statistics New Zealand Integration: Integration of the Balance of Payments (BOP) Directory and Surveys with the Business Frame: 1. When the Business Frame (BF) was developed in December 1996, a separate frame called the Balance of Payments Directory existed for Balance of Payments Surveys. Statistics New Zealand's Balance of Payments surveys were one of the major set of business surveys that were not fully integrated with the Business Frame. Price Index surveys and Building Activity collections are the only remaining sets of surveys not integrated wit the frame. 2. The Balance of payments directory along with the Balance of Payments Surveys were on different servers and used different software. At the end of 1997, a project was established to merge the Balance of Payments directory with the Business Frame, and to implement a frame maintenance strategy based on revised questionnaire design, and extended use of tax data. This project was combined with an IT project that was underway to integrate Balance of Payments surveys into the general Business Surveys management system. 3. Prior to 1998, the BF questionnaires asked only a few general overseas transaction questions. These questions determined whether a business was involved in overseas transactions. If the business on the BF indicated for the first time it was involved in overseas transactions, it was added to the Balance of Payments Directory. A second questionnaire was then sent out to determine what type overseas transactions the business was involved in. The detailed overseas transaction information from the second questionnaire was then captured on the Balance of Payments Directory. 4. The Balance of Payments Directory and Balance of Payments Survey Management System (SMS) had no mechanism for detecting when a business began to engage in new types of Balance of Payments transactions. Although the BF surveyed business that were likely to be involved in overseas transactions annually, the overseas transaction questions were not detailed enough to detect new types of Balance of Payments transactions for businesses that were already involved in overseas transactions. 5. To enable the BF to act as a frame for Balance of Payments surveys, additional overseas transaction variables were added to the Business Frame, and to all the BF questionnaires. 6. These questionnaires now contain specific questions on overseas transactions. As it is no longer necessary to send a second questionnaire, this has reduced both respondent 2 load and post out costs. Additional overseas transaction information is also captured on the BF system. By capturing much more detailed Balance of Payments transactions annually, the Balance of Payments Survey Management System is now able to use the BF to detect when a business has begun new types of Balance of Payments transactions. 7. In addition, Statistics New Zealand has been using annual tax returns to identify businesses that have overseas transactions so that they can be added to the Business Frame and the Balance of Payments Surveys. 8. As the final stage of development the Balance of Payments Survey Management System has been redeveloped, so that it is part of the new Business Surveys System. The Business Surveys System is on the same server as the BF and uses the same operating language as the BF. All surveys that use the Business Surveys System use the Business Frame as their survey frame. The integration of BF data into the new Survey Processing Template (SProceT) 9. A new economic survey processing system has been developed in Lotus Notes software to run in a client server environment. It was developed as part of our overall upgrade of Information Technology designed to move all SNZ functions to a clientserver environment. The Quarterly Economic Survey of Manufacturing was the first survey to use SProceT. Previously, mainframe COBOL systems, SAS, Word and Excel were used to process, run and analyse data. Now all the data capture, editing, imputation and treatment of outliers can be done on the one system. One of the main features of this processing system is the link to the BF system. This integration of the two systems has allowed direct comparison between survey system values and BF values. The flexibility of the BF system and its integration with SProceT, has eliminated the need for costly and detailed 'hard copy' reconciliations of survey respondents and the frame and has streamlined the sample maintenance procedures. 10. The main advantages of SProceT are that it is integrated, fast and very easy to use. On the old mainframe, running estimates would take around two or three hours. Lotus Notes can process the same information in minutes. The application design team are currently working on the second survey to utilise SProceT - the November 1998 Quarterly Employment Survey. There is a two year plan for migrating most business surveys to the SproceT environment. Frame Coverage Issues: Ceased Business Investigation 11. Following a comparison of business details reported by householders in our Household Labour Force Survey with businesses on the Business Frame (BF), it was found that the BF contained a number of business that were “ceased“ on the BF but were still operating (either as employers or non-employing sole traders). 12. Most of these businesses were correctly ceased at the time, as they had either sold an existing business activity or had ceased their business activity. However, the existing frame maintenance procedures, which are based on surveying new tax (GST) registrations and annually surveying live business on the BF, are unable to detect 3 business restarts. For example, a motel owner may sell their motel and keep their GST registration active, as they intend to start a new business activity within 6 months. Because the GST registration is kept active and re-registration is unnecessary, if they start a new business activity, such as building a new motel, the new business activity may not be detected. 13. To find those businesses that were ceased on the BF but were still operating, all businesses that were ceased on the BF were checked to see if they were still registered for GST (a VAT tax) and were still paying the GST tax. 14. After allowing for timing differences, and business winding down, etc. it was found that the head-count under coverage of business on the BF was 4.6%. An investigation into the ‘character’ of these under-covered businesses has been completed, and a number of strategies to improve coverage in this area have been implemented. Respondent Load: Key Word Strategy 15. The Key Word strategy represents an effective use of the tax database enabling Statistics New Zealand to reduce both post out costs and Respondent Load. The strategy has reduced the post out size of the Business Frame's 'Birth' questionnaire by 25-30% monthly. The strategy works by matching a predetermined industry descriptor with the industry descriptor recorded on the tax database. If there is a perfect match, the new tax registration can be assigned an industry code, and the business is not sent a questionnaire. The Key Word strategy does not apply to those businesses that are employing or importing and exporting. It also excludes businesses in the agricultural, accommodation, manufacturing, retail and wholesale industries. It is anticipated that further Key Word initiatives will create even more respondent and cost savings for SNZ in the future. Other Business Frame Developments: Taxation Data: 16. When the Business Frame was developed, the Taxation Office Registration Records and the Goods And Service Tax (GST) information was able to be viewed from the Business Frame. At the end of 1997, additional data, in the form of the Annual Salaries and Wages tax deduction return and the Annual Accounts return (a profit and loss account form with selected balance sheet items) was added to the BF system. This has enabled survey staff to carry out more detailed analysis of BF units, to make use of this data directly in compiling quarterly compensation of employees component of GDP and to monitor changes in business size, offering potential to further reduce the level of annual surveying required for frame maintenance. Geocoder 17. Up until March 1998 the Business Frame (BF) was still using a mainframe (off-line) based location coding tool called autocac. As all our systems are expected to be removed from the mainframe by December 1998, the BF needed a location coding 4 system that is able to operate in the BF's client server environment. As the BF had already successfully used Search Software America (SSA) “Names“ for its name and address matching system, the SSA names was used again, as the bulk and on-line location coding tool. The new Geocoder: is able to assign a location code from a given address to a high degree of accuracy (76% match compared with the 64% with the previous system), in both on-line and bulk operations; is intelligent enough to return location codes on unformatted addresses and cope with minor misspellings; and screen displays all required information - road side indicator, address, meshblock (location code), area unit, narrative, urban area, regional council & post code. Business Frame International 18. SNZ has supplied its Business Frame system to the Irish Central Statistical office last year. The system is due to go live in December 1998. Further expressions of interest have been received from the South African and Fijian statistical offices. A project team will arrive in South Africa to assist Statistics South Africa undertake an evaluation of the SNZ system. The Fijian project will involve a scaled-down version of the system designed to run in the Fiji environment. Development of a Strategy to assign classifications to smaller businesses 19. A project is underway to assign industry and sector classifications to businesses that are not 'maintained' by the BF, but are held on the tax database. One of the aims of the ‘tax data coding project’ is a fully coded data set that is able to be consistently coded over time. 20. Extending use of taxation data requires consistent comprehensive classification. If appropriate classifications are able to be applied to the tax database it may be possible to: 21. Create Inter-(population) Census estimates of internal migration and population growth Investigate Small Business penetration rates Create Owner Profiles for Small Business Owners (AGE, SEX) Create Household Income Estimates, etc. The project is still in its early stages.