FCR(1999-2000)13

For discussion
on 21 May 1999

CAPITAL WORKS RESERVE FUND
HEAD 710 - COMPUTERISATION
Census and Statistics Department
New Subhead "Computer Equipment and Services for the 2001 Population Census"
HEAD 26 - CENSUS AND STATISTICS DEPARTMENT
Subhead 001 Salaries

Members are invited to approve -

1. a new commitment of $56,288,000 for acquiring the computer equipment and services for the 2001 Population Census; and

2. an increase in the ceiling placed on the total notional annual mid-point salary value of all non-directorate posts in the permanent establishment of the Census and Statistics Department in 1999-2000 from $483,744,000 by $5,156,040 to $488,900,040 for creating nine non-directorate posts for the implementation of the proposed project.

PROBLEM

The existing computer system developed for the 1996 Population By-census (96BC) in the Census and Statistics Department (C&SD) cannot provide the necessary support for the fieldwork operations and the data processing work of the 2001 Population Census (2001C).

PROPOSAL

2. The Commissioner for Census and Statistics, in consultation with the Director of Information Technology Services, proposes to re-develop the 96BC computer system to provide the necessary support required.

JUSTIFICATION

Computer support for the 2001C

3. In conducting the 2001C, we will adopt a Long Form questionnaire on detailed characteristics for a sample of one-in-seven of the entire population and a Short Form questionnaire on basic characteristics for the remaining of the population. The 2001C is a large-scale operation involving recruitment of about 23 000 temporary field workers to visit more than 2 million households and processing of the resulting completed questionnaires under a very tight time schedule. It is impossible to conduct the 2001C without efficient computer support under the various operational constraints and the pressing data dissemination requirements. For instance, the recruitment and training of all temporary field workers has to take place within a four-month period before the census enumeration in March 2001. The entire enumeration operation should also be concluded within a 13-day period to minimise inconvenience to the public. These constraints can only be overcome with efficient computer support in various activities such as preparation of fieldwork assignments, recruitment of temporary field workers, conduct and control of fieldwork, processing of questionnaires and dissemination of census results with a multi-media approach. The computer will also expedite the release of 2001C results in the earliest instance in accordance with the United Nation recommendation on the release of census data and to meet the increasing call for timely population statistics from the community at large.

Inadequacy in the 96BC computer system

4. The computer system developed and used for the 96BC was based on a single mode operation involving just one Long Form questionnaire. It was largely tailor-made to meet the requirements, in terms of functions and capacity, of the By-census only. The C&SD has recently conducted an in-house study to ascertain the computer requirement of the 2001C, in particular the feasibility of using the 96BC computer system to support the 2001C. The study concluded that the 96BC system could not be re-used to meet the requirement of the 2001C due to the following limitations and deficiencies -

1. it cannot support the two-mode operation under the 2001C involving both Long Form and Short Form questionnaires;

2. its design is geared towards the Long Form questionnaires being used in the 96BC and does not allow for the processing of additional data items in both Long Form and Short Form questionnaires;

3. its capacity cannot support the scale of the 2001C which is much larger than that of the 96BC¹;

4. many sub-systems in the 96BC were developed on the VAX computer platform, which is a proprietary system not conforming to the established Government System Architecture (GSA) standard introduced in 1991². Furthermore, non-conformity to the GSA also increases difficulty in recruiting and deploying staff with the skills tied to that proprietary platform;

5. it is not Year 2000 (Y2K) compliant;

6. to maximise the utilisation of the available resources, a significant number of the hardware items used in the 96BC have already been re-deployed for use in other applications after the 96BC. It is very difficult, if not impossible, to revert these items without disrupting the operation of the applications concerned; and

7. owing to the fast changing information technology, many hardware and software items used in the 96BC have become obsolete.

The proposed computer system and its benefits

5. The in-house study recommends that a new 2001C computer system should be established using a combination approach involving enhancement of one sub-system and re-development of 11 sub-systems of the 96BC, and development of one new sub-system. This combination approach ensures that the most cost-effective solution is adopted. These sub-systems and their major functions are set out in Enclosure 1. The estimated non-recurrent expenditure of the new system by individual sub-systems is set out in Enclosure 2.

6. The new system will have the following major benefits -

1. apart from being able to support the two-mode operation and capturing new data items in both Long Form and Short Form questionnaires, it will also provide the following additional functions to facilitate pre-fieldwork activities -
   • preparation and allocation of fieldwork assignments;

   • printing of voluminous householder letters and assignment lists; and

   • recruitment, selection, training, deployment and payment of temporary field workers;

2. it will have the necessary capacity to support the scale of the 2001C;

3. its platform will conform to the established GSA standard and thus facilitates re-deployment of computer equipment after the 2001C and future development of statistical work;

4. it will be fully Y2K compliant;

5. it will have built-in control mechanism to control and check the amount and quality of outputs of temporary field workers;

6. it will have functions for editing and validating the Long Form and Short Form questionnaires to take account of their different requirements, as well as a special function to facilitate amalgamation of data collected from the Long Form and Short Form questionnaires for compilation and statistical analysis;

7. it will allow changes to the questionnaire, which is currently not allowed in the 96BC system, to meet the latest needs of data users;

8. it will facilitate the use of digital maps to partly replace the existing paper maps, which are maintained manually to support the Register of Quarters. This will lead to permanent staff savings in the maintenance of maps; and

9. it will enhance the efficiency in various 2001C activities through more efficient system design and using faster computers. For example, with the new system, the time lag in releasing 2001C results will be shortened to better meet the increasing demanding expectation of users for timely statistics.

Future use of the 2001C computer system

7. To maximise the return of our investment in the 2001C system, C&SD has taken into account as far as possible the computing requirements of future population censuses / by-censuses and other similar household surveys in designing the capability and functions of the 2001C system. In this connection, several of the sub-systems as listed below will be retained for use after the 2001C operation -

1. Register of Quarters Sub-system (RQS), which facilitates continuous updating of a complete list of quarters and building details in Hong Kong, will be retained to serve as the sampling frame for future population censuses / by-censuses and household surveys;

2. Sampling Sub-system (SAMS) will be retained for sample selection for future household surveys;

3. Digital Mapping Sub-system (DMS) will be retained for handling the use of digital maps; and

4. Statistics Dissemination Sub-system (SDS) will be retained for the on-going dissemination of the 2001C results for various planning and analytical uses.

8. C&SD will re-deploy the equipment and facilities of other sub-systems for various useful purposes after the completion of the 2001C with a view to maximising the utilisation of the resources available.

Cost and benefit analysis

9. Given the large scale of the 2001C operation (involving the enumeration of the entire Hong Kong population) and its stringent time frames (completing the operation within a 13-day period and making available its basic results within six months), it is impossible to conduct the census without efficient computer support. While it is generally accepted from experiences around the world that an efficient computer system is essential for such a large statistical operation, it is often not possible to quantify precisely the productivity gains and savings arising directly from the computer system as the alternative of conducting a large census operation without an efficient computer system will either mean a physical non-starter or a project requiring astronomical number of manpower resources. The normal cost and benefit analysis with an estimated pay-back period is therefore not entirely applicable in this case. The proposed system will ensure that the 2001C is conducted smoothly and statistics of good quality are available for use by the large number of users in both the public and private sectors.

10. The proposed 2001C computer system has been designed to meet all the operational, timing and data quality requirements of the 2001C project that are essential for its success. It is designed using a combination approach as explained in paragraph 5 to ensure maximum cost-effectiveness. The 2001C computer system, through its enhanced RQS and use of digital maps made available from the DMS will improve efficiency in maintaining the Register of Quarters. This will give rise to an annual staff cost saving of $856,000 due to the deletion of one Census and Survey Officer and one Assistant Census and Survey Officer posts in the C&SD. In addition, the proposed computer system will achieve an annual notional saving of $234,000 in terms of fractional staff and accommodation cost reductions which we will redeploy to meet other service requirements.

FINANCIAL IMPLICATIONS

Non-recurrent cost

11. The Commissioner for Census and Statistics estimates that implementation of the proposal will incur a non-recurrent cost of $72,552,000 over a three-year period from 1999-2000 to 2001-2002. This comprises $56,288,000 for the acquisition of computer hardware and software and related services and $16,264,000 for the cost of additional in-house staff responsible for the planning and implementation of the project. Detailed breakdowns are as follows -

This item seeks Members' approval of the non-recurrent commitment of $56,288,000 for items (a) to (f) and an increase in the Department's notional annual mid-point salary (NAMS) ceiling for the additional in-house staff needed.

12. As regards paragraph 11(a), the cost is for the acquisition and installation of computer hardware, software and networking equipment. This includes four mid-range computers, 15 Local Area Network with some 300 personal computer workstations, some 30 notebooks and handheld computers and miscellaneous peripheral equipment for scanning, printing and data communication to support the work in the five user sections, the Central Processing Area (CPA) and the 53 field centres.

13. As regards paragraph 11(b), the cost is for the preparation of the CPA and five user sections supporting different decentralised processing functions. A note on the organisation and functions of the CPA and the various user sections concerned is given in Enclosure 3.

14. As regards paragraph 11(c), the cost is for the installation of personal computers in field centres, data capturing services and implementation services for the DMS and the SDS.

15. As regards paragraph 11(d), the cost is for hiring contract staff services to supplement the Analyst/Programmer grade staff as well as the Statistical Officer grade staff in implementing the proposed system.

16. As regards paragraph 11(e), the cost is for training of the project team members and users, acquisition of consumables and backup tapes, as well as the connection charges and rental of communication lines for the implementation.

17. As regards paragraph 11(f), the cost represents a 5% contingency on the cost items set out in paragraphs 11(a) to (e).

18. As regards paragraph 11(g), the cost of $16,264,000 is for non-recurrent staff cost of 24 man-months of Senior Systems Manager, 56 man-months of Systems Manager, 116 man-months of Analyst/Programmer I and 29 man-months of Assistant Clerical Officer to provide support in project planning and monitoring, liaison with users, other Government departments and vendors, system design, procurement, implementation, data conversion and user training. To enable the creation of nine non-directorate posts required in 1999-2000, it is necessary to increase the ceiling of the NAMS value of C&SD from $483,744,000 by $5,156,040 to $488,900,040. All these posts will be filled by existing civil service staff. Details are set out in Enclosure 4.

Recurrent cost

19. It is estimated that the recurrent cost arising from the maintenance of the proposed system is $1,369,000 for 2000-2001, $2,739,000 for 2001-2002, and $2,519,000 for 2002-2003 and onwards. Detailed breakdowns are as follows -

20. As regards paragraph 19(a), the expenditure is for annual maintenance of the hardware and software items acquired starting 1999-2000.

21. As regards paragraph 19(b), the expenditure is for the acquisition of consumables and backup tapes, as well as the annual rental communication lines. Both paragraphs 19(a) and (b) are for various sub-systems including RQS, DMS, SAMS and SDS, which will continue to operate after the 2001C operation.

22. As regards paragraph 19(c), the cost is for the annual additional staff cost of one Analyst/Programmer I incurred from 2002-2003 and onwards to perform the on-going system support. This requirement will be met by internal redeployment in C&SD.

23. Subject to Members' approval of this item, we will include the provision for the purpose of paragraph 19(a) and (b) in future Estimates.

Implementation Plan

24. C&SD plans to implement the proposal according to the following schedule -

BACKGROUND INFORMATION

25. It is an established practice in Hong Kong to conduct a population census every ten years and a by-census in between two censuses for the purpose of obtaining up-to-date benchmark information on the social, economic and demographic characteristics of the population. Previous population censuses were conducted in 1961, 1971, 1981 and 1991 and by-censuses in 1966, 1976, 1986 and 1996. In April 1994, the Finance Committee approved the project for the installation of the 96BC computer system.

Financial Services Bureau
May 1999

Enclosure 1 to FCR(1999-2000)13

Enclosure 2 to FCR(1999-2000)13

Enclosure 3 to FCR(1999-2000)13

A 2001 Population Census (2001C) Office has been set up in April 1999 under the Social Statistics Division of the Census and Statistics Depatment to co-ordinate and monitor the entire 2001C project. In additional to the normal office space, 2001C Office will have a Central Processing Area (CPA), which is a centralized area for the accommodation of most of the equipment to support the operation of 2001C, and acts as a store for the completed 2001C questionnaires after the enumeration. The organization structure of the 2001C Office will comprise the following sections -

Enclosure 4 to FCR(1999-2000)13

Type of post		1999-2000		2000-2001		2001-2002
	NAMS value(1) $	Number of posts	NAMS value $	Number of posts	NAMS value $	Number of posts	NAMS value $
	(a)	(b)	(c) = (a) x (b)	(d)	(e) = (a) x (d)	(f)	(g) = (a) x (f)
Senior Systems Manager	985,260	1	985,260	1	985,260	1	985,260
Systems Manager	722,280	2	1,444,560	2	1,444,560	2	1,444,560
Analyst/ Programmer I	508,860	5	2,544,300	5	2,544,300	4	2,035,440
Assistant Clerical Officer	181,920	1	181,920	1	181,920	1	181,920
Total		9	5,156,040	9	5,156,040	8	4,647,180

(1) Based on the NAMS value as at 1 April 1998.

(2) Based on the Staff Cost Ready Reckoner No. 98/1.

1. The number of questionnaires to be processed in 2001C is estimated to be 2.8 million as compared to 0.4 million in the 96BC and the number of temporary field workers required will surge from 7 800 to 23 000. The capacity of the 96BC computer system will not be able to handle the 2001C workload.

2. The VAX platform was chosen for the 96BC to minimise the development cost of the 96BC computer system by rationalising relevant sub-systems previously developed for the 1991 Population Census. That approach was possible because the processing requirements of a by-census are far less complicated than those of a full scale census. Further development on this VAX platform now will defeat the purpose of having a GSA standard which would bring in benefits such as data and function sharing, improved efficiency in communication and devising mutually supporting disaster recovery plans.

3. Tendering for procurement of the required non-bulk equipment will be completed by February 2000, and tendering for acquiring vendor service to automate the data entry of questionnaires will be completed by May 2000.

4. System analysis and design / implementation for each related sub-system will be completed by phase. Coverage of sub-systems in the different phases is set out in Enclosure 1.

5. Post-implementation monitoring and support for each related sub-system will be completed by phase from February 2001 to October 2001 to cope with the scheduled production date. For the sub-systems of RQS, SAMS, DMS and SDS, on-going maintenance and support is required.