Jump to the beginning of content

LCQ11: Using Reference Class Forecasting method to estimate costs of infrastructure projects

Following is a question by the Dr Hon Yiu Chung-yim and a written reply by the Secretary for Development, Mr Eric Ma, in the Legislative Council today (April 12):


As pointed out in an article published in a civil engineering academic journal in November last year, the Development Bureau had conducted, in collaboration with some academics, a feasibility study (with an agreement number of CE11/2012(CE)) in 2012 on the use of Reference Class Forecasting (RCF) method to estimate the costs of major road works in Hong Kong. According to the relevant information, RCF helps enhance the accuracy of the cost estimation of works projects, thereby reducing the risk of cost overruns. In this connection, will the Government inform this Council:

(1) whether the Development Bureau has made public the report of the aforesaid feasibility study; if so, of the time when and the channel through which the report was made public; if not, the reasons for that;

(2) given that a number of infrastructure projects have experienced significant cost overruns at present, whether the authorities used the RCF method in estimating the costs of such projects; if so, of the details (including the relevant data); if not, the reasons for that; and

(3) whether the authorities have any plan to use the RCF method to estimate the costs of infrastructure projects in future; if so, of the implementation timetable; if not, the reasons for that?



The Government has been implementing public works projects in an appropriate and orderly manner with a view to improving people's quality of living, enhancing the long-term competitiveness and promoting the economic development of Hong Kong.

Notwithstanding that there have been instances of cost overruns and delays in the delivering of certain mega projects in recent years due to unforeseeable circumstances that arose in the course of project implementation, we have maintained consistently good performance in cost estimation for projects under the Capital Works Programme as a whole. In retrospect, the Finance Committee (FC) approved a total of about 650 Category A works projects with a total provision of $770 billion in the past 10 years. Among them, about 70 projects required application to the FC for additional funding, which totalled around $60 billion. In other words, additional funding was required in approximately 10 per cent of the projects and the amount represented some eight per cent of the total provision.

In addition, although there were projects that required additional funding owing to individual circumstances, we generally managed to complete the projects under the Capital Works Programme within the original Approved Project Estimates (APE) overall and even with surplus. For example, about 850 Category A projects had the final accounts settled in the past 10 years. Their original approved estimates totalled about $240 billion as compared with the total final expenditure of about $210 billion. Though some projects needed to apply for additional provisions from the FC, the surplus from other projects were not only able to offset the cost overruns but also managed to leave behind a balance of $30 billion. In short, the total expenditures of these projects at final settlement accounted for only about 85 per cent of their original APE.

One of the tasks of the Project Cost Management Office (PCMO) established under the Development Bureau in June last year is to enhance the project management standard for public works projects, including the performance in project cost estimation. The PCMO is working on ways to enhance the methodologies for formulation of project cost estimation.

My reply to the three parts of the Dr Hon Yiu's question is as follows:

(1) To enhance the performance in cost estimation for public works projects, the Development Bureau commissioned a world renowned scholar, Professor Bent Flyvbjerg of the University of Oxford in 2012, to study the feasibility of applying Reference Class Forecasting (RCF), which has been adopted in the United Kingdom, the United States, Denmark, etc., in major roadworks projects in Hong Kong and to establish reference classes for them. The study also made benchmark comparisons between the performance in cost estimation for major roadworks projects completed over the past 20 years or so in Hong Kong and those of around 1 000 similar projects in over 30 countries. The findings indicated that the cost overruns in other countries averaged about 20 per cent of the initial project estimates as compared with about 11 per cent for Hong Kong. The total actual expenditure for the projects in Hong Kong were also 1 per cent less than the total approved project estimates. In consideration of the research findings, the project cost estimation method has performed well for the major roadworks projects in Hong Kong overall. Professor Flyvbjerg also prepared a report on the proposals for optimising budget management for the Capital Works Programme. We have planned to report on the work progress of the PCMO, including the progress of the feasibility study on RCF, to the Panel on Development within the first half of this year. The above-mentioned report will also be submitted to the Panel for Members' reference before the meeting.

(2) and (3) As mentioned above, Professor Flyvbjerg has successfully established reference classes for major roadworks projects in Hong Kong. Other works departments, including the Architectural Services Department, the Drainage Services Department and the Water Supplies Department, are also conducting feasibility studies on the application of the RCF method. Upon completion of all these studies, we will consider ways to apply this forecasting method to the projects under the Capital Works Programme, and review the current estimation methods with a view to ensuring compatibility and complementarity, as well as to achieving better cost estimation for works projects with due regard to the actual situation of Hong Kong. Our present target is to complete the work in the first half of next year.

Ends/Wednesday, April 12, 2017
Issued at HKT 11:30