Desviaciones de costos

...

b In van Wee (2007).

c In Odeck (2004).

d Projects include: Road and bridge construction and rehabilitation projects; maintenance projects, with road maintenance and resurfacing contracts; Traffic and traffic maintenance contracts.

e In Siemiatycki (2009).

the variation in sample size. If the sample size is small, outliers may have a large influence on the results. Fourthly, the differ- ences can be explained by the differences in the geographical area that is covered (different economies), and the project types that are included (different project dynamics and complexity).

Table 1 shows furthermore that the studies by Nijkamp and Ubbels (1999), Odeck (2004), Bordat et al. (2004); Ellis et al. (2007) found rather small cost overruns—up to 20%—whereas Morris

(1990) and the Auditor General of Sweden (in Odeck) found enormous cost overruns of 164% and 86% respectively. Due to the probability that there are large differences in cost overruns among countries and across time, cost overruns in country specific studies might be quite different (either lower or higher) from the average cost overruns in the worldwide study. However, the abovemen- tioned findings do not support this expectation; some country specific studies have smaller average cost overruns and others higher average cost overruns compared to the worldwide study by Flyvbjerg et al. (2003a). Ideally country specific studies should have the same methodology as this worldwide study in order to compare its results. Without a general tendency in country specific studies towards lower or higher average cost overruns than in the world- wide study, it remains difficult to make inferences about individual countries based on the worldwide database. This therefore supports the need for further research into country specific cost performance of transport infrastructure projects.

The objective of this research is to determine the character- istics of cost overruns in Dutch large-scale transport infrastruc- ture projects. This concerns the frequency and the magnitude of cost overruns and whether cost estimates have improved over time. This research was financed by the Dutch Ministry of Infrastructure and the Environment and the Netherlands was therefore chosen as the country under scrutiny.

We apply the same methodology as used for the worldwide research regarding project types, the way data and prices are used. A database of 78 Dutch large-scale transport infrastructure projects was created and statistical analyses were used to determine the frequency and magnitude of cost overruns in the Netherlands and to examine whether cost estimates have become more accurate over time.

In the literature on cost overruns (see table 1) hardly any attention is given to the project phases (with the exception of Odeck (2004)). Until now literature has focused on identifying

---------------------------------------------------------------

cost overruns but the moment when projects are most vulnerable to cost increases has not been studied. This is however of the utmost importance because it could improve our understanding of cost overruns considerably. Whether cost increases are incre- mental over different project phases or extreme in certain project phases will help to distinguish between different explanations of cost overruns. It is essential to look more closely when cost increases occur to actually be able to deal with them. A second objective of this research is therefore to investigate whether projects are more vulnerable to cost increases during different project phases and if so which phase this concerns. A distinction was made between 2 phases (data did not allow to distinguish between more phases): 1. the pre-construction phase (the period between the formal decision to build and the start of construc- tion) and 2. the construction phase (the period between the start of construction and the start of operation (opening)). By addressing project phases in relation to project performance this paper fills a gap in literature on cost overruns and project management.

The structure of the paper is as follows. Section 2 describes the project selection, data collection, representativeness of the data- base, and the main methodological issues. Section 3 presents the cost performance in the Netherlands, focussing on the character- istics of cost overruns. In section 4 a comparison is made between the cost overruns in the pre-construction phase and in the construction phase. Finally, section 5 discusses the main conclu- sions and section 6 presents several areas for further research.

- Project selection, data collection and methodology

- Definition large-scale projects

Large-scale projects are often defined as major infrastructure projects that cost more than US$1 billion (Flyvbjerg et al., 2003a). However, past studies have often included a wider range of projects, both smaller sized projects costing several million dollars and large-scale projects (see e.g. Flyvbjerg et al., 2003b, where the smallest project cost US$ 1.5 million, and Odeck, 2004, which included projects costing less than 15 million NOK ~ US$

12.3 million). In addition to the size of the project in terms of costs, large-scale projects attract a high level of public attention or political interest because of substantial direct and indirect

---------------------------------------------------------------

C.C. Cantarelli et al. / Transport Policy 22 (2012) 49–56 51

impacts on the community, environment, and budgets (FHWA in Capka, 2004). Therefore, the definition of a large-scale project can also depend on the context, that is, the size of the project in relation to the size of the city (or country). Based on project size, their impact and context, projects that cost more than about h 20 million are considered large-scale projects in the Netherlands1. Regarding transport infrastructures, we adopt the definition of van Wee (2007): ‘‘Transport infrastructures include roads, rail lines, channels, (extensions of) airports and harbours, bridges and tunnels. Of these projects the ‘hardware’ is considered, excluding ‘software’; projects that are not related to the construction of infrastructure but are related to policies of deregulations, liberal- isation, privatisation, and so forth’’. In line with previous studies, the project types that are included in this research are road, rail, tunnel and bridge