Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 31

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 30
APPENDIX A FTA Capital Cost Database The starting percentages and numerical adjustments established in this Guidebook were developed from univariate and multivariate regression analyses based and calibrated on detailed cost and project data for 59 past transit capital projects. The 59 projects in this database represent a wide range of rail projects constructed in the United States over the past four decades, with detailed costs roughly conforming to the SCC structure developed from FTA Capital Cost Databases. The projects: Comprise 29 light rail and 30 heavy rail projects; Have construction dates ranging from 1974 to 2008; Have capital costs ranging from around $50 million to $2 billion in the year of construction, equivalent to a range of $90 million to over $5 billion in constant 2008 dollars; and Are new rail lines, extensions of existing networks, and rehabilitation projects. Because this dataset contains soft costs for a broad distribution of projects, it provides a rea- sonable statistical basis for the estimation of future rail projects based on the analysis of actual, as-built soft costs for completed projects. Analytical Approach The analytical process applied to examine these past projects to develop a new soft cost esti- mation methodology is briefly summarized below. For a more detailed description, please see the Final Report in Part 2, which follows this Guidebook. First, the projects were plotted on a frequency distribution of soft costs as a percentage of construction costs, resulting in several projects being rejected as outliers due to extraordinarily high costs or other circumstances. Please refer to the Final Report for further details. Second, a set of characteristics was gathered for the projects, including the following: 1. Physical attributes, such as alignment length, profile (below grade, at grade, aerial, etc.), number of stations, or whether the project initiated new service or extended an existing line; 2. Installation conditions, such as whether the project interacted with other active rail transit lines; 3. Schedule information, including major milestones in the project lifecycle; 4. Characteristics of the project sponsor, such as experience level, internal policies on capital costs, and use of outside contractors; and 5. The context of the project development process, such as the level of public involvement, delivery method, or whether a significant redesign was necessary. For these last two characteristics, the definition and determination of values required some judgment based on knowledge of the project development process. 32

OCR for page 30
FTA Capital Cost Database 33 Exhibit 27. Multivariate regression results on soft costs as a percentage of hard costs. Third, many additional measures were derived from this primary dataset that were intended to capture other project characteristics, such as project magnitude (e.g., construction costs per linear foot), complexity (e.g., percent of alignment below grade), unique circumstances (e.g., real estate acquisition costs, project occurred prior to certain federal requirements), and many others. Fourth, this research analyzed each indicator's statistical ability to predict the project's actual soft costs, in total and as individual components. After several hundred univariate and multi- variate regressions, a single multivariate regression was developed that can explain approximately 60% of the change in soft cost percentages by variations in the projects' characteristics (R2 = 0.58). Exhibit 27 shows the resulting coefficients from this regression, where the dependent variable is total soft costs as a percentage of construction costs. Using the projects contained in this FTA Capital Cost Database, the strongest correlation that could be produced is the regression described above. After testing many combinations of explana- tory independent variables, these nine could best predict the relationship between soft and hard costs. Although the strength of this correlation is not ideal, the relationship does highlight the importance of judgment in cost estimation. In addition, as more projects are included in this cost database, it may be possible to perform analysis with stronger cost relationships. Fifth, alternative multivariate regressions were examined that used different actual soft cost components (rather than total soft costs) as the dependent variable. The coefficient from the overall soft cost analysis was distributed to the soft cost components that correlated to the project characteristics in a statistically significant way. For example, alignment length showed an overall coefficient of around 1.4% per 10,000 linear feet regressed against overall soft costs, and this relationship was strongest when regressed against project management and other soft costs, so this Guidebook recommends adjusting the percentage estimate for those two components to a total of 1.4% per 10,000 linear feet. Finally, the starting points and recommended percentage adjustments were validated against the original projects to gauge how far off this Guidebook's new methodology would have been. Some minor adjustments to the coefficients were made to minimize the sum of each component's root mean square error (defined in the Glossary in Appendix C) for all projects.