ArticlePDF Available

Abstract and Figures

Cost overrun in rail transit projects in the United States is a pervasive and constant phenomenon, yet there has been a lack of rigorous statistical analysis on project costs. This has been partly due to the difficulty of obtaining reliable cost data and the small size of the sample. In addition, the Federal Transit Administration (FTA) has required formal risk assessment on transit projects since 2003. Whether this strategy has improved the accuracy of cost estimates is not known. To fill these gaps, this study has collected the cost data of 81 US rail transit projects built during the last 40 years, the largest sample of its kind. It is found that the accuracy of cost estimates at the decision-making phase has improved over time. Also, major projects have higher percentage cost overruns than smaller projects. Further, the accuracy of cost estimates decrease as project implementation durations increase. The utilized statistical methods yielded no evidence to suggest that cost overruns differ between projects designed before and after 2003, when the FTA risk assessment program started. However, since project size and implementation duration have increased over time, the positive impact of the FTA risk assessment program can be inferred and acknowledged to some extent. This research can serve as a reference point for studies on the accuracy of conceptual cost estimates and help in budgeting future projects. It can also help in understanding the causes of cost overruns in capital projects.
Content may be subject to copyright.
Case Study
Cost Overruns and Formal Risk Assessment
Program in US Rail Transit Projects
Nan Gao, S.M.ASCE1; and Ali Touran, Ph.D., P.E., F.ASCE2
Abstract: Cost overrun in rail transit projects in the United States is a pervasive and constant phenomenon, yet there has been a lack of
rigorous statistical analysis on project costs. This has been partly due to the difficulty of obtaining reliable cost data and the small size of the
sample. In addition, the Federal Transit Administration (FTA) has required formal risk assessment on transit projects since 2003. Whether this
strategy has improved the accuracy of cost estimates is not known. To fill these gaps, this study has collected the cost data of 81 US rail transit
projects built during the last 40 years, the largest sample of its kind. It is found that the accuracy of cost estimates at the decision-making phase
has improved over time. Also, major projects have higher percentage cost overruns than smaller projects. Further, the accuracy of cost
estimates decrease as project implementation durations increase. The utilized statistical methods yielded no evidence to suggest that cost
overruns differ between projects designed before and after 2003, when the FTA risk assessment program started. However, since project
size and implementation duration have increased over time, the positive impact of the FTA risk assessment program can be inferred and
acknowledged to some extent. This research can serve as a reference point for studies on the accuracy of conceptual cost estimates and help in
budgeting future projects. It can also help in understanding the causes of cost overruns in capital projects. DOI: 10.1061/(ASCE)CO.1943-
7862.0001827.© 2020 American Society of Civil Engineers.
Author keywords: Cost overruns; Federal Transit Administration (FTA) risk assessment program; Rail transit projects; Project size; Project
implementation duration.
Introduction
Cost overrun is a pervasive and constant phenomenon in the trans-
portation construction industry (Cantarelli et al. 2010). Previous
research has provided statistical evidence on this phenomenon
(Flyvbjerg et al. 2003b;Love et al. 2014;Odeck 2017). Internation-
ally, among different kinds of transportation infrastructure projects,
rail projects are most prone to cost overruns (Flyvbjerg et al. 2004;
Huo et al. 2018;Andri´c et al. 2019). According to Flyvbjerg et al.
(2002), capital costs in US rail transit projects had an average cost
overrun of 40.8%; while the average in Europe was 34.2%. Pickrell
(1989), in the first comprehensive evaluation of transit project
capital cost trends, reported that the average cost overrun was 52%
using data from 10 US urban rail projects completed before 1989.
Dantata et al. (2006) conducted a comparative study of actual and
estimated costs for 16 federally funded rail projects completed
between 1990 and 2004 in the US to examine whether the cost es-
timation accuracy had improved, concluding that cost overruns had
become smaller compared to what Pickrell (1989) found, though
without statistically significant proof due to the small sample size.
A more recent effort in investigating the trends of cost overrun in
the US was published by the Federal Transit Administration (FTA)
in 2008. This report (FTA 2008) stated that the average cost overrun
for 21 projects completed between 2003 and 2007 was 40.2%.
However, since no statistical test has been conducted on the differ-
ence of these average cost overruns from various studies, the ques-
tion whether cost estimates became more accurate is not explicitly
answered.
In addition, cost overruns are generally studied along with major
project characteristics. Questions like whether the cost overrun is a
phenomenon specific to certain types of projects or a prevailing
issue among projects with different features are also of high interest
in current literature (Flyvbjerg et al. 2004;Cantarelli et al. 2012).
In order to investigate cost overruns in different contexts, the au-
thors also studied project size and project implementation duration.
Relations between cost overruns and these two major project
characteristics were analyzed.
The substantial and pervasive cost underestimation reported in
extant studies and, for that matter various media, caused public con-
cern. This phenomenon is by no means limited to the US. Flyvbjerg
et al. (2003b,2004) have discussed the issue of cost performance in
transportation projects across the globe and their work has received
broad attention. The worldwide credibility concerns raised by
underestimation have driven several institutions to proactively look
for approaches to cope with cost underestimation and closely mon-
itor trends in cost overruns of infrastructure projects. One example
of trying to address the budget underestimation is the enforcement
of reference class forecasting in the United Kingdom (UK) (British
Department for Transport 2004). According to this directive,
mandatory percentage cost uplift should be added to all major UK
transport infrastructure projects. The cost underestimation of these
projects is claimed to be abated to some extent (Flyvbjerg et al.
2016). Another example is the FTAs requirement of formal risk
assessment in 2003 [FTA PG-22 (FTA 2003b); FTA PG-40
(FTA 2007)]. Since 2003, all New Start transit projects requesting
funding from the FTA have to go through a formal probabilistic risk
assessment. The objective of this requirement is to minimize the
inherent bias in estimates of project cost and ensure that the grant-
eescost estimates for the budget requests are reasonable, realistic,
1Graduate Research Assistant, Dept. of Civil and Environmental
Engineering, Northeastern Univ., 400 SN, Boston, MA 02115. Email:
gao.n@husky.neu.edu
2Professor, Dept. of Civil and Environmental Engineering, Northeastern
Univ., 400 SN, Boston, MA 02115 (corresponding author). Email:
a.touran@northeastern.edu
Note. This manuscript was submitted on May 16, 2019; approved on
November 1, 2019; published online on March 12, 2020. Discussion period
open until August 12, 2020; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Construction En-
gineering and Management, © ASCE, ISSN 0733-9364.
© ASCE 05020004-1 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
and achievable. Arguably, the FTAs activity should have been able
to mitigate cost overrun risks and lead to better cost estimation
for projects started since 2003, when this requirement was imple-
mented. To date, no study has been undertaken for a comparative
analysis between projects that started before the FTAs risk assess-
ment requirement and projects starting after. So the efficacy of the
FTA risk assessment program has not been verified.
All in all, to fill aforementioned gaps and update understanding
on the state of investments in capital transit projects, this study col-
lected transit case data from relevant sources including the FTA and
published literature. The objective and corresponding contributions
of the paper are twofold:
1. Study the issue of cost overruns in US capital transit projects in
the last four decades and investigate the effect of project size and
duration on cost overruns; and
2. Evaluate the FTAs risk assessment requirement that started in
2003 vis-à-vis providing more realistic project cost estimates in
capital transit projects in the US.
The research also has several practical implications. The issue of
transit project cost overruns has been a major concern for both
policy makers and the public with regard to consideration of new
projects. The overall positive trend found by this study could pro-
vide some optimism for establishing reasonable budgets for future
transit projects. Furthermore, the before and after analysis method-
ology developed and used in this paper to evaluate the efficacy of
risk assessment is not limited to US transit projects but can be used
in examining other policies and approaches by agencies and project
owners.
Causes of Cost Overrun and Methods of Mitigation
Many scholars have endeavored to uncover the root causes of cost
overruns in transportation projects (Pickrell 1992;Flyvbjerg et al.
2004;Cantarelli et al. 2010;Love et al 2014). Different approaches
have been used to categorize these causes. Flyvbjerg et al. (2003a)
and Cantarelli et al. (2010) developed a framework including four
classes: technical, economic, psychological, and political. Techni-
cal causes are considered honest errors deriving from a persons
bounded rationality, for example, inadequate project planning,
scope changes, unpredicted price rises, and incomplete estimates.
Psychological causes are also considered honest errors since they
involve cognitive bias, such as optimism bias. Both economic and
political causes fall within the overall category characterized by
deliberate cost underestimation and forecast manipulation. Love
et al. (2015) argued that both outside and inside perspectives are
required to understand the cost overrun phenomenon. The former
emphasizes the optimism bias and strategic misrepresentation while
the latter recognizes the underlying complexity and dynamics
embedded in the project development process. Siemiatycki (2009)
compared the academic and independent government auditorsex-
planations regarding the occurrence of cost overruns and concluded
that auditors have mainly emphasized technical explanations
while academic literature cites the political and psychological
explanations.
A few studies looked into the causes for cost overruns in US rail
transit projects specifically. For example, Pickrell (1992) pointed
out that the cost overruns were attributed to factors including fore-
casting techniques, the structure and appraisal process of transit
grant programs, and the cost control skills. Booz Allen Hamilton
report (2005) developed a methodology for tracking down cost
variation in each project phase and attributed the variation to three
major cost drivers: (1) inflation adjustment, (2) scope change,
and (3) schedule change. Before and after studies published by the
FTA (2019b) are another source of information on the causes
of project cost overrun and schedule delay. According to these re-
ports, the main reasons for cost overruns are (1) unanticipated in-
flation, (2) lengthy project implementation, (3) scope changes, and
(4) incomplete estimation.
Methods to cope with cost and schedule overrun in transit
projects have been proposed by various authors and institutions;
some of the better-known methods, such as the use of reference
class forecasting and probabilistic risk analysis, were previously
discussed in this paper. Flyvbjerg et al. (2002) and Siemiatycki
(2009) have suggested mitigation strategies such as the use of per-
formance specifications, enhanced regulations and monitoring of
projects, and involvement of external parties. Love et al. (2015)
proposed the use of Alternative contracting methods (ACMs), such
as integrated project delivery (IPD), and emerging technologies
such as building information modeling (BIM). IPD provides a more
desirable delivery environment for BIM-enabled projects, while
BIM can support collaboration among parties and locations by
accommodating all information into virtual models (Azhar 2011).
In addition, BIM could lessen change orders resulting from poor
design documentation and improve constructability, thus reducing
cost variation during design and construction (Eastman et al. 2011).
It should be noted that currently IPD is not yet employed in US
rail projects but ACMs such as design-build (DB), construction
management at risk (CMAR), or publicprivate partnerships (PPP)
have been successfully used in delivering transit projects (Touran
et al 2009;GAO 2019). There are also studies on comparing the
effectiveness of the different methods used to cope with cost over-
runs. For example, Bakhshi and Touran (2009) compared the FTAs
risk assessment method with the reference class forecasting method
used by the UK, based on a sample of rail projects completed be-
fore 2005. More research with updated data is needed in this area to
come to a tenable conclusion.
It should be noted that the literature on the causes of cost over-
runs are based on interviews, surveys, reviews of specific project
history, and empirical observations. The data required for investi-
gating the statistical relationship between some of the root causes
and cost overruns are beyond those that can be captured by total
estimated and actual cost figures (Siemiatycki 2009;Love et al.
2015). Since the main focus of the current paper is to investigate
the magnitude of cost overruns in US rail transit projects statisti-
cally, the authors only studied two of the project characteristics
against cost overrun based on the information availability. In addi-
tion, the effectiveness of the risk assessment method used by FTA
to cope with cost overruns has also been evaluated.
Definition of Cost Overruns
Cost overruns are usually defined as the percentage difference be-
tween actual costs and cost estimates [Eq. (1)]. To specify the base-
line estimates used for comparison, the corresponding milestones
of each cost estimate should be clarified first
cost overrun¼actual costcost estimate
cost estimate ×100% ð1Þ
According to the FTA guidelines (FTA 2016), the New Starts
project development process can be summarized as in Fig. 1. There
are three categories of projects under the Capital Investment Grant
(CIG) program: New Starts, Small Starts, and Core Capacity.
According to the FTA Capital Investment Grant Fact Sheet
(FTA 2019c), New Starts projects are new, fixed guideway projects
with a total estimated capital cost larger than $300 million or they
are seeking more than $100 million in CIG funds. Most rail transit
© ASCE 05020004-2 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
projects are in the category of New Starts, while Small Starts in-
cludes most bus rapid transit projects and Core Capacity projects
are corridor-based investment projects.
Fig. 1shows the steps in the approval process and construction
of a new transit project according to the FTA process. The FTA
evaluates and rates the project at multiple project milestones as
highly recommended, recommended, or not recommended. These
project milestones include approval for entry into project develop-
ment, engineering, and immediately prior to an award of a full
funding grant agreement (FFGA). Approved projects usually get
a substantial amount of funding (around 50% of total capital costs)
from the FTA. One of the evaluation and rating criteria is the
cost effectiveness measure, which is computed as the annualized
capital cost plus annual O&M cost of the project divided by the
annual number of forecasted trips on the project(FTA 2016,
p. 18). Based on Fig. 1and Duff et al. (2010), important cost es-
timation milestones for a New Starts project include alternative
analysis (AA)/draft environmental impact statement (DEIS), pre-
liminary engineering (PE)/final environment impact statement
(FEIS), and FFGA. The FFGA is a unique contractual obligation
employed by the FTAwhen investing a significant amount of New
Starts funding into a locally developed fixed guideway transit
project (FTA 2002). In other words, it establishes terms and con-
ditions of federal funding.
With the multiple cost estimates in different project milestones
in mind, the choice of a specific cost estimate as the baseline will
depend on the research purpose (Flyvbjerg et al. 2018). In order to
answer the question of whether cost estimates have become more
accurate over time and investigate how formal risk analysis impacts
the accuracy of cost estimates, two cost estimates are of high in-
terest in this study. First, to evaluate the reliability of cost forecasts
for decision-making, cost estimates at the AA/DEIS phase were
selected as the baseline for comparison. During this phase, several
alternative solutions are evaluated and a locally preferred alterna-
tive (LPA) is selected. For those projects that are not selected,
project sponsors would no longer conduct any further planning
work. Hence, the most critical decision to build a rail project is
made at the AA/DEIS phase and the corresponding cost estimate
is used to decide whether a project will be implemented (Pickrell
1989;Dantata et al. 2006).
The cost estimate established for the FFGA is also selected in
this study. The FFGA represents the final decision of FTAs funding
for a project. Cost estimate in the FFGA serves as the basis for the
FTA to allocate funding for the project. Most FTA funded projects
use the design-bid-build method. For these projects, effective risk
analysis is generally carried out in the final design phase consid-
ering the availability of project information. To evaluate the impact
of risk analysis on the accuracy of cost estimation, the cost estimate
in the FFGA is used as baseline for comparison.
Apart from the timing of cost estimation, it is also important to
clarify the impact of inflation by specifying the year of the estimate.
Research shows that inflation is one of the major influential factors
causing differences between estimated and actual costs (Booz
Allen Hamilton 2005). Some previous research has escalated both
forecast and actual costs to a specific years price to account for
the effect of inflation. In this study, when the projectspercentage
cost overruns were compared, there was no need to adjust estimates
and observed actual costs to the same year. The reasons are three-
fold. First, cost estimates are expressed in terms of the year-of-
expenditure (YOE) dollars, which already accounts for inflation
rates and expenditure schedules in estimating costs. Second, when
adjusting YOE prices and observed costs to a specific year, original
inflation indices and expenditure schedules used for each project
would be required, and such information is generally not available.
Third, since cost indices vary substantially across different geo-
graphic locations, sectors, and periods, calculating and comparing
percentage overruns based on unadjusted dollars is more realistic
and accurate. However, when absolute cost values are compared
across different projects and periods, the inflation-adjusted costs
need to be used even though they are approximate rather than exact.
Therefore, when examining cost overruns in projects of different
size, project actual costs were discounted to 1984 levels using
the ENR (2019) Building Cost Index.
Data Collection
To collect capital cost data of rail transit projects, a thorough search
was conducted in all relevant sources including websites, engineer-
ing spot reports, and published literature. Cost estimates and the
estimation date data are extracted and compiled into the same for-
mat in order to serve the specific research objective of this study.
Data sources are listed in Table 1. All data points have been cross-
checked. If a specific project is reported in several sources, the
source that provided the most detailed data is used. For example,
Prior to Project
development
Project
development
Early planning/scoping
Initiate the National
Environmental Policy
Act (NEPA) process
Request
entry into
PD
Engineering
Request
entry into
engineering
Construction
Full
Funding
Grant
Agreement
Select locally preferred
alternatives (LPA)
Complete NEPA process
Obta in commitments of
at least 30% of non-
Capital Investment
Grants (CIG) funding
At least 30% design &
engineering
Project management plan
(PMP)
Obta in commitments of
all non-CIG funding
Complete sufficient
engineering & design
Right-of-way (ROW)
Acquisition
Update PMP, including
risk and contingency
mana gement Plan Legend = FTA
evaluate and appro val
AA/DEIS FFGA
Fig. 1. FTA New Starts project development process.
© ASCE 05020004-3 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
some projects are reported by both Booz Allen Hamilton (2005)
and the FTA (2003a). In such cases, data from the FTA is used
because it has more detailed time information for projects. After
screening all data, 81 rail transit projects were kept for data
analysis. The Appendix presents the cost data as well as the cor-
responding time point for each project. No other published work
has been found that contained more data.
These 81 projects were completed between 1984 and 2018, with
actual capital costs ranging from $58.1 million to $7,968.0 million.
On average, actual costs exceeded the costs estimated in the
AA/DEIS phase by 31.2% and FFGA phase by 5.8%. The mean
AA/DEIS estimates and actual costs are $608.1 and $826.5 million,
respectively. More detailed information can be found in Table 2.
As provided in the table, several variables have missing values.
For example, AA/DEIS year information is only available for
68 projects, therefore, the sample size for AA cost overrun trend
analysis is 68.
Data Population Size
Over the years, inaccuracy of transit captial cost estimates has
caused the FTA to increase its role in technical assistance and
project oversight for these projects (Voulgaris 2017). Specifically,
the FTA has required the performance of formal risk assessment to
mitigate cost overrun risks since 2003 [FTA PG-22 (FTA 2003b)].
Fig. 1shows that formal risk assessment usually takes place around
the FFGA phase. Therefore, a specific interest of this study is to
compare the group of projects with FFGA year before 2003 and
the group of projects with FFGA year after 2003. Among the 81
projects, only 20 projects fall into the latter group. To decide the
effective sample size for the after 2003 group, the corresponding
population size was first estimated.
The FTA publishes annual reports on funding recommendations
for transit projects on their website. Since the target population is
made up of projects that have FFGA year after 2003 and have been
completed before the data collection date (i.e., March 2019) annual
funding reports during 2004 and 2019 were reviewed. Each annual
report was reviewed to extract the number of newly recommended
rail transit projects. After screening out the repetitive ones across
different years, the population of projects with formal risk analysis
program was obtained, as provided in Table 3. A declining trend of
the CIG program appropriation was obeserved in the fiscal year
(FY) 2018, 2019, and 2020 reports. As suggested by GAO (2018),
this reflects the direction that the CIG program might be phased out
and future new transit projects would be locally funded without
federal financial assistance.
Sample size can be calculated using the following formula
[Eq. (2)]:
n¼z2pð1pÞ
h2þz2pð1pÞh2
N
ð2Þ
According to Israel (1992), three parameters matter in determin-
ing sample size: confidence level, degree of variability, and the
level of precision. Confidence level (represented by the correspond-
ing Z-score) is derived from the central-limit theorem, which rep-
resents the proportion of the sample that have the true population
value. The degree of variability (p) measures the heterogeneity of
the population. 50% denotes the highest variability in a population
while a normally distributed population has a level of variability
of either 20% or 80%. The level of precision (h) is the absolute
percentage that the sample estimated value deviates from the pop-
ulations true value. In this case, the population size (N) is 24. If the
confidence level, degree of variability, and level of precision are
designated as 90%, 50%, and 10%, respectively, then the sample
size (n) should be 18 [Eq. (3)]. Therefore, it is concluded that the
20 projects can adequately represent the population of federally
funded rail transit projects with formal risk analysis
n¼1.6520.5ð10.5Þ
0.12þ1.6520.5ð10.5Þ0.12
24
18 ð3Þ
Data Analysis
AA/DEIS Capital Cost Estimates Have Improved
Over Time
In this section, the trend of cost overruns (percentage of AA) over
the years is explored. As was explained previously, the AA/DEIS
estimate is made at the time of making the decsion to move forward
with the project or not, so the accuracy of this estimate has impor-
tant consequences for the project. First, the use of an F-test is pro-
posed to verify that the percentage cost overrun has not remained
the same over the years. An F-test is used to compare means of
Table 2. Descriptive statistics of 81 projects
Parameter Number Minimum Maximum Mean Standard deviation
AA/DEIS estimate (millions YOE$) 81 36.0 4,352.0 608.1 704.2
FFGA estimate (millions YOE$) 71 73.3 4,866.6 685.0 776.7
Actual cost (millions $) 81 58.1 7,968.0 826.5 1,105.4
AA/DEIS year 68 1,969 2,009 1,992 9.4
FFGA year 54 1,984 2,012 2,000 6.9
Opening year 81 1,984 2,018 2,001 8.6
Project implementation duration (years) 68 4 22 10 3.6
Cost overrun (percentage of AA) 81 32.1% 123.5% 31.2% 33.6%
Cost overrun (percentage of FFGA) 71 20.7% 74.0% 5.8% 17.3%
Table 1. Number of projects from different data sources
Studies Counts
Before and after studies published by the FTA (2019b)13
Annual Report on Fundingrecommendations
published by the FTA (2019a) and project websites
8
OToole (2015)2
FTA (2008)17
Dantata et al. (2006)7
Booz Allen Hamilton (2005)18
FTA (2003a)7
Pickrell (1989)9
Total 81
© ASCE 05020004-4 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
populations across different years. The null hypothesis is that the
mean of cost overruns (percentages of AA) does not change over
the years. For the data tested, the hypothesis is rejected at the sig-
nificance level of 0.01 (F¼2.534,p¼0.004) with a sample size
of 68 projects. In engineering analysis, p-value between 0.1 and
0.01 have been used commonly, with p¼0.01 being at a high level
of significance (Walpole et al. 2011). Thus, the year of AA/DEIS
cost estimation does have an effect on cost overruns. Second, cost
overruns (percentage of AA) are plotted against the corresponding
estimation date, that is, year of AA/DEIS cost estimate (as shown in
Fig. 2). It can be observed that the accuracy of cost estimation has
improved over time.
The mean cost overrun (percentage of AA) in this study is com-
pared with corresponding means in other publications. Results are
given in Table 4. It is clear that the sample size in this study is the
largest of its kind. Cost overruns in the past were high (reached
52% once) but overall cost overruns in the last four decades stayed
at around 30%, which also implies an improvement in recent cost
estimation accuracy. The improvement in AA/DEIS cost estimate
accuracy could be partly explained by the change of perspective by
local project promoters on the purpose of the early cost estimation.
Voulgaris (2017) found that New Starts polices have assisted in
aligning the goals of federal and local agencies. For example, as
shown in Fig. 1, the local agency is required to obtain commitments
of at least 30% of local funding share to get FTAs approval to enter
into the engineering phase (FTA 2016). Such risk-sharing mecha-
nisms might prompt the local decision makers to focus more on the
feasibility of the proposed projects rather than simply obtaining the
federal funding.
Major Projects Have Significantly Higher Cost
Overruns
In this section, the authors examine the effect of project size on the
level of cost overrun. Project size measured by total cost has been
cited as a cause of cost overruns in many studies (Flyvbjerg et al.
2004;Odeck 2004;Cantarelli et al. 2012;Huo et al. 2018). Accord-
ing to SAFETEA-LU guidelines (109th Congress 2005), projects
with a total cost of $500 million or greater are designated as major
projects. Discounted actual costs (in 1984 dollars) were used to
measure project size in this study. The projects were grouped into
two classes: small projects and major projects. Fig. 3shows that
cost overruns of major projects are higher than small projects.
In addition, in the group of small projects, a mild decreasing trend
of cost overrun can be observed as the project becomes larger,
while in the group of major projects, cost overruns increase as
the project becomes larger. Three projects with discounted actual
costs larger than $1,500 million were identified as statistical out-
liers and excluded from the regression analysis. The regression line
for the rest of the major projects is y¼27.647 þ0.079x(R2¼
0.356,p¼0.007), where yis the percentage cost overrun and xis
the total project costs in $ millions expressed in 1984 dollars.
Thus, in the group of larger projects, the cost overrun increases
by 0.08% per $ million.
The observation in small projects echoes the results of Odeck
(2004) and Huo et al. (2018), in which smaller projects appear to
suffer more from cost overruns than larger ones in the sample of
road projects with total costs lower than $500 million. The obser-
vation in major projects corresponds to the results of Flyvbjerg et al.
(2004), in which larger projects tend to have larger percentage cost
overruns for bridges and tunnels with total costs greater than
$500 million in general. Based on this, it is proposed that there
may exist a threshold in project size, and the relationship between
project size and cost overruns reverses in the two sides of the
threshold. Since current research examining the relationship be-
tween project size and cost overruns remains inconclusive (Love
et al. 2015), the existence of threshold in project size can shed
insight into such kind of studies.
Fig. 2. Trend in accuracy of AA/DEIS capital cost versus AA/DEIS
year. The curve is the local regression fitting line. Each smoothed value
is given by a weighted quadratic least squares regression over the span
of values of the y-axis scattergram criterion variable. This method does
not specify a function to fit a global model to the data, only to fit seg-
ments of the data. The darker shading displays 0.95 confidence interval
around the regression fitting line. There is a total of 68 projects avail-
able for this analysis.
Table 3. Newly recommended projects with FFGA year after 2003 and
opening year before 2019
Source Number Project name
FY 2004 4 Chicago Ravenswood line extension
Las Vegas Resort corridor fixed Guideway
Minnesota Minneapolis Northstar corridor rail
project
VA, Norfolk The tide light rail project
FY 2005 3 Phoenix Central phoenix/east valley LRT corridor
Utah, salt lake city mid-Jordan light rail project
Pittsburgh North shore LRT connector
FY 2006 1 Charlotte South corridor LRT
FY 2007 6 CO, Denver, West corridor LRT
OR, Portland, South corridor I-205/Portland mall LRT
OR, Washington county, Wilsonville to Beaverton
commuter rail
TX, Dallas, Northwest/Southeast LRT MOS
Dulles Corridor Metrorail project extension to
Wiehle avenue (Phase 1)
UT, Salt Lake City, Weber County to Salt Lake City
Commuter rail
FY 2008 2 NY, New York/Second avenue subway phase I
WA, Seattle/University Link LRT extension
FY 2009 0
FY 2010 3 Sacramento, South Corridor phase 2
FL, Orlando, Central Florida commuter rail transit
TX, Houston, North Corridor LRT
FY 2011 1 MN, St. Paul-Minneapolis, Central Corridor LRT
FY 2012 4 OR, Portland, Portland-Milwaukie light rail project
UT, Salt Lake County, Draper transit corridor
NC, Charlotte, LYNX blue line extension Northeast
corridor
Central mesa light rail extension; Mesa, Arizona
Total 24
Note: FY = fiscal year.
© ASCE 05020004-5 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
To further test the difference between cost overruns of major
projects and small projects, an independent-samples t-test was
performed. As given in Table 5, average cost overruns for major
projects and small projects are 48.5% and 24.8%, respectively,
using the cost estimates at the AA/DEIS phase. This difference
(48.5% 24.8% ¼23.7%) is statistically significant at the sig-
nificance level of 0.01 (t¼2.962,p¼0.004). Therefore, it can
be concluded that major projects (those costing in excess of
$500 millions) have significantly higher cost overruns than smaller
projects (those costing less than $500 millions).
Cost Overruns Increase as Project Implementation
Durations Increase
The project implementation duration is defined as the period
between a projects AA/DEIS phase and the projects opening.
Similar to project size, the implementation period is an important
project characteristic and has been proved to be closely correlated
with cost overruns (Flyvbjerg et al. 2003b,2004;Cantarelli et al.
2012;Huo et al. 2018). Fig. 4shows that cost overruns increase as
the project implementation duration becomes longer. This aligns
with Flyvbjerg et al.s(2004) result that the cost overrun grows sig-
nificantly as the length of project implementation phase increases.
The project implementation duration can be further broken down
into the preconstruction period (project development duration) and
the construction period (Cantarelli et al. 2012;Huo et al. 2018). In
this study, FFGA year is used as a proxy for the starting point of
construction. The project preconstruction period was calculated as
the duration between the AA/DEIS year and FFGA year, while the
project construction period is the difference between the FFGAyear
and project opening year. The regression results showed that in the
sample of US rail transit projects, the construction period is a better
predictor of cost overruns (p¼0.023) compared to the precon-
struction period (p¼0.368). This aligns with Touran and Lopezs
(2006) finding that it is difficult to make an accurate forecast when
the construction period is longer.
Efficacy of the FTAs Risk Assessment Program
To enhance the oversight of a transit projects design and construc-
tion and aid in reliable cost estimation, the FTA has implemented
a formal risk assessment program since 2003 and published oper-
ating guidance and oversight procedures for risk assessment and
mitigation [FTA PG-22 (FTA 2003b); FTA PG-40 (2007)]. The pur-
pose of these guidelines is to support the FTAs programmatic de-
cisions made under uncertainty, especially to evaluate the reliability
of the grantee project cost and schedule estimate (FTA 2007).
Table 4. Comparison of average AA cost overruns in different publications
Parameter This study FTA (2008)
Dantata
et al. (2006)
Booz Allen
Hamilton (2005)FTA(2003a) Pickrell (1989)
Range of project completion year 19842018 20022007 19902004 19842002 19892002 19861989
Sample size 81 23 16 28 21 10
Average cost overrun (AA) 31.2% 40.2% 30.0% 36.3% 20.9% 52.0%
Note: There are repeated projects across the sample used by these publications.
Fig. 3. Cost overruns versus discounted actual capital cost. For scale
reasons, the Washington Metro project, Atlanta Metropolitan HRT
project, and New York Second Avenue Subway Phase I project were
removed from this graph and 78 projects were used. The vertical line
denotes the discounted actual capital cost of $500 million (in 1984
dollars).
Table 5. Group statistics of cost overruns (percentage of AA)
Parameter
Project size
group Number
Mean
(%)
Standard
deviation
(%)
Standard
error
mean (%)
Cost overrun
(AA)
Major projects 22 48.5 34.8 7.4
Small projects 59 24.8 31.1 4.0
Note: Standard error mean is the standard deviation of the sample divided
by the square root of the sample size.
Fig. 4. Cost overruns versus project implementation durations. The
Minneapolis Hiawatha Line project was excluded from this figure since
its duration (22 years) exceeded three SD from the mean of project
implementation durations. 67 projects are used in this figure.
© ASCE 05020004-6 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
All New Starts transit projects requesting funding from the FTA
have to go through a formal risk assessment. The output of the risk
assessment includes a risk register that lists risk factors contributing
to the variation in each cost element, that is, standard cost categories
(SCC). Each SCC represents a major cost component such as track-
work, stations, right-of-way, etc. According to the top-down model
recommended in FTA PG-40 (2007), each SCC will be modeled as
a random variable with a reasonable statistical distribution. All
SCCs shall be added up to obtain the probability distribution of
the total project cost. In addition, there are guidelines to account
for modeling the correlations between the SCCs. This is a substan-
tial effort usually undertaken by project management oversight
(PMO) consultants working for the FTA and the transit agencies.
Based on the result of this risk assessment, which usually utilizes
probabilistic methods, the FTA may require a revision of estimated
project budget or duration. For example, in the FTA (2011, p. 109),
it is stated:
METRO completed PE on the Houston Southeast Corridor
LRT project in early 2009. FTA completed a risk assessment
of the projects budget, schedule and scope in April 2009. As
a result, the projects capital cost estimate was revised to re-
flect the higher level of design and to include an increased
level of contingency. The financial plan was revised to reflect
the updated capital cost estimate, including an adjusted esti-
mate of finance charges.
Thus, it is expected that projects with a FFGA year later than
2003 (the year that risk assessment program was implemented)
have more accurate cost estimates. In order to verify this, the cost
overrun value (as a percentage of FFGA estimate) was plotted
against the FFGA year (Fig. 5). As can be seen, it is difficult to
ascertain a definite trend. The average cost overrun for projects
before 2003 was calculated as 6.7% and the average cost overrun
for projects after 2003 was calculated as 3.1% (Table 6). The
independent-samples t-test results shows that the difference be-
tween average cost overruns of projects before and after 2003
(3.6% as shown in Table 6) is not significant.
Although the mean FFGA cost overruns of both groups are rel-
atively low, their standard deviations (SDs) are high. For projects
before 2003, the SD is almost three times the mean (Table 6), which
reveals the projectscost estimation performance is not consistent
and needs to be improved. Therefore, except comparing means,
difference between the variances of the two groups before and after
2003 was also analyzed. Levenes test for equality of variances was
employed. The null hypothesis that the population variances are
equal is not rejected (F¼1.914,p¼0.172).
Based on the aforementioned analysis, there is no reason to con-
clude that cost overruns (percentage of FFGA) differ before and
after 2003. Previous results show that project size and project im-
plementation duration have negative impact on the cost estimation
performance. So, it is necessary to control the effect of these two
variables when investigating the effect of risk analysis. Table 7
shows that projects after 2003 are on average larger than projects
before 2003. Table 8shows that the implementation durations of
projects after 2003 are longer than those before 2003. Therefore,
it can be inferred that the risk analysis program has exerted certain
positive effects on the cost estimating performance to counteract
the negative effects caused by larger project size and longer imple-
mentation duration. Besides, the inherent optimistic bias from
project developers can lead to underestimation of the magnitude
of project risks, which undermines benefits and effectiveness of
the risk assessment program.
Conclusion and Discussion
Cost overruns in US rail transit projects have been recorded and
scattered in several studies in the last few decades. Thus, it is useful
and important to conduct a systematic and statistical study to ex-
amine the trend of cost overruns in different project contexts. In
addition, the FTA has enacted a formal risk analysis program since
2003 at considerable cost, yet there is no research to date investi-
gating the efficacy of such a program in curbing cost overruns. The
objective of this research was to address these two gaps. Specifi-
cally, cost overruns against two estimates were probed: (1) cost
estimates at the AA/DEIS phase (decision-making phase), and
(2) cost estimates at the FFGA phase (the point in time after risk
analysis is performed).
It was found that the accuracy of cost estimates at the decision-
making phase has improved significantly over time, which provides
reason for optimism regarding the accuracy of cost estimates in
future transportation projects. The relations between cost overruns
(percentage of AA) and broad project characteristics were also
Fig. 5. Trend in accuracy of FFGA capital cost versus FFGA year.
Table 6. Group statistics of cost overruns (percentage of FFGA)
Parameter
FFGA
year group Number
Mean
(%)
Standard
deviation
(%)
Standard
error
mean (%)
Cost overrun
(FFGA)
After 2003 20 3.1 13.0 2.9
Before 2003 34 6.7 19.0 3.2
Table 7. Group statistics of discounted actual cost
Parameter FFGA year Number
Mean
(millions 1984$)
Standard
deviation
Discounted
actual cost
After 2003 20 464.4 444.7
Before 2003 34 345.9 331.3
Table 8. Group statistics of project implementation duration
Parameter FFGA year Number
Mean
(years)
Standard
deviation
Project implementation
duration
After 2003 20 10.4 3.4
Before 2003 34 9.3 3.8
© ASCE 05020004-7 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
probed, including project size and project implementation duration.
In general, major projects (discounted actual costs in 1984 dollars
greater than 500 million) have significantly higher cost overruns
(23.7%) than small projects. Moreover, in the group of major proj-
ects, cost overruns increase by 0.08% as the capital costs increase
by $1 million, while a mild opposite trend appears in the group of
small projects. This poses an area for future research in examining
the existence of a threshold value where the relation between
project size and cost overrun could change. Such research may rec-
oncile the current contradictory results that some studies concluded
percentage cost overrun increases as project size grows while others
concluded that percentage cost overrun decreases as project size
grows. It was also found that the accuracy of cost estimates de-
creases as project implementation durations increase. The project
implementation duration was further broken down into the precon-
struction period and the construction period and the result showed
that the construction period was a better predictor of cost overruns
compared to the preconstruction period in the sample of US rail
transit projects.
It should be noted that apart from these two project character-
istics, there exist plenty of factors that account for cost overrun, as
mentioned in the section Causes of Cost Overrun and Methods of
Mitigation.Since the selection criteria and regulation policies used
for federally funded rail projects are the same, these projects make
up a special sample. It is worthy of further study to collect more
project information and uncover the causes of cost overruns in FTA
funded US rail transit projects. Potential cost influential factors for
US rail transit projects can include bias in estimating, the project
development process with various permit procedures, and the use of
a variety of project delivery methods.
In order to evaluate the impact of the FTAs formal risk analysis
program on cost overruns, projects were divided into two groups:
projects with FFGA year before 2003 and after 2003. The
independent-samples t-test and Levenes test for equality of varian-
ces showed that the difference regarding accuracy of FFGA cost
estimates is not significant. In other words, the performance of cost
estimation stays the same before and after 2003. Giving consider-
ation of the negative effects caused by larger project size and longer
implementation duration, it can be concluded that the risk assess-
ment program has exerted a positive impact to some extent. An-
other plausible explanation is that the successful assessment of
project risks requires the minimization of inherent bias. However,
the effect of the optimistic bias of project developers cannot be
eliminated, which leads to underestimation of the magnitude of
project risks.
In conclusion, this study has performed statistical analysis on
the cost overrun trend of US rail transit projects over the last
40 years and revealed the overall positive trend. Longer overall
project implementation duration is associated with higher percent-
age cost overruns. In addition, based on the observed sample data,
construction phase duration might be a better predictor for cost
overruns in US rail transit projects. The positive effects of the FTAs
risk assessment program on the cost estimating performance have
also been confirmed to some extent, considering that projects built
since 2003 are generally larger and of longer duration. Through
focusing exclusively on rail transit projects rather than using a
mix of all kinds of transportation projects, the results of this study
can be a more specific reference point for practitioners and policy
makers, both within and outside the US.
Appendix. Cost and Schedule Data of the 81 Case
Study Projects
This table is a valuable dataset containing cost data on US rail
transit projects completed in the last 40 years. The AA/DEIS cost
estimate, actual cost, and opening year data are available for all
projects, while other variables may have missing data. The table
contributes substantially to studies on the cost estimation topic
since collecting data is a primary undertaking and sample size is
a major restriction in such research; forecast and actual cost and
the corresponding date information are usually scattered in several
reports and not readily available.
ID Case study project
AA/DEIS
estimate
(millions
YOE$)
FFGA
estimate
(millions
YOE$)
Actual
cost
(millions)
Cost
overrun
(percentage
of AA)
Cost
overrun
(percentage
of FFGA)
AA/DEIS
year
FFGA
year
Opening
year
Project
implementation
duration
(Years)
1 Atlanta Metropolitan Atlanta HRT $1,723.0 $2,720.0 57.9% 1971 1987 16
2 Atlanta North Line Extension $439.5 $352.0 $472.7 7.6% 34.3% 1990 1994 1999 9
3 Baltimore BWI, Hunt Valley,
Penn Station Ext.
$81.9 $109.5 $116.2 41.9% 6.1% 1990 1994 1997 7
4 Baltimore Central LRT
Double-Tracking
$150.5 $154.4 $151.6 0.7% 1.8% 2000 2001 2006 6
5 Baltimore Ext. to Johns Hopkins $313.7 $310.5 $353.0 12.5% 13.7% 1984 1988 1995 11
6 Baltimore Phase I Rapid Transit $804.0 $1,289.0 60.3% 1972 1987 15
7 BART Extension to SFO $1,193.9 $1,185.7 $1,551.6 30.0% 30.9% 1995 1997 2003 8
8 Boston Old Colony Rehabilitation $447.2 $551.9 $565.0 26.3% 2.4% 1990 1997 7
9 Buffalo Minimum LR Rapid Transit $478.0 $722.0 51.0% 1977 1989 12
10 Central Florida Commuter Rail
Transit Initial Operating Segment
$361.5 $357.2 $357.2 1.2% 0.0% 2004 2011 2014 7
11 Central Mesa Light Rail Extension $198.5 $196.7 $196.7 0.9% 0.0% 2009 2012 2015 6
12 Charlotte South Corridor Light Rail $370.9 $426.8 $462.8 24.8% 8.4% 1999 2005 2007 8
13 Charlotte, LYNX Blue Line
Extension Northeast Corridor
$1,205.5 $1,160.1 $1,159.0 3.9% 0.1% 2002 2012 2018 16
14 Chicago Douglas Branch $441.7 $473.2 $440.8 0.2% 6.8% 2000 2001 2005 5
15 Chicago Southwest Extension $453.0 $350.9 $474.6 4.8% 35.3% ——1989
16 Chicago SW Transitway $604.0 $438.4 $522.0 13.6% 19.1% 1982 1987 1993 11
17 Dallas North Central LRT $332.7 $460.8 $437.3 31.4% 5.1% 1996 1999 2002 6
© ASCE 05020004-8 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
Appendix. (Continued.)
ID Case study project
AA/DEIS
estimate
(millions
YOE$)
FFGA
estimate
(millions
YOE$)
Actual
cost
(millions)
Cost
overrun
(percentage
of AA)
Cost
overrun
(percentage
of FFGA)
AA/DEIS
year
FFGA
year
Opening
year
Project
implementation
duration
(Years)
18 Dallas Northwest-Southeast
Light Rail Project
$1,151.4 $1,406.2 $1,406.2 22.1% 0.0% 2000 2005 2010 10
19 Dallas South Oak Cliff $325.4 $377.0 $360.0 10.6% 4.5% 1990 1993 1996 6
20 Dallas South Oak Cliff Extension $347.1 $517.2 $437.2 26.0% 15.5% ——2002
21 Denver West Rail Line LRT $350.0 $709.8 $707.0 102.0% 0.4% 2001 2005 2013 12
22 Denver Southeast Corridor $585.0 $867.8 $850.8 45.4% 2.0% 1997 2000 2006 9
23 Denver Southwest Line $149.6 $176.9 $177.7 18.8% 0.5% 1995 1999 2000 5
24 Detroit Central Automated Transit $144.0 $215.0 49.3% 1980 1988 8
25 Dulles Corridor Metrorail
Project Extension to Wiehle
Avenue (Phase 1)
$1,500.0 $3,142.5 $2,900.0 93.3% 7.7% 2003 2009 2014 11
26 Jacksonville Skyway $85.8 $142.0 $137.3 60.0% 3.3% 1982 1994 2000 18
27 Largo Metrorail Extenstion $375.0 $412.6 $426.4 13.7% 3.3% 1996 2000 2004 8
28 Los Angeles Blue Line $561.0 $877.0 56.3% 1984 1990 6
29 Los Angeles Metro Gold Line
Eastside Extension Project
$759.5 $898.8 $899.1 18.4% 0.0% 1995 2002 2009 14
30 Los Angeles Red Line MOS 1 $695.8 $960.3 $1,490.1 114.2% 55.2% 1983 1993 10
31 Los Angeles Red Line MOS 2 $1,116.6 $1,524.6 $1,921.6 72.1% 26.0% ——1999
32 Los Angeles Red Line MOS 3 $1,113.3 $1,345.6 $1,227.6 10.3% 8.8% ——2000
33 Memphis Med Center LRT $36.0 $73.3 $58.1 61.4% 20.7% 1997 2000 2004 7
34 Metra North Central $204.8 $224.8 $216.8 5.9% 3.6% 1998 2001 2006 8
35 Metra SW Corridor $178.7 $191.0 $185.3 3.7% 3.0% 1998 2001 2006 8
36 Metra UP West $98.8 $128.1 $106.1 7.4% 17.2% 1998 2001 2006 8
37 Miami Dade County HRT $1,008.0 $1,341.0 33.0% 1978 1988 10
38 Miami DPM Starter Line $84.0 $175.0 108.3% 1980 1988 8
39 Miami Omni &Brickell Ext. $221.2 $161.3 $228.0 3.1% 41.4% 1987 1989 1995 8
40 Minneapolis Hiawatha Line $581.0 $675.4 $715.3 23.1% 5.9% 1982 2001 2004 22
41 Minneapolis Northstar Corridor
Rail Project
$265.1 $317.4 $308.5 16.4% 2.8% 2000 2006 2009 9
42 Minneapolis St. Paul Central
Corridor LRT
$581.0 $956.9 $957.0 64.7% 0.0% 2006 2011 2013 7
43 Mission Valley East LRT Extension $386.6 $426.6 $506.2 30.9% 18.7% 1997 2000 2005 8
44 New Jersey Hudson-Bergen
MOS 1 & 2
$930.4 $1,842.0 $1,756.2 88.8% 4.7% 1992 2000 2006 14
45 New York 63rd Street Connector $488.4 $645.0 $632.3 29.5% 2.0% ——2001
46 Newark Rail Link MOS-1 $181.3 $215.4 $207.7 14.6% 3.6% 1997 2000 2006 9
47 Norfolk The Tide Light Rail Project $194.6 $231.6 $314.6 61.7% 35.8% 2000 2006 2011 11
48 New York Second Avenue Subway
Phase I
$3,880.0 $4,866.6 $4,450.0 14.7% 8.6% 1999 2007 2017 18
49 Oceanside Sprinter Light Rail
(San Diego Oceanside/Escondido)
$213.7 $332.3 $477.6 123.5% 43.7% 1995 2000 2008 13
50 Pasadena Gold Line $997.5 $693.9 $677.6 32.1% 2.3% ——2003
51 Phoenix, Central Phoenix/East
Valley LRT Corridor
$1,076.0 $1,412.0 $1,405.0 30.6% -0.5% 1998 2003 2008 10
52 Pittsburgh North Shore Connector $326.7 $435.0 $510.4 56.2% 17.3% 1999 2003 2012 13
53 Pittsburgh South Hills
Reconstruction
$699.0 $622.0 11.0% 1976 1989 13
54 Portland Airport MAX Extension $125.0 $125.0 $127.0 1.6% 1.6% ——2001
55 Portland Banfield Corridor $214.0 $286.6 $246.8 15.3% 13.9% ——1984
56 Portland Green Line Light Rail
Project
$495.4 $576.0 $576.0 16.3% 0.0% 1994 2005 2009 15
57 Portland Interstate MAX $313.4 $314.9 $349.4 11.5% 11.0% 1996 2000 2004 8
58 Portland Westside/Hillsboro MAX $559.3 $886.5 $964.0 72.4% 8.7% 1982 1994 1998 16
59 Portland-Milwaukie Light Rail
Project
$1,389.1 $1,490.4 $1,490.0 7.3% 0.0% 2002 2012 2015 13
60 Sacramento South LRT (Phase 1) $201.6 $219.7 $218.6 8.4% 0.5% 1996 1997 2003 7
61 Sacramento South LRT (Phase 2) $226.3 $270.0 $260.0 14.9% 3.7% 2005 2012 2015 10
62 Sacramento Starter Line
(Stage I LRT)
$165.0 $188.0 13.9% 1981 1988 7
63 Salt Lake City Mid-Jordan Light
Rail Project
$521.8 $535.4 $509.8 2.3% 4.8% 2005 2008 2011 6
64 Salt Lake North-South Line $276.9 $312.0 $311.8 12.6% 0.1% 1990 1999 9
65 San Diego, Calif., East Urban
Corridor
$114.4 $100.4 $102.7 10.2% 2.3% 1985 1986 1989 4
© ASCE 05020004-9 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
Appendix. (Continued.)
ID Case study project
AA/DEIS
estimate
(millions
YOE$)
FFGA
estimate
(millions
YOE$)
Actual
cost
(millions)
Cost
overrun
(percentage
of AA)
Cost
overrun
(percentage
of FFGA)
AA/DEIS
year
FFGA
year
Opening
year
Project
implementation
duration
(Years)
66 San Francisco, Colma BART
Station
$112.5 $171.6 $179.9 59.9% 4.8% 1988 1993 1996 8
67 San Francisco, SFO Airport Ext. $1,175.3 $1,167.0 $1,550.2 31.9% 32.8% ——2003
68 San Jose Tasman West $451.2 $346.1 $325.2 27.9% 6.0% 1991 1996 1999 8
69 San Jose, Calif., Guadalupe
Corridor
$257.7 $395.2 $380.3 47.6% 3.8% 1981 1984 1991 10
70 San Juan Tren Urbano $1,085.6 $1,280.6 $2,228.4 105.3% 74.0% 1995 1996 2005 10
71 Santa Clara Capitol Line $147.1 $159.8 $162.5 10.5% 1.7% ——2003
72 Santa Clara Tasman East Line $197.7 $275.1 $276.2 39.7% 0.4% ——2001
73 Santa Clara Vasona Line $269.1 $313.6 $316.8 17.7% 1.0% ——2004
74 Seattle Central Link Initial Segment
and Airport Link Project
$1,858.0 $2,668.0 $2,558.0 37.7% 4.1% 1999 2001 2009
75 Seattle/University Link LRT
Extension
$1,645.9 $1,947.7 $1,745.7 6.1% 10.4% 2005 2009 2016 11
76 St. Louis Saint Clair Corridor $337.3 $339.2 $336.5 -0.2% 0.8% ——2000
77 St. Louis Metrolink $379.7 $455.8 $464.0 22.2% 1.8% 1984 1995 1993 9
78 The Weber County to Salt Lake
Commuter Rail Project
(FrontRunner North
$408.0 $611.7 $614.0 50.5% 0.4% 2001 2005 2008 7
79 Washington and Multnomah
Counties Wilsonville to Beaverton
Commuter Rail Project (Westside
Express Service)
$84.8 $117.3 $162.0 91.0% 38.1% 2001 2004 2009 8
80 Washington Largo Extension $397.1 $433.9 $456.0 14.8% 5.1% 1996 2004 8
81 Washington METRO $4,352.0 $7,968.0 83.1% 1969 1986 17
Data Availability Statement
All data, models, and code generated or used during the study
appear in the published article. Information about ASCEs data
sharing policy can be found here: https://ascelibrary.org/page
/dataavailability.
Notation
The following symbols are used in this paper:
h= level of precision;
N= population size;
n= sample size;
p= degree of variability; and
z=Z-score for a confidence level.
References
109th Congress. 2005. Safe, accountable, flexible, efficient transportation
equity act: A legacy for users. Washington, DC: United States
Government Publishing Office.
Andri´c, J. M., A. M. Mahamadu, J. Wang, P. X. Zou, and R. Zhong. 2019.
The cost performance and causes of overruns in infrastructure develop-
ment projects in Asia.J. Civ. Eng. Manage. 25 (3): 203214. https://
doi.org/10.3846/jcem.2019.8646.
Azhar, S. 2011. Building information modeling (BIM): Trends, benefits,
risks, and challenges for the AEC industry.Leadersh. Manage.
Eng. 11 (3): 241252. https://doi.org/10.1061/(ASCE)LM.1943-5630
.0000127.
Bakhshi, P., and A. Touran. 2009. Comparison of current probabilistic
approaches for budget estimating for transportation projects.In Proc.,
7th Int. Probabilistic Workshop, edited by P. Van Gelder, D. Proske, and
H. Vrijling, 479. Delft, Netherlands: TU Delf.
Booz Allen Hamilton. 2005. Managing capital costs of major federally
funded public transportation projects. TCRP Project G-7. Washington,
DC: Transportation Research Board.
British Department for Transport. 2004. Procedure for dealing with
optimism bias in transport planning. London: British Department for
Transport.
Cantarelli, C. C., B. Flybjerg, E. J. Molin, and B. Van Wee. 2010. Cost
overruns in large-scale transportation infrastructure projects: Explana-
tions and their theoretical embeddedness.Eur. J. Transp. Infrastruct.
10 (1): 3041.
Cantarelli, C. C., B. van Wee, E. J. Molin, and B. Flyvbjerg. 2012. Differ-
ent cost performance: Different determinants?: The case of cost over-
runs in Dutch transport infrastructure projects.Transp. Policy 22 (Jul):
8895. https://doi.org/10.1016/j.tranpol.2012.04.002.
Dantata, N. A., A. Touran, and D. C. Schneck. 2006. Trends in US rail
transit project cost overrun. Washington, DC: Transportation Research
Board.
Duff, D., E. J. Gill Jr., G. K. Woodman, and T. Coburn. 2010. Legal
handbook for the New Starts process. Washington, DC: Transportation
Research Board.
Eastman, C., P. Teicholz, R. Sacks, and K. Liston. 2011. BIM handbook:
A guide to building information modeling for owners, managers, de-
signers, engineers and contractors. Hoboken, NJ: Wiley.
ENR (Engineering News-Record). 2019. Building cost index historical
data.Accessed March 27, 2019. http://enr.construction.com
/economics/.
Flyvbjerg, B., et al. 2018. Five things you should know about cost
overrun.Transp. Res. Part A: Policy Pract. 118 (Dec): 174190.
https://doi.org/10.1016/j.tra.2018.07.013.
Flyvbjerg, B., N. Bruzelius, and W. Rothengatter. 2003a. Megaprojects and
risk: An anatomy of ambition. Cambridge, UK: Cambridge University
Press.
© ASCE 05020004-10 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
Flyvbjerg, B., M. S. Holm, and S. Buhl. 2002. Underestimating costs
in public works projects: Error or lie?J. Am. Plann. Assoc. 68 (3):
279295. https://doi.org/10.1080/01944360208976273.
Flyvbjerg, B., C. K. Hon, and W. H. Fok. 2016. Reference class forecast-
ing for Hong Kongs major roadworks projects.Proc. Inst. Civ. Eng.
169 (6): 1724. https://doi.org/10.1680/jcien.15.00075.
Flyvbjerg, B., M. K. Skamris Holm, and S. L. Buhl. 2003b. How common
and how large are cost overruns in transport infrastructure projects?
Transp. Rev. 23 (1): 7188. https://doi.org/10.1080/01441640309904.
Flyvbjerg, B., M. K. Skamris Holm, and S. L. Buhl. 2004. What causes
cost overrun in transport infrastructure projects?Transp. Rev. 24 (1):
318. https://doi.org/10.1080/0144164032000080494a.
FTA (Federal Transit Administration). 2002. Full-funding grant agree-
ments guidance.Accessed March 27, 2019. https://www.transit.dot
.gov/regulations-and-guidance/fta-circulars/full-funding-grant-agreements
-guidance.
FTA (Federal Transit Administration). 2003a. Predicted and actual impacts
of New Starts projects: Capital cost, operating cost and ridership data.
Washington, DC: FTA.
FTA (Federal Transit Administration). 2003b. Risk assessment and mitiga-
tion procedures. Washington, DC: FTA.
FTA (Federal Transit Administration). 2007. Risk management products
and procedures. Washington, DC: FTA.
FTA (Federal Transit Administration). 2008. The predicted and actual
impacts of New Starts projects: Capital cost and ridership. Washington,
DC: FTA.
FTA (Federal Transit Administration). 2011. Annual report on funding
recommendations.Accessed March 27, 2019. https://www.transit
.dot.gov/funding/grant-programs/capital-investments/proposed-allocation
-funds-fiscal-year-2011-capital.
FTA (Federal Transit Administration). 2016. Final interim policy guidance
federal transit administration capital investment grant program.
Accessed March 27, 2019. https://www.transit.dot.gov/sites/fta.dot.gov
/files/docs/FAST_Updated_Interim_Policy_Guidance_June%20_2016
.pdf.
FTA (Federal Transit Administration). 2019a. Annual report on funding
recommendations.Accessed March 27, 2019. https://www.transit.dot
.gov/funding/grant-programs/capital-investments/annual-report-funding
-recommendations.
FTA (Federal Transit Administration). 2019b. Before-and-after studies of
New Starts projects: Report to congress.Accessed March 27, 2019.
https://www.transit.dot.gov/before-and-after-studies.
FTA (Federal Transit Administration). 2019c. Capital investment grant fact
sheet.Accessed March 27, 2019. https://www.transit.dot.gov/sites/fta
.dot.gov/files/docs/5309_Capital_Investment_Grant_Fact_Sheet.pdf.
GAO (Government Accountability Office). 2018. Capital investment
grants program: FTA should address several statutory provisions.Ac-
cessed March 27, 2019. https://www.gao.gov/products/GAO-18-462.
GAO (Government Accountability Office). 2019. Federal transit admin-
istration could improve information on estimating project costs.
Accessed July 25, 2019. https://www.gao.gov/products/GAO-19-562.
Huo, T., H. Ren, W. Cai, G. Q. Shen, B. Liu, M. Zhu, and H. Wu. 2018.
Measurement and dependence analysis of cost overruns in megatran-
sport infrastructure projects: Case study in Hong Kong.J. Constr. Eng.
Manage. 144 (3): 05018001. https://doi.org/10.1061/(ASCE)CO.1943
-7862.0001444.
Israel, G. D. 1992. Determining sample size. Gainesville, FL: Univ. of
Florida.
Love, P. E., C. P. Sing, X. Wang, Z. Irani, and D. W. Thwala. 2014. Over-
runs in transportation infrastructure projects.Struct. Infrastruct. Eng.
10 (2): 141159. https://doi.org/10.1080/15732479.2012.715173.
Love, P. E., J. Smith, I. Simpson, M. Regan, and O. Olatunji. 2015. Under-
standing the landscape of overruns in transport infrastructure projects.
Environ. Plann. B: Plann. Des. 42 (3): 490509. https://doi.org/10
.1068/b130102p.
Odeck, J. 2004. Cost overruns in road constructionWhat are their sizes
and determinants?Transp. Policy 11 (1): 4353. https://doi.org/10
.1016/S0967-070X(03)00017-9.
Odeck, J. 2017. Variation in cost overruns of transportation projects:
An econometric meta-regression analysis of studies reported in the
literature.Transportation 46 (4): 13451368. https://doi.org/10.1007
/s11116-017-9836-5.
OToole, R. 2015. Rail transit overruns.Accessed March 27, 2019. http://
ti.org/antiplanner/?p=10036.
Pickrell, D. H. 1989. Urban rail transit projects: Forecast versus actual
ridership and costs. Washington, DC: Urban Mass Transportation
Administration.
Pickrell, D. H. 1992. A desire named streetcar: Fantasy and fact in rail
transit planning.J. Am. Plann. Assoc. 58 (2): 158176. https://doi
.org/10.1080/01944369208975791.
Siemiatycki, M. 2009. Academics and auditors: Comparing perspectives
on transportation project cost overruns.J. Plann. Educ. Res. 29 (2):
142156. https://doi.org/10.1177/0739456X09348798.
Touran, A., D. D. Gransberg, K. R. Molenaar, K. Ghavamifar, D. J. Mason,
and L. A. Fithian. 2009. A guidebook for the evaluation of project de-
livery methods. Transit Cooperative Research Program Rep. No. 131,
Washington, DC: Transportation Research Board.
Touran, A., and R. Lopez. 2006. Modeling cost escalation in large infra-
structure projectsJ. Constr. Eng. Manage. 132 (8): 853860. https://
doi.org/10.1061/(ASCE)0733-9364(2006)132:8(853).
Voulgaris, C. T. 2017. Crystal balls and black boxes: Optimism bias
in ridership and cost forecasts for new starts rapid transit projects.
Ph.D. dissertation, Dept. of Urban Planning, Univ. of California,
Los Angeles.
Walpole, R. E., R. H. Myers, S. L. Myers, and K. Ye. 2011. Probability &
statistics for engineers & scientists. 9th ed. London: Pearson.
© ASCE 05020004-11 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2020, 146(5): 05020004
Downloaded from ascelibrary.org by Northeastern University Library on 05/18/20. Copyright ASCE. For personal use only; all rights reserved.
... A review of the literature on the factors affecting the magnitude of overrun reveals that project type, for example, rail, road or fixed link (bridge and tunnel), is an important determinant (Cantarelli et al. 2012a(Cantarelli et al. , 2012bFlyvbjerg 2006;Flyvbjerg et al. 2002Flyvbjerg et al. , 2003Flyvbjerg et al. , 2004Morris 1990;Odeck 2004;Skamris and Flyvbjerg 1997). Many researchers have attempted to identify the most important factors affecting overrun either by sending out questionnaires to experts or by conducting statistical analysis on a large number of completed projects (Gao and Touran 2020;Huo et al. 2018;Johnson and Babu 2020;Larsen et al. 2016;Memon et al. 2011;Zewdu and Aregaw 2015). After examining a large sample of projects, Flyvbjerg et al. (2002) concluded that technical explanations, such as estimation methods, are not important; instead, they contended that the most likely causes of time and cost overruns are either political factors (i.e., deliberate underestimation) or economic motives, whereby certain parties benefit from the approval of particular projects. ...
... However, according to a number of recent studies, forecast accuracy has in fact improved over time (Gao and Touran 2020;Sarmento and Renneboog 2017), and cost overruns for U. S. rail transit projects have become smaller (Dantata et al. 2006). As for the biased distribution of overrun, Flyvbjerg et al. (2002) acknowledge the necessity and viability of accounting for different risk factors (such as geological, environmental and safety problems) in the time and cost estimation process and point out that proponents of technical explanations for overruns should explain why the forecasters ignored these risks over the long term. ...
Article
Full-text available
Probabilistic time estimation is an essential part of proper risk management in tunneling projects. In recent decades, several models have been developed for this purpose, one of which was developed by Isaksson and Stille (Rock Mech Rock Eng 38:373–398, 2005). In this paper, Isaksson and Stille’s probabilistic time and cost estimation model was improved and then applied to estimate the total tunneling time of the headrace tunnel in the Uri hydropower project in India. The improvements allow the user to more accurately account for different types of geological features and disruptive events. The result of the estimation is a distribution of tunneling time. The outcome illustrates how a proper understanding of the geological setting of the project and its effect on construction performance can contribute to effective risk management.
... Many researchers have worked on the phenomenon of cost overruns and time delays. A review of the literature shows that the majority of existing papers tried to find the main causes of this phenomenon either by conducting studies through sending questionnaires to the experts in the field (Johnson and Babu, 2020;Kaliba et al., 2009;Larsen et al., 2016;Mahamid and Dmaidi, 2013;Memon et al., 2011;Moura et al., 2007;Zewdu and Aregaw, 2015) or through statistical analysis of large numbers of completed project data (Gao and Touran, 2020;Huo et al., 2018;Lee, 2008;Torp et al., 2016). In both cases, the researchers identified many factors, including scope changes, project complexity, construction delays, unreasonable estimation, and many others to be the reasons for occurrence of cost overruns. ...
... Finally, according to some recent studies, forecast accuracy has in fact improved over time. For instance, Gao and Touran (2020) found that the accuracy of cost estimates at decision-making phase for US rail transit projects has improved over time. Sarmento and Renneboog (2017) concluded that cost deviations in Portuguese public infrastructural investment projects has improved over time. ...
... The preprocessing operations to identify outliers' data and their normalization are performed to reduce errors in the accuracy of learning algorithms (Gao and Touran, 2020). The collected projects had a different range of costs and execution times. ...
Design/methodology/approach - The cases were extracted by studying 68 water-related projects. This research employs earned value management (EVM) factors to consider time and cost features and economic, natural, technical, and project risks to account for uncertainties and supervised learning models to estimate cost overrun. Time-series algorithms were also used to predict construction cost indexes (CCI) and model improvements in future forecasts. Outliers were deleted by the pre-processing process. Next, datasets were split into testing and training sets, and algorithms were implemented. The accuracy of different models was measured with the mean absolute percentage error (MAPE) and the normalized root mean square error (NRSME) criteria. Purpose - The present study aims to develop a risk-supported case-based reasoning (RS-CBR) approach for water-related projects by incorporating various uncertainties and risks in the revision step. Findings - The findings show an improvement in the accuracy of predictions using datasets that consider uncertainties, and ensemble algorithms such as Random Forest and AdaBoost had higher accuracy. Also, among the single algorithms, the support vector regressor (SVR) with the sigmoid kernel outperformed the others. Originality/value - This research is the first attempt to develop a case-based reasoning model based on various risks and uncertainties. The developed model has provided an approving overlap with machine learning models to predict cost overruns. The model has been implemented in collected water-related projects and results have been reported.
... Hasil penelitian menunjukkan adanya hubungan timbal balik antara besaran anggaran proyek dengan persentase cost overrun. Gao et al., (2020) melakukan penelitian pada proyek transit kereta api yang dibangun selama 40 tahun terakhir di Amerika menunjukkan bahwa pembengkakan biaya berbeda terjadi antara proyek yang dirancang sebelum dan setelah tahun 2003, dimana proyek menjadi semakin besar dan komplek sehingga terjadi perbedaan asumsi design awal dan pada pelaksanaannya. Susanti & Nurdiana (2021) melakukan penelitian pada proyek strategis jalan tol di Indonesia, meskipun semua proyek ini diprioritaskan dalam proses konstruksinya, pelaksanaan proyek strategis nasional menghadapi masalah terutama terkait dengan tekanan tinggi dari pemerintah yang berdampak pada biaya proyek bahkan menimbulkan biaya. ...
Article
Full-text available
Biaya adalah komponen mendasar untuk setiap proyek konstruksi. Hasil pengamatan menunjukkan bahwa pembengkakan biaya (cost overrun) adalah fenomena utama dalam industri konstruksi di seluruh dunia. Oleh karena itu, penting untuk mengetahui faktor yang menyebabkan pembengkakan biaya. Tujuan penelitian ini adalah untuk mengidentifikasi faktor-faktor utama teratas terbaru penyebab pembengkakan biaya proyek konstruksi baik di negara maju dan berkembang (khususnya) menurut banyak variabel selama tiga dekade terakhir. Melalui tinjauan literatur yang komprehensif, yang paling umum dan penyebab pembengkakan biaya yang sering terjadi di seluruh dunia dan negara-negara berkembang khususnya, faktor-faktornya adalah: Kesalahan Estimasi, Penambahan item/VO, Finance, Partisipasi Tim dan lingkungan, masing-masing faktor diprioritaskan menurut kemunculannya dan disimpulkan bahwa faktor penyebab cost overrun sangat bervariasi di seluruh negara. Studi tentang faktor penyebab pembengkakan biaya dalam proyek konstruksi akan memberikan referensi untuk proyek lain yang mungkin dieksekusi dalam keadaan serupa dan juga akan memberikan informasi berharga bagi perusahaan internasional yang bermaksud untuk menyediakan proyek konstruksi.
... As compared to the planned cost, the most significant cost overrun (45%) has been noted in rail projects, whereas 20% for road projects. European projects' cost overruns were lower than North American's [21,42]. Several previous studies have analyzed the project cost performance, commonly for developed countries, but there are very limited studies focused on developing countries. ...
Article
Full-text available
Cost overrun is a common phenomenon observed in construction projects worldwide. It is one of the biggest challenges experienced in the construction industry that leads to overstretched budgets, and directly affects the country's gross domestic product (GDP). This study aims to identify the factors that significantly contribute to cost overrun for road network projects in Egypt during the implementation phase. A survey was conducted to determine the most critical factors affecting the cost overrun for road network construction projects. Results showed that the main factors causing cost overrun are inaccurate cost estimates, design modifications, quantity changes, variation orders, political interference, inflation, specification changes, and change in the scope of work. This study will help decision-makers identify the factors that may affect the costs of future road projects and provide guidelines to mitigate the adverse effects observed during the implementation stage.
... 4 These findings correspond to other findings based on smaller samples, which we take to attest to the robustness of results (see, e.g., Albalate and Bel 2014;Altshuler and Luberoff, 2003;Ansar et al. 2014;Bain 2009;Bain and Wilkins 2002;Cantarelli et al. 2012;Dantata et al. 2006;Federal Transit Administration 2003Flyvbjerg et al. , 2004Fouracre et al. 1990;Gao and Touran 2020;Huo et al. 2018, Leavitt et al. 1993Lee 2008;National Audit Office 1992;Nijkamp and Ubbels 1999;Pickrell 1990Pickrell , 1992Riksrevisionen 2011;Riksrevisionsverket 1994;Walmsley and Pickett 1992;World Bank 1994). 5 We ran the same tests with similar results for a subsample of 327 investments for which data were available for both cost overrun and benefit overrun for each investment. ...
Preprint
Full-text available
Most cost-benefit analyses assume that the estimates of costs and benefits are more or less accurate and unbiased. But what if, in reality, estimates are highly inaccurate and biased? Then the assumption that cost-benefit analysis is a rational way to improve resource allocation would be a fallacy. Based on the largest dataset of its kind, we test the assumption that cost and benefit estimates of public investments are accurate and unbiased. We find this is not the case with overwhelming statistical significance. We document the extent of cost overruns, benefit shortfalls, and forecasting bias in public investments. We further assess whether such inaccuracies seriously distort effective resource allocation, which is found to be the case. We explain our findings in behavioral terms and explore their policy implications. Finally, we conclude that cost-benefit analysis of public investments stands in need of reform and we outline four steps to such reform.
Article
Full-text available
Timely completion is main concern for any project though it is a global issue basically for infrastructure construction projects. It is keen area of interest for Local and International professional and Researchers still it is a problem yet to overcome. The study confined to the Literature review followed by key informant experience through discussion with a view to strengthen the depth analysis of existing knowledge in the area to produce contextual content for smooth operation of infrastructure construction projects.The study revealed that the uniqueness of projects is the main cause of projects delay. We are attempting to reach at society 5.0 where machine will be molding our activities. Our needs and wants would be cared by machine through data produced and sensitized. In this regard, the uniqueness of projects should be decoded after knowing the exact distribution of data on the basis of long-term data assessment. Still, it is one area where several research are needed to convert uncertainty into risk through appropriate risk probability allocation. Archrivals of projects should be systematically stored in project bank for future utilization. This research draws the attention for further empirical research and share solution on similarity of problems existing in infrastructure projects.
Article
Evaluation of time-related risks often leads to changes and adjustments in construction schedules, but this influential process is not extensively mapped in literature because of the unstructured data. This absence causes information deficiency and confusion in communication, which, in turn, affects project planning and control, and hinders knowledge sharing and management. This research aims to investigate the obstacles in project execution with the assumption that a decision tree model can map user needs and trace workflows. To understand user needs, we first recorded the focus group discussions followed by semi-structured interviews, then transcribed them into a combined script of 120,000 words. Meanwhile, the prevailing standards of project execution were diagrammed to define ten interconnected categories and derive the workflow schemes to obtain a novel decision tree model, which was used together with the interview data to validate the probabilities and the weighted significance values. The next step was to quantify and prioritize the exact user needs and data specifications of the industry professionals during project execution. After the verification with the invited expert panel, the results show that the most predominant urgencies of time-related risk management are (1) the decision support tools using automated functions to check schedule logic and evaluating risks, (2) preparation assistance for risk communication, and (3) generation of risk mitigation scenarios. The findings can help to line up software development steps for scheduling programs and improve the quality of the risk response strategies for construction projects. One limitation of the current work is the sample size of interviewees, which could be enhanced by designing a questionnaire based on the findings and survey the construction industry.
Chapter
Full-text available
Most cost-benefit analyses assume that the estimates of costs and benefits are more or less accurate and unbiased. But what if, in reality, estimates of costs and benefits are highly inaccurate and biased? Then the assumption that cost-benefit analysis is a rational way to improve resource allocation would be a fallacy. Using the largest dataset of its kind, we test the assumption that cost and benefit estimates of infrastructure capital investments are accurate and unbiased. We find this is not the case with overwhelming statistical significance. We document the extent of cost overruns, demand shortfalls, and forecasting bias in infrastructure investments. We further assess whether such forecasting inaccuracies seriously distort effective resource allocation, which is found to be the case. We explain our findings in behavioral terms and explore their implications for infrastructure economics and policy. Finally, we conclude that cost-benefit analysis stands in need of reform, and we outline four steps to such reform.
Article
Full-text available
Infrastructure plays a major role in the economic development of countries, especially in Asia which has experienced tremendous growth in recent years. The procurement of infrastructure continues to be characterized by cost overruns resulting in significant academic interest and theoretical propositions on the influential factors. This study contributes to this issue through adoption of pragmatic research methodology involving deterministic statistical analysis of real project data from reports as well as a qualitative analysis of these reports to unearth underlying issues from a thematic analysis. Furthermore, the study design takes a multi-country view towards establishing the role of contextual and geographical influences on cost overrun. An evaluation of 102 major infrastructure projects was performed covering railways, roadways and energy sectors in different regions in Asia. Findings reveal that differences in the propensity for cost overrun are mostly dependent on a type of infrastructure with rail projects being the most likely to overrun budget. Theoretically, propositions on the influence of project contextual factors are tested highlighting the influence of the project size, project type, geographical locations, and the length of implementation period of a project as well as factors related to political, economic, strategic, and competence in infrastructure delivery which vary across countries.
Article
Full-text available
This paper uses an econometric meta-regression analysis framework to examine how the mean percentage cost overruns (MPCOs) of transportation projects reported in the literature vary by study specific characteristics, such as whether it is a road or rail project, the world region where the study was conducted; and the year of publication of the study. The motivation of the study is that there are no other studies in the literature on MPCOs of transportation projects using meta-regression analysis; a method well respected in other fields for summarizing findings from different studies. The data used comprise 127 data points retrieved from 48 studies from across the continents. The results show that (1) cost overruns rather than underruns are prevalent across all types of studies, and the typical cost overrun is, on average, 34% regardless of the transportation infrastructure under consideration; (2) recent studies show significantly lower MPCOs as compared to earlier studies; (3) published studies in scientific journals or conference proceedings show higher MPCOs than unpublished reports; (4) in comparison with Asian studies and those of the rest of the world, European and North American studies show significantly lower MPCOs, but Australian studies are not significantly different; (5) road projects show significantly lower MPCOs than rail projects regardless of the region of study; and (6) European and North American roads show higher MPCOs than rails in Asia and the rest of the world. We conclude that the magnitudes of MPCO results from individual studies should not be generalized as the study specific characteristics matter. Insights offered by this study can, therefore, serve as a reference when reporting results from individual studies.
Article
Full-text available
Reference class forecasting is a method to remove optimism bias and strategic misrepresentation in infrastructure projects and programmes. In 2012 the Hong Kong government’s Development Bureau commissioned a feasibility study on reference class forecasting in Hong Kong – a first for the Asia-Pacific region. This study involved 25 roadwork projects, for which forecast costs and durations were compared with actual outcomes. The analysis established and verified the statistical distribution of the forecast accuracy at various stages of project development, and benchmarked the projects against a sample of 863 similar projects. The study contributed to the understanding of how to improve forecasts by de-biasing early estimates, explicitly considering the risk appetite of decision makers, and safeguarding public funding allocation by balancing exceedance and under-use of project budgets.
Article
Full-text available
Back cover text: Megaprojects and Risk provides the first detailed examination of the phenomenon of megaprojects. It is a fascinating account of how the promoters of multibillion-dollar megaprojects systematically and self-servingly misinform parliaments, the public and the media in order to get projects approved and built. It shows, in unusual depth, how the formula for approval is an unhealthy cocktail of underestimated costs, overestimated revenues, undervalued environmental impacts and overvalued economic development effects. This results in projects that are extremely risky, but where the risk is concealed from MPs, taxpayers and investors. The authors not only explore the problems but also suggest practical solutions drawing on theory and hard, scientific evidence from the several hundred projects in twenty nations that illustrate the book. Accessibly written, it will be essential reading in its field for students, scholars, planners, economists, auditors, politicians, journalists and interested citizens.
Article
This study investigates the characteristics of the cost performance of megatransport infrastructure projects in Hong Kong and performs dependence analysis of cost overruns. It is based on 57 samples of road, rail, bridge, and tunnel projects over the period of 1985-2015 and pursues the following two objectives: (1) it measures the characteristics of cost overruns in Hong Kong megatransport infrastructure projects using case data and statistical analysis method; and (2) it examines three independent explanatory variables (i.e., project type, size of the project, and length of the project implementation period) and their statistical relationship with cost overruns to determine whether there exist differences between Hong Kong infrastructure projects and the worldwide findings. The results indicate that, in Hong Kong, projects on average have a cost escalation of 39.18% and the average cost overrun is 34.83% for rail, 32.52% for road projects, and 37.48% for fixed-link (i.e., bridges and tunnels) projects. The year of the decision to build has no effect on cost escalation. The major findings concerning the dependence of cost escalations are as follows. In terms of project type, rail projects are most prone to cost increase, followed by fixed-links, whereas road projects are the least vulnerable to cost escalations. Cost overruns have no significant relationship with project size, but for road projects, smaller scale projects tend to be more prone to larger cost overruns. The cost overruns strongly depend on the length of the implementation phase; cost escalations increase 1.35% with each additional year of preconstruction period. This indicates that the preconstruction phase length can be viewed as a better indicator of cost escalations compared to the construction length. This can be an essential contribution to the existing literature on cost overruns as the stage in which the infrastructures are most prone to cost escalations is considerably narrowed down. The practical implications of this study are that decision makers should focus on this period and pay more attention to delays and longer preconstruction phases if they want to determine the causes and solutions of cost overruns.
Article
Building information modeling (BIM) is one of the most promising recent developments in the architecture, engineering, and construction (AEC) industry. With BIM technology, an accurate virtual model of a building is digitally constructed. This model, known as a building information model, can be used for planning, design, construction, and operation of the facility. It helps architects, engineers, and constructors visualize what is to be built in a simulated environment to identify any potential design, construction, or operational issues. BIM represents a new paradigm within AEC, one that encourages integration of the roles of all stakeholders on a project. In this paper, current trends, benefits, possible risks, and future challenges of BIM for the AEC industry are discussed. The findings of this study provide useful information for AEC industry practitioners considering implementing BIM technology in their projects.