Optimal Productivity in Labor-Intensive Construction

Operations: Pilot Study

Krishna P. Kisi 1; Nirajan Mani 2; Eddy M. Rojas 3; and E. Terence Foster 4
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Optimal productivity is the highest sustainable productivity level achievable under good management and typical working
conditions. Accordingly, optimal productivity provides the foundation for determining the absolute efficiency of construction operations
because an accurate estimate of optimal productivity enables the comparison between actual versus optimal rather than actual versus historical
productivity. This research contributes to the current body of knowledge by introducing a two-prong strategy for estimating optimal pro-
ductivity in labor-intensive construction operations and by applying this strategy to a pilot study on the replacement of electrical lighting
fixtures. The first prong, or top-down approach, estimates the upper limit of optimal productivity by introducing system inefficiencies into the
productivity frontier—the productivity achieved under perfect conditions. This study uses a qualitative factor model to identify this upper
limit. The second prong, or bottom-up approach, estimates the lower limit of optimal productivity by taking away operational inefficiencies
from actual productivity—productivity recorded in the field. A discrete-event simulation model provides this lower-limit value. An average of
the upper and lower limits yields the best estimate of optimal productivity. This paper reviews relevant literature, presents the details of both
the top-down and bottom-up approaches, analyzes the data from a pilot project, evaluates the feasibility of this two-prong strategy, and finally
provides a novel framework for project managers who want to accurately estimate the optimal productivity of their labor-intensive
construction operations. DOI: 10.1061/(ASCE)CO.1943-7862.0001257. © 2016 American Society of Civil Engineers.
Author keywords: Productivity; Construction management; Simulation; Methodology; Labor and personnel issues.

Introduction Economists defined labor productivity as the ratio of total

product output to total labor input (Hanna et al. 2005) or simply
Productivity is one of the most important and influential basic var- the ratio of output to input (Rojas and Aramvareekul 2003a).
iables governing economic production activities (Singh et al. 2000; The U.S. Bureau of Labor Statistics (2016) defined productivity
Tangen 2005). Higher productivity levels allow constructors to si- as the real output per hour worked. In contrast, productivity has
multaneously increase profitability, improve competitiveness, and also been defined traditionally as the ratio of input to output, where
pay higher wages to workers while completing activities earlier input refers to an associated resource (usually, but not necessarily,
(Rojas 2008). Though commonly used, the term productivity is expressed in person-hours) and output refers to the real output in
often poorly defined or used interchangeably with profitability and creating economic value (Dozzi and AbouRizk 1993). There is
performance, which explains why productivity has become an no consensus on which ratio (output/input or input/output) defines
ambiguous concept that seems to depend on the reviewer’s point productivity, which explains why the term lacks a standard defini-
of view and the context in which it appears (Pekuri et al. 2011). tion (Thomas and Mathews 1985).
Jarkas (2010) argued that construction productivity mainly depends Park (2006) outlined the two forms of productivity: the first
on human effort and performance. However, Haskell (2012) stated form—i.e., output/input—appears widely in the construction in-
that “productivity in the construction industry is largely unmeas- dustry and the existing literature; the second form—i.e., input/
ured, and those measures that do exist are contradictory and con- output—usually applies to estimations. Yi and Chan (2014) ex-
flicting.” Haskell also argued that even if productivity is defined as plored the implications of the different definitions of construction
aggregate output divided by aggregate input, the term remains one labor productivity and concluded that hourly output provided the
of the most-controversial and least-understood concepts. most-reliable measurement of productivity for construction activ-
ities. For this reason, the current paper uses hourly output as an
1
Assistant Professor, Dept. of Engineering Technology, Missouri appropriate unit of measurement for productivity analysis.
Western State Univ., 4525 Downs Dr., St. Joseph, MO 64507 (correspond- Many studies have assessed the performance of the construction
ing author). E-mail: [email protected]
2 industry, primarily from a labor productivity perspective (Allen
Assistant Professor, Dept. of Industrial Technology, Fitchburg State
Univ., 160 Pearl St., Fitchburg, MA 01420. E-mail: nmani@fitchburgstate 1985; Allmon et al. 2000; Rojas and Aramvareekul 2003a, b;
.edu Thomas et al. 1990; Yi and Chan 2014). However, given construc-
3 tion’s diverse and unique operations, labor productivity becomes
Dean, School of Engineering, Univ. of Dayton, 300 College Park,
Dayton, OH 45469. E-mail: [email protected] extremely difficult to measure due to the heterogeneity of the indus-
4
Associate Director, Durham School of Architectural Engineering try’s outputs and inputs. Unfortunately, the lack of reliable means
and Construction, Univ. of Nebraska-Lincoln, 1110 S. 67th St., Omaha, for evaluating the efficiency of labor-intensive construction opera-
NE 68182. E-mail: [email protected]
tions makes it more difficult for the construction industry to im-
Note. This manuscript was submitted on May 19, 2016; approved on
August 22, 2016; published online on October 14, 2016. Discussion period prove productivity. As Drucker (1993) articulated, “If you can’t
open until March 14, 2017; separate discussions must be submitted for measure it, you can’t manage it.” Studies such as those by Alinaitwe
individual papers. This paper is part of the Journal of Construction En- et al. (2007), Jarkas and Bitar (2012), Rojas and Aramvareekul
gineering and Management, © ASCE, ISSN 0733-9364. (2003b), Thomas and Yiakoumis (1987), and Thomas et al. (1999)

© ASCE 04016107-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 have all identified and analyzed factors that affect productivity; they idea when they presented several factors involved in the processes
found management factors, labor characteristics, site conditions, of construction change over time, namely, that productivity cannot
and weather conditions are all primary factors affecting labor pro- be easily judged by the same data or information that was docu-
ductivity. Similar studies have focused on the relationship among mented a decade or more ago—tools, materials, and methods
factors affecting labor productivity and have presented methods for change greatly with each innovation. Without a method for evalu-
measuring their effects (Sonmez and Rowings 1998; Tam et al. ating productivity against an objective standard, the practice of
2002; Thomas et al. 1990; Thomas and Sakarcan 1994). These stud- benchmarking against historical averages will continue to remain
ies used qualitative, quantitative, or mixed-method approaches to commonplace in the industry, regardless of how flawed the process
address the diverse nature of construction productivity measure- is acknowledged to be.
ment (Panas and Pantouvkis 2010). Optimal productivity, defined as the highest sustainable produc-
Studies have also presented theoretical models to forecast con- tivity achievable in the field under good management and typical
struction labor productivity (Dissanayake et al. 2005; Fayek and field conditions (Son and Rojas 2011), has the potential to provide
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

Oduba 2005; Lu et al. 2000; Srinavin and Mohamed 2003). These an objective benchmark for gauging performance. An accurate es-
models take advantage of a variety of techniques, including simu- timation of optimal labor productivity allows project managers to
lation, artificial intelligence, expert systems, factor models, and determine the efficiency of their labor-intensive construction oper-
statistical and regression approaches. Each technique has its own ations by comparing actual versus optimal rather than actual versus
merits and shortcomings. For example, Srinavin and Mohamed historical productivity. However, to date, no substantive model for
(2003) developed a model using regression analysis for qualitative estimating optimal productivity has been proposed in the construc-
evaluation of the impact of different factors on construction labor tion domain.
productivity. However, since a regression equation is limited to This study builds a novel framework and its fundamental con-
certain variables, the approach did not allow for the subjective cepts by introducing a two-prong strategy for estimating optimal
evaluation of qualitative factors. In response to this limitation, ex- productivity in labor-intensive construction operations; then, the
pert systems have been widely used to quantify this kind of sub- text reports on a pilot study performed to evaluate this framework’s
jective evaluation. Yi and Chan (2014) performed a critical review feasibility. Determining the feasibility of the methodology is the
of the labor productivity research published in construction journals first step toward a future tool that will apply to field practitioners.
and claimed that expert systems are superior to statistical models In particular, this paper identifies major factors affecting labor pro-
because of their flexibility in adapting to different project contexts. ductivity, classifies them by affinity groups, presents the effects of
Combined, these studies provided insights into the qualitative mea- inefficiencies on optimal productivity, applies a qualitative factor
sure of productivity forecasting. model and simulation techniques to estimate the magnitude of those
Many studies have used probability approaches to model con- inefficiencies, and evaluates the feasibility of the proposed two-
struction data regarding productivity analysis (Huang and Hsieh prong strategy for estimating optimal labor productivity through
2005; Rustom and Yahia 2007; Smith 1998). Smith (1998) used a pilot study. The outcomes of this study reveal that the proposed
discrete-event simulation to model construction operations accord- methodology is adequate when applied to a simple construction
ing to the probability distribution of each event involved in a operation with sequential tasks.
construction activity. Hamm et al. (2011) presented an optimiza-
tion framework to determine efficient construction schedules by
linking discrete-event simulations with optimization concepts. Theoretical Framework
Zhang (2013) presented an alternative discrete-event simulation
This study is an extension of the research performed by Son and
method for estimating construction emissions by addressing un-
Rojas (2011), who identified some basic productivity concepts, as
certainties and randomness as well as complex interactions. These
shown in Fig. 1. All of the definitions of the terms shown in Fig. 1
studies provided insights into the quantitative measure of produc-
stem from Son and Rojas (2011). Specifically, the figure graphi-
tivity modeling.
cally depicts the dynamic relationships among productivity levels
Even with all of these productivity studies, there is a general
at different states of time t. Once the project achieves its steady-
consensus that current construction data do not provide an adequate
state phase, the levels within the approach can be estimated. The
or accurate measure of productivity (BFC 2006). In an attempt to
productivity frontier is the theoretical maximum productivity that
evaluate the efficiency of labor-intensive construction operations,
would be achieved under perfect conditions. Perfect conditions in-
project managers typically compare actual with historical produc-
clude ideal weather, a highly motivated and productive workforce
tivity for equivalent operations. However, such an approach toward
examining productivity only provides a relative benchmark for
efficiency and may lead to the characterization of operations as ob-
System Inefficiency
jectively efficient when in reality such operations may be only com-
Productivity of an Activity

Productivity Frontier
parably efficient. For example, actual productivity equal to 95% of
average historical productivity does not necessarily mean that the Optimal Productivity
operation is efficient but only that the efficiency of the operation is Actual Productivity
in line with historical averages, which may be well below optimal Operational Inefficiency
productivity (Kisi et al. 2014). Therefore, if a company begins from
an inefficient benchmark and compares current productivity with
such historical data, the result is simply productivity checked
against inefficient work. Learning Phase Steady State Phase
Song and AbouRizk (2008) stated that there is currently no sys-
tm tn
tematic approach for measuring and estimating labor productivity, Time
an assertion that implies that there are no benchmarks or standards
Fig. 1. Productivity dynamics (adapted from Son and Rojas 2011,
to validate historical data as suitable for either estimating or evalu-
© ASCE)
ating productivity. Liberda et al. (2003) further complicated this

© ASCE 04016107-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 with perfect artisanship, optimal materials, optimal equipment, undisturbed work period, their approach is very hard to implement
experiencing no design errors or interference from other trades, and in practice. Furthermore, even though the types of productivity
enjoying a perfect understanding of design intent, among others inefficiencies are distinct, none of the established techniques di-
(Son and Rojas 2011; Mani et al. 2014). Below the productivity vide inefficiencies into multiple categories. This lack of separa-
frontier appear optimal productivity and actual productivity. Opti- tion makes calculations challenging since some inefficiencies
mal productivity is the maximum sustained productivity obtained (e.g., system inefficiencies) by definition need qualitative evaluation
under good management and typical working conditions, while ac- whereas others (e.g., operational inefficiencies) can be evaluated
tual productivity is the productivity achieved in the field. All three quantitatively.
of these levels provide the basic foundation for the two-prong strat- With these considerations in mind, this paper tackles the chal-
egy for estimating optimal productivity in labor-intensive construc- lenge of properly estimating the value of optimal productivity by
tion operations. demonstrating how this estimation can be achieved using two in-
Optimal productivity falls between system inefficiencies and dependent but complementary approaches. While the long-term
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

operational inefficiencies (Fig. 1). System inefficiencies refer to goal of this work is a direct application in the field, this paper rep-
the collection of factors outside the control/influence of project resents a first step in the process of developing such a productivity-
managers such as environmental conditions (high humidity or cold monitoring methodology. The first prong, or top-down approach,
or hot temperatures), workers’ health, absenteeism driven by health estimates loss due to system inefficiencies and incorporates them
or family issues, interference from other trades, and design errors, into the productivity frontier. The result of this top-down approach
among others; these inefficiencies decrease productivity but are not gives the upper limit estimation of optimal productivity (OPUL ).
the result of poor management/labor shortcomings. Operational in- The second prong, or bottom-up approach, determines loss due to
efficiencies are the result of project managers’ suboptimal strategies operational inefficiencies by quantifying and subtracting detrimen-
regarding issues such as sequencing, crew size and composition, tal elements from actual field productivity and thereby increasing
construction methods, and quality assurance/control, among others the limit of productivity. This bottom-up approach results in the
(Alinaitwe et al. 2007; Dai et al. 2009; Enshassi et al. 2007; Kadir lower limit estimation of optimal productivity (OPLL ). An average
et al. 2005; Rivas et al. 2011); these inefficiencies reveal places of the upper and lower limits provides the best estimate of optimal
where management can improve workers’ productivity. As shown productivity for labor-intensive operations.
in Fig. 1, actual productivity is typically below optimal productivity The top-down and bottom-up approaches can be easily under-
(though these two levels can overlap occasionally), whereas the stood from Fig. 2, which graphically depicts different productivity
productivity frontier is above optimal productivity because, by def- levels once the steady-state condition is reached for a construction
inition, the productivity frontier is the theoretical maximum value, operation. As illustrated (Fig. 2) and described by Son and Rojas
and system inefficiencies in construction operations always exist. (2011), optimal productivity (OP) lies between the productivity
The distance between actual and optimal productivity and between frontier (PF) and actual productivity (AP). The difference between
optimal productivity and the productivity frontier quantify the ex- OP and PF quantifies the system inefficiencies (Δsi ), while the
tent of the operational and system inefficiencies, respectively. difference between OP and AP quantifies the operational inefficien-
Son and Rojas (2011) explained optimal productivity using both cies (Δoi ). Since inefficiencies cannot be measured, they must be
optimistic and conservative perspectives: (1) productivity is opti- estimated. The estimate of the system inefficiencies is represented
mistically forecasted when the estimated values are higher than by Δsi0 , while Δoi0 represents the estimate of the operational ineffi-
the optimal productivity and (2) productivity is conservatively fore- ciencies. The upper limit of optimal productivity (OPUL ) results
casted when the estimated values are lower than the optimal pro- from subtracting the estimate of system inefficiencies, Δsi0 , from
ductivity. These forecasting perspectives will often depend upon the PF; concurrently, the lower limit of optimal productivity
historical averages and personal judgment. However, this paper in- (OPLL ) emerges from summing the estimate of operational ineffi-
troduces a distinct perspective on the topic and argues that optimal ciencies, Δoi0 , with AP
productivity depends upon the presence of system and operational
OPUL ¼ PF − Δsi0 ð1Þ
inefficiencies.
The diverse types of inefficiencies result in loss of productivity.
While quantifying lost labor productivity remains challenging, OPLL ¼ AP − Δoi0 ð2Þ
widely used approaches include measured mile (Loulakis and
Santiago 1999; Zink 1986, 1990), and baseline method (Thomas While these equations provide the means of calculating the op-
et al. 1999, 2002, 2003a; Thomas and Sanvido 2000a, b). Measured timal productivity limits, the parameters of each of these values
mile compares periods of a project that have been impacted by requires an innovative quantification, described subsequently.
change to those that have not. This process requires a production
period free of productivity disruptions, which can be challenging
to find in some complicated projects. Alternatively, the baseline Methodology
method uses a period of time when the contractor performs his
or her best productivity. While these approaches find favor among Fig. 2 helps readers visualize the process of estimating optimal
many researchers and managers, the assumptions made in these two labor productivity by using a two-way approach: a top-down ap-
methods have several drawbacks. For instance, measured mile re- proach illustrates steps to solve the relationship developed in
quires that the referenced productivity comes from a continuous, Eq. (1), whereas a bottom-up approach illustrates steps to solve
uninterrupted period, and the baseline method assumes an arbitrary Eq. (2). Fig. 3 shows a methodology for these two approaches
10% as the basis of measuring average productivity. Subsequently, by illustrating the work-flow of the processes and the methods
Ibbs and Liu (2005) discussed the drawbacks, compared the mea- and techniques involved.
sured mile and baseline method, and introduced a k-means statis- The method for estimating loss due to inefficiencies is a funda-
tical clustering technique as a new approach for baseline calculation mental decision that governs which data should be gathered and
of inefficiencies. Despite the advantages of their clustering tech- which modeling techniques may be used. Data may be directly
nique, as a purely statistical method that requires data from an measured in the field and quantified easily, or they may be of a

© ASCE 04016107-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 factors with similar behavior/nature appear under the same cat-
PF (Productivity Frontier) egory during factor classification.
Factors causing system inefficiencies are subjective in nature
OPUL (Upper Limit of and therefore require qualitative analysis. Models have been devel-
Optimal Productivity) oped to measure these types of qualitative factors (Thomas and
Productivity

OP (Optimal Productivity) Sakarcan 1994; Christian and Hachey 1995; Kindinger and Darby
2000; Srinavin and Mohamad 2003; Dai et al. 2009); based on
OPLL (Lower Limit of these different methods of measuring qualitative factors, this study
Optimal Productivity)
developed a qualitative factor model (QFM) to evaluate loss due to
AP (Actual Productivity) system inefficiencies. The QFM technique, as shown in Eq. (3),
estimates loss due to system inefficiencies
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

tn tn+1 n X
X m  
Si P i
Time Δsi0 ¼ ΔðPF−OP
0
LL Þ
× Wz ð3Þ
z¼1 i¼1
TSi
Fig. 2. Different productivity levels

where Δsi0 = estimate of productivity loss due to system inefficiency;

0
ΔðPF−OP = estimate of difference between productivity frontier
LL Þ
subjective nature that needs to be evaluated qualitatively. For ex- and the lower limit of optimal productivity—OPLL is a quantitative
ample, losses such as sitting idle, wasting time on discussions, and value described later; n = number of parameters (work zones for
frequent breaks all fall under operational inefficiencies and can z series or productivity factors for i series); m = numbers of pro-
easily be measured in the field. However, factors affecting workers’ ductivity factors; z = work zone (classrooms, lockers, corridor/
performance—such as change in productivity due to temperature, hallway, etc.); i = number of system inefficiency factors in zone z;
humidity, time of day, and interference from other trades—are part Si = severity score of individual factor I; Pi = probability of indi-
of system inefficiency losses that are difficult to measure in the vidual factor I; TSi = total severity score (sum of severity scores of
field. Therefore, this research uses two modeling techniques, one all factors in zone z); and W z = relative weights of each zone.
for evaluating system inefficiencies and the other for measuring Qualitative definitions of severity for each of the factors are de-
operational inefficiencies. termined by a severity ranking score from 0 to 5, with 0 = no im-
This paper extracted lists of system inefficiency factors pact; 1 = very low impact; 2 = low impact; 3 = medium impact;
from papers published since 1985 in four top engineering and 4 = high impact; and 5 = very high impact. Probabilities are used
management-focused journals, namely the Journal of Construction to establish the likelihood of factors being present. As an example,
Engineering and Management, Journal of Management in Engi- a severity score of 5 with a 0.4 probability means that the factor has
neering, Journal of Civil Engineering Management, and Construc- a probability of occurring 40% of the time, and when it occurs, it
tion Management and Economics. Dozens of factors that affect has a very high impact on labor productivity. Depending upon the
labor productivity appear in existing papers; to consolidate the var- nature of the work environment, the severity score may vary across
iables at play in this research, this research team selected those work zones. In addition, the number of tasks (e.g., number of bulbs
factors mentioned repeatedly in the four journals. Many of the fac- installed during a lighting installation) may also differ from one
tors mentioned in papers have similar nomenclature; as such, analo- zone to another. A relative weight for each zone is calculated based
gous factors are represented as a single factor for convenience, and on how many tasks are completed in a particular zone.

Fig. 3. Research methodology

© ASCE 04016107-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 The two final inputs for Eq. (3) are the value of the productivity the simultaneous involvement of workers on multiple activities
frontier (PF) and the lower limit of optimal productivity (OPLL ). all represent noncontributory actions (Heizer and Render 1996).
The equation uses the difference PF − OPLL as a multiplying factor These noncontributory actions are operational inefficiencies, and
to estimate Δsi0 because in extreme cases, the optimal productivity the length of these activities quantifies the operational inefficiencies
upper limit fluctuates from PF to OPLL . For instance, if the system impacting productivity. Therefore, by removing these inefficiencies
inefficiencies are absent (Fig. 2), the OPUL aligns with PF. How- from the synthetic simulation, the resulting productivity solves
ever, if the system inefficiencies are 100% present, then OPUL Eq. (2) and reveals the estimation of the lower limit of optimal
merges with OPLL ; though this scenario is unlikely in construction productivity.
activities, the equation still considers the extreme cases. Therefore, As shown in Fig. 2, the average of the upper and lower limits of
the accuracy of optimal productivity depends on the estimate of optimal productivity yields a best estimate of optimal productivity.
PF and OPLL .
To determine the productivity frontier, this project uses the tech-
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

nique described in Mani et al. (2014) and Mani (2015). Specifi- Pilot Study
cally, one must first determine the length of time involved in
the productivity frontier. The observed duration approach to calcu- This paper evaluated the feasibility of the proposed methodology
lating this time requires users to break down an activity into tasks for estimating optimal productivity through a pilot study. A large
and each task into actions, measure the minimum duration of each electrical constructor conducted an electrical lighting replacement
action, and calculate the shortest overall duration by reassembling project at Omaha, Nebraska. The project involved repetitive proc-
the actions into the task and the tasks into the activity. Since ob- esses during the replacement of ceiling lighting fixtures inside a
served durations may not include the shortest possible duration for school building. Given the fact that feasibility studies evaluate a
a given action, another approach to determining the shortest pos- new methodology against an exemplar task, the repetitive nature
sible duration of an activity is to fit a probability distribution curve of the lighting replacement work suited this paper’s test of the fea-
to the data and estimate the shortest durations for each action using sibility of the proposed methodology; however, as this paper’s con-
the resulting distribution function. Regardless of which of these clusions discuss, this methodology will need further development
two options—reassembled observed shortest durations or statisti- to apply to nonrepetitive processes.
cally estimated shortest durations—yields the shortest duration During this pilot study, data were recorded from five differ-
of the activity, the resulting value becomes the time estimate for ent zones: classrooms, locker room, corridors/hallways, weight/
the productivity-frontier. Users then divide the production rate training room, and family consumer science room. This project in-
by the activity’s shortest duration to find the productivity-frontier cluded multiple tasks such as removal of the existing main frame
value. for the lighting fixtures, removal of the old T-12 fluorescent bulbs,
One of the benefits of determining the productivity frontier removal of the ballast, installation of new Type-2 ballasts, instal-
according to the methodology defined by Mani (2015) is the fact lation of T-8 fluorescent bulbs, and closure of the main outer cover.
that the value represents the theoretical maximum productivity Two electrical workers, namely a veteran and a novice, participated
that would be achieved under perfect conditions. Accordingly, if in the project.
inefficiencies appear in productivity due to such challenges as
fatigue due to time-of-day—which Dai et al. (2009), Rojas and Top-Down Approach
Aramvareekul (2003b), and Thomas et al. (1992) have identified
as detrimental to productivity—these decreases in the productivity Identification of Productivity Factors
values of the workers become subsumed into the system inefficien- The first step of the top-down approach is to identify the major
cies that lower the optimal productivity upper limit. Subsequent factors affecting labor productivity in the construction field for
sections in this paper outline such a defined system and operational the type of construction operation under consideration. A detailed
inefficiencies determined during the performed literature review. literature review from the four top engineering and management-
To then determine the optimal productivity lower limit, one focused journals discussed above identified more than 50 different
must assess the effects of operational inefficiencies on actual pro- factors applicable the installation of lighting fixtures in the pilot
ductivity. This process requires the use of a discrete-event simula- study. These factors are identified and classified in the following
tion (DES) model. DES evaluates an activity by breaking down the section.
activity into tasks and each task into time-measurable actions—
similar to the productivity frontier duration calculation described Classification of Productivity Factors
earlier. These measurable actions require a probability function The identified factors were filtered into manageable groups by ap-
as an input to execute the DES. Therefore, the repetitive nature of plying the affinity grouping technique. This technique creates and
each action is fitted with probability distribution curves that satisfy clusters factors into categories on the basis of their similarity. The
a goodness-of-fit test, and the expression of these curves works as following affinity groups are frequently used in the literature:
an input for the DES, where each action is treated as an event. The • Technical factors, such as uncoordinated, incomplete, and ille-
simulation models the workflow of an activity, and the output of the gible drawings, and complex designs of unusual shapes and
entire simulation can then be compared with actual field values heights (Arditi 1985; Dai et al. 2009; Herbsman and Ellis 1990;
to establish validity. For the purpose of this study, the simulation Rivas et al. 2011; Thomas et al. 1992);
model is considered valid when the output deviation lies within a • Management factors, such as inadequate supervision, manage-
5% margin of error. ment control/ project team, incompetent supervisors, inspection
After validation, all noncontributory actions are removed from delays, overstaffing, and management practices (Alinaitwe
the DES model to generate a synthetic simulation model. Noncon- et al. 2007; Arditi 1985; Dai et al. 2009; Enshassi et al. 2007;
tributory actions are those that are counterproductive or that do not Herbsman and Ellis 1990; Rojas and Aramvareekul 2003b;
contribute to the progress of the tasks at hand. For example, un- Sanders and Thomas 1991; Thomas et al. 1992).
scheduled breaks, frequent relaxation, late starts and early quits, • Site conditions, such as site access, site layout, congestion/
waiting idle, engagement of workers in personal discussions, and interferences, and material handling (AbouRizk et al. 2001;

© ASCE 04016107-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 Makulsawatudom et al. 2004; Rivas et al. 2011; Thomas and Table 1. Severity and Probability Analysis for Productivity Factors
Yiakoumis 1987); Probability Product of
• Environmental conditions, such as cold or hot temperatures, Severity of severity and
high or low humidity, and winter storms (Koehn and Brown score occurrence probability
1985; Thomas et al. 1999; Thomas and Yiakoumis 1987); Zone Factors (Si ) (Pi ) (Si Pi )
• Scheduling issues, such as schedule acceleration, overcrowding Classrooms High humidity 2 0.4 0.8
and/or overmanning, scheduled overtime, shift work, and out- Low temperature 2 0.3 0.6
of-sequence work (Chang et al. 2007; Dai et al. 2009; Hanna Low luminance 2 0.3 0.6
et al. 2005, 2008; Sanders and Thomas 1991); High noise level 2 0.6 1.2
• Coordination issues, such as poor coordination and poor Space congestion 4 0.8 3.2
communication (Arditi 1985; Dai et al. 2009; Koehn and Brown Lockers High humidity 3 0.4 1.2
1986); Low temperature 2 0.5 1.0
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

• Changes and omissions; such as rework and change orders Low luminance 2 0.4 0.8
High noise level 4 0.3 1.2
(AbouRizk et al. 2001; Alinaitwe et al. 2007; Borcherding et al.
Restricted access 2 0.6 1.2
1986; Rivas et al. 2011); Space congestion 3 0.6 1.8
• Project characteristics, such as ownership type, work type, and Corridor/hallway High humidity 1 0.2 0.2
project goals (AbouRizk et al. 2001; Rojas and Aramvareekul Low temperature 2 0.3 0.6
2003b; Thomas et al. 1992); High luminance 2 0.3 0.6
• Labor characteristics, such as labor/worker quantity, quality of High noise level 4 0.4 1.6
craftsmanship, absenteeism, fatigue and health issues (Dai et al. Space congestion 1 0.3 0.3
2009; Koehn and Brown 1986; Rojas and Aramvareekul 2003b; Weight High humidity 2 0.3 0.6
Thomas et al. 1992), craft turnover, skills, experience, motiva- room/training Low temperature 2 0.3 0.6
tion, and worker shortages (Arditi 1985; Enshassi et al. 2007; room Low luminance 2 0.3 0.6
High noise level 3 0.6 1.8
Rivas et al. 2011; Rojas and Aramvareekul 2003b);
Space congestion 4 0.7 2.4
• External conditions, such as project location and external Family consumer High humidity 2 0.3 0.6
conditions, government, economic activity, availability of science room High temperature 2 0.3 0.6
skilled labor, and job availability (Dai et al. 2009; Rojas and High luminance 2 0.3 0.6
Aramvareekul 2003b); High noise level 3 0.4 1.2
• Nonproductive activities, such as waiting idle, working slowly, Space congestion 4 0.6 2.4
doing ineffective work, frequent relaxation, and late starts and
early quits (Borcherding et al. 1986; Dai et al. 2009);
• Tools and equipment, such as unavailability of suitable equip-
ment, lack of tools, and maintenance of power tools (Arditi they were present. Many of the factors from the affinity groups
1985; Dai et al. 2009; Herbsman and Ellis 1990; Sanders and did not manifest in this pilot study since the project was conducted
Thomas 1991); in a controlled environment. For simplicity’s sake, the factors that
• Material factors, such as shortage of materials, difficulty in have either a severity score of zero or a probability of zero do not
tracking material, and poor material quality (AbouRizk et al. appear in Table 1 because the product of the severity and proba-
2001; Arditi 1985; Dai et al. 2009; Enshassi et al. 2007; Sanders bility (Si Pi ) column in the table would be zero and thus would have
and Thomas 1991); and no impact in Eq. (3) for estimating losses due to system inefficien-
• Safety factors, such as lack of site safety resources, incidents, cies. For example, in this pilot study, Pi of factors such as rainfall,
and accidents (Arditi 1985; Dai et al. 2009; Thomas et al. 1992). excessive hot temperature, and other severe weather conditions
These affinity groups provided the basis for classifying ineffi- were absent, which means that the product of Si Pi is zero. There-
ciencies, which helped in generating a questionnaire for collecting fore, Table 1 is a simplified version excluding all the factors that
experts’ opinions on severity and probability scores. These identi- have product values of zero.
fied factors were also grouped according to the kind of inefficiency Although the impact due to external weather condition was null
they represent: system inefficiencies (Δsi ) or operational ineffi- (not present in Table 1 with null values) there were certain varia-
ciencies (Δoi ). Factors such as weather, humidity, temperature, tions in temperature and humidity among different zones inside the
etc.—which are not under the control or influence of project building. For example, due to students taking showers, humidity
managers—equated to system inefficiencies. On the other hand, was higher in the locker room than in other rooms. These variations
factors such as unscheduled breaks, site congestion, poor inspec- were considered during the analysis. However, given that this pilot
tion, etc.—which are under the control of project managers— study took place in an inside environment, the severity score for
signified operational inefficiencies. temperature, humidity, and lighting are relatively low, as indicated
in Table 1.
Estimation of Upper Limit of Optimal Productivity During the pilot study, classes were running at the school.
Estimation of the upper limit of optimal productivity involves the Therefore, the probability and severity score for noise level were
quantification of losses due to system inefficiency factors. These observed to be high due to the presence of students. As expected,
factors are qualitative in nature, so their quantification is challeng- the probability and severity score for space congestion are high in
ing. As per the methodology shown in Fig. 3, this research imple- classrooms because the working space was furnished, which caused
mented a QFM technique to overcome this challenge. obstruction. This reality caused the worker to stretch his arms to
In this pilot study, five experts—three researchers, one supervi- move materials, which impacted his movement even though he
sor, and a worker—listed the factors causing system inefficiencies. maintained the same steps in replacing the light fixtures.
Experts then provided probability of occurrence (Pi ) and severity In addition to the severity product analysis, two other inputs for
scores (Si ) for each factor depending on how likely the factors are Δsi0 in Eq. (3) are the value of the productivity frontier (PF) and
and how severely the factors would impact productivity if indeed the lower limit of optimal productivity (OPLL ). To determine the

© ASCE 04016107-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 productivity frontier, this study uses a time and motion study per-
formed by Mani (2015), which uses the same dataset as this paper.
The methodology and results of the productivity frontier are dis-
cussed in Mani (2015), but in summary, the estimated productivity
frontier for the pilot study’s lighting installation was 22.32 stations
per hour, which was estimated based on several days of data col-
lection. This value represents the theoretical highest productivity
achievable in the field if the conditions were perfect.
On the other hand, the estimate of the lower limit of optimal
productivity that assesses the effects of operational inefficiencies
is obtained from simulation, which is explained in the later
“Bottom-Up Approach” section. Though the use of OPLL in the
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

quantification of Δsi0 may appear to confound the results, this ap-

proach is best because Eq. (3) holds true in extreme cases of system
inefficiencies. The remaining variables at play in Eq. (3) offset the
extreme case contained within the use of OPLL and provide a rea- Fig. 4. Hierarchical structure for lighting fixtures replacement activity
sonable estimation of the OPUL .

Bottom-Up Approach The hierarchical structure was analyzed using the video footage.
Using the video data, a time and motion study similar to Shahidul
Data Collection and Shazali (2011) and Finkler et al. (1993) revealed the time spent
Canon XF100 (Oita, Japan) camcorders recorded every detail of the on each contributory and noncontributory action. Each action con-
lighting-replacement activity. One or more camcorders were uti- sists of sequential movements and includes an expected duration to
lized per space availability to capture movements from different sufficiently accomplish the action. The explanations of assorted
angles. One of the benefits of these video recording is that the data steps involved in accomplishing the eight actions are described next
can be reviewed whenever required. This paper’s team also took for better understanding.
field notes to document workers’ start times, break times, finishing The process of the fluorescent bulb replacement task proceeds
times, etc. with the glass frame removal action. Removing the cover consists
In this replacement of electrical lighting fixtures project, differ- of unscrewing or unlocking one edge of the outer cover of the
ent types of ballasts and fluorescent bulbs were used. For consis- ceiling light fixture, letting it open to one side, and subsequently
tency, only activities involving Type-2 ballast and T8 fluorescent allowing the other end to hang, all while permitting enough space to
bulbs were considered in this study. Type-2 ballast can supply continue onto the second action. The second sequential action is old
power for up to two fluorescent bulbs, so the tasks associated with bulb (T-12) removal, whose duration includes reaching for the
this activity included each bulb. bulbs, twisting the bulbs to unlock, and then dumping them into
Two workers, a veteran and a novice, participated in the project. the collection box that are hung on either side of the scaffold.
They completed the majority of the work independently. In order to The third sequential action is ballast cover removal, which involves
maintain the homogeneity and consistency of the data, the duration reaching out hands to the ballast cover, unscrewing or unlocking
of tasks performed by the veteran worker were recorded in greater the cover, removing the cover, and safely placing that cover over
detail for the analysis. In particular, the veteran worker completed the scaffold so that it is readily available. The fourth sequential ac-
62 stations at five different zones. Each station included replacing tion is old ballast removal, which includes the worker reaching out
one Type-2 ballast and two T-8 fluorescent bulbs. Video data from hands for the ballast, disconnecting circuit wires, inserting push-in
the 62 stations—i.e., data representing the installation of 62 Type-2 wire connectors, unscrewing all screws, removing the old ballast,
ballasts and 124 T8 fluorescent bulbs—were captured for a time and discarding the old ballast into a collector bin placed over the
and motion study at different zones. scaffold. The unscrewing involves use of manual and powered tools
according to access in the given location. Relative to the duration
Data Analysis of other actions, removing old ballast has the longest duration.
The analysis was based exclusively on the data collected for the Similarly the new ballast installation, ballast cover closure, new
veteran worker. These data particularly document the worker func- bulb (T-8) installation, and frame cover closure—as depicted in
tioning under analogous working conditions—e.g., using similar Fig. 4—proceeds in reverse order to finish the installation task.
tools and equipment, working at similar heights, and installing the The duration of all actions were recorded in a spreadsheet
same type of ballast and bulbs. and analyzed. The recorded data included contributory and non-
A hierarchical structure broke down activities into tasks and ac- contributory actions, so each action was classified as to whether
tions. Fig. 4 shows the hierarchical analysis used. This pilot study it fell into the contributory or noncontributory category. Then,
selected the activity of replacement of electrical lighting fixtures for simulations were performed to estimate the lower limit of optimal
analysis given its homogeneity across the construction project. The productivity.
activity broke down into four tasks: (1) site preparation, (2) fluores-
cent bulb replacement, (3) waste management, and (4) documenta- Estimation of Lower Limit of Optimal Productivity
tion. The task of fluorescent bulb replacement received further Fig. 5 shows the workflow of the actions involved in the fluorescent
analysis given its consistency and the number of repetitions avail- bulb replacement task that was modeled in a DES. The model is
able; to accomplish this analysis, the task broke down further into very simple and consists of only sequential actions involved in
eight actions: (1) glass frame removal, (2) old bulb (T-12) removal, the task. For instance, entities arrive at the station where light fix-
(3) ballast cover removal, (4) old ballast removal, (5) new ballast tures need to be replaced; in the model, entities are new bulbs and
installation, (6) ballast cover closure, (7) new bulb installation, and new ballasts. The veteran worker processes the actions. When the
(8) glass frame closure. worker finishes the replacement task, the worker moves to the next

© ASCE 04016107-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 Parts number of observations and the y-axis represents the duration.
Probability Mathematical The mathematical expressions represent the best curve based on
Arrival
Distributions Expression
significance p-values, square errors, and visual inspection. Fig. 5
shows fitted curves along with the mathematical expression that
Frame Cover represent each curve as it was found using Arena’s Input Analyzer
Exponential
Removal
2.5+EXPO(2.11)
tool.
Recorded data were modeled into the DES to represent process
workflow. The model was then verified by comparing the sequen-
Old Bulb T12 Weibull ces of the actions in the model with the actual sequences in the
Removal 9.5+WEIB( 7.41, 1.17) field. After verifying sequences, the simulation was run under two
scenarios: actual (including noncontributory actions) and synthetic
(excluding noncontributory actions). The actual scenario provided
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

Ballast Cover
Weibull model validation while the synthetic scenario estimated the lower
Removal 4.5 + WEIB(10.4, 1.94) limit of the optimal labor productivity.
Simulation results from the actual scenario were compared
Old Ballast against field data to calculate the deviation and see if the deviation
Weibull is within a reasonable limit (5%). The veteran worker completed
Removal 70+WEIB( 28.2,1.38 )
62 stations in the 5 different work zones in 279 min, which equals
13.33 stations per hour. The simulation results from the actual
New Ballast scenario show a completion rate of 13.07 stations per hour. These
Weibull
Installation 51.5+WEIB( 20.8, 1.14) values represent less than 2% deviation from the recorded field val-
ues. Thus, the simulation model was validated with face validity;
Ballast Cover
the technique is used in determining if the logic in the conceptual
Gamma
Closure 9.5+GAMM(7.18, 1.49) model is correct and if a model’s input–output relationships are
reasonable (Lucko and Rojas 2010; Sargent 2010).
After model validation, noncontributory actions were eliminated
New Bulb T8 Gamma
from the model. The model was run again for the synthetic sce-
Installation 12.5+GAMM(5.59,2.25) nario. The simulation results for the synthetic scenario show a com-
pletion rate of 14.32 stations per hour (4.19 min per station). This
outcome implies that the loss due to operational inefficiency (Δoi0 )
Frame Cover Weibull is 1.25 stations per hour. The results of the lower-limit optimal pro-
Closure 2.5+WEIB( 2.05, 1.24) ductivity show a possible 7.4% improvement in productivity over
the actual scenario.
The mean values from these two models were compared to de-
Departure
termine if they were statistically different. Using Arena’s output
analyzer and a 95% confidence interval, a paired-t-means com-
parison test of the null hypothesis that both means were equal
Fig. 5. Discrete-event simulation model of fluorescent bulb replace- concluded that the means were different.
ment task The result from this modified simulation (synthetic scenario)
represents the estimate of the lower limit of optimal productivity,
rather than the optimal productivity itself. The reason for this
choice is that even when noncontributory actions are excluded, a
station. The time taken for the worker to complete each action
simulation model that relies on field data for validation can neither
is recorded.
eliminate all of the operational inefficiencies embedded in a con-
In order to get a realistic model, the process is based on actual
struction task nor adjust for the system inefficiencies that play a
field data; the more real data that are collected, the more realistic
the model becomes (Smith 1998). Each action shown in Fig. 5 has role in optimal productivity. Thus, the simulation output consider-
62 repetitions of field data in 5 different work zones. The duration ing only contributory actions (direct and indirect work) represents
of each action was recorded in spreadsheets by playing the video the lower limit of optimal productivity (OPLL ). The exclusion of
several times and observing the time. To cross validate the times, a any additional operational inefficiencies buried within the simula-
researcher also used a stopwatch to measure action lengths. tion outcome would only raise the OPLL value and are therefore
Once the durations of each action were recorded, this study de- contained within the range between OPLL and OPUL .
termined which probability distribution fit the sample data. There
are many techniques available to fit distributions to the sample data; Estimation of Optimal Productivity
these are usually goodness-of-fit tests or heuristic graphical tech-
niques. This study used Arena simulation software, which supports Substituting all the required parameters in Eq. (3), the estimate
a wide variety of probability distributions including uniform, of productivity loss due to system inefficiencies (Δsi0 gives 2.98
normal, lognormal, beta, gamma, Weibull, and Erlang (Kelton et al. stations per hour, which when subtracted from the productivity
2010). Input Analyzer in the Arena software easily plots distribu- frontier, as in Eq. (1), gives 19.34 stations per hour as an estimate
tion curves for a given sample. It provides square error and signifi- of the upper limit of optimal productivity (OPUL ). The loss due to
cance p-value for chi-square tests and Kolmogorov-Smirnov tests, operational inefficiencies (Δoi0 ) is 1.25 stations per hour, which
which serve as goodness-of-fit tests. when added to actual productivity, as shown in Eq. (2), gives 14.32
The 62 observations of each action appear in the curves gener- stations per hour as an estimate of lower limit of optimal produc-
ated by the Arena simulation (Fig. 5); the x-axis represents the tivity (OPLL ). Therefore, the average of the upper and lower limit of

© ASCE 04016107-8 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 Table 2. Tabular Synopsis of Optimal Productivity Estimation
Problem identification Breakdown of problem Solution approach Required value Modeling technique Model output Estimation result
Total inefficiency System inefficiency Top-down Productivity Qualitative Upper limit of Optimal
approach frontier factor model optimal productivity productivity
Operational Bottom-up Actual Discrete-event Lower limit of optimal
inefficiency approach productivity simulation productivity

optimal productivity results in 16.83 stations per hour. This value While the benefits of such an objective metric for productivity
represents the estimate of optimal productivity (OP). are broad, the authors acknowledge that the calculation of this
Compared to average actual productivity—which is 13.33 sta- metric can be cost prohibitive; only such key activities that are
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

tions per hour—the estimate of optimal productivity may seem either very expensive, have no available historical data, or are very
high. However, the actual field data show that at one point, the repetitive—and thereby could receive the greatest amount of pro-
veteran worker completed one station in 3.4 min, which is equiv- ductivity improvement when innovations emerge—would make the
alent to 17.64 stations per hour if such productivity were sustained. costs associated with this current methodology a reasonable invest-
This duration demonstrates that the estimate of 16.83 stations per ment. Nevertheless, this pilot study demonstrates that the suggested
hour is challenging but not necessarily out of reach. methodology for estimating optimal labor productivity is adequate
In summary, given these values, this project determined that dur- when applied to a simple electrical operation with a single worker
ing the pilot study, the fluorescent bulb replacement tasks achieved and sequential tasks. The impacts of factors that affect labor
79.2% efficiency—a quantitative evaluation metric objectively productivity in this pilot study were normal due to the study’s con-
generated without the need for historic averages. trolled environment—which, though not standard among construc-
tion settings, at least demonstrates the feasibility of the approach.
The adaptability within this approach could foreseeably transform
Conclusions the construction industry by obviating uncertainty within produc-
tivity metrics and priming the industry for greater innovation in
In practice, efficiency of construction operations is usually deter- labor-intensive operations. Future research into such pathways
mined by comparing actual versus historical productivity. How- would be valuable.
ever, this comparison only provides a relative measure of efficiency. As this research represents an early stage in the development
An accurate estimate of optimal productivity would allow project of a process for determining optimal productivity that is robust
managers to determine the efficiency of their labor-intensive con- enough to be beneficial to industry, there are many future advances
struction operations by comparing actual versus optimal rather than the research must undertake to become applicable to a construction
actual versus historical productivity. site. In particular, (1) the results are based on a single worker’s per-
This study contributes to the current body of knowledge in con- formance; future work will be done in multiple workers, (2) the
struction engineering and management by introducing a two-prong analysis was performed at the action level; future work will be per-
strategy for estimating optimal productivity in labor-intensive con- formed at the movement level, and (3) the analysis was performed
struction operations and by reporting on a pilot study performed to in a more-controlled environment; future work will target evalu-
evaluate the approach’s feasibility during a simple electrical light- ation of the approach in an uncontrolled environment. Further
ing fixtures replacement. Table 2 shows a brief summary of how the research is in progress to study the feasibility of this methodology
problem of estimating optimal productivity is broken down, which in complex construction operations.
solution approaches are performed on each type of inefficiency,
which techniques are implemented, and how the different paths
converge to estimate optimal labor productivity. Acknowledgments
In particular, total inefficiencies are separated into system and
The authors acknowledge the assistance of the Commonwealth
operational inefficiencies (major components shown in Fig. 2),
Electric Company and Omaha South Magnet High School and ex-
which in turn are quantified via QFM and DES modeling, respec-
press their special gratitude to the manager, site supervisor, and
tively. The top-down approach estimates the upper limit of optimal
workers of the company for providing access to their jobsite.
productivity by taking away system inefficiencies from the produc-
Commonwealth Electric is the premier electrical contractor rooted
tivity frontier. The QFM was found to be effective in modeling
in the Midwest of the United States. Without their support, this pilot
system inefficiency, an abstract concept difficult to measure in the
study would not have been possible.
field. The bottom-up approach determines the lower limit of opti-
mal labor productivity by taking away noncontributory events from
actual field productivity. The DES was found effective at modeling References
operational inefficiency. An average of the upper and lower limits is
taken as the best estimate of optimal labor productivity. AbouRizk, S., Knowles, P., and Hermann, U. R. (2001). “Estimating labor
As there is currently a vacuum within the realm of optimal production rates for industrial construction activities.” J. Constr. Eng.
productivity estimation, the success of this research yields a novel Manage., 10.1061/(ASCE)0733-9364(2001)127:6(502), 502–511.
means with which the construction industry could accurately exam- Alinaitwe, H. M., Mwakali, J. A., and Hanson, B. (2007). “Factors affect-
ing the productivity of building craftsmen-studies of Uganda.” J. Civ.
ine and improve labor-intensive operations. Since the proposed
Eng. Manage., 13(3), 169–176.
two-prong approach does not depend upon past productivity data Allen, S. G. (1985). “Why construction industry productivity is declining.”
for assessing current operations, it has the potential to create a dy- Rev. Econ. Stat., 67(4), 661–65.
namic means by which project managers could measure and assess Allmon, B. E., Haas, C. T., Borcherding, J. D., and Goodrum, P. M. (2000).
productivity for any type of labor-intensive operation, regardless of “U.S. construction labor productivity trends, 1970–1998.” J. Constr.
whether managers possess historical productivity data. Eng. Manage., 10.1061/(ASCE)0733-9364(2000)126:2(97), 97–104.

© ASCE 04016107-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 Arditi, D. (1985). “Construction productivity improvement.” J. Constr. Jarkas, A. M., and Bitar, C. G. (2012). “Factors affecting construction labor
Eng. Manage., 10.1061/(ASCE)0733-9364(1985)111:1(1), 1–14. productivity in Kuwait.” J. Constr. Eng. Manage., 10.1061/(ASCE)CO
Arena Simulation 14 [Computer software]. Rockwell Automation, .1943-7862.0000501, 811–820.
Wexford, PA. Kadir, M. R. A., Lee, W. P., Jaafar, M. S., Sapuan, S. M., and Ali,
BFC (Building Futures Council). (2006). Measuring productivity and A. A. A. (2005). “Factors affecting construction labour productivity for
evaluating innovation in the U.S. construction industry, Alexandria, Malaysian residential projects.” Struct. Surv., 23(1), 42–54.
VA. Kelton, W. D., Sadowski, R. P., and Sadowski, D. A. (2010). Simulation
Borcherding, J. D., Palmeter, S. B., and Jansma, G. L. (1986). “Work force with ARENA, Vol. 5, McGraw-Hill, New York.
management programs for increased productivity and quality work.” Kindinger, J. P., and Darby, J. L. (2000). “Risk factor analysis-a new
EEI Construction Committee Spring Meetings, Edison Electric Institute, qualitative risk management tool.” Proc., Project Management Institute
Washington, DC. Annual Seminars and Symp., Project Management Institute, Newton
Bureau of Labor Statistics. (2016). “Labor productivity and costs.” 〈http:// Square, VA.
www.bls.gov/lpc/〉 (Jan. 1, 2016). Kisi, K. P., Mani, N., and Rojas, E. M. (2014). “Estimating optimal labor
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

Chang, C. K., Hanna, A. S., Lackney, J. A., and Sullivan, K. T. (2007). productivity: A two-prong strategy.” Proc., Construction Research
“Quantifying the impact of schedule compression on labor productivity Congress, ASCE, Reston, VA, 757–766.
for mechanical and sheet metal contractor.” J. Constr. Eng. Manage., Koehn, E., and Brown, G. (1985). “Climatic effects on construction.” J.
10.1061/(ASCE)0733-9364(2007)133:4(287), 287–296. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(1985)111:2(129),
Christian, J., and Hachey, D. (1995). “Effects of delay times on production 129–137.
rates in construction.” J. Constr. Eng. Manage., 10.1061/(ASCE) Koehn, E., and Brown, G. (1986). “International labor productivity fac-
0733-9364(1995)121:1(20), 20–26. tors.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(1986)112:
Dai, J., Goodrum, P. M., and Maloney, W. F. (2009). “Construction craft 2(299), 299–302.
workers’ perceptions of the factors affecting their productivity.” J. Liberda, M., Ruwanpura, J. Y., and Jergeas, G. (2003). “Construction
Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(2009)135:3(217), productivity improvement: A study of human, managerial and external
217–226. factors.” Construction Research Congress: Wind of Change: Integra-
Dissanayake, M., Robinson Fayek, A., Russell, A., and Pedrycz, W. tion and Innovation, ASCE, Reston, VA.
(2005). “A hybrid neural network for predicting construction labour Loulakis, M. C., and Santiago, S. J. (1999). “Getting the most out of your
productivity.” Proc., Int. Conf. on Computing in Civil Engineering, ‘measured mile’ approach.” Civ. Eng., 69(11), 69.
ASCE, Reston, VA, 1–12.
Lu, M., AbouRizk, S. M., and Hermann, U. H. (2000). “Estimating labor
Dozzi, S. P., and AbouRizk, S. M. (1993). Productivity in construction, productivity using probability inference neural network.” J. Comput.
Institute for Research in Construction, National Research Council,
Civ. Eng., 10.1061/(ASCE)0887-3801(2000)14:4(241), 241–248.
Ottawa.
Lucko, G., and Rojas, E. M. (2010). “Research validation: Challenges
Drucker, P. (1993). Management: Tasks, responsibilities, practices, Harper
and opportunities in the construction domain.” J. Constr. Eng. Manage.,
Business, New York.
10.1061/(ASCE)CO.1943-7862.0000025, 127–135.
Enshassi, A., Mohamed, S., Abu Mustafa, Z., and Mayer, P. (2007). “Fac-
Makulsawatudom, A., Emsley, M., and Sinthawanarong, K. (2004).
tors affecting labour productivity in building projects in the Gaza strip.”
“Critical factors influencing construction productivity in Thailand.”
J. Civ. Eng. Manage., 13(4), 245–254.
J. King Mongkut’s Inst. Technol. North Bangkok, 14(3), 1–6.
Fayek, A. R., and Oduba, A. (2005). “Predicting industrial construction
Mani, N. (2015). “A framework for estimating labor productivity frontiers.”
labor productivity using fuzzy expert systems.” J. Constr. Eng.
Ph.D. thesis, Univ. of Nebraska, Lincoln, NE.
Manage., 10.1061/(ASCE)0733-9364(2005)131:8(938), 938–941.
Mani, N., Kisi, K., and Rojas, E. (2014). “Estimating labor productivity
Finkler, S. A., Knickman, J. R., Hendrickson, G., Lipkin, M., and
frontier: A pilot study.” Proc., Construction Research Congress, ASCE,
Thompson, W. G. (1993). “A comparison of work-sampling and
time-and-motion techniques for studies in health services research.” Reston, VA, 807–816.
Health Serv. Res., 28(5), 577–597. Panas, A., and Pantouvakis, J. P. (2010). “Evaluating research methodology
Hamm, M., Szczesny, K., Nguyen, V., and Konig, M. (2011). “Optimiza- in construction productivity studies.” Built Human Environ. Rev., 3(1),
tion of construction schedules with discrete-event simulation using an 63–85.
optimization framework.” Proc., Int. Workshop on Computing in Civil Park, H. S. (2006). “Conceptual framework of construction productivity
Engineering, ASCE, Reston, VA, 682–659. estimation.” KSCE J. Civ. Eng., 10(5), 311–317.
Hanna, A. S., Taylor, C. S., and Sullivan, K. T. (2005). “Impact of extended Pekuri, A., Haapasalo, H., and Herrala, M. (2011). “Productivity and
overtime on construction labor productivity.” J. Constr. Eng. Manage., performance management—Managerial practices in the construction
10.1061/(ASCE)0733-9364(2005)131:6(734), 734–739. industry.” Int. J. Perform. Measure., 1, 39–58.
Hanna, A. S. M., Chang, C., Sullivan, K. T., and Lackney, J. A. (2008). Rivas, R. A., Borcherding, J. D. M., González, V., and Alarcón, L. F.
“Impact of shift work on labor productivity for labor intensive contrac- (2011). “Analysis of factors influencing productivity using craftsmen
tor.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(2008)134: questionnaires: Case study in a Chilean construction company.”
3(197), 197–204. J. Constr. Eng. Manage., 10.1061/(ASCE)CO.1943-7862.0000274,
Haskell, B. P. H. (2012). “Construction industry productivity: Its history 312–320.
and future.” J. Des.-Build Inst. Am., 6(1), 5–10. Rojas, E. M. (2008). Construction productivity: A practical guide for
Heizer, J., and Render, B. (1996). “Production and operations management: building and electrical constructors, J. Ross Publishing, Fort
Strategic and tactical decisions.” Series in decision sciences, Prentice Lauderdale, FL.
Hall, Upper Saddle River, NJ. Rojas, E. M., and Aramvareekul, P. (2003a). “Is construction labor produc-
Herbsman, Z., and Ellis, R. (1990). “Research of factors influencing tivity really declining?” J. Constr. Eng. Manage., 10.1061/(ASCE)
construction productivity.” Constr. Manage. Econ., 8(1), 49–61. 0733-9364(2003)129:1(41), 41–46.
Huang, R. Y., and Hsieh, B. C. (2005). “A construction simulation system Rojas, E. M., and Aramvareekul, P. (2003b). “Labor productivity drivers
(COMSim) with object-oriented modeling elements.” J. Chin. Inst. and opportunities in the construction industry.” J. Manage. Eng.,
Eng., 28(2), 267–280. 10.1061/(ASCE)0742-597X(2003)19:2(78), 78–82.
Ibbs, W., and Liu, M. (2005). “Improved measured mile analysis Rustom, R. N., and Yahia, A. (2007). “Estimating productivity using sim-
technique.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364 ulation: A case study of Gaza beach embankment protection project.”
(2005)131:12(1249), 1249–1256. Constr. Innovation, 7(2), 167–186
Jarkas, A. (2010). “Critical investigation into the applicability of the Sanders, S. R., and Thomas, R. (1991). “Factors affecting masonry-labor
learning curve theory to rebar fixing labor productivity.” J. Constr. Eng. productivity.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364
Manage., 10.1061/(ASCE)CO.1943-7862.0000236, 1279–1288. (1991)117:4(626), 626–644.

© ASCE 04016107-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107

 Sargent, R. G. (2010). “Verification and validation of simulation principle.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(2003)
models.” Proc., 2010 Winter Simulation Conf., B. Johansson, S. Jain, 129:3(251), 251–261.
J. Montoya-Torres, J. Hugan, and E. Yücesan, eds., IEEE, Piscataway, Thomas, H. R., Maloney, W. F., Horner, M. W., Smith, G. R., Handa, V. K.,
NJ, 166–183. and Sanders, S. R. (1990). “Modeling construction labor productivity.”
Shahidul, M. I., and Shazali, S. (2011). “Determinants of manufacturing J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(1990)116:4(705),
productivity: Pilot study on labor-intensive industries.” Int. J. Prod. 705–726.
Perform. Manage., 60(6), 567–582. Thomas, H. R., and Mathews, C. T. (1985). An analysis of the methods of
Singh, H., Motwani, J., and Kumar, A. (2000). “A review and analysis measuring construction productivity, Construction Industry Institute,
of the state-of-the-art research on productivity measurement.” Ind. Austin, TX.
Manage. Data Syst., 100(5), 234–241 Thomas, H. R., Riley, D. R., and Sanvido, V. E. (1999). “Loss of labor
Smith, S. D. (1998). “Concrete placing analysis using discrete-event productivity due to delivery methods and weather.” J. Constr. Eng.
simulation.” Proc. Inst. Civ. Eng. Struct. Bldgs., 128(11), 351–358. Manage., 10.1061/(ASCE)0733-9364(1999)125:1(39), 39–46.
Son, J., and Rojas, E. (2011). “Impact of optimism bias regarding organi- Thomas, H. R., and Sakarcan, A. S. (1994). “Forecasting labor productivity
Downloaded from ascelibrary.org by GADJAH MADA UNIVERSITY on 05/03/19. Copyright ASCE. For personal use only; all rights reserved.

zational dynamics on project planning and control.” J. Constr. Eng. using factor model.” J. Constr. Eng. Manage., 10.1061/(ASCE)
Manage., 10.1061/(ASCE)CO.1943-7862.0000260, 147–157. 0733-9364(1994)120:1(228), 228–239.
Thomas, H. R., Sanders, S. R., and Bilal, S. (1992). “Comparison of labor
Song, L., and AbouRizk, S. M. (2008). “Measuring and modeling labor
productivity.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364
productivity using historical data.” J. Constr. Eng. Manage., 10.1061
(1992)118:4(635), 635–650.
/(ASCE)0733-9364(2008)134:10(786), 786–794.
Thomas, H. R., and Sanvido, V. E. (2000a). “Quantification of losses
Sonmez, R., and Rowings, J. E. (1998). “Construction labor productivity
caused by labor inefficiencies: Where is the elusive measured mile?”
modeling with neural networks.” J. Constr. Eng. Manage., 10.1061
Constr. Law Bus., 1(3), 1–14.
/(ASCE)0733-9364(1998)124:6(498), 498–504.
Thomas, H. R., and Sanvido, V. E. (2000b). “Role of the fabricator in labor
Srinavin, K., and Mohamed, S. (2003). “Thermal environment and productivity.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364
construction workers’ productivity: Some evidence from Thailand.” (2000)126:5(358), 358–365.
Build. Environ., 38(2), 339–345. Thomas, H. R., and Yiakoumis, I. (1987). “Factor model of construction
Tam, C. M., Tong, K. L., and Tse, L. (2002). “Artificial neural networks productivity.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364
model for predicting excavator productivity.” Eng. Constr. Archit. (1987)113:4(623), 623–639.
Manage., 9(5/6), 446–452 Yi, W., and Chan, A. P. C. (2014). “Critical review of labor productivity
Tangen, S. (2005). “Demystifying productivity and performance.” Int. J. research in construction journals.” J. Manage. Eng., 10.1061/(ASCE)
Prod. Perform. Manage., 54(1), 34–46. ME.1943-5479.0000194, 214–225.
Thomas, H. R., Horman, M. J., DeSouza, U. E. L., and Zavrski, I. (2002). Zhang, H. (2013). “Discrete-event simulation for estimating emissions
“Reducing variability to improve performance as a lean construction from construction processes.” J. Manage. Eng., 01014034.
principle.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364(2002) Zink, D. A. (1986). “The measured mile: Proving construction inefficiency
128:2(144), 144–154. costs.” Cost Eng., 28(4), 19–21.
Thomas, H. R., Horman, M. J., Minchin, R. E., and Chen, D. (2003a). “Im- Zink, D. A. (1990). “Impacts and construction inefficiency.” Cost Eng.,
proving labor flow reliability for better productivity as lean construction 32(11), 21–23.

© ASCE 04016107-11 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(3): 04016107