Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 196 (2017) 706 – 713

Creative Construction Conference 2017, CCC 2017, 19-22 June 2017, Primošten, Croatia

Measuring Project Performance Inspired by Stock Index

Richard C. Thompson, Jr.a, Yi Sua, Gunnar Luckoa*
a
Department of Civil Engineering, Catholic University of America, 620 Michigan Avenue NE, Washington, DC 20064, USA

Abstract

A fundamental precept of management is that projects must be measured to be controllable. But existing approaches to capture
project performance suffer from various problems – being proprietary, measuring time performance in dollar terms, reaching zero
(work left) or one (work done) upon completion, and erasing any notion of progression. Therefore the concept for a unified,
generalizable, and scalable performance metric is presented. It can function at levels from individual activities to entire industry
sectors. Its inspiration is gleaned from modern portfolio theory, which has long been tracking and successfully comparing highly
different companies. An analogy-based methodology will adapt and adopt the financial index beta and related concepts and test
their functioning on a hypothetical schedule with known progress deviations. Such indicator has the potential to become a vital
tool to measure and identify production efficiency, competitiveness, and ultimately the propensity to complete work on time.
© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2017 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review
Peer-review under
under responsibility
responsibility ofscientific
of the the scientific committee
committee of the Creative
of the Creative Construction
Construction Conference
Conference 2017 2017.

Keywords: Projects; performance; measurement; stocks; beta.

1. Introduction

According to the U.S. Bureau of Labor Statistics, over the past half century the construction industry has
experienced an average productivity decline of approximately ½ % annually versus an average increase in productivity
of 1¾ % for non-construction industries [1, 2], while a more recent study indicates a very minor growth, extracted
adverse factors, and confirmed lagging behind other industries [3]. Schedule performance is of primary importance,
because labor consistently accounts for more than half of all construction costs. Yet while a multiplicity of metrics
and processes focus on project pricing and acquisition, the construction industry suffers from an inability to measure
the schedule aspect, i.e. its ability to complete individual tasks and projects on time, to a similar depth.

* Corresponding author. Tel.: +1-202-319-4381; fax: +1-202-319-6677.

E-mail address: lucko@cua.edu

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the Creative Construction Conference 2017
doi:10.1016/j.proeng.2017.07.237
 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713 707

2. Literature Review

2.1. Schedule Performance Measurement in Construction Industry

Schedule performance, which tracks actual versus planned completion times, is a vital indicator of production
efficiency, organizational maturity, competitiveness, and ultimately the ability to complete work in a timely manner.
Performance is at the core of all managerial functions of decision makers; planning how projects should progress,
controlling how they perform, and, if needed, deploying remedies to bring lagging ones back on track. Yet despite, or
perhaps because of the size of the construction industry as a major contributor to the U.S. economy, its enormous
number of projects, and their variety, no single measure has emerged to provide such a snapshot-in-time assessment.
Unless delay claims require a forensic schedule analysis, performance is seldom considered after completion. Nor is
it easily possible to forecast expected performance from past projects of a given contractor. Why, because competitive
bidding is price-focused to be awarded the next job. This is compounded by a glaring lack of a measure of schedule
performance to benchmark the collective performance of contractors across the construction industry. A wealth of
literature focuses on pricing and acquisition measures. But performance assessment appears to concentrate on financial
aspects of projects within an industry that by some sources experiences a steady decline in productivity [4, 5]. In light
of this, schedule performance is of primary importance, because construction is such a labor-intensive industry [6]
and it typically encompasses more than half of the entire cost; even more for individual labor-intensive trades.
Currently only a pair of simplistic metrics exists, the Schedule Performance Index (SPI) and Schedule Variance
(SV) of the Earned Value Management (EVM) method. With roots in 1960s U.S. Government project management,
the earned value divided by planned value is SPI = EV / PV; their difference is SV = EV - PV [7], but they are only
indicative and suffer from problems: The SPI simply lists the percentage of total work as ‘value’ that a project has
achieved. Its metrics converge to one (SPI work completed) or zero (SV work remaining) at the project end, so that
the mere fact that a project was eventually finished makes it appear successful; any problems that have occurred are
forgotten. This gives an overly optimistic, incomplete, and non-predictive view. Worse, by definition EV and PV are
expressed rather counterintuitively in monetary terms (dollars, not days), while SPI at least is unitless and scale-free.
Construction practice uses a noteworthy predictive measure, which is unrelated to schedules, but captures safety
performance: The Experience Modification Rate (EMR) [8]. It is defined as a moving average ratio of actual
occurrences to expected within a three-year period. This determines the insurance premiums that individual
contractors must pay. Different from even EVM, the EMR is widely accepted and an integral part of contractor safety
quantification.

2.2. Inspiration from Stock Performance Index (β) in Finance

Several early studies contributed to conceiving the capital asset pricing model (CAPM) [9], including original work
in Treynor’s [10] unpublished, but widely cited manuscript on how to establish prices for risky assets within a market
and Sharpe’s [11] formal CAPM derivation, for which he and other researchers later received a Nobel prize [12].
Further groundbreaking work was contributed by Black and Scholes [13] on pricing options for volatile scenarios.
Their premise that historic behavior of the stock market becomes an indicator of its future performance has spawned
many ideas. Among them are the efficient market hypothesis [14], groupthink [15], and other irrational behavior, but
also empirical statistics-based formulas for price and performance expectations of markets, classes or groups of shares,
or individual share values. Within the body of knowledge of financial management the foremost seminal theoretical
contribution to describe stock performance in an exchange remains the aforementioned CAPM.
The essence of CAPM is that is provides a mathematical expression for comparing the average expected return of
an individual share of financial stock with the average expected value of holding a portfolio that comprises the entire
market. Averages are typically established over a longer period; for a market it can use large portfolio-based indices,
e.g. the S&P 500 [16]. Both of them are measured above the baseline of the so-called risk-free rate of return, which is
assumed as a constant and represented by debt obligations that the U.S. Government routinely issues. Beta is the
scaling factor for the variable component of the model that links these two measurements [17]. Black et al. [18] have
mathematically defined it in Equation 1 (written with the commonly used variable symbols per [19, p. 240]):
708 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713

E~
r  rf  E ˜ E~
rm   rf > @ (1)

“where E ~ r  = expected return on asset; rf = risk-free rate; and E ~

rm  = expected market return” and β “is the
expected return on the market in excess of the risk-free rate” [20, p. 30]. In other words, Equations 1 quantifies the
relationship of risk and expected return for an individual stock. Risk is expressed via a ‘premium’ with which an
investor expects to be compensated for assuming said risk. The term in the rectangular brackets is the risk premium
when acquiring a well-diversified investment portfolio (essentially ‘buying a small share of the entire market’), not
just the baseline of U.S. Treasury bonds. As Equation 2 shows, beta describes the relationship of the return for an
asset (stock) with that of the financial market [21], which in practice can be represented by the S&P 500 index [16].
Beta thus is the volatility of one stock [22], which is more risky than the overall market (whose risk is spread across
many stocks). Thus the expected return of a particular stock may actually exceed that of the market, E~ r  ! E~
rm  .

Cov~
r ,~rm 
E (2)
Var rm 
~

As beta represents “the influence of the overall market’s return on an individual stock” [23, p. 176], it indicates the
surplus risk (in relation to expected reward) that one acquires when a single risky stock is added to an existing stock
portfolio. Market risk is considered to be non-diversifiable ‘systematic risk’ as opposed to the ‘specific risk’ of one
stock. Each investor’s personal preference and utility determines what risk is acceptable; thus beta is a vital tool in
investment management. Beta is an asset-specific historic coefficient of degree to which one stock moves within its
market. For example, a hypothetical stock with β = 0.5 should increase or decrease in value about 50% as fast as the
entire market. These concepts will be explored in analogy to intrinsic risk of construction project participants. Their
relationships and impacts on schedule performance (schedule risk as the probability of failing to meet the as-planned
duration) can be matched with CAPM, which expresses the correlation of an individual stock to its market. This
inspiration for modeling the risk for each activity enables establishing new theory to (a) identify and correlate schedule
risk and (b) create a functional performance measure. Toward this goal, two Research Objectives are set:
1. Create a beta definition for schedule performance whose capabilities are analogous to those of beta in finance;
2. Perform simulations of sufficiently complex network schedules to determine and interpret realistic beta values.

3. Methodology

3.1 Development of a Schedule Performance Beta Index

The goal of this research is to overcome a major problem of the construction industry, as explained in the literature
review, by developing a schedule performance metric that will function similar to beta from finance. The new βi per
~ ~
Equation 3 uses the individual duration d i of an activity i and the total project duration d Total as substitutions for the
aforementioned individual stock return and the market return. Covariance here is defined as the scaled sum-product
of all individual durations compared to (as differences) their expected activity durations E(di) with all total project
durations compared to their expected project durations E(dTotal). In construction practice, these values are the as-built
and as-planned activity and project durations, respectively. They can be obtained from the initial baseline schedule
and the updated progress schedule. Variance here is defined as the scaled sum of squares of total project durations
compared to their expected project durations E(dTotal). This new definition fulfills Research Objective 1.

Ei

~ ~
Cov d i , d Total  ~ ~
    
˜ ¦ d i  E d i  ˜ d Total  E d Total 
1 ni ~ ~ ~
  1 ni ~
 
˜ ¦ d Total  E d Total  (3)
2
~

Var d Total 
, where Cov d i , d Total
ni i 1
and Var d Total
ni i 1
 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713 709

3.2 Simulation and Probability

Figure 1 presents the proposed analysis. Step 1 of the flowchart captures input data, including activity names,
sequential relations, and a unique probability distribution function (PDF) for each activity duration (e.g. triangular,
beta) as have been used in the literature. This research uses 50 randomly generated variations of the “J30” network
schedule from the Project Scheduling Problem Library (PSPLIB) [24]. This widely accepted database offers known
schedules of various sizes in plain computer files for benchmarking purposes. A semi-automatic conversion of the
PSPLIB format into plain text files has been created for this research, so that the simulation can read them. Resource
input is not needed and has been deleted. For realism, distributions like the Figure 2 examples have been randomly
added. Table 1 lists parameters that remain fixed across projects. This is a realistic assumption, because performance
is an inherent characteristic of firms (but may also be impacted by external events). The L, M, H are low, mode, and
high values for the triangular or beta distributions, the latter of which can be converted to an alternative parameter
form of mean and standard deviation following [25], and μ and σ are mean and standard deviation for the normal.

Step 1: Schedule Input and CPM

Activity name {e.g. A, B, C, D}, duration PDF
(e.g. triangular, beta), and relations (FTS, STS,
FTF, STF), perform CPM using PDF modes

Step 2: Monte Carlo Simulation (a) Activity 3 (b) Activity 5

Randomize durations {dA1, dB1, dC1, dD1}, …,
{dAn, dBn, dCn, dDn} in n different schedules,
perform CPM calculation for each schedule,
record total project durations , …,

Step 3: Performance Index

Calculate {βA, βB, βC, βD} as βi = Cov( , )
/ Var( ) across n schedules and complexity (c) Activity 7 (d) Activity 24

Fig. 1. Methodology Flowchart Fig. 2. PDF for Selected Activities

Table 1. Simulation Input

Activity Distribution Parameter Activity Distribution Parameter
1 Dummy 0 17 Triangular (5, 6, 7)
2 Triangular (L, M, H) (7, 8, 12) 18 Triangular (4, 5, 7)
3 Beta (L, M, H) (3, 4, 8) 19 Beta (2, 3, 7)
4 Beta (L, M, H) (5, 6, 10) 20 Beta (5, 7, 10)
5 Beta (L, M, H) (2, 3, 6) 21 Triangular (1, 2, 3)
6 Triangular (L, M, H) (7, 8, 12) 22 Triangular (6, 7, 11)
7 Triangular (L, M, H) (4, 5, 6) 23 Triangular (1, 2, 3)
8 Beta (L, M, H) (8, 9, 12) 24 Triangular (2, 3, 5)
9 Triangular (L, M, H) (1, 2, 3) 25 Triangular (2, 3, 5)
10 Triangular (L, M, H) (6, 7, 9) 26 Beta (6, 7, 9)
11 Triangular (L, M, H) (7, 9, 12) 27 Triangular (7, 8, 12)
12 Beta (L, M, H) (2, 4, 8) 28 Triangular (1, 3, 4)
13 Beta (L, M, H) (5, 6, 10) 29 Beta (5, 7, 12)
14 Triangular (L, M, H) (1, 3, 4) 30 Beta (1, 2, 5)
15 Triangular (L, M, H) (8, 9, 14) 31 Triangular (1, 2, 3)
16 Triangular (L, M, H) (9, 10, 15) 32 Dummy 0
710 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713

Overall the simulation executes 50 schedules that represent projects with different performance. Each schedule
has a somewhat different network structure and contains 30 activities (plus a dummy start and finish activity to tie all
link ends together), which are performed by different subcontractors. Pairings remain the same between projects,
which is a realistic assumption, because subcontractors specialize in tasks that occur at particular stages of projects.
Modes are the expected activity durations E(di) to calculate expected project durations E(dTotal) with CPM in Step 1.
Table 2 lists E(dTotal) for the 50 projects, ranging from 33 to 76 days with mean 47.16 and standard deviation 7.82. In
~
Step 2 a Monte Carlo (MC) simulation samples randomized durations d i for all activities in each schedule (here n =
~
50) while applying the critical path method (CPM). Their total schedule durations d Total from a second CPM
~
calculation – this time for the probabilistically sampled durations d i, not fixed modes M or μ – are recorded. Step 3
finally calculates the new performance beta of Table 3 for each subcontractor, i.e. activity, per Equation 3. Note than
~
Var( d Total) is constant at 74.1599, because it is calculated across all 50 schedule variations.

Table 2. Project Output

Project E(dTotal) RT Project E(dTotal) RT Project E(dTotal) RT Project E(dTotal) RT Project E(dTotal) RT
1 40 0.4133 11 39 0.3871 21 51 0.4778 31 66 0.5121 41 49 0.4012
2 42 0.3690 12 37 0.3609 22 51 0.4012 32 53 0.4698 42 43 0.3992
3 44 0.4355 13 39 0.375 23 46 0.4315 33 51 0.4012 43 38 0.4052
4 47 0.4093 14 38 0.4052 24 76 0.5544 34 62 0.5343 44 42 0.375
5 45 0.4012 15 53 0.4435 25 54 0.4254 35 52 0.4113 45 46 0.4395
6 56 0.4577 16 49 0.4274 26 52 0.4556 36 52 0.4516 46 36 0.381
7 55 0.4435 17 42 0.377 27 38 0.3831 37 49 0.4254 47 33 0.3387
8 42 0.3710 18 43 0.3992 28 46 0.4032 38 51 0.4657 48 49 0.4194
9 43 0.4274 19 45 0.373 29 55 0.4274 39 46 0.4032 49 42 0.4093
10 45 0.4153 20 43 0.4536 30 51 0.4536 40 47 0.4214 50 43 0.3972

Table 3. Simulation Output

~ ~ ~ ~
Activity Cov( d i, d Total) Beta Activity Cov( d i, d Total) Beta
1 0 0 17 0.0040 0.0001
2 8.5039 0.1147 18 2.6926 0.0363
3 11.2726 0.1520 19 13.1190 0.1769
4 14.4838 0.1953 20 2.8265 0.0381
5 9.7876 0.1320 21 0.0892 0.0012
6 7.5933 0.1024 22 9.8028 0.1322
7 0.3624 0.0049 23 -0.9588 -0.0129
8 8.0298 0.1083 24 2.7722 0.0374
9 0.1287 0.0017 25 4.0203 0.0542
10 1.7508 0.0236 26 4.0516 0.0546
11 3.2337 0.0436 27 8.2287 0.1110
12 7.7618 0.1047 28 -2.1553 -0.0291
13 13.5672 0.1829 29 14.0136 0.1890
14 -3.3478 -0.0451 30 6.8060 0.0918
15 11.2089 0.1511 31 0.2629 0.0035
16 11.0879 0.1495 32 0 0

The simulation can also calculate the complexity of each of the 50 selected schedules to examine how varied they
are in terms of their connectivity. Figure 3 shows an example network. Measures of complexity for network schedules
in the literature include various indices that are calculated from the number of nodes (i.e. activities) and arcs (i.e. links)
in a network [26]. The density (also called order strength) is the number of actual links divided by the number of
possible links, which is n · (n - 1) / 2, where n are non-dummy activities [27], and is constant at 0.0968 for all 50
schedules due to the manner in which PSPLIB has generated them. The so-called restrictiveness estimator (RT) is a
 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713 711

meaningful index [28], because it remains unaffected by redundant links, is 0 for purely parallel schedules, and 1 for
serial ones. It can be calculated with a so-called reachability matrix. Figure 4 plots the RT of all 50 schedules, which
range from 0.3387 to 0.5544, i.e. have medium complexity. This is realistic, because real-world construction schedules
are neither purely parallel nor purely serial. Future research could refine this comparison by characterising actual
schedules in terms of their complexity when studying their planned versus actual performance.

0,60
0,55
0,50
0,45
0,40
0,35
0,30
0 10 20 30 40 50

Fig. 3. Network Graph (Schedule j301_10) Fig 4. Complexity RT Values for 50 Projects

4. Beta Behavior

Figure 5 plots all beta values for the 32 different activities. Of course, the dummy activity 1 and 32 have a beta of 0.
Of course, in an ideal case beta for a non-dummy activity would be 0 (or very close to it, e.g. for a user-defined band
from 0 to 0.05), because this means that it exhibits near-perfect performance and represents no more than 5% risk
compared to the overall schedule (13 out of 30). These activities – as is intuitively understood – contribute some
quantifiable, but small risk when added to the project. Most activities (27 out of 30) have a positive beta, where activity
4 has the maximum of 0.1953. Yet a few activities (numbers 14, 23, and 28) even have a negative beta. This means
~
that the fluctuation of the activity duration d i - E(di) changes opposite to the fluctuation of the total project duration
~
d Total - E(dTotal). Recall that beta measures the ratio of an individual activity duration fluctuation with the total project
duration fluctuation. It expresses the contribution of one to the other. But it cannot establish causation, because
schedule performance depends both on the intricate network structure and random activity durations. It thus is possible
that the activity is negatively correlated with the total schedule performance [29]. Such phenomenon is also found in
the stock market, where precious metals, especially gold, provide ‘hedging’ against other forms of investments like
stocks, and due to their counteracting trends have negative betas that are determined by regression analysis [30]. These
simulations and the interpretation of the resulting values for beta fulfill Research Objective 2.

0,08
0,06
0,04
0,02
0
0 5 10 15 20 25 30 35
-0,02
-0,04
-0,06

Fig.5. Beta and Activity Name

712 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713

5. Conclusions

Construction schedules have lacked a performance index to reveal the contribution of individual activities to the
total project duration. Taking the analogy from the stock market, which has solved this challenge long ago, beta can
fill this gap in scheduling. Simulating a set of schedules of realistic complexity has allowed exploring the behavior of
beta values for different hypothetical subcontractors. The magnitude and sign of beta can give a clear indication of
the propensity to finish as-planned and can be a de facto measure of the risk posed within a project schedule.

6. Future Research

Technology provides a mechanism for the construction industry to model potential performance of future work, as
well as to statistically evaluate that of completed projects. Future research on the applicability of beta as a new
performance index and living evaluation tool for the construction industry should explore this powerful ability. It
should focus on moving from the theoretical demonstration of its meaning and determination of its mathematics via
hypothetical example schedules to case studies of actual construction projects. From such a retrospective review of
completed projects the resulting betas could be calculated to establish guidelines and expectations for their values and
their corresponding validity; and ultimately, culminate with drafting an industry standard to define beta and explain
its appropriate use for the construction industry. Research is currently underway to realize this ambition.

Acknowledgements

The support of the National Science Foundation (Grant CMMI-1536005) for portions of the work presented here
is gratefully acknowledged. Any opinions, findings, and conclusions or recommendations expressed in this material
are those of the authors and do not necessarily represent the views of the National Science Foundation.

References

[1] P. C. Teicholz, Labor-productivity declines in the construction industry: Causes and remedies (another look). AECbytes Viewpoint 67, 2013,
ed. L. Khemlani, Union City, CA, <http://www.aecbytes.com/viewpoint/2013/issue_67.html>, January 13, 2015.
[2] S. C. Vereen, W. Rasdorf, J. E. Hummer, Development and comparative analysis of construction industry labor productivity metrics. Journal
of Construction Engineering and Management, 2016: 04016020(9).
[3] L. Sveikauskas, S. Rowe, J. Mildenberger, J. Price, A. Young, Productivity growth in construction. Journal of Construction Engineering and
Management, 2016: 04016045(8).
[4] E. M. Rojas, P. Aramvareekul, Is Construction labor productivity really declining? Journal of Construction Engineering and Management,
129(1), 2003: 41-46.
[5] E. Allmon, C. T. Haas, J. D. Borcherding, P. M. Goodruum, U.S. construction labor productivity trends, 1970-1998. Journal of Construction
Engineering and Management, 126(2), 2000: 97-104.
[6] K. P. Kisi, N. Mani, E. M. Rojas, E. T. Foster, Optimal Productivity in labor-intensive construction operations: Pilot Study. Journal of
Construction Engineering and Management, 2016: 04016107(11).
[7] H. L. Chen, Improving forecasting accuracy of project earned value metrics: Linear modeling approach. Journal of Construction Engineering
and Management 30(2), 2014: 135-145.
[8] J. E. Wilson, E. Koehn, Safety management: Problems encountered and recommended solutions. Journal of Construction Engineering and
Management 126(1), 2000: 77-79.
[9] K. L. Benson, R. W. Faff, E. Abacus: A Journal of Accounting, Finance and Business Studies 49(Supplement), 2013): 24-31.
[10] J. L. Treynor, Toward a theory of market value of risky assets. Unpublished manuscript, Social Science Research Network, 1962.
<http://papers.ssrn.com/sol3/papers.cfm?abstract_id=628187>, January 13, 2015.
[11] W. F. Sharpe, Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance 19(3), 1964: 425-442.
[12] Royal Swedish Academy of Sciences, The prize in economics 1990 – press release. Nobelprize.org. – The Official Website of the Nobel
Price, Nobel Media AB, Nobel Foundation Rights Association, Stockholm, Sweden, 1990. <http://nobelprize.org/nobel_
prizes/economics/laureates-/1990/press.html>, January 13, 2015.
[13] F. Black, M. S. Scholes, The pricing of options and corporate liabilities. Journal of Political Economy 81(3), 1973: 637-654.
[14] D. Alajbeg, Z. Bubaš, V. Šonje, The efficient market hypothesis: Problems with interpretations of empirical tests. Financial Theory and
Practice 36(1), 2012: 53-72.
[15] I. L. Janis, Victims of groupthink; A psychological study of foreign-policy decisions and fiascoes. Houghton Mifflin, Boston, MA, 1972.
[16] Standard and Poor’s, U.S. equity: Market cap. S&P Down Jones Indices, McGraw-Hill Financial, New York City, NY, 2015. <http://us.
spindices.com/index-family/us-equity/style>, January 13, 2015.
[17] R. C. Thompson, G. Lucko, Modeling measures of risk correlation for quantitative float management of construction projects. Proceedings
 Richard C. Thompson et al. / Procedia Engineering 196 (2017) 706 – 713 713

of the 5th International Conference on Construction Engineering and Project Management, ed. H. Kim, Anaheim, CA, January 9-11, 2013,
University of North Carolina, Charlotte, NC: 8 pp.
[18] F. Black, M. C. Jensen, M. S. Scholes, The capital asset pricing model: Some empirical tests. Studies in the Theory of Capital Markets, ed.
M. C. Jensen, Praeger Publishers, New York City, NY, 1972.
[19] A.Wibowo, CAPM-based valuation of financial government supports to infeasible and risky private infrastructure projects. Journal of
Construction Engineering and Management 132(3), 2006: 239-248.
[20] E. F. Fama, K. R. French, The capital asset pricing model: Theory and evidence. Journal of Economic Perspectives 18(3), 2004: 25-46.
[21] M. Levinson, The Economist: Guide to financial markets. 5th ed., John Wiley and Sons, Hoboken, NJ, 2009.
[22] K. Pukthuanthong-Le, N. Visaltanachoti, Idiosyncratic volatility and stock returns: A cross country analysis. Applied Financial Economics
19(16), 2009: 1269-1281.
[23] B. M. Smith, A history of the global stock market: From Ancient Rome to Silicon Valley. University of Chicago Press, Chicago, IL, 2003.
[24] R. Kolisch, A. Sprecher, PSPLIB – a project scheduling library. European Journal of Operational Research, 96(1), 1996: 205-216.
[25] R. Davis, Teaching project simulation in Excel using PERT-beta distributions. Informs Transactions on Education, 8(3): 139-148.
[26] G. Lucko, R. C. Thompson, Y. Su, Simulating balanced allocation of project float to the critical path in network schedules. Proceedings of
the Construction Research Congress, eds. J. L. Perdomo-Rivera, A. Gonzáles-Quevedo, C. López del Puerto, F. Maldonado-Fortunet, O. I.
Molina-Bas, San Juan, Puerto Rico, May 31 – June 2, 2016, American Society of Civil Engineers, Reston, VA: 749-758.
[27] E. Demeulemeester, M. Vanhoucke, W. S. Herroelen, RANGEN: A random network generator for activity-on-the-node networks. Journal of
Scheduling, (1), 2003: 17-38.
[28] C. Schwindt, ProGen/max: A new problem generator for different resource-xonstrained project scheduling problems with minimal and
maximal time lags. Report WIOR-449, Institut für Wirtschaftstheorie und Operations Research, Universität Karlsruhe, Germany, 1995: 57 pp.
[29] R. C. Thompson, G. Lucko, Modeling measures of risk correlation for quantitative float management of construction projects.” Proceedings
of the 5th International Conference on Construction Engineering and Project Management, ed. H. Kim, Anaheim, CA, January 9-11, 2013,
University of North Carolina, Charlotte, NC: 8 pp.
[30] Investopedia, Beta: Gauging price fluctuations. Website, Investopedia, <http://www.investopedia.com/articles/01/102401.asp>,
March 1, 2017.