Professional Documents
Culture Documents
Firat Dogu Akin, Atilla Damci, David Arditi & Sevilay Demirkesen
To cite this article: Firat Dogu Akin, Atilla Damci, David Arditi & Sevilay Demirkesen (18 Jan
2024): An advanced decision-support system for ranking unbalanced bids, Construction
Management and Economics, DOI: 10.1080/01446193.2024.2303158
Article views: 91
CONTACT Atilla Damci damcia@itu.edu.tr Department of Civil Engineering, Istanbul Technical University, Maslak, Istanbul, Turkey
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
� 2024 Informa UK Limited, trading as Taylor & Francis Group
2 F. D. AKIN ET AL.
owner’s legal prerogative to reject the bid, in which engineer’s estimate or the average line-item prices
case the bidder cannot undertake the project and con quoted by all bidders in the project as benchmarks.
sequently loses potential profit in addition to absorb However, Arditi and Chotibhongs (2009) assert that
ing the cost of bid preparation. In addition, having a considering only the engineer’s estimated prices as a
bid rejected for unbalancing is likely to have a nega benchmark may mislead the decision-maker because
tive effect on the bidder’s reputation, a major problem engineers can make inaccurate predictions. Arditi and
in future biddings. On the other hand, from the own Chotibhongs (2009) also claim that even if an owner
er’s perspective, if a contract is awarded to a contrac uses the average line item prices as a benchmark,
tor who has unbalanced the bid, the final cost of the more than one unbalanced bid in the same procure
project may end up being much higher than had the ment process may cause these prices to be skewed.
bidder not unbalanced the bid (Arditi and In sum, the aforementioned studies reveal that the
Chotibhongs 2009). These outcomes are not consistent benchmark used to detect unbalanced bids greatly
with an owner’s objective to select the most appropri affects the performance of unbalanced bid detection
ate contractor who can complete the construction models. However, these researchers also cite the many
project on schedule, within budget, and at the desired drawbacks of using the existing benchmarks, which
quality (Mehta et al. 2009). For that reason, unbal implies that there is still a critical need for a bench
anced bids are not welcome by owners in the con mark that improves the performance of unbalanced
struction industry. bidding detection models.
In a bid evaluation process, detecting unbalanced The second challenging issue cited by Li et al.
bids is not an easy task. Indeed, Skitmore and Cattell (2023) in unbalanced bid detection is the extent to
(2013) assert that the number of cases in which own which the bidder manipulated the unit prices of the
ers mistakenly identify balanced bids as unbalanced
line items. They claim that since the detection of
bids is too great to ignore. More recent studies (e.g. Li
unbalanced bids is solely based on comparing the bid
et al. 2021, 2023) in the unbalanced bidding literature
ders’ offers against a benchmark, using a method that
agree that it is still challenging to identify unbalanced
can assess how much a bidder has manipulated the
bids accurately. After reviewing the extant literature
unit prices of the line items can help owners to decide
thoroughly, Li et al. (2023) state that two critical issues
which bid is more desirable than the others. In other
are quite challenging in detecting unbalanced bids.
words, the bidders can be ranked by the extent to
The first issue they cite is the inadequacy of the proc
which they manipulated their unit prices, a high score
esses used so far to set an evaluation benchmark for
(e.g. VIKOR Index Qi ) prompting the owner’s engineer
the unit price of line items. Li et al. (2023) assert that
to examine the bid in detail to ensure a fair competi
the nature of the benchmark used to detect unbal
tive environment. So far, construction researchers did
anced bids is still a major problem that should be
resolved to improve the existing unbalanced bidding not take full advantage of decision-making methods
detection models. As a solution to this problem, they such as TOPSIS and VIKOR that can provide owners
suggest a benchmark that is based on a simple mov with more robust solutions. This may be caused by
ing average of the unit prices collected from previous the scarcity of studies that use sophisticated and
projects, even though collecting data from previous advanced decision-making models in studies that
projects is not an easy task for owners. Actually, even investigate bidding practices (e.g. Su et al. 2020a, Wu
if such a database exists, moving averages may not and Xu 2021, Li et al. 2023). Another reason can be
reflect the most recent unit prices at the time of ana that the traditional decision-making methods are
lysis, especially in times of economic turbulence. In designed to determine the superior alternative (i.e. the
addition, according to a number of studies (e.g. alternative that represents the solution that is closest
Stramarcos and Cattell 2013, Cattell et al. 2007, to the positive ideal value), whereas detecting unbal
Skitmore and Cattell 2013), using a benchmark based anced bids requires the determination of the inferior
on a moving average can mislead the decision-maker alternative (i.e. the alternative that represents the solu
because the benchmark itself can be unbalanced tion that is farthest from the positive ideal value). To
when the majority of bids in the database are unbal put a finer point on it, in unbalanced bidding, the
anced. In other words, unbalanced bids may appear to inferior alternative represents the bid that the highest
be balanced and vice versa when the database is not likelihood of being unbalanced, whereas the superior
representative of balanced bids. Some of the existing alternative represents the bid that has the lowest like
unbalanced bid detection models prefer to use the lihood of being unbalanced.
CONSTRUCTION MANAGEMENT AND ECONOMICS 3
In sum, the literature indicates that there is a need or low compared to the owner’s or other bidders’ unit
for unbalanced bid detection models that make use of prices, then the bid can be considered to be mathem
advanced decision-making methods to see how much atically unbalanced.
bidders manipulated the unit prices of the line items An owner should be aware of that a mathematically
in their offer. Modifying the traditional decision-mak unbalanced bid may or may not increase the cost to
ing methods for use in such models can create a the owner. Therefore, an owner must examine a math
more robust model and can allow construction owners ematically unbalanced bid to see whether or not it is
to use these models with more confidence. Therefore, also materially unbalanced. The FHWA guidelines
there is still a need for further research for the devel (page A-113, 1988) define a materially unbalanced bid
opment of unbalanced bid detection models including as, “A bid is materially unbalanced if there is a reason
the ones based on advanced decision-making meth able doubt that award to the bidder submitting the
ods. The study presented in this paper is initiated in mathematically unbalanced bid will result in the low
response to not only the lack of modified advanced est ultimate cost to the Government.” If the detailed
decision-making methods developed over the years examination of the mathematically unbalanced bid
for use in unbalanced bid detection models but also uncovers that it may increase the cost to the owner,
to propose an approach to overcome the limitation then the owner can conclude that it is materially
related to the selection of a benchmark. This study’s unbalanced. On the other hand, it should be noted
main objective is to develop an advanced decision- that mathematically unbalanced bids are not always
making model that overcomes the aforementioned materially unbalanced.
shortcomings. The model proposed in this study uti An example is provided below to explain the mean
lizes the geometric mean of the unit prices provided ing of mathematically and materially unbalanced bids,
by all the bidders for each line item as a benchmark, and to describe how bids are tested to see whether
and an advanced decision-making model that involves they are mathematically and materially unbalanced.
the combined use of the Information Entropy and the Table 1 shows the information about the offers sub
Modified VIKOR methods to rank the bids relative to mitted by three bidders in a project that involves six
the likelihood that a bid has been unbalanced. The line items. Even though Bidder 1’s total offer is the
proposed model shows that the detection of unbal lowest one among the submitted offers, when the
anced bids can be efficiently performed using the pro owner compares Bidder 1’s unit prices to the unit pri
posed method. Armed with such an advanced tool, it ces submitted by Bidders 2 and 3, the owner sees that
may be easier for owners to prevent unbalanced bids. the unit prices of some line items (i.e. line items A, B,
Indeed, the results of this research reveal that the pro C, and E) submitted by Bidder 1 are significantly differ
posed model can be used as an effective tool to ent from the unit prices offered by Bidders 2 and 3.
detect unbalanced bids, create a fair competitive The owner concludes that Bidder 1’s offer is mathem
environment, and prevent future losses to construc atically unbalanced. However, the owner does not
tion owners. know at this stage whether awarding the contract to
Bidder 1 will cost the owner more (or less) than
awarding the contract to another bidder. Therefore,
Literature review the owner examines Bidder 1’s offer in detail to see
The review of the literature indicates that bids are first whether Bidder 1’s unit prices reflect reasonable costs,
reviewed to see whether they are mathematically anticipated profit, overheads, and other indirect costs.
unbalanced, and later to see whether they are materi If the owner finds that Bidder 1’s unit prices do not
ally unbalanced too. The Federal Highway reflect reality and if for this reason a contract awarded
Administration (FHWA) guidelines (FHWA 1988, page to Bidder 1 is likely to be completed at an overall cost
A-113) define a mathematically unbalanced bid as “A
Table 1. The data for the example.
mathematically unbalanced bid is one containing
Unit prices ($)
lump sum or unit line items which do not reflect rea
sonable actual costs plus a reasonable proportionate Line item Quantity Bidder 1 Bidder 2 Bidder 3
A 61 days $2100/day $510/day $505/day
share of the bidder’s anticipated profit, overhead B 11 days $1,250/day $810/day $815/day
costs, and other indirect costs, which he/she antici C 423 yards $35/yard $16/yard $18/yard
pates for the performance of the items in question.” D 12,100 feet2 $10/feet2 $13/feet2 $13/feet2
E 655 yard3 $155/yard3 $255/yard3 $245/yard3
To put a finer point on it, if an owner detects that the F 10,100 feet2 $13/feet2 $14/feet2 $15/feet2
unit prices of some line items are exceptionally high Total Project Cost $510,480 $512,513 $516,659
4 F. D. AKIN ET AL.
to the owner that is higher compared to the other that checks whether the bid is mathematically unbal
bidders’ overall costs, then the owner makes the anced and if it is, they added a second step to the
determination that Bidder 1’s offer is not only math process where they calculate the present worth of all
ematically by also materially unbalanced, in which payments using an appropriate discount rate and a
case the owner has the legal right to reject Bidder 1’s simple work schedule developed with the limited
offer. information available at the time the bidding takes
If, on the other hand, the owner finds that Bidder place. It should be noted that these NPV calculations
1’s unit prices do not reflect reality and yet a contract are based on an approximate schedule that may dras
awarded to Bidder 1 is likely to be completed at the tically change along the project. It should also be
lowest possible overall cost to the owner compared to noted that the time value of money is significant only
the cost of the other bidders, then the owner makes if the duration of the project is several years long and
the determination that Bidder 1’s offer is not materi if the discount rate is rather high. For example, con
ally unbalanced and the owner concludes that award tractors who undertake highway construction projects
ing the contract to Bidder 1 is the best alternative. for State Departments of Transportation are tempted
to unbalance their bid because most of these projects
Previous studies on detection and prevention involve unit-price contracts that are only 8–12 months
models of unbalanced bids long, in which case NPV calculations are quite redun
dant. In addition, if the interest rate is around 2%–3%
Unbalanced bidding in the construction industry has
as it has been for many years in the U.S., the impact
been a focus of debate since the 1960s. Unbalanced
of the interest rate on NPV calculations is trivial. If the
bids are not prohibited in the construction industry
duration of the project and the discount rate are both
and are often used as a legitimate business practice,
low, there is no need to perform NPV analysis.
but studies exist (e.g. Stramarcos and Cattell 2013,
Unlike other studies that recommend rejecting
Doran 2004) that claim that most researchers and pro
unbalanced bids, Hyari (2016) suggests a model that
fessionals in the construction industry identify unbal
rebalances unbalanced bids as an alternative to reject
anced bidding as an unethical activity. Even though
ing them. The main objective of Hyari’s (2016) model
some research aimed to improve a contractor’s cash
is to readjust all unit prices without changing the pro
flow by using unbalanced bidding, the use of unbal
vided bid prices in order to provide fair competition.
anced bidding leads to serious problems for construc
To improve Hyari’s (2016) model, Hyari et al. (2016)
tion owners (Arditi and Chotibhongs 2009). Since the
purpose of this study is to develop a model that can add a feature that utilizes the divergences from the
detect unbalanced bids, the goal of this literature estimated quantities in previous projects to evaluate
review is to outline the efforts to prevent it, rather each bid’s likelihood of being the truly lowest bid.
than to summarize the studies about how to unbal The aforementioned studies ignore the use of a vis
ance a bid without getting caught. ual way to detect unbalanced bids. To fill this gap,
A person who evaluates the bids received by the Nikpour et al. (2017) propose an approach that uses
owner needs guidelines to figure out the acceptable graphs generated using the Bid Markup Distribution
difference between the unit prices proposed by the Index. The assertion of the authors is that using these
bidder for any line item and a benchmark. Wang graphs makes it easier to detect unbalanced bids.
(2004) presents an “electronic-facilitated bid evaluation Another approach that detects unbalanced bids via a
model” to detect unbalanced bids. This model speeds different point of view is proposed by Polat et al.
up the process because the unit prices are obtained (2019, 2020) who developed a model that uses several
from all the bidders electronically. Wang et al. (2006) grading systems. The decision of identifying a bid as
uses a ratio to measure the deviation between the an unbalanced bid is based on its score computed by
unit prices considered by the owner and submitted by these grading systems. The model has also a feature
the lowest bidder. This ratio helps to highlight the that allows owners to assign different weights to these
unreasonable unit prices and to decide whether or grading systems while computing each bid’s final
not the bid is unbalanced. Even though Wang et al.’s score. Even though having these models is an advan
(2006) approach provides great value in detecting tage for owners in detecting unbalanced bids, the
unbalanced bids, the time value of money has not results of Skitmore and Cattell’s (2013) study reveal
been considered in this study. Arditi and Chotibhongs’ that users of these models should be aware that if the
study (2009) fills this gap as they developed a model majority of the bids used as a database are
CONSTRUCTION MANAGEMENT AND ECONOMICS 5
unbalanced then balanced bids may appear to be up if researchers have a better understanding of the
unbalanced and vice versa. barriers that cause such resistance. One of the reasons
Su et al. (2020a) propose an unbalanced bid detec for not preferring the traditional decision-making
tion model using VIKOR, one of the popular decision- methods is the fact that decision-making methods are
making methods. Their model uses the engineer’s designed to determine the superior alternative (i.e. the
estimated prices as the evaluation criterion for compari alternative that represents the solution that is closer
son. However, Arditi and Chotibhongs (2009) assert to the ideal values) rather than determining the infer
that considering only the engineer’s estimated prices as ior alternative (i.e. the alternative that represents the
an evaluation benchmark might mislead the decision- solution that is farthest from the ideal values). It
maker because engineers can also make inaccurate pre should be noted that detecting unbalanced bids
dictions. On the other hand, they also state that even if requires the determination of the undesirable alterna
an owner uses average line item prices for comparison, tives. Therefore, the traditional decision-making meth
more than one unbalanced bid in the same procure ods should be modified to make they fit the problem
ment process may cause the distribution to be skewed. at hand, i.e. the detection of unbalanced bids.
Su et al. (2020b) improve their previous model by utiliz Unfortunately, even though these studies attempt to
ing Grey rational analysis theory and fuzzy set theory detect unbalanced bids by comparing the engineer’s
to determine unbalanced bids. The Grey rational theory estimated prices against the prices specified by the
is used to determine the deviation between the engi bidder, they ignore the fact that engineers can make
neer’s estimate and the unit price provided by the bid inaccurate predictions. When studies compare the
der. Fuzzy set theory is utilized to rank the bids average line item prices against the prices specified by
according to their level of being unbalanced. Wu and the bidder, they ignore the fact that more than one
Xu (2021) utilize unascertained measurement theory unbalanced bid in the same procurement process may
and analytic hierarchy process-fuzzy comprehensive cause average line item prices to be skewed (Arditi
evaluation method to detect quantity error exploitation and Chotibhongs 2009). Indeed, Li et al. (2023) also
type of unbalanced bidding in railway construction assert that the accuracy of the benchmark used in
projects. Thus, their model is different in the sense that detecting unbalanced bids is still a problem that
it does not consider only the risk of deviation in unit should be resolved. Therefore, the objective of the
prices and quantities. study presented in this paper is to modify a traditional
In sum, the literature review on unbalanced bidding decision-making model to overcome the limitation
shows that most researchers do not take full advan related to benchmark issues and the shortcomings of
tage of decision-making methods when researching existing models. Only then can the complexity of
how to detect unbalanced bids. In other words, the detecting unbalanced bids be overcome.
models that use rigorous decision-making methods
are very rare in the unbalanced bidding literature, the
only examples being Su et al.’s (2020a), Wu and Xu’s
Research methodology
(2021) and Li et al.’s (2023) models. Nevertheless, Detecting unbalanced bids is most important in creat
researchers and professionals in the construction ing a fair and competitive bidding process. Therefore,
industry are familiar with these methods. These most this study intends to develop a decision-making model
important advantage of these methods is their help in that allows owners to detect unbalanced bids effi
selecting the most appropriate alternative among a ciently and rapidly. To achieve this goal, an unbal
set of a large number of alternatives while considering anced bid detection model is proposed that makes
several criteria regardless of whether these criteria are combined use of the Information Entropy and VIKOR
contradictory or not. These methods make it easier to methods described in the following sections.
deal with the complexity of decision-making and the
tedious computational problems. Since the detection
Information entropy weight method
of unbalanced bids is solely based on the selection of
an alternative among a set of alternatives, using The information entropy theory was introduced by
advanced decision-making methods can indeed make Shannon in 1948 to measure knowledge uncertainty
it easier to deal with the challenge of detecting the by using probability principles. This theory can be
unbalanced bid accurately. used to calculate the weights of the criteria used in a
The number of studies where decision-making decision-making problem. The most important aspect
methods are used to detect unbalanced bids can go of this theory is that, unlike methods such as expert
6 F. D. AKIN ET AL.
The five steps of the VIKOR method are presented the distance between alternative i and the positive
below (Opricovic and Tzeng 2004): ideal solution a�j ) and Ri represents the minimum indi
vidual regret for each alternative i (i.e. the distance
Step 1: Development of the decision matrix A with n between alternative i and the negative ideal solu
alternatives ði ¼ 1, : . . . , nÞ and m criteria tion a−j ).
ðj ¼ 1, : . . . , mÞ:
2 3
a11 a21 � � � an1 Step 5: Calculation of the VIKOR index Qi for the ith
6 a12 a22 � � � an2 7 alternative.
6 7
A ¼ ðaij Þnxm ¼ 6 .. .. .. .. 7 (5) � � � �
4 . . . . 5 Si − Sþ Ri − Rþ
a1m a2m � � � anm Qi ¼ v − þ ð1 − v Þ − (10)
S − Sþ R − Rþ
where aij represents the value of the ith alternative’s
where Sþ ¼ min Si , S− ¼ max Si , Rþ ¼ min Ri , R− ¼
jth criterion.
max Ri and v corresponds to a subjective assessment
Step 2: Identification of the positive and negative of the ratio of the weights of group utility divided by
ideal solutions of n alternatives with respect to each individual regret. The practical range of v is between 0
criterion j: If A ¼ ðaij Þnxm is considered as the deci and 1, where “0” signifies a complete focus on individ
sion matrix, then the positive (a�j Þ and negative ða−j Þ ual regret, “1” signifies a complete focus on group util
solutions for beneficial criteria are: ity, and “0.5” signifies an equal emphasis on both
(Anvari et al. 2014). The selection of v should be deter
a�j ¼ maxi faij g (6)
mined by the context of the decision, the decision-
a−j ¼ mini faij g maker’s preferences, and the specific requirements of
the positive (a�j Þ and negative ða−j Þ solutions for non- the problem under consideration (Opricovic and
beneficial (i.e. cost) criteria are: Tzeng 2007).
a�j ¼ mini faij g (7)
a−j ¼ maxi faij g Step 6: Rank the alternatives in ascending order based
on their VIKOR indices Qi :
If any of these conditions are not met, then a list of bidders may unbalance their bids by jacking up the
compromise solutions is proposed, which includes the unit price of some line items and reducing the unit
following: price of other line items, hence creating a distribution
that includes some outlier unit prices. It is well known
� If Condition 1 is not met, then Condition 2 is not that mathematically unbalanced bids are not always
met either due to the nature of the calculations of materially unbalanced. For example, a bidder may
the VIKOR method. Therefore, the compromise plan to use a low-cost solution in a line item that was
solution consists of A1 , A2 , A3 . . . AK : AK for max developed over the years as a company-initiated nov
imum K is calculated using Equation 12. elty. Typically, the low unit price in this line item
would be used by the bidder to create a legitimate
1 competitive advantage over the other bidders. At the
QðAK Þ − QðA1 Þ � (12)
n−1 sight of the unusually low unit price in this line item,
the owner’s engineer may suspect an unbalanced bid,
but the owner always performs a detailed examination
If only Condition 2 is not met, the compromise solu to determine whether the bid is materially unbalanced
tion consists of A1 and A2 : or not.
sffiffiffiffiffiffiffiffiffiffiffi
Y n
n
The alternative with the lowest VIKOR Index Qi value ðxij Þgeomean ¼ xij (13)
is regarded as the best alternative. i¼1
proposed model intends to detect unbalanced bids, ðv ¼ 0:5Þ in the “consensus’’ approach, or ðv < 0:5Þ
and therefore the inferior alternative should be deter in the “minimum individual regret” approach.
mined rather than the superior alternative. In other If any of these conditions are not met, then a set of
words, while the lowest VIKOR Index Qi is used to solutions is proposed, which includes the following:
identify the superior alternative, the largest VIKOR
Index Qi is preferred to identify the inferior alternative. � If Condition 1 is not met, the solutions consist of
Thus, in this study, the alternatives are ranked in A1 , A2 , A3 . . . AK , where. AK is calculated using
descending order of their VIKOR Indices Qi and the Equation (12) for maximum K defined as the num
alternative with the largest VIKOR Index Qi is reviewed ber of alternatives n minus the alternative that
to see if it meets the two fundamental conditions of does not meet Condition 1.
the VIKOR method. The steps to perform the VIKOR � If only Condition 2 is not met, the solutions con
method are as follows: sist of A1 and A2 :
Step 1: The decision matrix constructed in the first The bid alternative with the largest VIKOR Index Qi
module is also used in this step. The values in the is regarded as the worst alternative.
decision matrix are normalized before being pre Implementation of the model. The usefulness of the
sented in matrix form, model proposed in this study can be best shown
jxij − ðxij Þ j
where aij ¼ ðxij Þ geomean , where aij represents the nor through a case example. The example that was used
geomean
malized value of the unit price in the ith bid for the jth by Arditi and Chotibhongs (2009) is considered for
line item, xij represents the unit price in the ith bid for this purpose. It consists of five bids submitted by five
qualified bidders (i ¼ 1,2,.,5; where i represents the bid
the jth line item, and ðxij Þgeomean represents the bench
submitted by each individual bidder), each bid con
mark for the jth line item. During the normalization
taining 17 line items (j ¼ 1,2,.,17; where j represents
process, positive and negative ideal solutions are iden
each line item included in the bid). The information
tified in the matrix using Equations (6) and (7).
for each line item is presented in Table 2 and involves
ID number, quantity, bidders’ unit prices, bidders’ total
Step 2: The maximum group utility Si and the min
bid prices, and the engineer’s estimated unit prices.
imum individual regret Ri are calculated for each
These values are input into the model by the owner’s
alternative i using Equations (8) and (9), respectively.
engineer.
Step 3: The VIKOR Indices Qi are calculated for each
It can be seen in Table 2 that even though many of
alternative i using Equation (10).
the prices provided by bidders for a line item are rela
Step 4: The alternatives are ranked in descending
tively close to each other, the unit prices of some line
order of their VIKOR Indices Qi when v ¼ 0:5:
items may differ significantly from each other. For
example, in Bid 1 the price of line item 104 is
The alternative ðA1 Þ with the largest VIKOR Index Qi
$600,000, whereas in Bid 5 it is $300,000, i.e. half of
is the worst solution if all of the following conditions
Bidder 1’s offer. The difference reveals that there is a
are met:
reasonable doubt that Bid 1 may be unbalanced. It
should be noted that the greater the difference
Condition 1. “Acceptable advantage rate”
1 between the unit prices submitted by bidders for a
QðA1 Þ − QðA2 Þ � (14) line item, the more likely it is that the unit prices have
n−1
been manipulated unless the bidders have legitimate
where QðA1 Þ is the largest VIKOR Index for the bid
explanations for overpriced or underpriced line items
alternative on the list, QðA2 Þ is the second largest
such as using different markups for different line items
VIKOR Index for the bid alternative on the list, and n is
in line with the different levels of risk associated with
the number of alternatives.
different line items or having developed novel low-
cost solutions relative to some line items. Once the
Condition 2. “Acceptable stability in decision making”
owner suspects that a bid is mathematically unbal
A1 must also be the first ranked alternative anced, it is in the owner’s interest to run the model to
with respect to the maximum group utility Si and the determine whether the bids are materially unbalanced
minimum individual regret Ri for each alternative i. or not.
This solution is considered to be stable if ðv > 0:5Þ When the first module of the proposed model is
in “the maximum group utility approach”, or run, it automatically calculates the benchmark price
10 F. D. AKIN ET AL.
Table 3. The normalized values in the ith bid for the jth line Table 4. The result matrix of the entropy method.
item.
Information Deviation of
Bidi entropy for the essential information Entropy weight
jth criterion for the jth criterion of the jth criterion
Line itemj Bid1 Bid2 Bid3 Bid4 Bid5 Line item Ej Gj wj
100 0.3615 0.1465 0.1386 0.1536 0.1998 100 0.0477 0.9523 0.0223
101 0.1908 0.2963 0.1544 0.1859 0.1726 101 0.0177 0.9823 0.0083
102 0.2719 0.2185 0.0993 0.3542 0.0561 102 0.1022 0.8978 0.0477
103 0.5333 0.1125 0.1799 0.0444 0.1299 103 0.1966 0.8034 0.0918
104 0.4373 0.1373 0.1129 0.1497 0.1628 104 0.0926 0.9074 0.0433
105 0.3988 0.1191 0.0841 0.3140 0.0841 105 0.1301 0.8699 0.0608
106 0.2481 0.1611 0.2471 0.1673 0.1763 106 0.0117 0.9883 0.0055
107 0.4892 0.0148 0.4303 0.0223 0.0436 107 0.3811 0.6189 0.1780
108 0.3943 0.0880 0.1378 0.1179 0.2620 108 0.0948 0.9052 0.0443
109 0.3026 0.5344 0.0047 0.0675 0.0907 109 0.3032 0.6968 0.1416
110 0.2833 0.2307 0.1089 0.1539 0.2232 110 0.0308 0.9692 0.0144
111 0.1921 0.3207 0.0696 0.0636 0.3540 111 0.1240 0.8760 0.0579
112 0.2781 0.1929 0.1712 0.1867 0.1712 112 0.0115 0.9885 0.0054
113 0.0487 0.1767 0.3402 0.0053 0.4292 113 0.2477 0.7523 0.1157
114 0.2898 0.1628 0.3483 0.0652 0.1339 114 0.0872 0.9128 0.0407
115 0.4101 0.1048 0.0796 0.2979 0.1076 115 0.1276 0.8724 0.0596
116 0.3011 0.1765 0.0050 0.2736 0.2437 116 0.1347 0.8653 0.0629
for each line item using Equation (13) and the bidders’ utility Si (Equation (8)), the minimum individual regret
total bid prices. Then, the proposed model plots the Ri (Equation (9)), and the VIKOR Index Qi (Equation
matrix of normalized values (Table 3) to implement (10)) for each alternative i to rank the alternatives (see
the Information Entropy method (Table 4). The results Table 6). It should be noted that due to the nature of
are used to calculate the weights of the line items. the problem of detecting unbalanced bids, all criteria
The weight of a line item indicates the likelihood that (i.e. 17 line items) should be evaluated as non-benefi
its unit price is unbalanced. cial criteria, where lower values are always preferred.
After calculating the weights of the line items, the When all the required calculations are performed, the
second module of the proposed model should be run second module automatically ranks the bids in
by the owner’s engineer. The decision matrix (see descending order according to their VIKOR Indices Qi
Table 2) constructed in the first module is also used in scores. The bids are ranked in the following descend
the second module of the proposed model. However, ing order: Bid 1, Bid 3, Bid 2, Bid 5, and Bid 4. The
in this module, the normalization process of the deci higher the VIKOR Index Qi score, the more likely it is
sion matrix (Figure 1) also includes determining posi that the bid is unbalanced. A high VIKOR Index Qi
tive and negative ideal solutions (Table 5). Then the score prompts the owner’s engineer to examine the
module automatically calculates the maximum group bid in detail to see whether it is also materially
CONSTRUCTION MANAGEMENT AND ECONOMICS 11
Figure 1. The normalized decision matrix (aij represents the normalized value of the unit price in the ith bid for the jth line item).
Table 5. The positive and negative ideal solutions from the think that Bid 1 (i.e. the lowest bid) is the most advan
matrix. tageous choice. However, the proposed model shows
Positive solutions for Negative solutions for that the significant difference between the unit prices
non-beneficial criteria non-beneficial criteria
Line item a�j ¼ mini faij g a−j ¼ maxi faij g of some line items in Bid 1 and the unit prices of the
100 0.2371 0.6186 same line items in other bids supports the earlier
101 0.4075 0.7823
102 0.0083 0.0522
assertion that Bid 1 is mathematically unbalanced and
103 0.0188 0.2258 it should be inspected to see whether or not it is also
104 0.1479 0.5732
105 0.0213 0.1012
materially unbalanced.
106 0.2704 0.4163 The example presented in Arditi and Chotibhongs’s
107 0.0016 0.0526 (2009) study is also used by Su et al. (2020a). Su
108 0.0228 0.1022
109 0.0020 0.2313 et al.’s model (2020a) considers the engineer’s esti
110 0.0283 0.0735 mates of the unit prices of each line item in their ana
111 0.0118 0.0660
112 0.5034 0.8178 lysis. However, using the geometric mean of the unit
113 0.0010 0.0855 prices offered by the bidders for a line item may be a
114 0.0115 0.0613
115 0.0168 0.0867 more realistic approach than using the engineer’s esti
116 0.0017 0.1047 mate for that line item particularly because this pro
cedure eliminates the effect of extreme variations in
unbalanced, which should result in an early detection the distributions (i.e. the effect of unbalanced line
of unbalanced bids, hence ensuring a fair competitive items that some of the bidders may have used) while
environment. doing away with the inaccuracies that are known to
be in the engineer’s estimates (Beeston 1975, Clark-
Carter 2010). In addition, Arditi and Chotibhongs
Findings and discussion
(2009) also state that considering the engineer’s esti
The results in Table 6 show that Bid 1 has the highest mates as an evaluation benchmark might mislead the
VIKOR Index Qi score which means that Bid 1 is math decision-maker because according to Beeston (1975),
ematically unbalanced. If one evaluates the bids by engineers can sometimes make inaccurate predictions.
considering their total bid prices without paying any While Su et al. (2020a) identified Bid 2 as mathematic
attention to the unit prices of the line items, one may ally unbalanced (using the engineer’s estimates), the
12 F. D. AKIN ET AL.
to calculate the weights of the line items and the Cattell, D.W., Bowen, P.A., and Kaka, A.P., 2007. Review of
VIKOR method to rank the bids according to their like unbalanced bidding models in construction. Journal of
construction engineering and management, 133 (8), 562–
lihood of being unbalanced. The proposed advanced
573.
model is a robust tool that may help owners to detect Cattell, D.W., Bowen, P.A., and Kaka, A.P., 2008. A simplified
unbalanced bids with much confidence. It is also unbalanced bidding model. Construction management and
anticipated that the model will encourage construc economics, 26 (12), 1283–1290.
tion owners to be more sensitive and reactive to Chatterjee, P., and Chakraborty, S., 2016. A comparative ana
unbalanced bidding, and will motivate researchers to lysis of VIKOR method and its variants. Decision science let
ters, 5 (4), 469–486.
conduct more research about detecting and prevent Clark-Carter, D., 2010. Measures of central tendency. In:
ing unbalanced bids. This research can also help con P.L. Peterson, E. Baker, and B. McGaw, eds. International
tractors to revisit and revise their bidding strategies in encyclopedia of education. New York: Elsevier Ltd,
response to owners who may be more cognizant and 264–266.
knowledgeable about the detection and prevention of Ding, X., et al., 2017. Fuzzy comprehensive assessment
method based on the entropy weight method and its
unbalanced bids. application in the water environmental safety evaluation
Not considering the time value of money in the of the Heshangshan drinking water source area, Three
detection of unbalanced bids is a limitation of the pro Gorges Reservoir area, China. Water, 9 (5), 329.
posed model. In addition, not using a real life project Dong, J.Y., Yuan, F.F., and Wan, S.P., 2017. Extended VIKOR
for demonstration purposes is also a limitation of the method for multiple criteria decision-making with linguis
tic hesitant fuzzy information. Computers and ındustrial
study. In future studies, developing a sufficiently reli
engineering, 112, 305–319.
able work schedule at the bidding phase of a project Doran, D., 2004. FMI/CMAA survey of construction ındustry
can be explored that allows the owner to perform a ethical practices. Raleigh, NC: FMI Corporation.
realistic NPV analysis, hence allowing the owner to FHWA. 1988. Bid analysis and unbalanced bids. Available
detect unbalanced bids with more confidence. Also, from: http://www.fhwa.dot.gov/programadmin/contracts/
ta508046.cfm [Accessed 12 June 2023].
the performance of decision-making methods other
Ghaleb, A.M., et al., 2020. Assessment and comparison of
than VIKOR (e.g. TOPSIS) can be explored. The imple various MCDM approaches in the selection of manufactur
mentation of a model that considers multiple bench ing process. Advances in materials science and engineering,
marks simultaneously would be a potential 2020, 1–16.
improvement to this study. Hyari, K.H., 2016. Handling unbalanced bidding in construc
tion projects: Prevention rather than detection. Journal of
construction engineering and management, 142 (2),
Disclosure statement 04015060.
Hyari, K.H., Tarawneh, Z.S., and Katkhuda, H.N., 2016.
No potential conflict of interest was reported by the Detection model for unbalanced pricing in construction
author(s). projects: a risk-based approach. Journal of construction
engineering and management, 142 (12), 04016078.
Hyari, K.H., 2017a. The controversy around unbalanced bid
Data availability statement ding in construction: seeking a fair balance. Journal of pro
The authors confirm that the data supporting the findings of fessional ıssues in engineering education and practice, 143
this study are available within the article and its supplemen (1), 04016015.
tary materials. Hyari, K.H., 2017b. Owner’s countermeasures to skewed bid
ding in construction projects: Review of current practices
and proposal for new countermeasures. Journal of man
References agement in engineering, 33 (3), 04016053.
Hyari, K.H., and Alamayreh, T., 2023. Unbalanced bidding in
An, X., et al., 2018. Identification and prevention of construction projects: a contractors’ perspective.
unbalanced bids using the unascertained model. Journal International journal of construction management, 23 (12),
of construction engineering and management, 144 (11), 2058–2066.
05018013. Lan, S., Yang, C., and Huang, G.Q., 2017. Data analysis for
Anvari, A., Zulkifli, N., and Arghish, O., 2014. Application of a metropolitan economic and logistics development.
modified VIKOR method for decision-making problems in Advanced engineering ınformatics, 32, 66–76.
lean tool selection. The ınternational journal of advanced Li, H., et al., 2021. Detecting unbalanced bidding to achieve
manufacturing technology, 71, 829–841. economic sustainability using fuzzy logic approach.
Arditi, D., and Chotibhongs, R., 2009. Detection and preven Construction ınnovation, 21 (2), 164–181.
tion of unbalanced bids. Construction management and Li, H., et al., 2023. The framework of data-driven and multi-
economics, 27 (8), 721–732. criteria decision-making for detecting unbalanced bidding.
Beeston, D., 1975. One statistician’s view of estimating. Engineering, construction and architectural management,
Chartered surveyor, 2 (4), 49–54. 30 (2), 598–622.
14 F. D. AKIN ET AL.
Mehta, M., Scarborough, W., and Armpriest, D., 2009. Su, L., et al., 2020a. Multi-criteria decision making for identifi
Building construction: principles, materials, and systems. cation of unbalanced bidding. Journal of civil engineering
2nd ed. Columbus, Ohio, USA: Pearson Prentice Hall. and management, 26 (1), 43–52.
Nikpour, B., Senouci, A., and Eldin, N., 2017. Detection tool Su, L., et al., 2020b. Identification of unbalanced bids based
for unbalanced bids. Open journal of civil engineering, 7 on grey-fuzzy evaluation method. Canadian journal of civil
(03), 409. engineering, 47 (3), 272–278.
Opricovic, S., and Tzeng, G.H., 2004. Compromise solution by Taheriyoun, M., Karamouz, M., and Baghvand, A., 2010.
MCDM methods: A comparative analysis of VIKOR and Development of an entropy-based fuzzy eutrophication
TOPSIS. European journal of operational research, 156 (2),
index for reservoir water quality evaluation. Journal of
445–455.
environmental health science and engineering, 7 (1), 1–14.
Opricovic, S., and Tzeng, G.H., 2007. Extended VIKOR method
Wang, W.C., 2004. Electronic-based procedure for managing
in comparison with outranking methods. European journal
of operational research, 178 (2), 514–529. unbalanced bids. Journal of construction engineering and
Polat, G., Turkoglu, H., and Damci, A., 2018. Structural system management, 130 (3), 455–460.
selection using the ıntegration of multi-attribute-decision- Wang, W.C., et al., 2006. Unit-price-based model for evaluat
making (MADM) methods. Periodica Polytechnica ing competitive bids. International journal of project man
Architecture, 49 (1), 38–46. agement, 24 (2), 156–166.
Polat, G., Turkoglu, H., and Damci, A., 2019. A grading sys Wu, G., et al., 2017. Integrated sustainability assessment of
tem-based model for detecting unbalanced bids during public rental housing community based on a hybrid
the tendering process. Periodica polytechnica architecture, method of AHP-entropy weight and cloud model.
50 (2), 139–147. Sustainability, 9 (4), 603.
Polat, G., et al., 2020. Detecting unbalanced bids via an Wu, X., and Xu, F., 2021. Detection model for unbalanced
improved grading-based model. Organization, technology bidding in railway construction projects: considering the
and management in construction, 12 (1), 2072–2082. risk of quantity variation. Journal of construction engineer
Shannon, C.E., 1948. A mathematical theory of communica ing and management, 147 (7), 04021055.
tion. Bell system technical journal, 27 (3), 379–423. Yin, Y. L., Qiao, L., and Li, B. (2010). Research on the owner’s
Shemshadi, A., et al., 2011. A fuzzy VIKOR method for sup
tactics to unbalanced bid under the mode of Code of
plier selection based on entropy measure for objective
Valuation with Bill Quantity of construction works. 17th
weighting. Expert systems with applications, 38 (10),
12160–12167. International Conference on Industrial Engineering and
Skitmore, M., and Cattell, D., 2013. On being balanced in an Engineering Management, Xiamen, China, 249–252.
unbalanced world. Journal of the operational research soci Zhang, X., et al., 2014. Assessment model of ecoenvironmen
ety, 64 (1), 138–146. tal vulnerability based on ımproved entropy weight
Stramarcos, A., and Cattell, D., 2013. The ethics of item pric method. The scientific world journal, 2014, 1–7.
ing. 38th Australian University Building Educators Zhu, Y., Tian, D., and Yan, F., 2020. Effectiveness of entropy
Association Conference. New Zealand: University of weight method in decision-making. Mathematical prob
Auckland, 1–14. lems in engineering, 2020, 1–5.