REGRET
AND INFORMATION AT FIRST PRICE AUCTIONS: A STUDY WITH BRAZILIAN DECISION-MAKERS
Glaucia
Miura Ota
Instituto
Federal de São Paulo - Campus Suzano, Brazil
E-mail: mglaucia@hotmail.com
Adriano
Maniçoba da Silva
Instituto
Federal de São Paulo - Campus Suzano, Brazil
E-mail:
adrianoms@ifsp.edu.br
Submission: 2/20/2020
Revision: 2/29/2020
Accept: 3/6/2020
ABSTRACT
Studies point out that the utility of the
decision maker in the results depends on the profit. However, acting rationally
in an auction environment can be difficult, and the reasons may be due to
cognitive aspects such as regret. The "Winner's Regret" bias is
characterized by the regret that the winner suffers for bidding too high
compared to his opponent. The "Loser’s regret" comes from realizing
that his very low bid made him lose. Therefore, the objective of this study was
to investigate the effect of regret and information on first price auctions in
Brazilian individuals and compare with studies conducted abroad, checking if
there is a difference in bias amplitude. Initially, a literature review on
behavioral operations, auctions and the effect of regret in auctions was
performed. Regarding the methodology, the sample applied was non-probabilistic
and for convenience. The experiment design had four treatments: Winner's
Regret, Loser's Regret, Both and None, manipulated in a 4x1 experiment with
college students. Regarding the winner's regret, the results were similar to
studies abroad and in the hypothesis of the loser's regret the results were
different.
Keywords: Behavior, Regret in auctions, Production.
1.
INTRODUCTION
Studies related to the Behavioral Operations
area that investigate the subjective aspects that occur in decision making have
been developed abroad. These characteristics impact the operational performance
of organizations.
Organizations use the purchase
function to operationalize their processes. Auctions are one of the ways to
obtain raw materials, elements, finished products and services, and through
them, a large number of deals are closed by companies.
By combining two characteristics,
the auctions are classified as: bid information and the amount to be paid by
the winner. The types of auctions found are: open ascending price or English
auction; open descending auction or Dutch auction; first price closed auction
and second price closed auction.
Although research carried out in an
auction environment shows that the utility of the decision maker in the results
depends on the profit, acting rationally can be difficult, considering that
several factors must be considered, and it is not an easy task to define the
ideal bid. Thus, decision makers do not always maximize their results, and the
cause of this behavior may be due to cognitive aspects. As an example, when
making a decision, a person may regret not having made another; that’s when the
feeling of regret occurs.
Studies related to behavioral
biases present in auctions allowed to identify an individual behavior called
"Winner's Regret", which occurs in first price auctions, and is
characterized by the regret that the winning bidder suffers from having offered
a bid too high in comparison to its opponent. On the other hand, the “Regret of
the Loser” was identified, whose regret occurs when he realizes that his very
low bid made him lose the opportunity to win (ENGELBRECHT-WIGGANS; KATOK,
2008).
Thus, the objective of this
research will be to investigate the regret and information in first price
auctions in Brazilian individuals and to compare it with studies carried out
abroad, checking if there is a difference in the spectrum of the bias.
Initially, a literature review was
made on behavioral operations, an approach to auctions and the effect of regret
at auctions.
2.
LITERATURE REVIEW
In this section, behavioral operations, auctions and the effect of regret at
auctions will be covered.
2.1.
Behavioral Operations
Behavioral Operations (OC)
refers to a sub-area belonging to Production Management, whose research
programs are aimed at studying the subjective aspects of the subjects in
decision making, and the impact caused on the operational performance of
organizations (SILVA, 2015). Although this area of knowledge
developing abroad, in Brazil it is still non-existent (DA SILVA, 2015).
As found by the work
produced by Bendoly, Donohue and Schultz (2006), the study of OC has a record
of articles published since 1985 in renowned journals on this theme.
The authors Lee, Seo and
Siemsen (2018) researched articles that carried out experimental studies in the
laboratory in the context of behavioral operations in the period comprehended
between 2006 to 2016, and found as a result that experiments in the context of
more relevant operations in number of average annual citations and total
citations were three: the newsvendor model studied by Bolton and Katok (2008),
another study on supply chain contracts (LOCH; WU, 2008) and the purchase
auction experiment (ENGELBRECHT-WIGGANS; KATOK, 2008). Still according to the
authors, the experiment of buying auctions allows the behavioral study of the
individual bias when relating to opponents played by the computer.
A study produced nationwide
on the newspaper vendor model can be found in Ota and Da Silva (2017), where
the work carried out by Bolton and Katok (2008) and Feng, Keller and Zheng
(2011) was reproduced, with in order to make a comparison of research conducted
abroad with their study. The experiment in supply chain contracts was also
carried out nationwide by authors Ota and Da Silva (2019). However, research
made from national databases did not present studies related to the theme of
the first price auction experiment.
Thus,
this study addresses the experiment regret and information in first price
auctions.
2.2.
Auctions
In the context of a supply
chain, the purchase function is a process in which supplies, materials,
finished products, services and goods are obtained by organizations in order to
operationalize their processes (CHOPRA; MEINDL, 2011; DA SILVA, 2008), and this
activity characterizes an element important in achieving the objectives of a
company (CAXITO, 2014). To this end, it must then decide whether to manufacture
the product internally or to outsource, in order to increase competitiveness
and minimize costs (CAXITO, 2014). One way to obtain these goods is through
auctions.
Companies commonly use this
method of buying and selling to purchase supplies, or subcontract services. A
large volume of negotiations is held through auctions (KLEMPERER, 1999).
These auctions are
classified by combining two characteristics: bid information and the amount to
be paid by the winner. In the first category, the auction can be opened, which
occurs when everyone has access to the bid amounts, which are communicated by
the auctioneer; or closed, where bids are sent to the auctioneer through sealed
envelopes. In the second category, although the winner is always the one who
launched the best offer, the amount to be paid by the winner can either be the
value of the best bid or the second best offer offered, depending on the
context in which it occurs.
Auction types in the
business-to-business (B2B) context, that is, trade between companies, are a
combination of these four mechanisms:
•
Open rising price or English
auction - starts with a low price and increases successively;
•
Open descending auction or Dutch
auction - works through values that gradually decrease;
•
First price closed auction - bid
proposals are sent via a sealed envelope, and the highest value wins and;
•
Second price closed auction - in
this type of auction, also known as the Vickrey auction, the highest bidder is
the winner, but he pays the second highest bid (DALY; NATH, 2005).
A
problem that can occur in the auctions of the first sealed price is known as
the winner's curse, and this phenomenon does not happen in any open auction,
much less in the auction for the second price, and is characterized by the
regret bias (CHOPRA; MEINDL, 2011).
2.3.
Effect of Regret at Auctions
The understanding that
emotions such as Regret can play a relevant role in situations where decision
making occurs under uncertainty has a long history (ENGELBRECHT-WIGGANS; KATOK,
2008).
Therefore, the subjects do
not always make their decisions maximizing profits, and the reasons point that
the cognitive aspects may be responsible for this behavior (BELL, 1982). In
this way, a person, after making a decision and knowing the results, can
perceive that another choice could have been more favorable, and left with a
feeling of regret or loss (BELL, 1982). Thus, it can be difficult to act
rationally in an auction, whose environment has factors to be considered and
deciding the ideal bid is not a simple task (THALER, 1988).
According to Bell (1983),
decision-makers, in an environment under uncertainty, are afraid to make a
wrong decision, and to avoid the consequences and annoyances that can result
from regret for that decision, they may be prone to paying a financial premium.
His study also signals the likelihood of aversion to regret being part of risk
aversion behavior.
On the other hand, in
addition to the risk-aversion behavior observed in the competitors of an
auction, in his model, Morgan et al (2003) considered the presence of a grudge,
assuming that the bidder, when winning an auction, is not concerned only with
his excess, but also, in case of losing, with the excess of its opponents.
Although most auction model
studies claim that the bidder's utility in the outcome depends on profit,
Engelbrecht-Wiggans (1989) suggests that the various types of regret suffered
in this context also affect bidding decisions.
Experiments performed in
behavioral operations research made it possible to study and identify an
individual bias in closed price acquisition auctions (LEE; SEO; SIEMSEN, 2018).
One of the detected behaviors was the “Winner's Regret”, which occurs when the
successful bidder of a first price auction regrets having placed a bid too high
compared to his closest opponent. Another behavior observed was the “regret of
the Loser”, which occurs when the participant loses the opportunity to win by
having made a very low bid (ENGELBRECHT-WIGGANS; KATOK, 2008).
In the Engelbrecht-Wiggans
and Katok (2008) experiment, a first price auction environment was simulated,
with four treatments, and the participants made bids against two opponents
represented by the computer. Computerized participants assigned values
from 1 to 100 distributed evenly and according to the Nash
equilibrium. Each participant made 20 decisions for each of five resale values
(50, 60, 70, 80 and 90), totaling 100 bids. Information was
provided separately for treatments. In two of them, the “winning bid” and
“profit if winning” were informed; in two other treatments, the “Value more
than the second bid” and the “Second highest bid” were announced. In another
treatment, no feedback information was communicated.
The model formulated by the
authors (ENGELBRECHT-WIGGANS; KATOK, 2008), consists of the Winner's Regret
hypothesis and the Loser's Regret hypothesis.
In the first case, the
Winner's Regret hypothesis predicts that:
·
Average bids in Treatment Both
are lower than in Loser's regret Treatment;
·
Average bids on the Winner's
Regret Treatment were lower than in the No feedback treatment
Whereas in the case of the
Loser's Regret it is predicted that:
·
Average bids on Treatment Both
are relatively higher than on Winner's Regret Treatment and;
- The average bids in the Loser Regret Treatment are higher than in
the No feedback treatment
3.
METHODOLOGY
In order to achieve the
objectives proposed in this study, whose variables have already been studied by
previous studies, the sample applied was of the non-probabilistic and
convenience type.
Experiments similar to
those conducted by Engelbrecht-Wiggans and Katok (2008), and Lee, Seo and
Siemsen (2018) were carried out, in order to identify a behavioral bias of
decision makers in an environment of acquisition by auction of the first sealed
bid price, and thus, compare the results collected with those obtained in
studies conducted abroad.
The design of the
experiment had four treatments: None, Regret of the Winner, Regret of the Loser
and Both (that is, Regret of the winner and Regret of the loser), which were
manipulated in a 4x1 experiment, with students of higher education. Students
from a higher technology course at a public higher education institution in the
state of São Paulo participated in the research. In total, 53 students
participated in the survey, 52.83% were male, and the average age was 22 years.
In this study on decision
making in first price auctions, each participant made bids against two computer
simulated opponents, N = 3. The computer opponents' bid values
were integers evenly distributed from 1 to 100, as well as the
values of the students who participated in the event. Resale
values were set at 50, 60, 70, 80 and 90. Each session consisted
of 1000 auctions in blocks of 10. Students made 100 bid decisions, and each
decision was used in 10 consecutive auctions. The resale value was the same for
20 consecutive decisions (200 auctions). The calculation of the profit in the
auction was done as follows:
When the auction did not
win, the profit was 0 (zero).
Like the studies conducted
by Engelbrecht-Wiggans and Katok (2008), and Lee, Seo and Siemsen (2018), this
work verified the effect of providing feedback information to participants
regarding winner’s and loser’s regret. According to previous studies, offering
certain information can influence decision making, because for a participant
who lost the round, being informed of the winning bid gives him the opportunity
to understand which bidding decision would make him win at a favorable price.
However, by not being notified of the winning bid, you can underestimate that
amount.
In the model, bids in
different treatments were compared, where information that can produce the
winner's regret and the loser's regret was displayed simultaneously,
alternately or not communicated.
The following feedback
information was determined:
•
In the Loser's Regret treatment,
the loser's regret value was calculated and informed to the participant, which
was the “Profit if he won” and also the value of the “Winning bid”;
•
In the Winner's Regret treatment,
the winner's Regret value was calculated, which is the “Value over the second
bid”, that is, the difference between the winning bid amount and the second
highest bid of the round, and in addition to the winner's Regret amount, the
“second highest bid” was announced;
•
In treatment “Both”, all the
above information was communicated to the participants in this treatment.
This design consists of a
replica of the study by Engelbrecht-Wiggans and Katok (2008) and Lee, Seo and
Siemsen (2018), in order to verify whether there was a significant difference
in the behavior of Brazilian decision-makers.
4.
RESULTS ANALYSIS
As in the studies by
Engelbrecht-Wiggans and Katok (2008) and Lee, Seo and Siemsen (2018), the
impact of the behavioral bias "Winner's Regret" and "Loser's
Regret" on the different treatments of this experiment was verified. The
values, decisions and evolution during the rounds were also checked.
Graph1:
Average bid / Average value resale for 20 treatment decisions
Source: Prepared by the
study authors
Graph 1 shows the average
bid / average resale value on the 20 decisions that participants made with the
same resale value. Each point on the graph corresponds to the average bid /
average resale value for participants and five values. The graph provides an
insight into how bids have evolved over time in response to the information
provided.
The average obtained from
the average bid / average resale value in each treatment is as follows:
Treatment Both: 0.6107, Treatment None: 0.7857, Winner's Regret Treatment:
0.6429 and Loser's Regret Treatment: 0.7143. In the same way as the original
study, a regression analysis was performed to verify if there was a decrease in
the value of the average bids in the Winner's Regret Treatment, however the
results indicated the model's non-significance. The analysis to find support
for these results is based on the average bid calculations in the 20 decisions
(as shown in Table 1).
Table
1. Summary of theoretical predictions and experimental results.
|
|
Treatment both |
Winner's Regret
Treatment |
Loser's Regret
Treatment |
Treatment none |
|
|
|
Decisions 1-20 average bid / value (standard deviation) |
0,6107 (0,0245) |
0,6429 (0,0293) |
0,7143 (0,7142) |
0,7857 (0,7143) |
P Value |
|
|
Decisions 11-20 average bid / value (standard deviation) |
0,6071 (0,0278) |
0,6286 (0,027) |
0,7214 (0,0331) |
0,7857 (0,0139) |
Decisions |
|
|
Hypotheses |
Forecast |
|
|
|
|
1-20 |
|
1: Winner's Regret |
Treatment both < Loser's Regret Treatment |
Was lower |
|
Was lower |
|
p<0,01 not rejected |
|
Winner's Regret Treatment< Treatment None |
|
Was lower |
|
Was lower |
p<0,01 not rejected |
|
|
2: Loser's Regret |
Treatment both > Winner's Regret Treatment |
Was not higher |
Was not higher |
|
|
Rejected |
|
Loser's Regret
Treatment > Treatment None |
|
|
Was not higher |
Was not higher |
Rejected |
|
Source: Prepared by the
study authors
The average bid / average
resale value can be seen in Table 1 as well as the corresponding standard
deviation (shown in parentheses and in line 2). The bottom four lines refer to
a comparison between the two treatments, as shown in the second column. The
p-values presented in the last two columns relate to the result
of the t test, which compares the average purchase value / value in the two
treatments. The observation unit is the average bid / average resale value for
an individual participant in all 20 rounds in column 7.
5.
DISCUSSION
Unlike the theory presented
and the result obtained in previous studies, the average bid / average resale
value was above the neutral Nash balance only in two of the four treatments
(Loser Regret Treatment and Treatment None), and the bids remained high even in
last half of the decisions.
Table 1 points out that the
Winner's Regret hypothesis is supported, which has two predictions. As in the
studies by Engelbrecht-Wiggans and Katok (2008) and Lee, Seo and Siemsen
(2018), both the prediction that the bids would be lower in Treatment Both than
in Treatment Loser’s Regret (not rejected with p <0,01), regarding the
prediction that the bids would be lower in the Winner's Regret Treatment and in
the Treatment None (not rejected with p <0.01), both converged according to
forecasts.
In the case of the Loser's
Regret hypothesis, which in one prediction would result in higher bids in
Treatment Both than in the Winner's Regret Treatment, and in the other, the
bids would be higher in the Loser's Regret Treatment than in Treatment None,
both hypotheses in this study contradict the predictions of Engelbrecht-Wiggans
and Katok (2008).
In addition, another theoretical
prediction predicts that the average bids in the Winner's Regret Treatment
would result in lower average bids, while in the Loser's Regret Treatment the
average bids would be higher. In this study, the average bids on the Winner's
Regret Treatment were lower compared to the Loser's Regret Treatment, and not
on the four treatments, and the average bids on the Loser's Regret Treatment
were higher compared to the Winner's Regret Treatment, and not the largest
among the four treatments.
6.
CONCLUSIONS
Although most studies of
auction models state that the bidder's utility in the result depends on profit,
there is an understanding that emotions such as the various types of regret
suffered in this context also affect bidding decisions (ENGELBRECHT-WIGGANS,
1989). Therefore, the subjects do not always make their decisions maximizing
profits (BELL, 1982).
The present study aimed to
verify the effect of the behavioral bias known as "Winner's Regret"
and "Loser's Regret" in decision making in the context of an auction.
According to the results of
the experiment, it was found that the subjects deviate from the amount that
maximizes profit due to the effects of this bias.
The biggest differences
found contradict the two predictions of the Loser Regret hypothesis:
•
a forecast that the average bids
for Treatment Both (0.6107) were no higher than for the Winner's Regret
Treatment (0.6429) and;
•
another forecast in which the
average bids in the Loser Regret Treatment (0.7143) were not higher than in the
Treatment None (0.7857).
In addition, the average
bids on the Winner's Regret Treatment were lower compared to the average bids
on the Loser's Regret Treatment, and not compared to the four treatments. The
same happened with the average bids of the Loser's Regret Treatment, which were
higher only in relation to the Winner's Regret Treatment, and not among the
four treatments, as the hypothesis predicts.
Furthermore, the average
bids in the Winner's Regret Treatment and Treatment Both are below the neutral
Nash equilibrium, diverging from the original study, where all treatments had
average bids with values above the neutral Nash equilibrium.
This study can be
considered innovative in the national context, due to the fact that it was the
first to analyze regret and information in first price auctions and comparing
it with other works carried out abroad.
As a limitation to the
study, the number of decisions that were made stands out, which may have
dispersed the participants' concentration during the experiment, as it is very
numerous; another limiting factor was the distribution and number of students
recruited in the survey.
Thus,
as a suggestion for future research, it is recommended to replicate this study
with a larger number of participants.
REFERENCES
BELL, D. E. (1982) Regret in decision
making under uncertainty. Operations
research, v. 30, n. 5, p. 961-981.
BELL, D. E.
(1983) Risk premiums for decision regret. Management Science, v. 29, n. 10, p. 1156-1166.
BENDOLY,
E.; DONOHUE, K.; SCHULTZ, K. L. (2006) Behavior
in operations management: Assessing recent findings and revisiting old
assumptions. Journal of operations management, v.
24, n. 6, p. 737-752.
BOLTON, G.
E.; KATOK, E. (2008) Learning
by doing in the newsvendor problem: A laboratory investigation of the role of
experience and feedback. Manufacturing & Service Operations
Management, v. 10, n. 3,
p. 519-538.
CAXITO, F. A.
(Coord.) (2014)
Logística: um enfoque prático, 2 ed. São Paulo:
Saraiva.
CHOPRA, S.; MEINDL, P. (2011) Gestão da cadeia de suprimentos: estratégia, planejamento e operações, 4 ed. São Paulo:
Pearson.
DALY, S.
P.; NATH, P. (2005) Reverse
auctions for relationship marketers. Industrial Marketing Management,v.
34,
n. 2, p. 157-166.
DA SILVA,
A. M. (2008) Distribution channel structure: an overview of determinants. Revela Journal, v. 2, n. 3, p. 1-13.
DA SILVA,
A. M. (2015) Operaçöes
Comportamentais: Oportunidades de Pesquisa no Contexto Brasileiro. In: Seminários em Administraçâo, 18, Säo Paulo, Anais...São Paulo: SemeAd, 2015.
ENGELBRECHT-WIGGANS,
R. (1989)
The effect of regret on optimal bidding in
auctions. Management Science, v. 35, n. 6, p. 685-692.
ENGELBRECHT-WIGGANS,
R.; KATOK, E. (2008) Regret
and feedback information in first-price sealed-bid auctions. Management
Science, v. 54, n. 4,
p. 808-819.
FENG, T.;
KELLER, L. R.; ZHENG, X. (2011) Decision
making in the newsvendor problem: A cross-national laboratory study. Omega, v.
39, n. 1, p. 41-50.
KLEMPERER,
P. (1999)
Auction theory: A guide to the literature. Journal
of economic surveys, v. 13,
n. 3, p. 227-286.
LEE, Y.
S.; SEO, Y. W.; SIEMSEN, E. (2018) Running
behavioral operations experiments using Amazon's Mechanical Turk. Production
and Operations Management, v. 27,
n. 5, p. 973-989.
LOCH, C. H.; WU, Y. (2008)
Social
preferences and supply chain performance: An experimental study. Management Science, v. 54, n. 11,
p. 1835-1849.
MORGAN, J.; STEIGLITZ, K.; REIS, G. (2003)
The spite
motive and equilibrium behavior in auctions. Contributions in Economic Analysis & Policy, v. 2, n. 1.
OTA, G. M.; DA SILVA, A. M. (2017) Efeito centralização no comportamento de decisores
brasileiros. In: Congresso de Logística, 3, Suzano, Anais...Suzano: IFLOG, 2017.
OTA, G.
M.; DA SILVA, A. M. (2019) Social preferences and supply chain transactions: an analysis with
Brazilian decision makers. Independent Journal of Management &
Production, v. 10, n. 4,
p. 1288-1304.
SILVA, G. (2015) Operaçöes comportamentais: perspectivas
teóricas. In: Seminários em
Administraçâo, 18, Säo Paulo, Anais...São
Paulo: SemeAd, 2015.
THALER, R.
H. (1988) Anomalies: The winner's curse. Journal
of Economic Perspectives, v. 2, n. 1, p. 191-202.