Gender differences in experimental wage negotiations.
Dittrich, Marcus ; Knabe, Andreas ; Leipold, Kristina 等
I. INTRODUCTION
Gender wage gaps have been the subject of extensive research in
labor economics. Although this gap has narrowed in the last decades,
Blau and Kahn (2003) report that women still receive lower wages than
men in many countries. Traditionally, differences in the accumulation of
human capital and discrimination are seen as the main factors
responsible for this wage gap (Bertrand 2011). However, even after
controlling for various measures of human capital, the wages of men and
women differ significantly and this wage gap persists over time.
Recently, alternative explanations have been analyzed more closely, such
as gender differences in preferences (Croson and Gneezy 2009), risk
attitudes (Eckel and Grossman 2008), and responses to competitive
situations (Gneezy, Leonard, and List 2009).
In many labor relationships, wages are determined by bilateral
bargaining. Hence, another source of the persistent gender wage gap
could be differences in bargaining behavior between women and men.
Babcock and Laschever (2003) find that 57% of male graduate business
students bargain over their starting salaries, while only 7% of women do
so. Consequently, male starting salaries are 7.6% higher than those
obtained by women. Investigating the salary negotiating behaviors and
starting salary outcomes of graduating Master of Business Administration
(MBA) students, Gerhart and Rynes (1991) find that women obtain lower
salary increments when negotiating than men. Wood et al. (1993) report
significant male-female pay differences of law school graduates, both 1
year and 15 years after graduation. Apart from wages, team production,
hiring, and promotion decisions are further domains where significant
gender differences emerge (Sutter et al. 2009). At least some of these
differences might also be related to differences in the bargaining
behavior of men and women.
This paper studies whether men and women behave differently in wage
negotiations. (1) The analysis is based on a laboratory experiment with
face-to-face alternating-offers wage bargaining. Allowing experimental
subjects to see each other in face-to-face negotiations, and thus to
easily identify their bargaining partner's gender, allows us to
analyze gender differences. Our results suggest that gender and gender
pairing play an important role in bargaining processes. In our
experimental wage negotiations, male employees are able to negotiate
higher wages than female employees. Moreover, male employers pay lower
wages to female employees than female employers pay to male employees.
We find that men acting as employees achieve more favorable outcomes
than female employees because they already demand a larger share for
themselves when making their first offers. When acting as employers, we
do not find behavioral differences between men and women. Overall, these
results provide further evidence that part of the gender wage gap can be
attributed to behavioral differences in wage negotiations.
The paper is organized as follows. Section II provides a brief
review of the literature on gender differences in bargaining
experiments. In Section III, we present our experimental design. Section
IV discusses the effects of gender and gender pairing in the bargaining
process. Section V concludes.
II. GENDER DIFFERENCES AND BARGAINING
A number of studies have examined differences in the behavior of
women and men in bargaining situations. In a controlled field study,
Ayres and Siegelman (1995) analyze the bargaining outcomes of men and
women negotiating the price of a new car. They report that car sellers
offer male buyers lower prices than female buyers. Ayres (1991) finds
that women obtain worse bargaining outcomes than men negotiating with a
female seller. Save-Soderbergh (2009) finds that female university
graduates propose lower wage bids than male university graduates and
also receive lower wage offers than male students. Conducting field
experiments in Lima (Peru), Castillo et al. (2013) examine whether male
taxi drivers differ in their treatment of male and female passengers.
Using a group of trained buyers and holding their behavior fixed,
Castillo et al. (2013) analyze whether the seller's negotiation
path differs between customers of both genders. In contrast to previous
studies, women obtain better bargaining outcomes than men, that is, they
are quoted lower initial prices and are less likely to be rejected by
the drivers. Castillo et al. (2013) argue that taxi drivers perceive men
to be passengers with a high willingness to pay. The observed gender
differences, however, are also consistent with taxi drivers having a
relative preference for female passengers and engaging in taste-based
discrimination.
Gender effects on bargaining behavior have also been studied in
laboratory experiments where driving forces of men's and
women's behavior can be isolated (for an overview see Croson and
Gneezy 2009; Eckel and Grossman 2008). In a dictator game, Bolton and
Katok (1995) and Carpenter, Verhoogen, and Burks (2005) find no
differences in the behavior of men and women. Other studies report that
women are more inequality averse and less selfish than men in dictator
games (Eckel and Grossman 1996, 1998; Fehr, Naef, and Schmidt 2006;
Selten and Ockenfels 1998). Andreoni and Vesterlund (2001) shed light on
the mixed results by varying the price of being altruistic. Women tend
to equalize payoffs and are more generous when giving is expensive while
men are more generous when giving is rather cheap (see also Dickinson
and Tiefenthaler 2002). Ben-Ner, Kong, and Putterman (2004) report no
difference in giving between men and women when the gender of the
receiving person is unknown. However, they find that women give less
when giving to a woman than to a man.
Gender effects have also been examined in ultimatum games, as a
stylized representation of bargaining. Eckel and Grossman (2001) apply a
repeated design where partners face each other. Players are matched with
partners of both the same and the opposite gender. Similarly, Solnick
(2001) applies a one-shot game with two treatments using the strategy
method, where the proposer makes an offer and the receiver
simultaneously notes the minimum acceptable offer. In the first
treatment, the partner is unknown. In the second treatment, players know
the partner's first name but they do not face each other. Neither
Eckel and Grossman (2001) nor Solnick (2001) find significant
differences in overall mean offers between men and women. Concerning the
gender pairing, they find lower offers proposed to women than to men. In
both studies, the rate at which women reject an offer depends on the
proposers' gender. Eckel and Grossman (2001) observe the lowest
rejection rates for women paired with women, whereas Solnick (2001)
reports the highest rejection rates for this group. These conflicting
results could be because of differences in the experimental design
(Eckel and Grossman 2001), driven by effects of visual expression
(Sutter et al. 2009) or due to the larger context sensitivity of women
(Croson and Gneezy 2009). Guth, Schmidt, and Sutter (2007) conduct a
three-person ultimatum newspaper experiment and find that women prefer
the equal split solution more often than men. Garcia-Gallego,
Georgantzis, and Jaramillo-Gutierrez (2012) apply an ultimatum game
framed as a firm-worker bargaining situation and focus on risk
attitudes. Their results suggest that both gender and risk-related
effects co-exist, but differences in risk-attitudes cannot explain
gender effects. Castillo and Cross (2008) find, in a series of ultimatum
and dictator games, that male behavior exhibits more variation than
female behavior. These gender differences are due not only to social
preferences, as e.g., altruism, but also to beliefs about other
players' strategic behavior.
Apart from dictator and ultimatum games, gender effects have also
been studied in the trust game. In this game, the first player can
transfer money to a second person. The sum sent is multiplied by three.
After that, the second player can send any amount back to the first
player. While some studies report that men send more money than women,
other studies report the opposite result or find no differences (for an
overview, see Croson and Gneezy 2009).
Sutter et al. (2009) conduct a power-to-take game in order to study
gender pairing effects. This bargaining game is played by two persons
over two stages. In the first stage, a "take authority"
defines the take rate, that is, the proportion of the responder's
endowment that the take authority receives at the end of the game. In
the second stage, the responder decides how much of the endowment he
will destroy. Sutter et al. (2009) find no significant differences
between women and men concerning take and destruction rates or frequency
of destruction. When bargaining with the same gender, however, take
authorities claim more from responders and destruction rates are higher.
Sutter et al. (2009) note that these gender pairing results are in line
with predictions from evolutionary psychology as human behavior complies
with survival and reproductive success. Persons with the same gender see
themselves as competitors for the opposite gender, which explains the
observed competitive behavior in the experiment towards the same gender.
Summing up, the evidence on gender differences in laboratory
bargaining experiments does not provide a clear-cut picture (Camerer
2003). Croson and Gneezy (2009) suppose that mixed results might occur
because women are more risk-averse than men. A second explanation is the
higher context-sensitivity of women, that is, their social preferences
vary stronger within experimental treatments and between different
experiments than those of men (Croson and Gneezy 2009). Moreover, men
and women differ in their inclination to engage in competitive
interactions, i.e., women have a smaller propensity to enter competitive
situations (Gneezy, Leonard, and List 2009; Gupta, Poulsen, and Villeval
2013; Niederle and Vesterlund 2007) which has a strong impact on
bargaining behavior (Croson and Gneezy 2009).
In this paper, we provide more evidence for the existence of gender
differences in bargaining behavior. In contrast to the previously used
dictator, ultimatum, and trust games, we apply a multistage
alternating-offer bargaining game in a labor market context. With
multistage bargaining, we aim to represent real-life bargaining which is
typically characterized by a series of offers and counteroffers and
step-by-step bilateral convergence. The multistage design allows testing
a variety of labor market and bargaining aspects. In addition to
bargaining outcomes, we also examine first offers of employers and
employees, their counteroffers, and the impact of an external breakdown
on the bargaining process.
III. EXPERIMENTAL DESIGN
A. Experimental Wage Negotiations
In our experiment, a "worker" and a "firm"
bargain over the division of a rent. Participants are told that the firm
is offered a contract by some customer worth 300 experimental currency
units ("tokens"). To fulfill the contract, the firm needs to
hire a worker. There is only one worker available, with whom the firm
has to agree on a wage before hiring. The rent of 300 tokens can be
split between the firm and the worker via bargaining over the
worker's wage w. If there is a bargaining agreement, the worker is
hired by the firm and the contract is fulfilled. The worker receives the
bargained wage w and the firm's profit is 300-w. If both parties
are unable to agree on a wage (i.e., bargaining breaks down), the firm
is assumed to have some alternative production possibility. This outside
option, for which employing the worker is not needed, generates a payoff
of 110 tokens for the firm. The worker stays unemployed and receives
nothing, that is, his outside option is assumed to be zero.
In contrast to other bargaining experiments by, for example,
Castillo and Cross (2008), Garcia-Gallego, Georgantzis, and
Jaramillo-Gutierrez (2012), or Sutter et al. (2009), the wage is
determined via alternating-offers bargaining (as in the study of
Rubinstein 1982). In the first round, the firm offers a wage which can
be accepted or rejected by the worker. If the worker accepts the offer,
he will be hired, the contract will be fulfilled, and the game ends. If
the worker rejects, he can make a counteroffer in round 2 which can then
be accepted or rejected by the firm. Bargaining is over if an offer is
accepted in any round or if bargaining breaks down exogenously after an
offer has been rejected. The probability of such a breakdown is 20% for
each bargaining round in which an offer is rejected. To give the
participants the chance to make their first offer and counteroffer
without having to fear the breakdown of negotiations, rejections could
only lead to breakdowns after round 3.
B. Subjects and Procedures
The experiments were conducted on a single day at the Technical
University Dresden with a total of 122 participants (72 men, 50 women).
All participants were undergraduate students with various majors. Before
the start of the experiment, half of the subjects were randomly assigned
to the role of a worker and the others to the role of a firm. These
roles remained fixed for the whole experiment. Each participant
bargained in 20 separate negotiations ("periods"). The right
to make the first offer alternated after a negotiation concluded, that
is, in period 1 the firm made the first offer, in period 2 the worker
made the first offer, and so on. Subjects were randomly matched into
bargaining pairs (one worker, one firm) in each period and could only
sign a wage agreement within their respective pair. Participants were
randomly rematched after each period. (2) During the experiment, the
bargaining partners were facing each other directly (face-to-face
bargaining) but were not allowed to communicate except in writing via a
prepared form. This setup ensured that each subject was able to identify
its bargaining partner's gender through visual contact.
To make sure that the subjects understood the bargaining procedures
and the payoff consequences of their actions, each subject was given a
detailed set of instructions before the experiment started. (3) The
players were asked to read through the written instructions. Afterwards,
each participant had the possibility to ask questions. These were
answered privately by the experimenter. No additional questions were
answered and any form of uncontrolled communication was made impossible
during the actual experiment. Whenever the players had to act, they had
to fill out prepared forms. The basic setup was as follows: In round 1,
the first player offered a wage by entering the offer into the prepared
form. Then the form was passed to the other player and round 2 started.
Now the other player could either put a check mark in a specific box if
he accepted the wage offered or he could enter a new wage in the form.
The form was passed back to the first player. This procedure went on
until an agreement was reached or bargaining broke down (with a
probability of 20% starting after round 3). (4)
C. Treatments
We conducted five treatments directly one after another, where each
treatment was played for four periods, respectively. In each period, the
worker and the firm bargained over the wage via alternating offers as
described above. The various treatments only differed in the level of
the legal minimum wage. Treatment 1 (periods 1-4) had no minimum wage,
that is, the range of permissible wages was 0 [less than or equal to] w
[less than or equal to] 300. In treatment 2 (periods 5-8), a minimum
wage of 70 tokens was introduced. Even though wages below the minimum
could be offered, the eventual agreement had to obey the wage constraint
70 [less than or equal to] w [less than or equal to] 300. Agreements
below the minimum wage were regarded as breakdowns. However, the minimum
wage of 70 was a non-binding restriction for almost all wage agreements
in treatment 1. In treatment 3 (periods 9-12), the minimum wage was
raised to 95. Again, this minimum was non-binding for most wage
agreements in the previous periods. In treatment 4 (periods 13-16), the
minimum wage was raised to 160, which constituted a binding restriction
for most wage agreements in previous periods. Finally, the minimum wage
was abolished in treatment 5 (periods 17-20). (5)
D. Payments
The exchange rate between tokens and real money was 50 tokens = EUR
1. At the end of the experiment, one out of the four bargaining
agreements in each treatment was randomly chosen. These agreements were
relevant for the payoffs in the respective treatment. The experiment
lasted approximately 90 minutes and subjects earned on average EUR
17.50, including a show-up fee of EUR 6. (6) All participants were paid
in cash directly after the experiment.
IV. EXPERIMENTAL RESULTS
Table 1, Column 1, shows mean negotiated wages, that is, bargaining
outcomes, across all five minimum wage treatments. Our results
illustrate the general impact of outside options and the minimum wage on
negotiated wages. First, employers earn, on average, a larger share of
the total rent. The main reason for this inequality is that employers
have a positive outside option, whereas employees would earn nothing if
negotiations failed. While a wage of 150 would ensure an equal split of
the entire rent among both players, actually negotiated wages are, on
average, below this level. The only exception is treatment 4, where the
minimum wage of 160 ensures that employees receive more than half of the
total rent. Our second observation is that the introduction of a minimum
wage and its subsequent increases lift up negotiated wages. This even
holds for very low--in fact, nonbinding--levels of the minimum wage (see
Dittrich, Knabe, and Leipold 2011, for an in-depth discussion of this
effect).
To analyze gender differences in wage negotiations, Table 1
provides information on mean negotiated wages for different gender
pairings. In all treatments, male employees are able to negotiate higher
wages than female employees, both when bargaining with male employers
(columns 2 and 3) and female employers (columns 4 and 5). Wage
differences between male and female employers when bargaining with male
employees (columns 2 and 4) or female employees (columns 3 and 5) are
less pronounced and vary in sign. Strong differences appear when
comparing the wage outcomes in cases where a male employer negotiates
with a female employee to the wage outcomes when a female employer
negotiates with a male employee (columns 3 and 5). The latter
constellation results in higher wages than the former gender pairing in
all treatments. These results suggest that men bargain more successfully
than women, which is consistent with empirical findings (Save-Soderbergh
2009). Gender differences seem to be more pronounced when comparing the
outcomes of male and female employees than when comparing the outcomes
of male and female employers.
There are two (not mutually exclusive) potential explanations for
the observed differences in negotiated wages. The first potential
explanation is that men and women have different negotiation skills.
Given the same initial conditions, that is, the same first offers and
counteroffers, it could be that women are less able to use the process
of alternating offers to obtain favorable outcomes. The second
explanation is that initial conditions differ. Men might enter
negotiations with offers or demands that are more in their own favor
than women.
The first explanation is not supported by our data. Weaker
negotiation skills could mean, for example, that women are more
risk-averse and are more ready to accept their bargaining partner's
offer. In our experiment, however, 53% of all bargaining rounds with
mixed gender pairs are terminated by an acceptance by the male partner.
There is no significant difference in this share between pairs in which
men are employers (52% male acceptance) and in which women are employers
(54% male acceptance). Another manifestation of weaker bargaining skills
could be that women are able to obtain only a smaller share of the
initial bargaining range (the distance between the initial offers and
demands made by employers and employees, respectively). We find only
weak support for this type of gender bias. In the bargaining rounds
where the employer makes the first offer, the final bargaining outcome
allows male employers to obtain 60% of the initial bargaining range,
while female employers obtain only 57%. When the employees make the
first move, male employees receive 69% of the bargaining range, female
employees receive only 63%. In both cases, however, the observed gender
differences are not statistically significant (p> .1).
We also checked whether the size of the concession that a party
makes (the absolute difference between a party's current and
preceding proposals) has an influence on the probability with which the
counterparty accepts the proposal. On the one hand, one could expect
that a larger concession causes the acceptance rate to rise because it
moves the bargaining outcome closer to some target outcome that the
reacting party might have. On the other hand, a large concession could
raise the other party's hopes for even more concessions in later
rounds, thereby reducing its willingness to accept the current proposal.
We ran probit regressions to estimate how the size of the
counterparty's last concession, together with the level of the
proposed wage, and the gender pairing are related to the rate at which a
party accepts the proposal. In general, we do not find statistically
significant evidence that the counterparty's concession has any
impact on a party's willingness to accept. The only statistically
significant effect is found for women acting as employers who appear to
be less likely to accept a proposal accompanied by a large concession.
Another important question is whether the levels of employers'
and employees' initial offers and counteroffers matter for
subsequent bargaining rounds. To analyze this question, we will look at
proposals made in the third and fourth round, that is, the first
concession made after the first offer and the corresponding counteroffer
have been observed (round 3) and the concession made after the
concession of the first-mover has been observed (round 4). In the third
round, one can formulate two different hypotheses how concessions are
made. The first hypothesis is that a party observes the bargaining range
bounded by the first offer and counteroffer and then expects the
eventual bargaining outcome to be in the middle of this bargaining
range. If this bargaining outcome is to be reached within a given number
of alternating offers, the size of the concession the party makes should
depend positively on the size of the bargaining range. A second,
alternative hypothesis is that each party has a target wage that it
wants to reach through alternating offers and that is determined already
before first offers and counteroffers are exchanged. The closer the
counterparty's proposal is to a party's own target, the more
willing this party is to make a concession. This means that if the
employer made a first offer and the employee made a high counteroffer,
the employer reacts by making only a small concession to maintain the
distance to its target wage. Vice versa, a high counteroffer made by an
employer to the employee's first offer prompts the employee to make
a large concession in round 3. Our data show that, for all
gender-pairings combined, there is statistically significant evidence
that concessions are made according to the first hypothesis. Separate
analyses for the different gender pairings support either this finding
or produce statistically insignificant results.
In round 4 and later rounds, one can again formulate two hypotheses
about the behavior of bargainers. The first hypothesis is that they play
"tit-for-tat," that is, they react to a small (large)
concession of the other party by making a small (large) concession
oneself. The alternative hypothesis is that bargainers react in the
opposite direction of their negotiation partner. If the other side made
only a small concession, one makes a large concession to increase the
chance to conclude the negotiation successfully. If the other
side's last concession was large, one reacts by making only a small
concession. For all gender-pairings combined, participants in our
experiment played "tit-for-tat." When we conduct separate
analyses for the different gender-pairings, we either find significant
evidence for "tit-for-tat" or insignificant associations
between the magnitudes of subsequent concessions.
Our findings from the analysis of the proposals made in round 3 and
subsequent rounds highlight two important points. First, we do not find
evidence for behavioral differences between men and women in the process
of alternating offers after first offers and counteroffers are made.
Second, our finding that the bargaining parties use
"tit-for-tat" strategies when making concessions strengthens
the importance of initial offers and counteroffers for the determination
of eventual bargaining outcomes.
The second potential explanation for the observed gender
differences in bargaining outcomes was that initial offers and
counteroffers differ between the different gender pairings. In half the
rounds of our experiment, employers had the chance to make the first
offer. In the other half, employees made the first move. Table 2
presents descriptive statistics on average first wage offers or demands.
When employers made the first move, the average first wage offer made by
male employers to male employees was higher than offers to female
employees in all five treatments (columns 1 and 2). This is suggestive
of wage discrimination because the bargaining partner's gender
appears to affect wage offers even before he or she has actively
participated in the negotiation. Female employers do not seem to
systematically discriminate in their first offers (columns 3 and 4).
Average first offers made to men are lower than those made to women in
the first two treatments and higher in the last three treatments. A
similar pattern can be observed in employees' first demands in
those negotiation rounds in which they could make the first move. Male
employees state higher initial wage demands when facing a female
employer than a male employer in four of the five treatments (columns 5
and 7). Female employees demand higher wages in their first offers when
negotiating with a female employer than with a male employer only in
three of the five treatments (columns 6 and 8). The results presented in
Table 2 indicate that part of the reason why men achieve more favorable
bargaining outcomes is that they already start the negotiation process
by making offers that are more favorable to themselves when negotiating
with a female partner. The results presented in Table 2 provide no
evidence that women also behave in such a discriminatory way.
Because of the multiplicity of gender combinations and treatments,
the number of observations in each cell in Tables 1 and 2 is rather
small, so the observed gender differences are typically not statistical
significant. To get around this problem, we assume that the relationship
between observed negotiated wages (or initial wage offers and first
replies, respectively) follows the following parametric functional form:
(1) [w.sub.i] = [alpha] + [gamma]'[S.sub.i] + [beta]'
[P.sub.i] + [[epsilon].sub.i]
where [w.sub.i] is the wage (bargaining outcome, initial wage
offer, or first reply, respectively) in the ith observed negotiation,
[S.sub.i] is a vector of control variables for the participants'
age as well as dummy variables indicating the minimum-wage treatment
which applied during this negotiation. [P.sub.i] is a vector of dummy
variables for the gender pairing of employers and employees, and
[[epsilon].sub.i] is an error term.
We estimate Equation (1) for bargaining outcomes, first offers, and
first replies by employers and employees using ordinary least squares
(OLS) with clustered standard errors. (7) Table 3 presents the
regression results. The treatment indicators show that minimum wages
exert a positive influence on negotiated wages. A positive effect of
minimum wages is also found on employers' initial wage offers, but
not on employees' initial wage demands (cf. Dittrich, Knabe, and
Leipold 2011). Negotiated wages are significantly smaller for female
employees than for male employees, independently of the employer's
gender. The wages paid by female employers do not differ significantly
from those paid by male employers to employees of a given gender. These
findings support the asymmetry of gender differences between the roles
of employers and employees. Women obtain worse outcomes than men only
when they take on the role of employees, but not when they act as
employers.
When employers make the first move (column 2), male employers offer
significantly lower wages to female employees than to male employees.
Female employers also seem to offer lower wages to female employees, but
this difference is not statistically significant. In column 3, we
examine the respective counteroffers made by employees. After being
offered lower wages by male employers than their male counterparts,
female employees seem to react by also demanding lower wages than male
employees (even though the difference is statistically not significant).
When bargaining with female employers, female employees demand lower
wages than male employees in their counteroffers. Thus, columns 2 and 3
suggest that female employees are offered lower wages, but also demand
lower wages than male employees. In their role as employers, however, we
do not see significant systematic differences in the behavior of men and
women.
In cases where employees are first-movers, wage demands by male
employees are significantly higher when negotiating with a female than
with a male employer (column 4). Interestingly, female employers do not
give in to these higher demands, but instead react by offering
(insignificantly) lower wages to male employees than are offered by male
employers. In the same way as when they acted as employers, men make
less generous proposals to women when they act as employees. Contrary to
their behavior as employees, however, women as employers do not accept
these unfavorable offers and instead make counteroffers that are
unfavorable to the male employees. There is no significant difference
between the wages demanded by female employees from male or female
employers. Both male and female employers react, however, by offering
lower wages to female employees than to male employees (statistically
significant for male employers).
These regression results support our findings that there are gender
differences in bargaining behavior. Men propose wages that are more in
their own favor when bargaining with women and also obtain low wages
when they are acting as employers and the women as employees. Women,
however, are willing to give in to the counterparty's demands and
obtain worse outcomes than men only when they are acting as employees.
As employers, women do not seem to accept the unfavorable proposals of
male employees and are able to reach bargaining outcomes that are not
worse than those of male employers.
We also want to briefly discuss the number of offers and
counteroffers it took before an agreement was reached and how often
negotiations broke down unsuccessfully. Overall, 22% of all negotiations
were exogenously interrupted and concluded unsuccessfully. The highest
breakdown rate was observed for negotiations with female employers and
female employees (25.2%), the lowest for male employers and female
employees (19.3%). The differences between the different gender pairings
are statistically insignificant. As it could be the case that gender
pairings are not equally distributed over negotiations with different
randomly determined break-off rounds, we ran a probit regression to
explain the probability of a breakdown by gender pairings, the break-off
round, as well as the age of the participants. We did not find
significant differences between gender pairings. We also looked at the
number of offers it took to reach an agreement in those negotiations
that concluded successfully. On average, it took 4.4 offers to agree on
a wage. Interestingly, when looking at these numbers separately for
different gender pairings, one finds a statistically significant
difference (p< .1) between negotiations with male employers and
female employees (4.2 rounds) and female employers and male employees
(4.6 rounds). This corresponds to our abovementioned findings about
bargaining behavior. Since women appear to be more willing to accept the
aggressive proposals made by men when they are in the role of employees,
these negotiations are concluded relatively fast. When women take on the
employer's role they resist such proposals, which improves their
bargaining outcomes but also prolongs the negotiation process.
Our results suggest an asymmetry of gender differences between the
roles of employers and employees. Gender differences are more pronounced
when comparing the outcomes of male and female employees than when
comparing the outcomes of employers. This suggests that women's
bargaining behavior seems to be sensitive to the experimental framing.
These results are in line with previous findings from economics and
social psychology that women are more context-sensitive than men (Croson
and Gneezy 2009). A recent study by Ellingsen et al. (2013) finds that
women behave more cooperatively in a one-shot prisoners' dilemma
experiment than men if the experiment is framed as a "community
game," while there are no gender differences if the experiment is
framed as a "stock market game." Social psychological studies
report similar findings implying that women are highly sensitive to the
social frame, while men are not (Kay and Ross 2003; Liberman, Samuels,
and Ross 2004). Thus, the asymmetry of gender differences in our results
may be explained by the different social frames and norms that women
feel subjected to when they are assigned to the roles of employers or
employees in our experiment. One could speculate which norms are
responsible for our observation that women behave differently in the
roles of employers and employees. A potential explanation could be that
women are more respectful of hierarchical or status arrangements than
men. If employers are perceived as being higher up on the hierarchical
scale or having more status, this may explain the asymmetry of gender
differences in our results. (8)
V. CONCLUSION
This paper examines gender and gender pairing effects in an
experimental labor market. In a face-to-face situation framed as wage
bargaining between an employer and an employee, we find that the
bargaining outcome differs between different gender pairings. When male
employers bargain with female employees, they end up paying lower wages
compared to bargaining with male employees. This result also holds for
female employers. When bargaining with female employees, they are on
average able to press wages down more than when bargaining with male
employees. Female employers, however, obtain the same bargaining
outcomes as male employers when both face either male or female
employees. Overall, women obtain worse bargaining outcomes than men when
they take on the role of employees, but not when they act as employers.
The results have some implications for labor market policies that
aim at reducing the gender wage gap. We provided evidence for behavioral
differences in bargaining between women and men. Hence, at least a part
of the gender wage gap might be attributable to these differences. Since
we did not find that women's bargaining skills in the process of
making offers and counteroffers were worse than those of men, our
findings suggest that it is not necessarily the lack of bargaining
skills that leads to less favorable outcomes for women. Gender
differences rather occur in how women and men approach the negotiation
(see also Babcock and Laschever 2003; Save-Soderbergh 2009). Therefore,
policy recommendations should not necessarily focus on improving
women's bargaining skills, but rather on encouraging women to
negotiate in the first place and not just to accept what is offered to
them. Moreover, women could be encouraged to be more demanding and tough
in how they approach a negotiation. Our finding that women are able to
change their bargaining behavior very quickly when they are placed in a
different social role suggests that one way, among many, to narrow the
gender wage gap is to change women's self-perception of their role
in the workplace. This could, at least partly, help to reduce the gender
wage gap.
ABBREVIATIONS
MBA: Master of Business Administration
OLS: Ordinary Least Squares
doi: 10.1111/ecin. 12060
REFERENCES
Andreoni, J., and L. Vesterlund. "Which Is the Fair Sex?
Gender Differences in Altruism." Quarterly Journal of Economics,
116(1), 2001, 293-312.
Ayres, I. "Fair Driving: Gender and Race Discrimination in
Retail Car Negotiations." Harvard Law Review, 104(4), 1991, 817-72.
Ayres, I., and P. Siegelman. "Race and Gender Discrimination
in Bargaining for a New Car." American Economic Review, 85(3),
1995, 304-21.
Babcock, L., and S. Laschever. Women Don't Ask: Negotiation
and the Gender Divide. Princeton: Princeton University Press, 2003.
Ben-Ner, A., F. Kong, and L. Putterman. "Share and Share
Alike? Intelligence, Socialization. Personality, and Gender-pairing as
Determinants of Giving." Journal of Economic Psychology, 25(5),
2004, 581-89.
Bertrand, M. "New Perspectives on Gender," in Handbook of
Labor Economics, volume 4B, edited by O. Ashenfelter and D. Card.
Amsterdam: Elsevier, 2011, 1543-90.
Blau, F., and L. Kahn. "Understanding International
Differences in the Gender Pay Gap." Journal of Labor Economics,
21(1), 2003, 106-44.
Bolton, G., and E. Katok. "An Experimental Test for Gender
Differences in Beneficent Behavior." Economics Letters, 48(3-4),
1995, 287-92.
Camerer, C. Behavioral Game Theory. Experiments in Strategic
Interaction. Princeton: Princeton University Press, 2003.
Carpenter, J., E. Verhoogen, and S. Burks. "The Effects of
Stakes in Distribution Experiments." Economics Letters, 86(3).
2005, 393-98.
Castillo, M., and P. Cross. "Of Mice and Men: Within Gender
Variation in Strategic Behavior." Games and Economic Behavior,
64(2), 2008, 421-32.
Castillo, M., R. Petrie, M. Torero, and L. Vesterlund. "Gender
Differences in Bargaining Outcomes: A Field Experiment on
Discrimination." Journal of Public Economics, 99, 2013, 35-48.
Croson, R., and U. Gneezy. "Gender Differences in
Preferences." Journal of Economic Literature, 47(2), 2009, 448-74.
Dickinson, D., and J. Tiefenthaler. "What Is Fair?
Experimental Evidence." Southern Economic Journal, 69(2), 2002,
414-28.
Dittrich, M., A. Knabe, and K. Leipold. "Spillover Effects of
Minimum Wages: Theory and Experimental Evidence." CESifo Working
Paper No. 3576, 2011.
Eckel, C., and P. Grossman. "The Relative Price of Fairness:
Gender Differences in a Punishment Game." Journal of Economic
Behavior and Organization, 30(2), 1996, 143-58.
--. "Are Women Less Selfish than Men? Evidence from Dictator
Experiments." Economic Journal, 108(448), 1998, 726-35.
--. "Chivalry and Solidarity in Ultimatum Games."
Economic Inquiry, 39(2), 2001, 171-88.
--. "Differences in the Economic Decisions of Men and Women:
Experimental Evidence," in Handbook of Experimental Economics
Results, volume 1, edited by C. Plott and V. Smith. Amsterdam: Elsevier.
2008, 509-519.
Ellingsen, T., M. Johannesson, J. Mollerstrom, and S. Munkhammar.
"Gender Differences in Social Framing Effects." Economics
Letters, 118(3), 2013, 470-72.
Fehr, E., M. Naef, and K. Schmidt. "The Role of Equality,
Efficiency, and Rawlsian Motives in Social Preferences: A Reply to
Engelmann and Strobel." American Economic Review, 96(5), 2006,
1912-17.
Garcia-Gallego, A., N. Georgantzis, and A. Jaramillo-Gutierrez.
"Gender Differences in Ultimatum Games: Despite Rather Than Due to
Risk Attitudes." Journal of Economic Behavior and Organization, 83,
2012, 42-49.
Gerhart, B., and S. Rynes. "Determinants and Consequences of
Salary Negotiations by Male and Female MBA Graduates." Journal of
Applied Psychology, 76(2), 1991, 256-62.
Gneezy, U., K. Leonard, and J. List. "Gender Differences in
Competition: Evidence from a Matrilineal and a Patriarchal
Society." Econometrica, 77(5), 2009, 1637-64.
Gupta, N. D., A. Poulsen, and M. C. Villeval. "Gender Matching
and Competitiveness: Experimental Evidence." Economic Inquiry,
51(1), 2013, 816-35.
Guth, W., C. Schmidt, and M. Sutter. "Bargaining Outside the
Lab--A Newspaper Experiment of a Three-Person Ultimatum Game."
Economic Journal, 117(518), 2007, 449-69.
Kay, A. C., and L. Ross. "The Perceptual Push: The Interplay
of Implicit Cues and Explicit Situational Construals on Behavioral
Intentions in the Prisoner's Dilemma." Journal of Experimental
Social Psychology, 39, 2003, 634-43.
Liberman, V., S. M. Samuels, and L. Ross. "The Name of the
Game: Predictive Power of Reputations versus Situational Labels in
Determining Prisoner's Dilemma Game Moves." Personality and
Social Psychology Bulletin, 30, 2004, 1175-85.
Niederle, M., and L. Vesterlund. "Do Women Shy Away from
Competition? Do Men Compete Too Much?" Quarterly Journal of
Economics, 122(3), 2007, 1067-101.
Rubinstein, A. "Perfect Equilibrium in a Bargaining
Model." Econometrica, 50(1), 1982, 97-109.
Save-Soderbergh, J. "Are Women Asking for Low Wages? Gender
Differences in Competitive Bargaining Strategies and Ensuing Bargaining
Success." Working Paper, Stockholm University, 2009.
Selten, R., and A. Ockenfels. "An Experimental Solidarity
Game." Journal of Economic Behavior and Organization, 34(4), 1998,
517-39.
Solnick, S. "Gender Differences in the Ultimatum Game."
Economic Inquiry, 39(2), 2001, 189-200.
Sutter, M., R. Bosman, M. Kocher, and F. van Winden. "Gender
Pairing and Bargaining--Beware the Same Sex!" Experimental
Economics, 12(3), 2009, 318-31.
Wood, R., M. Corcoran, and P. Courant. "Pay Differences among
the Highly Paid: The Male-Female Earnings Gap in Lawyers'
Salaries." Journal of Labor Economics, 11(3), 1993,417-41.
(1.) Even though our experiment was originally designed to analyze
how minimum wages affect wage negotiations, we will only briefly comment
on the impact of minimum wages in this paper. A more detailed analysis
of results that are not gender-related can be found in a companion paper
(Dittrich, Knabe, and Leipold 2011).
(2.) After each round, half of the subjects (those in the role of
employers) stayed at their desks, while the other half (the employees)
quickly moved to a new partner assigned by the experimenters. No
communication between participants took place during the re-matching.
(3.) Translated versions of the experimental instructions and the
negotiation forms are provided in an online annex.
(4.) The timing of the breakdown was determined by a computer
program. Based on the 20% breakdown probability, the program calculated
for each round (after round 3) whether bargaining breaks down or not.
(5.) Due to time restrictions when conducting the experiment, 30
students participated in treatments 3 to 5 only for two instead of four
periods. We checked that our results are robust to excluding this group
from the analysis and also to restricting our analysis to the periods in
which all subjects participated.
(6.) The comparable market wage for undergraduate students in
Dresden is approximately EUR 8 per hour.
(7.) To take into account that observations within and across
treatments are obtained from a limited number of participants, we
calculated clustered standard errors at the individual level. First
offers and replies are clustered at the level of individual employees or
employers, depending on who made the proposal. Standard errors in the
regression for bargaining outcomes are clustered at the level of
individual employers. We checked that clustering of standard errors at
the level of employees yields qualitatively equivalent results.
(8.) We are thankful to one of the referees for pointing us to this
interpretation. However, a thorough analysis of these issues is beyond
the scope of this paper.
MARCUS DITTRICH, ANDREAS KNABE and KRISTINA LEIPOLD *
* Financial support from the Randstad Foundation is gratefully
acknowledged. We thank the editor and two anonymous referees for helpful
comments.
Dittrich: Assistant Professor Chemnitz University of Technology,
Department of Economics and Business Administration, 09107 Chemnitz,
Germany. Phone +49-371-53138278, Fax +49-371-531834197, E-mail
marcus.dittrich@wirtschaft.tu-chemnitz.de
Knabe: Professor Otto von Guericke University Magdeburg, P.O. Box
4120, 39016 Magdeburg, Germany. Phone +49-391 -6718546, Fax
+49-391-6711218, E-mail andreas.knabe@ovgu.de
Leipold: Research Associate, Technische Universitaet Dresden,
Faculty of Business and Economics, 01062 Dresden, Germany. Phone
+49-351-46332168, Fax +49-35146337285, E-mail
kristina.leipold@mailbox.tu-dresden.de
TABLE 1
Bargaining Outcomes by Employers' and Employees' Gender
Male Employer, Bargaining
with
A Male A Female
Min. Overall employee Employee
Treatment Wage (1) (2) (3)
1 -- 124.4 123.0 117.0
(SD = 21; (SD = 24.4; (SD = 18.8;
n = 161) n = 50) n = 20)
2 70 129.9 135.7 124.3
(SD = 18.4; (SD = 21.9; (SD = 22.2;
n = 155) n = 45) n = 30)
3 95 134.5 136.4 133.7
(SD = 16.3; (SD = 20.4; (SD = 17.7;
n = 150) n = 47) n = 27)
4 160 170.4 169.6 168.8
(SD = 6.4; (SD = 6.8; (SD = 5.7;
n = 189) n = 75) n = 28)
5 -- 128.7 132.6 125.7
(SD = 18.7; (SD = 23.4; (SD = 15.4;
n = 183) n = 61) n = 33)
Female Employer,
Bargaining with
A Male A Female
Employee Employee
Treatment (4) (5)
1 127.6 127.0
(SD = 28.1; (SD = 21.8;
n = 45) n = 27)
2 131.5 122.7
(SD = 27.5; (SD = 19.9;
n = 46) n = 24)
3 136.1 128.2
(SD = 23; (SD = 14.8;
n = 46) n = 19)
4 172.8 169.2
(SD = 7.5; (SD = 5.4;
n = 49) n = 25)
5 129.8 119.9
(SD = 20.2; (SD = 13.1;
n = 57) n = 24)
Note: Standard deviation (SD) and number of
observations (n) in parentheses.
TABLE 2
First Offers by Employers' and Employees' Gender
Employers' First Offer
Male Employer,
Bargaining with
Min. A Male A Female
Treatment Wage Employee Employee
(1) (2)
1 -- 78.6 76.7
(SD = 31.2) (SD = 27.5)
(n = 38) (n = 15)
2 70 87.6 76.7
(SD = 24.4) (SD = 30.4)
(n = 39) (n = 19)
3 95 105.5 101.5
(SD = 14.4) (SD = 30.9)
(n = 33) (n = 19)
4 160 162.5 162.0
(SD = 5.2) (SD = 3.1)
(n = 38) (n = 13)
5 -- 89.4 67.9
(SD = 46.2) (SD = 36.4)
(n = 30) (n = 18)
Employers' First Offer Employees' First Demands
Female Employer, Male Employer,
Bargaining with Bargaining with
A Male A Female A Male A Female
Treatment Employee Employee Employee Employee
(3) (4) (5) (6)
1 80.2 81.5 176.0 163.8
(SD = 37.5) (SD = 34.7) (SD = 20.4) (SD = 25.1)
(n = 28) (n = 24) (n = 32) (n = 16)
2 81.8 84.0 170.1 161.3
(SD = 39.5) (SD = 31.2) (SD = 31.4) (SD = 37.9)
(n = 32) (n = 20) (n = 35) (n = 20)
3 101.9 100.4 170.8 163.2
(SD = 39.4) (SD = 26.5) (SD = 30.1) (SD = 24.8)
(n = 31) (n = 14) (n = 37) (n = 19)
4 159.2 154.6 186.8 194.2
(SD = 4.5) (SD = 3.1) (SD = 17) (SD = 30.6)
(n = 27) (n = 13) (n = 39) (n = 18)
5 88.6 68.6 164.2 175.0
(SD = 47) (SD = 28.4) (SD = 23.9) (SD = 30.7)
(n = 30) (n = 11) (n = 33) (n = 15)
Employees' First Demands
Female Employer,
Bargaining with
A Male A Female
Treatment Employee Employee
(7) (8)
1 177.6 164.4
(SD = 36.1) (SD = 25.9)
(n = 33) (n = 17)
2 177.2 170.3
(SD = 49.4) (SD = 37.9)
(n = 34) (n = 20)
3 169.6 166.5
(SD = 31.2) (SD = 20.5)
(n = 30) (n = 13)
4 195.0 184.2
(SD = 38.2) (SD = 26)
(n = 28) (n = 13)
5 175.5 164.9
(SD = 41.9) (SD = 29.6)
(n = 28) (n = 14)
Note: Standard deviation (SD) and number
of observations (n) in parentheses.
TABLE 3
Regression Results
First-Mover: Employer
Wage
Demanded by
Wage Employee
Bargaining Offered (Counter-
Outcome by Employer offer)
(1) (2) (3)
Treatment 1 (no minimum wage) Reference Reference Reference
Treatment 2 (MW = 70) 5.87 *** 3.97 1.02
(2.00) (3.79) (2.68)
Treatment 3 (MW = 95) 10.16 *** 23.38 *** -3.77
(2.36) (3.63) (3.28)
Treatment 4 (MW = 160) 45.61 *** 80.33 *** 25.07 ***
(2.24) (4.24) (3.80)
Treatment 5 (no minimum wage) 4.05 * 2.66 -4.24
(2.37) (4.78) (3.36)
Gender pairing
Employer Employee
Male Male Reference Reference Reference
Male Female -5.00 ** -8.60 ** -6.47
(2.02) (3.35) (4.83)
Female Male -0.78 -2.10 4.74
(3.03) (4.76) (3.00)
Female Female -6.62 ** -5.42 -7.38
(2.73) (4.51) (4.83)
Employer's age -0.80 ** 0.01 -0.34
(0.38) (0.74) (0.44)
Employee's age -0.36 -0.76 0.10
(0.25) (0.65) (1.01)
Constant 150.97 *** 98.07 *** 168.46 ***
(8.80) (18.86) (21.98)
Number of observations 778 492 481
[R.sup.2] 0.538 0.567 0.203
First-Mover: Employee
Wage Wage Offered
Demanded by by Employer
Employee (Counteroffer)
(4) (5)
Treatment 1 (no minimum wage) Reference Reference
Treatment 2 (MW = 70) -1.49 7.93 ***
(2.76) 2.64)
Treatment 3 (MW = 95) -3.64 23.65 ***
(2.92) (3.70)
Treatment 4 (MW = 160) 18.19 *** 74.57 ***
(3.78) (4.11)
Treatment 5 (no minimum wage) -2.38 10.47 **
(3.72) (4.06)
Gender pairing
Employer Employee
Male Male Reference Reference
Male Female -1.71 -9.11 **
(5.41) (3.65)
Female Male 5.28 ** -4.83
(2.43) (5.27)
Female Female -3.13 -9.39 *
(4.88) (5.42)
Employer's age -0.02 -0.28
(0.61) (0.81)
Employee's age 0.96 -0.83
(1.15) (0.50)
Constant 151.71 *** 117.16 ***
(28.41) (20.35)
Number of observations 494 484
R.sup.2] 0.113 0.597
Note: Standard errors reported in parentheses
(clustered at the individual level) * p < .01,
** p < .05, *** p < .1.
