Content uploaded by Asad Zaman
Author content
All content in this area was uploaded by Asad Zaman on Sep 01, 2017
Content may be subject to copyright.
P I D E W O R K I N G P A P E R S
2013:90
PA KISTAN IN ST IT UT E O F D EV EL OP ME NT EC ON OM IC S
Saima Naeem
Asad Zaman
For Love or Money?
Motivating Workers
PIDE Working Papers
2013: 90
For Love or Money?
Motivating Workers
Saima Naeem
State Bank of Pakistan, Islamabad
and
Asad Zaman
International Islamic University, Islamabad
PAKISTAN INSTITUTE OF DEVELOPMENT ECONOMICS
ISLAMABAD
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means—electronic, mechanical, photocopying, recording or
otherwise—without prior permission of the Publications Division, Pakistan Institute of Development
Economics, P. O. Box 1091, Islamabad 44000.
© Pakistan Institute of Development
Economics, 2013.
Pakistan Institute of Development Economics
Islamabad, Pakistan
E-mail: publications@pide.org.pk
Website: http://www.pide.org.pk
Fax: +92-51-9248065
Designed, composed, and finished at the Publications Division, PIDE.
C O N T E N T S
Page
Abstract v
I.
Introduction 1
II. The Experiment 5
II.1. Subject Detail 5
II.2. Treatment Detail 5
II.3. Experimental Setup 6
III. Results and Discussion 6
III.1. Appreciation Works 7
III.2. Relative-Wage Concerns 8
III.3. Supportive Analysis through Quality and Shirking 10
III.4. Supportive Analysis Through the Post-Experimental
Questionnaire 11
IV.
Conclusions 12
V. Appendices 14
References 19
List of Tables
Table 1. Treatment Detail 5
Table 2. 7
Table 3. Average Output of All Treatment Groups 7
Table 4. Regression Analysis 10
Table 5. Shirking Time Over 30-minute Sessions 11
ABSTRACT
We conducted a field experiment and tested how employers can use
socioemotional resources, such as appreciation and recognition, in order to signal
intentions and create positive reciprocal relationships with employees. Results
showed that these resources led to a significant gain in productivity. The study
was extended to account for relative wage concerns both with and without
appreciation treatment. Efficiency gains with appreciation appeared to be robust
even after including information regarding relatively disadvantageous wage
discrimination. However, workers’ without socioemotional resources exhibited
strong resentment toward relatively lower wages by showing a significant
systematic decrease in their labour supply. Our results suggest that workers not only
compare their wages, as pointed out in previous literature, but also compare the
socioemotional resources provided by their employer. This provides important
evidence against one-dimensional comparisons of relative wages relevant to
worker productivity.
JEL classification: C93, M5, J31, J32, J53
Keywords: Appreciation, Recognition, Symbolic Gift Exchange, Wage
Comparisons
I. INTRODUCTION
*
People value their self-image and also how others perceive them. Others’
intentions sometimes influence our emotional states, and these emotions can
affect our subsequent actions [Cox, et al. (2007); Levine (1998)]. For purposes
of the current paper, we restricted our analysis to employment relationships: the
economic relationship between the principal and the agent is based on material
gains in the presence of conflicting interests. However, several studies indicate
that principal-agent relationships work beyond wage-effort exchanges;
organisational theory has argued that employees’ loyalty towards their employer
enhances job effort and profitability of the organisation [Rhoades and
Eisenberger (2002); Ellingsen and Johannesson (2008)]. Campbell and Kamlani
(1997) have provided survey evidence that employers care for workers’ loyalty
and commitment to the organisation. Similarly, employees also value
commitment from their employer in regard to respect, wages, quality work
environment, and career growth. Employees develop perceived organisational
support (POS) beliefs in order to meet their socio-emotional needs.
1
Employees
with higher POS beliefs are more likely to reciprocate their employer’s
commitment through higher job performance and reduced absenteeism [Armeli,
et al. (1998); Cropanzano and Mitchell (2005)]. Dur (2009) also argues that
attention and care from an employer is repaid with higher loyalty, which might
evolve into a symbolic gift exchange relationship between employer and
employee.
Conventional economic wisdom has exclusively focused on material
incentives as a tool for signalling positive intentions to help alter employees’
behaviour [Ellingsen and Johannesson (2008); Arbak and Kranich (2005)].
Research on non-monetary incentives remained scarce until very recently.
Indeed, the literature on socially integrated exchanges in the principal-agent
Acknowledgements:
Financial support from Social Research Unit, jointly working in
Higher
Education Commission of Pakistan, International Islamic University and Harvard
University, is
gratefully acknowledged. We thank Muhammad Naeem, Muhammad Amjad Malik
and Hisham Tariq for providing research assistance for execution of experiment, Muhammad Naeem for
designing software and all the university administration for providing technical help.
1
Data from Gallup surveys also confirm that employees who feel that their employers care
about them are more satisfied on their job and in turn provide higher job efforts [Wagner and Harter
(2006)].
2
context has found an association between stronger signalling power and hence
higher reciprocal responses on the one hand and non-monetary payments on the
other [Kube, et al. (2011); Mahmood and Zaman (2010)]. Dur (2009) extended
this argument to show that employers can signal good attitudes through care and
attention to employees in order to develop social relationships even when
providing lower wages. Other exceptions include research on non-material
incentives such as reward [Kosfeld and Neckermann (2010)], respect (Eriksson
and Villeval, 2010), trust [Dickinson and Villeval (2008)], and punishment
[Masclet, et al. (2003)]. Wiscombe (2002) argued that such non-monetary
motives improve workers’ morale without hurting an organisation’s cost
structure; recognition of employee’s contribution is among the strongest
motives.
In this paper, we analysed the ability of employers to provide care and
attention as a way to motivate their employees. This paper is mainly divided into
two parts. The first addresses two important questions: (i) Can employers signal
positive intentions through care and appreciation? (ii) How do these signals alter
employees’ behaviour in terms of higher job effort? The extent of a reciprocal
response by an employee relies on the strength and credibility of the positive
gestural signal; this issue might become even more important for a job with a
limited duration. To overcome this limitation, we used written forms of
appreciation (i.e., appreciation letters from a research team whom a group of
students knew).
2
These appreciation letters provided no information on
performance expectation; these letters just included a thank you statement and
encouragement to join a future project as a goodwill gesture. The letters were
provided ex ante (before starting the job) to all workers. We trusted that workers
would value this positive signal of care and appreciation as a credible socio-
emotional resource and reciprocate by increasing their job efforts. We consulted
previous work that investigated trust, respect, and appreciation as incentives
within a labour market context; however, our study differs from most of these in
two ways. First, Kosfeld and Neckermann (2010) and Eriksson and Villeval
(2010) used “award” within a performance-contingent environment (i.e., only
after observing the work effort), while our appreciation treatment was provided
unconditionally to everyone in a trust-based setting. Second, Falk and Kosfeld
(2006) and Dickinson and Villeval (2008) used ex ante trust signals within a
laboratory setting, while our treatment did not explicitly use trust (the letter did
not include any expectation for performance). Our treatment of appreciation
works as a token of expression of employer’s care and attention for workers and
deals with their contributions implicitly.
In the second part, we extended our research to analyse relative wage
concerns in the presence of socio-emotional resources. A large body of the
2
The Social Research Unit is jointly working with the Higher Education Commission,
International Islamic University, and Harvard University.
3
existing literature has been dedicated to relative wage concerns since the work
of Keynes (1936). The idea here is that reciprocity has its basis in beliefs about
fairness compared to a reference wage, mostly in regard to wages of individuals
with similar merit, status, gender, job, etc. Although several studies have
highlighted the importance of relative wage concerns for labour supply
decisions, the theoretical discussion on the effect of relative positions on
performance is somewhat inconclusive. While some theories propose lower job
performance resulting from frustration with low wages or positions [Clark, et al.
(2010); Torgler, et al. (2006)], others demonstrate the motivational aspects of
larger differences in achieving higher performance. However, most of these
concerns have been studied in terms of social and economic status. In contrast,
we investigated how employees respond to relative wage concerns if they are
provided with socio-emotional incentives. We argue that if employees value
these socio-emotional resources, they might substitute monetary incentives for
such resources, at least for smaller wage differentials.
3
If so, workers might
show fewer concerns for their relatively disadvantageous wage position
compared to when they are not provided with such incentives. This idea
contradicts the standard uni-dimensional wage comparisons cited as important
determinants of labour supply in many labour market textbooks. The idea of
adverse relative wage effects is based on the idea of fairness that implicitly
ignores heterogeneity among workers [Abeler, et al. (2009)] and employers’
intentions [Gachter and Thoni (2010)]. Abeler, et al. (2009) provide interesting
evidence for relative wage comparisons; they reported that employees not only
compare their absolute relative wages but also care about the effort workers
exert in their performance. Wage differentials do not always indicate
discrimination; information on peer effort and productivity can affect standard
gift exchange mechanisms through equity norms. Similarly, intentional wage
discrimination among workers is more detrimental than non-intentional
differentials. Gachter and Thoni (2010), in a three-player gift exchange game,
showed that workers reacted adversely to relative wage differentials if
employers intentionally discriminated between their workers, rather than merely
providing absolute wage differentials. We extend this argument in a similar
manner. For instance, we argue that absolute wage discrimination might not hurt
employees if they are compensated with other socio-emotional resources. As far
as we know, no previous study has analysed relative wage concerns in the
presence of socio-emotional incentives.
Finally, an interesting point is related to preference-decision mismatches
for monetary incentives [see Kube, et al. (2011); Mahmood and Zaman (2010);
3
This study is extended from doctoral research of Saima Naeem. In additional treatment
groups, we also tested information of relative wage concerns for larger wage differentials. The
introduction of higher wage differentials in this group decreased output significantly by 6 percent as
compared to AT, yet we found no productivity loss as in CI.
4
Hsee (1999)]. These studies assessed cross-comparisons of monetary and non-
monetary incentives and revealed a generally higher preference for money
during joint valuation. However, workers perform more efficiently when
provided non-monetary incentives in isolation. In addition, these studies
compared material non-monetary incentives with money, and a fondness for
money is explained by a preference-mismatch in case of non-monetary
incentives [Kube, et al. (2011)] and the high fungibility of money [Hsee (1999)].
Thus, no systematic study has been conducted to assess how employees value
non-material incentives.
In order to test our hypothesis, we hired five groups of twenty-five
students for a data entry project. Our first group was paid an announced wage
for standard data entry; this group was the benchmark for the other treatment
groups. Our main treatment group was an appreciation group that received an ex
ante appreciation letter along with an announced wage. The third group was
given an unexpected wage increase of 17 percent; this group served as a
benchmark for the final two treatment groups. To examine whether identical
information on similar wage differentials can trigger different reciprocal
responses, our final two treatment groups were either provided (or not provided)
with an appreciation letter.
The results of this study strongly supported our initial hypothesis. The
groups provided with appreciation letters were significantly more productive;
the observed gift exchange behaviour was even higher than that resulting from
an unexpected 17 percent wage increase. Results also confirmed that
information regarding a small wage differential had no effect on productivity
when workers were provided with appreciation letters. However, there was a
substantial loss in productivity for the same wage differentials when workers
were not provided with socio-emotional resources. Information regarding higher
relative wages for other groups resulted in lower job satisfaction, less positive
evaluation of employers’ kindness signals, and greater relative wage concerns.
However, such concerns were significantly decreased when participants were
provided with socio-emotional resources. These results should be interpreted
carefully. Workers might not be willing to substitute social incentives with
money and might react adversely to larger wage differentials. Similarly, the
credibility of the employer and socio-emotional signals can affect reciprocal
responses. However, our experiment does provide counter-evidence to
assessments of one-dimensional wage comparisons and suggests that workers
compare not only relative wages but also socio-emotional resources provided.
This is very important evidence that conflicts with uni-dimensional wage
comparisons cited as important determinants of labour supply in many labour
market textbooks.
The remainder of this paper is organised as follows. Section 2 presents
the experimental design and procedures. Section 3 provides an analysis of the
results, and Section 4 the conclusion.
5
II. THE EXPERIMENT
II.1. Subject Detail
The experiment was conducted during the first quarter of 2009 at the
International Islamic University. For the experiment, 150 students were
randomly selected through an advertisement displayed on students’ notice
boards, announcing a three-hour, one-time job opportunity. The wage was
announced as PKR 120 per hour. These subjects were hired without any
knowledge of being part of an experiment. The students were hired for a data
entry project to develop a comprehensive database for a personality test used in
a separate research study. Selected students belonged to different departments
(economics, business management, finance, and electronics). The age range of
our subjects was 19-25 years, with an average age of 21.4 years. Most of our
experimental subjects were male; only 11 female students volunteered.
II.2. Treatment Detail
Students were randomly divided into five groups of twenty-five students
each (i.e., control group (C0 ), small gift (C17 ), appreciation treatment (AT ),
appreciation-information (AI ), and control information (CI )).
The first group, C0, was given the announced protocol. They were provided
an announced wage for the standard data entry protocol; this served as benchmark
for our group comparisons. For the AT group, an appreciation letter was awarded
along with the announced wage (see Appendix D for the letter). Our third group was
paid an unexpectedly higher wage (17 percent higher); the unanticipated wage
increase was announced after a training session. The final wage payment for this
group, C17, was PKR 140 per hour. Both letters and wage increases were announced
after a training session in order to make sure both types of incentives were presented
at the same time. We then created two information treatment groups, one with socio-
emotional incentives and the other without. While the first information treatment
group, AI was given the appreciation letter, the project coordinator informed this
group that another group performing the same job would be given a 17 percent wage
increase instead of the letter. In the second information group, CI, students were paid
the announced wage (PKR 120 per hour) and were informed about the wage
increase for the C17 group.
Table 1
Treatment Detail
C0 C17 AT AI CI
Wage 120 140 120 120 120
% Wage Increase – – 17% – – –
Relative Wage Information No No No Yes Yes
Letter No No Yes Yes No
N 25 24 25 25 21
6
II.3. Experimental Setup
All students were asked to report to work during university hours on
separate days. Upon arrival, students were given a separate university office
with a computer for data entry. In the office, we placed some newspapers and
magazines (sports and fashion); similarly, an icon of a famous game was placed
close to the data entry portal icon on the computer. Students were provided 30
minutes of training to familiarise them with the data entry protocol. The first
fifteen minutes were reserved for demonstration, and the final fifteen minutes
were reserved for practice. Materials for the practice session were kept the same
for all workers. This practice in data entry was used to control for differences in
skills across the participants. To help keep the setting more formal, workers
were also paid for the training session. The experimenter made it clear that the
job should be taken very seriously. The task was made boring enough to be
affected by incentives only. After training, participants were exposed to the
aforementioned treatment manipulation.
During the experimental phase, workers were left unsupervised except for a
coordinator who was sitting in the next room in order to provide technical help, if
needed. Furthermore, to eliminate experimenter effects, we ensured that all workers
interacted with the same project coordinators. To observe shirking behaviour in the
participants, we recorded data entry times without their knowledge. Time not spent
entering data was used as a proxy for shirking behaviour.
Exactly after three hours, the project coordinator ended the session,
making sure that all data were saved. Each student filled out a post-experimental
questionnaire. Additional questionnaires included items on gender, subject
major, job satisfaction, fairness of their contract, employers’ kindness, and
preference for monetary and non-monetary incentives (Appendix B).
III. RESULTS AND DISCUSSION
We based our subsequent efficiency analysis on effective output.
Effective output (which we will refer to as output hereafter) was calculated as
the difference between total output and data entry errors; this was the dependent
variable for our efficiency analysis. Our experimental data produced two
additional variables for each subject during the allotted job time: (i) time
wastage and (ii) quality of observed output. Our main variable for efficiency did
not take into account the cost of incurring errors; thus, we evaluated the quality
of output for each worker. Quality of output is measured as the percentage of
effective output relative to total output. Workers in our experiment were left
unsupervised except that data entry time was recorded without their knowledge.
Times were recorded as “shirking” if the time duration between two data entries
exceeded two minutes. Since the data on these two variables showed only minor
differences across treatment groups, we have included a brief discussion to
support our main findings based on these two variables in Section 3.3.
7
III.1. Appreciation Works
Table 2 presents the development of output over time within all treatment
groups. A comparison of average hourly output suggests that workers in the AT
group entered 29.8 percent more output compared to the benchmark group.
4
This observation clearly supports our hypothesis that appreciation and
recognition of trust seems to be profitable for employers; the productivity
increase is significant at the 5 percent level. This productivity gain is even
higher than the efficiency increase due to an unexpected wage increase in the
C17 group; the announcement of an unexpected 17 percent increase in wages
resulted in a 21.4 percent increase in output compared to the C0 group.
5
Output
elasticity in the AT group was 1.4 in comparison to the C17 group. This positive
wage-effort trend is consistent with previous work on gift exchange and
incentive research [Berg, et al. (1995); Fischbacher, et al. (2001); Bewley
(1999); Fehr and Gächter (2000); Falk (2007)].
Table 2
Hour 1 Hour 2 Hour 3 Total
C0 67.99 65.93 65.35 66.42
C17 77.48 83.44 80.92 80.61
AT 80.82 90.64 87.19 86.22
AI 75.23 89.18 89.29 84.57
CI 59.90 60.20 61.94 60.68
The comparisons of average output between all treatment groups, using the
Wilcoxon rank-sum test, are summarised in Table 3. Results indicate significantly
stronger observed gift exchange effects within the AT as compared to the C17 group
(p = .01). This interesting result not only confirms our behavioural prediction that
socio-emotional resources create a symbolic exchange mechanism but also suggests
that money is an expensive source of motivation.
Table 3
Average Output of All Treatment Groups
C17 CI AI AT
C0 4.9*** 2.3** 5.1*** 5.1***
AT 2.6** 5.3*** 1.0 –
AI 1.9* 5.2*** – –
CI 5.2*** – – –
Values in parenthesis are p-value of Mann-Whitney U test.
* p<0.05, ** p<0.01, *** p<0.001.
4
However, we cannot rule out the possibility of psychological and social costs associated
with these incentives.
5
The elasticity of output with respect to associated cost is 1.26 in the C17 group.
8
Results are summarised as follows:
Result 1: Workers produced significantly higher amounts of output when
provided with socio-emotional resources, such as appreciation and recognition.
The observed gift exchange effect from the appreciation letter was significantly
stronger than the effect of a 17 percent wage increase.
III.2. Relative-Wage Concerns
We next discuss results from our two information treatments, AI
(appreciation letter with relative wage information) and CI (no appreciation
letter with relative wage information). Our results suggest that identical
information about relatively disadvantageous wage positions generated
systematically different reciprocal responses for the two information groups.
The information regarding wage differentials for the CI group resulted in a
significant 8.6 percent decrease in productivity compared to the benchmark
group (Wilcoxon rank-sum test: p = .02). Our results confirm the findings of
Fischer and Steiger (2009) that employees decide to reduce their labour supply
as a result of the frustration created by lower relative wages. However, Hennig-
Schmidt, et al. (2008) provide evidence against the adverse effects resulting
from relative wage comparisons.
The adverse effect of relative wage concerns is not severe for a
similar wage differential if workers are provided with an appreciation letter.
Output entered by the AI participants decreased by 2 percent compared to
the AT group (Wilcoxon rank-sum test: p > .05); nevertheless, there was a
remarkable productivity gain of 27 percent compared to the benchmark
group at no additional physical cost. The differences between productivity
gains in the AI and C17 groups remain non-significant at the .05 level.
While we cannot rule out the possibility that the size of the wage differential
affects relative wage concerns, there is still a clear indication that on
average, workers reacted less negatively to disadvantageous wage
discrimination in the AI than in the CI treatment group. These results can be
summarised as follows:
Result 2: Information on relative wages in the presence of socio-
emotional incentives does not significantly decrease productivity compared to
the original appreciation treatment. However, relative wage concerns in the
absence of an appreciation letter results in a significant loss in productivity.
Our observation of a decreased adverse reaction in the AI condition
seems to conflict with findings of Kube, et al. (2011), Hsee (1999), and
Mahmood and Zaman (2010); these studies reported that people tend to
overestimate money when jointly evaluating money with non-monetary
commodities. The differences in our results might be explained in two ways.
First, previous studies used non-monetary payments. In such cases, people
start to compare the monetary worth of both incentives. The lack of
9
relativity of such abstract, non-monetary incentives might hinder people
from making direct monetary comparisons. [Ariely (2008) gives a
comprehensive discussion over relativity problems].
6
This difficulty in
conducting a comparison removes this exchange from a conventional market
context, and workers engage in pro-social behaviour [for details, see
Heyman and Ariely (2004)]. Second, mismatched preferences might arise
when non-monetary incentives are used. Money, owing to its versatility in
satisfying human physical needs, dominates non-monetary commodities;
appreciation, trust, and recognition are usually ranked higher than physical
incentives in many motivational theories because of their ability to satisfy
psychological needs [Pink (2009); Herzberg (1959)].
To explain workers’ behaviour among different treatment groups, we
estimated random effects GLS models. Treatment dummies measured the
treatment effects. Part (1) of Table 4 provides the treatment effect at the
aggregate level using OLS regression. The results are consistent with those
presented in Table 3. We extended the model by including time effects and
the interaction of treatment effects with time (30 minutes and 15 minutes).
Part (2) of Table 4 provides estimates on the treatment effect after
incorporating a 30-minute temporal dimension by using panel data analysis.
Part (3) uses the 15-minute data for a similar analysis. The random effects
model with GLS specification is estimated after controlling for job specific
and socio-economic variables. Job related variables included job skill, job
satisfaction, perception of fairness with job contract, and kindness of the
employer. The socio-economic variables included gender, age, and major
subject at the university (summary statistics are provided in Appendix A,
Table 3). Results remained robust after these sets of variables were
controlled for.
All of the treatment effects remained significant in both the 15 and 30-
minute sessions, except for the CI condition. The significance of productivity
loss in this condition, especially of shirking time, decreased during the shorter
time span when socioeconomic and job variables were controlled for. This
clearly indicates that relative wage information in the CI condition resulted in
greater overall shirking on the job as a result of frustration. Interestingly, the
learning effect was not significant, but the interaction terms, especially with the
appreciation treatments (AT and AI ), remained significant. This indicated that
adaptation behaviour/learning over time seems to be related to these specific
treatments. Shirking behaviour also significantly affected the performance of
workers.
6
Relativity of comparative goods facilitates comparison; people tend to avoid difficult
comparisons [Ariely (2008), p. 8].
10
Table 4
Regression Analysis
Aggregate (1)
a
30 Minutes (2)
b
15 Minutes (3)
b
Variable Coeff. SE Coeff. SE Coeff. SE
Constant 797.1 20.05
∗∗∗
155.7 21.17
∗∗∗
76.2 10.47
∗∗∗
C17 170.3 28.36
∗∗∗
22.0 8.99
∗∗
12.0 4.25
∗∗∗
AT 226.8 28.36
∗∗∗
21.7 8.94
∗∗
12.2 4.21
∗∗∗
AI 237.5 28.66
∗∗∗
14.6 8.98
∗
8.9 4.24
∗∗∗
CI 217.7 28.65
∗∗∗
–11.6 8.10 –5.0 3.75
T 175.9 28.36
∗∗∗
–1.4 1.26 –0.3 0.3
T*C17 4.6 1.79
∗∗
1.1 0.43
∗∗
T*AT 6.9 1.80
∗∗∗
1.7 0.43
∗∗∗
T*AI 8.4 1.80
∗∗∗
2.0 0.43
∗∗∗
T*CI 2.3 1.87 0.5 0.44
Shirking –5.5 0.47
∗∗∗
–4.8 0.26
∗∗∗
Wald Statistics Wald p-value Wald p-value Wald p-value
C17 Vs. AT 2.3 0.22 0.1 0.0.965 0.1 0.950
C17 Vs. A1 1.6 0.104 0.9 0.333 0.9 0.376
C17 Vs. CI 8.0 0.000 3.5 0.000 3.8 0.000
AT Vs. AI 0.7 0.499 0.9 0.366 0.9 0.356
AT Vs. CI 10.1 0.000 3.5 0.000 3.9 0.000
AI Vs. CI 9.5 0.000 2.8 0.005 3.1 0.001
R
2
− Adj 58.3 38.6 30.0
F-Test 42.2 0.000 15.8 0.000 35.0 0.000
a
Column (1) contains aggregate results for the OLS estimates.
b
Columns (2) and (3) give results for the random effects model with GLS regression. There were no
significant differences when clustered error panel regression was used, so we have reported
random effects with GLS only. These estimates were obtained after controlling for certain job-
related and socio-economic variables. Job-related variables were skill level (calculated from the
practice period), job satisfaction, employers’ kindness signals, and fairness of the job contract.
Socio-economic variables were age, gender, and major subject of study. Summary statistics of
these variables are provided in Table 3, Appendix A.
III.3. Supportive Analysis through Quality and Shirking
We analysed two additional behaviour variables (quality of output and
shirking time) to test whether these two variables supported our findings from
the efficiency analysis. Our behavioural hypothesis was that if workers valued
the socio-emotional incentives, the relative wage concerns might not
significantly decrease their morale. Relative wage concerns in the group with no
socio-emotional resources should induce decreased motivation that could result
in higher shirking times and lower output quality.
We first considered the quality of output; workers’ faulty output induces
additional costs to the employer and cannot be ignored when assessing their
efficiency. However, our preceding analysis of effective output does not take
into account the cost of faulty output. The quality of output is defined as the
percentage of the effective output relative to the total output produced. Our
results showed negligible differences in quality among all treatment groups
11
except the CI condition. The quality of output varied from 99.24 percent (in C0
and C17 ) to 99.41 percent (in AI ). Quality in the CI group was 98.89 percent,
which was a significant decrease compared to the other treatment groups (Table
6 in Appendix A). Statistics for the quality analysis are in line with our main
findings.
With regard to shirking times, our regression analysis has already
highlighted the importance of shirking behaviour. Table 5 exhibits shirking time
over 30-minute sessions in all treatment groups. We observed a general
tendency for breaks after the first hour, especially in the C0 and C17 groups.
Similarly, workers in all groups, except for the C0 and AI groups, stopped
working before the three hours were complete; this resulted in decreased output
over the final session. Summary statistics of shirking times show that on
average, students in the two appreciation treatments wasted the least time (Table
6 in Appendix A). Conversely, lower morale in the C1 group resulted in
significantly more time wasted than in the other treatment groups. However, the
C17 group provided average output but also displayed significantly more time
wasted than both the appreciation treatment groups did (p < .10).
Table 5
Shirking Time Over 30-minute Sessions
Treatment Groups 1 2 3 4 5 6 Total
C0 1.7 1.6 3.1 1.3 2.0 1.7 1.9
C17 2.0 1.4 2.2 1.5 1.7 3.7 2.1
AT 1.0 1.1 1.5 1.6 1.8 2.8 1.6
AI 1.5 1.9 1.7 2.0 2.0 1.5 1.8
CI 2.9 2.6 2.9 2.6 2.2 3.1 2.7
Total 1.8 1.7 2.3 1.8 1.9 2.5 2.0
III.4. Supportive Analysis through the Post-Experimental Questionnaire
At the end of the job assignment, we asked participants to fill out a
questionnaire. The questionnaire asked them to rank their perception of fairness
of their job contract, employers’ kindness, and job satisfaction, and provide
socio-demographic information. These variables were selected to address our
hypothesis on the following theoretical basis. Many studies show that fairness
considerations are important (see, for example, the surveys conducted by Sobel
(2002); Camerer (2003); Fehr and Schmidt (2003)]. Concerns for fairness
strongly affect incentive properties of the contracts. This might affect labour
supply decisions because of an employee’s perception that the principal did not
fairly distribute the available incentive. Similarly, Kube, et al. (2011) and
Mahmood and Zaman (2010) have shown that non-monetary incentives send
stronger signals of kindness that facilitate more reciprocal behaviour. To test our
assumption that socio-emotional resources can also signal kind sentiments from
12
employers, we asked workers to report employers’ kindness on a Likert scale.
Finally, we included job satisfaction on the basis that the mutual exchange of
positive sentiments influences employees’ overall attitude toward their job [Witt
(1991)].
Before analysing any of these variables in detail, we ran a reliability
analysis on our three main variables; Cronbach’s alpha values were at least 0.75,
indicating that participants consistently answered all three questions.
7
The
summary statistics of these three variables are consistent and supportive of our
hypothesis. The data show lower levels of job satisfaction, employer kindness,
and greater concerns about fairness that resulted in the lowest reciprocal
behaviour within the CI group. We found no significant differences in fairness,
job satisfaction, and kindness within the AI or other “uninformed” groups. This
was reasonably consistent with our main findings.
IV. CONCLUSIONS
According to the basic idea of the “labour market as a partial GEM,”
worker loyalty is exchanged for high wages, and this loyalty can be translated
into high productivity through effective management [Akerlof (1984)].
However, non-cash incentives are not only cost effective but might also
substantially contribute to boosting morale; increasing productivity; and
improving work quality, safety standards, and customer service [Wiscombe
(2002)]. The results from our controlled field experiment support the view that
socio-emotional resources like appreciation and recognition can also be used to
induce reciprocal gift exchange to enforce incomplete contracts. We found
strong gift exchange effects within our appreciation treatment; efficiency gains
were even higher than the more costly high wage treatment, where we paid
wages that were 17 percent higher.
This study was extended to test relative wage concerns in the presence of
socioemotional resources. We provided two groups of workers—one with
appreciation and the other without it—the same information regarding relative
disadvantageous wage positions. The treatment group with a 17 percent higher
wage served as a reference point for our information treatments. Results clearly
showed that workers given appreciation showed systematically decreased
concerns for the relative wage differential. Additionally, there was only a slight
decrease in productivity compared to the pure appreciation treatment.
Nevertheless, workers in this group entered significantly more output compared
to our benchmark group. Workers with no appreciation responded more
adversely; we found significant losses in productivity compared to the
benchmark treatment. A parsimonious interpretation of our results can be
7
Cronbach’s alpha is one of the most popular reliability statistics [Cronbach (1951)]. It
determines the internal consistency, or average correlation among items. Along with proper
theoretical backing, a higher Cronbach’s alpha value indicates greater consistency among test items.
13
provided by the social gift exchange theory. Workers can positively judge the
fairness of a contract and employers’ kindness, and systematically provide
higher job effort only if they sufficiently value appreciation more than the
relative wage differentials. This observation provides important evidence against
results of studies using only one-dimensional wage comparisons.
Our results have several important implications, both theoretically and
practically. First, the adverse consequences of wage differentials are based on a
notion of fairness that implicitly violates worker heterogeneity [Abeler, et al.
(2011)]. We suggest that this argument ignores the social costs an employer
pays to create a socially acceptable relationship with the worker. These
efficiency-enhancing social ties, in turn, create stronger reciprocal behaviours
[Brandts and Sola (2010)]. We argue that workers compare the social bond with
their employer (and peer group), along with absolute wages; this observation
also explains decreased job turnover rates compared to existing wage
differentials. The results of this experiment should not be interpreted as evidence
of absolute substitutability of socio-emotional incentives for monetary
incentives. The size of the wage differential and credibility of social incentives
play important roles in determining the reciprocal effect. Furthermore, how
workers respond to social incentives in different institutional settings, their
framing of these incentives, and cultural and economic situations still need
additional research.
14
V. APPENDICES
V.1. Appendix A
Table 1
Workers’ Overview
Control Group C17
# Y Q T # Y Q T
1 729 99.45 15 1 1107 99.55 11
2 830 99.52 12 2 1056 98.97 16
3 820 98.56 13 3 946 99.89 19
4 873 99.77 15 4 870 98.64 14
5 707 98.88 3 5 814 99.39 11
6 799 99.25 6 6 936 99.05 16
7 741 99.20 7 7 712 98.34 16
8 765 98.97 7 8 821 99.15 13
9 929 99.25 7 9 839 99.06 15
10 823 98.80 3 10 986 99.60 4
11 772 99.87 16 11 1009 99.70 14
12 832 99.64 7 12 878 98.99 12
13 697 99.43 16 13 974 99.29 19
14 713 99.17 11 14 1038 98.67 8
15 785 99.12 9 15 959 99.17 12
16 835 99.64 9 16 1007 99.21 9
17 876 98.65 8 17 949 99.16 14
18 784 100.00 13 18 1118 99.38 11
19 721 98.50 11 19 960 99.59 12
20 941 98.74 6 20 1015 99.71 7
21 828 99.40 12 21 995 99.20 15
22 736 99.46 13 22 1031 99.52 8
23 711 98.61 15 23 1057 98.97 15
24 796 99.50 26 24 988 99.20 8
25 884 99.66 22 25 1119 99.56 12
Note: Y,Q and T denote output, Quality and shirking time respectively.
15
Table 1
Workers’ Overview (cont.)
AT AI CI
# y Q T # y Q T # y Q T
1 1024 98.65 14 1 995 99.10 8 1 686 98.14 10
2 1074 99.44 11 2 1153 99.57 7 2 818 99.88 20
3 1042 98.67 8 3 1141 99.56 6 3 805 98.77 14
4 1162 99.66 9 4 1077 99.17 7 4 784 99.49 9
5 1043 99.90 7 5 1040 99.43 9 5 746 98.68 17
6 1050 99.34 10 6 1012 99.41 14 6 659 98.80 11
7 1085 99.45 9 7 1015 99.71 12 7 875 99.89 8
8 1106 99.55 12 8 992 99.10 15 8 574 98.80 23
9 742 99.07 30 9 1150 98.97 12 9 803 98.41 11
10 1116 98.85 10 10 1153 99.74 12 10 735 98.79 8
11 767 99.35 12 11 1076 98.99 4 11 870 99.32 14
12 1166 99.83 5 12 1035 99.62 13 12 670 98.97 16
13 1107 99.82 8 13 993 99.90 13 13 737 99.06 14
14 910 99.34 13 14 1068 98.16 11 14 640 99.22 24
15 833 99.64 13 15 1048 99.71 11 15 496 98.02 35
16 919 99.89 8 16 1123 99.65 8 16 632 98.14 21
17 1027 99.42 6 17 1107 99.82 6 17 760 98.57 19
18 1054 99.62 9 18 1056 99.06 6 18 842 99.06 9
19 1057 99.06 11 19 946 99.68 10 19 763 99.09 21
20 1116 99.55 6 20 870 99.32 17 20 742 99.07 15
21 1141 99.83 7 21 814 99.39 16 21 655 98.50 23
22 1107 99.28 5 22 936 99.15 9
23 1166 99.32 6 23 734 99.73 15
24 1016 99.32 4 24 821 99.64 13
Note: Y,Q and T denote output, Quality and shirking time respectively.
Table 2
Summary Statistics of Variables in Experiment
Output Mean Std. Dev. Min. Max.
C0 797.1 69.36 697 941
C17 967.4 100.68 712 1119
AT 1034.6 118.60 742 1166
AI 1014.8 113.62 734 1153
CI 728.2 97.51 496 875
Shirk Mean Std. Dev. Min. Max.
C0 11.3 5.46 3 26
C17 12.4 3.71 4 19
AT 9.7 5.15 4 30
AI 10.6 3.59 4 17
CI 16.3 6.79 8 35
Errors Mean Std. Dev. Min. Max.
C0 0.5 0.98 0 6
C17 0.6 1.04 0 4
AT 0.5 0.98 0 5
AI 0.5 1.07 0 9
CI 0.7 1.12 0 5
Quality Mean Std. Dev. Min. Max.
C0 99.2 0.43 98.5 100.0
C17 99.2 0.37 98.3 99.9
AT 99.4 0.36 98.7 99.9
AI 99.4 0.39 98.2 99.9
CI 98.9 0.51 98.0 99.9
16
Table 3
Summary Statistics of Variables in Experiment
Age Mean Std. Dev. Min. Max.
C0 21.2 1.76 19 25
C17 21.2 1.76 19 25
AT 22.3 1.55 19 25
AI 20.9 1.23 19 23
CI 21.3 3.20 19 24
Satisfaction Mean Std. Dev. Min. Max.
C0 4.5 0.96 3 7
C17 4.6 1.12 3 7
AT 4.9 1.25 3 7
AI 4.3 1.23 3 7
CI 3.3 1.35 1 7
Fairness Mean Std. Dev. Min. Max.
C0 4.0 0.87 2 5
C17 4.3 1.03 2 6
AT 4.4 1.10 3 7
AI 3.9 1.42 1 7
CI 2.9 1.06 1 4
Kindness Mean Std. Dev. Min. Max.
C0 3.6 1.50 1 6
C17 4.9 1.30 2 7
AT 5.0 1.33 2 7
AI 4.7 1.20 2 7
CI 3.1 1.24 1 6
Skill Mean Std. Dev. Min. Max.
C0 44.4 16.78 17 75
C17 36.2 14.50 12 65
AT 39.6 14.72 12 65
AI 40.5 18.33 16 65
CI 46.2 13.83 21 65
Table 4
Cross Comparisons of Post Experiment Variables
Fairness Satisfaction Kindness
Co vs C17 1.12 0.20 2.90
(0.26) (0.84) (0.00)
Co vs AT 1.13 0.99 3.24
(0.26) (0.32) (0.00)
Co vs AI 0.49 1.26 2.45
(0.62) (0.21) (0.01)
Co vs CI 3.41 3.33 1.31
(0.00) (0.00) (0.19)
C17 vs AT 0.08 0.81 0.49
(0.93) (0.42) (0.63)
C17 vs AI 1.31 1.19 0.61
(0.19) (0.23) (0.54)
C17 vs CI 3.94 3.24 3.92
(0.00) (0.00) (0.00)
AT vs AI 1.39 1.91 1.12
(0.17) (0.06) (0.26)
AT vs CI 3.96 3.67 4.16
(0.00) (0.00) (0.00)
AI vs CI 2.44 2.47 3.73
(0.01) (0.01) (0.00)
Note: Test statistics reported is Wilcoxon Rank-Sum Test, values in parentheses are p-values.
17
Table 5
Cross Comparisons of Shirking Time and Quality
Shirk Quality
Co vs C17 –1.25 –2.11
(0.21) (0.04)
Co vs AT –1.35 –3.11
(0.18) (0.00)
Co vs AI –0.23 –3.21
(0.82) (0.00)
Co vs CI –2.59 –1.73
(0.01) (0.08)
C17 vs AT –2.97 –1.76
(0.00) (0.08)
C17 vs AI –1.69 –1.78
(0.09) (0.08)
C17 vs CI –1.88 –0.07
(0.06) (0.95)
AT vs AI –1.40 –0.03
(0.16) (0.98)
AT vs CI –3.71 –1.50
(0.00) (0.13)
AI vs CI –3.01 –1.39
(0.00) (0.17)
Note: Test statistics reported is Wilcoxon Rank-Sum Test, values in parentheses are p-values.
V.2. Appendix B
Fig. 1. Screen Shot of Data Entry Software
18
V.3. Appendix C
QUESTIONNAIRE
1. Name ________________________
2. Assigned Code _________________
3. Registration No. ________________________
4. Department: ____________________
5. University _____________________
6. Contact No. ____________________
7. Age: ________________years
8. Gender
Male ○
Female ○
9. Previous Job Experience
Yes ○
No ○
10. How many hours on average you spend using computer per week?
_______________________________hours
11. Please rate the following statements on a scale from 1 to 7, where “1” means
“fully Disagree” and 7 means “Fully Agree”.
11.1: I am satisfied with my Job
1 2 3 4 5 6 7
11.2: I am treated kindly by employer
1 2 3 4 5 6 7
11.3: I consider my job contract as “Fair”
1 2 3 4 5 6 7
19
V.4. Appendix D
Fig. 2. Example Letter
REFERENCES
Abeler, J., A. Falk, L. Götte, and D. Huffman (2009) Reference Points and
Effort Provision. American Economic Review 101:2, 470–92.
Akerlof, George (1984) Gift Exchange and Efficiency Wage Theory: Four
Views. American Economic Review 74, 79–83.
Ariely, Dan (2008) Predictably Irrational: The Hidden Forces That Shape Our
Decisions. Harper Collins.
Armeli, S., R. Eisenberger, P. Fasolo, and P. Lynch (1998) Perceived
Organisational Support and Police Performance: The Moderating Influence
of Socioemotional Needs. Journal of Applied Psychology 83:2, 288–297.
20
Arbak, Emrah and Laurence Kranich (2005) Can Wages Signal Kindness?
Groupe d’Analyse et de Théorie Economique (GATE), Centre national de la
recherche scientifique (CNRS), Université Lyon 2, Ecole Normale
Supérieure. (Working Papers 0511).
Berg, Joyce, John Dickhaut, and Kevin McCabe (1995) Trust, Reciprocity, and
Social History. Games and Economic Behaviour, Elsevier 10:1, 122–142.
Bewley, Truman (1999) Why Wages Don’t Fall During a Recession.
Cambridge, Massachusetts: Harvard University Press.
Brandts, J. and C. Sola (2010) Personal Relations and Their Effect on Behaviour
in an Organisational Setting: An Experimental Study. Journal of Economic
behaviour and Organisation 73, 246–53.
Camerer, C. F. (2003) Behavioural Game Theory: Experiments in Strategic
Interaction. Princeton: Princeton University Press.
Campbell, C. and Kanal Kamlani (1997) The Reason for Wage Rigidity:
Evidence from Survey of Firms. Quarterly Journal of Economics 112, 759–
89.
Clark, Andrew E., David Masclet, and Marie-Claire Villeval (2010) Effort and
Comparison Income: Survey and Experimental Evidence. Industrial and
Labour Relations Review 63:3, 407–426.
Cox, J. C., D. Friedman, and S. Gjerstad (2007) A Tractable Model of
Reciprocity and Fairness. Games and Economic Behaviour 59:1, 17–45.
Cronbach, L. J. (1951) Coefficient Alpha and the Internal Structure of Tests.
Psychometrika 16:3, 297–334.
Cropanzano, R., and M. S. Mitchell (2005) Social Exchange Theory: An
Interdisciplinary Review. Journal of Management 31, 874–900.
Dickinson, David and Marie-Claire Villeval (2008) Does Monitoring Decrease
Work Effort?: The Complementarity between Agency and Crowding-out
Theories. Games and Economic Behaviour 63:1, 56–76.
Dur, Robert (2009) Gift Exchange in The Workplace: Money or Attention? The
Journal of the European Economic Association 7:2-3, 550–560.
Ellingsen, Tore and Magnus Johannesson (2008) Pride and Prejudice: The Human
Side of Incentive Theory. American Economic Review 98:3, 990–1008.
Eriksson, Tor and Marie Claire Villeval (2010) Respect as an Incentive. Institute
for the Study of Labour (IZA). (IZA Discussion Papers 5200).
Falk, A. (2007) Gift Exchange in the Field. Econometrica 75:5, 1501–1511.
Falk, Armin and Michael Kosfeld (2006) The Hidden Costs of Control.
American Economic Review 96:5, 1611–1630.
Fehr, E. and Simon Gachter (2000a) Cooperation and Punishment in Public
Goods Experiment. American Economic Review 90, 980–994.
Fehr, E. and K. Schmidt (2003) Theories of Fairness and Reciprocity: Evidence
and Economic Applications. In M. Dewatripont, L. Hansen, and S. Turnovky
(Eds.) Advances in Economics and Econometrics. Cambridge: Cambridge
University Press. pp. 208–257.
21
Fischbacher, U., S. Gächter, and E. Fehr (2001) Are People Conditionally
Cooperative? Evidence from Public Goods Experiment. Economics Letters
71, 397–404.
Fischer, Sven and Eva-Maria Steiger (2009) Exploring the Effects of Unequal
and Secretive Pay. Friedrich-Schiller-University Jena, MaxPlanck-Institute
of Economics. (Jena Economic Research Papers 2009-107).
Gächter, Simon and Christian Thöni (2010) Social Comparison and
Performance: Experimental Evidence on the Fair Wage-Effort Hypothesis.
Journal of Economic Behaviour and Organisation 76:3, 531–43.
Hennig-Schmidt, Heike, Bettina Rockenbach, and Abdolkarim Sadrieh (2008)
In Search of Workers’ Real Effort Reciprocity—A Field and a Laboratory
Experiment. Journal of the European Economic Association 49:55, 1–16.
Herzberg, Frederick (1959) The Motivation to Work. New York: John Wiley and
Sons.
Heyman, James and Dan Ariely (2004) Effort for Payment: A Tale of Two
Markets. Psychological Science 15:11, 787–793.
Hsee, C. K. (1999) Value Seeking and Prediction-decision Inconsistency: Why
Don’t People Take What They Predict They’ll Like Most? Psychonomic
Bulletin and Review 6:4, 555–561.
Keynes, J. M. (1936) The General Theory of Employment, Interest, and Money.
London: Macmillan.
Kosfeld, Michael and Susanne Neckermann (2010) Getting More Work for
Nothing? Symbolic Awards and Worker Performance. Institute for the Study
of Labour (IZA). (IZA Discussion Papers 5040).
Kube, Sebastian and Michel André Maréchal and Clemens Puppe (2011) The
Currency of Reciprocity-gift-exchange in the Workplace. Karlsruhe Institute
of Technology (KIT), Department of Economics and Business Engineering.
(Working Paper Series in Economics 25).
Levine, D. (1998) Modelling Altruism and Spitefulness in Experiments. Review
of Economic Dynamics 1, 593–622.
Mahmood, Saima and Asad Zaman (2010) Monetary and Non-monetary Gift
Exchange. The Pakistan Development Review 49:4, 719–740.
Masclet, D., C. Noussair, S. Tucker, and M. C. Villeval (2003) Monetary and
Non-monetary Punishment in the Voluntary Contributions Mechanism.
American Economic Review 93, 366–380.
Pink, D. (2009) Drive: The Surprising Truth About What Motivates Us.
Riverhead Books.
Rhoades, L. and R. Eisenberger (2002) Perceived Organisational Support: A
Review of the Literature. Journal of Applied Psychology 87:4, 698–714.
Sobel, Joel (2002) Can We Trust Social Capital? Journal of Economic
Literature 40:1, 139–154.
22
Torgler, Benno, Sascha L. Schmidt, and Bruno S. Frey (2006) Relative Income
Position and Performance: An Empirical Analysis. Institute for Empirical
Research in Economics. (Working Paper No. 282).
Wagner, Rod and J. K. Harter (2006) 12 Elements of Great Managing. Gallup
Press; 1 edition.
Wiscombe, Janet (2002) Rewards Gets Results. Workforce 81, 42–47.
Witt, L. A. (1991) Exchange Ideology as a Moderator of Job Attitudes
Organisational Citizenship Behaviours Relationships. Journal of Applied
Social Psychology 21, 1490–1501.
