Journal of Economic Psychology 40 (2014) 220–233

Contents lists available at ScienceDirect

Journal of Economic Psychology

journal homepage: www.elsevier.com/locate/joep

The effects of income source, context, and income level on tax

compliance decisions in a dynamic experiment
Yvonne Durham a,⇑, Tracy S. Manly b,1, Christina Ritsema c,2
a
Department of Economics, Western Washington University, 516 High Street, Bellingham, WA 98225, USA
b
Collins College of Business, School of Accounting & MIS, University of Tulsa, 800 S. Tucker Dr., Tulsa, OK 74104, USA
c
Department of Economics, Management, and Accounting, Hope College, 41 Graves Place, Holland, MI 49423 USA

a r t i c l e i n f o a b s t r a c t

Article history: This study employs a laboratory experiment to explore the joint effect of income source
Received 30 December 2011 (earned versus endowed) and decision context (tax versus nontax) on tax compliance
Received in revised form 5 June 2012 behavior. During the experiment, subjects faced various income levels and made multiple
Accepted 24 September 2012
reporting decisions. The results indicate that overall compliance is not significantly
Available online 29 September 2012
affected by the interaction of income source and context. However, this joint effect influ-
ences the relationship between income level and compliance and how compliance behav-
JEL classification:
ior evolves over time. In both cases, the treatment group with earned income in a tax
H26
C91
context displays behavior that is distinct from the other three groups.
Published by Elsevier B.V.
PsycINFO classification:
2260

Keywords:
Tax compliance
Income source
Context
Income level
Experiments

1. Introduction

In 2002, the IRS launched its National Research Program to obtain up-to-date information on tax compliance. The infor-
mation collected from the program is used by the IRS to target compliance efforts towards those tax returns that are most
likely to have errors. In a speech to the IRS Research Conference, Senator Chuck Grassley commented, ‘‘the IRS has limited
resources. . . Research that assists the IRS to target its resources in an effective manner is absolutely critical.’’ (Grassley, 2004).
In addition to efforts by the IRS, academic research has devoted substantial attention to understanding the characteristics
and motivations of compliant versus noncompliant taxpayers. Although a variety of different research methods (archival–
empirical, opinion survey, experiment) have been used to investigate these variables, little consensus has been reached.
This study focuses on the use of laboratory experiments in tax compliance research. Experiments are one means of gain-
ing insight into tax compliance behavior. Being able to interpret the results from these studies and what they imply for actual

⇑ Corresponding author. Tel.: +1 360 650 2794 (O); fax: +1 360 650 6315.
E-mail addresses: yvonne.durham@wwu.edu (Y. Durham), tracy-manly@utulsa.edu (T.S. Manly), ritsemac@hope.edu (C. Ritsema).
1
Tel.: +1 918 631 3992 (O); fax: +1 918 631 2164.
2
Tel.: +1 616 395 7580 (O); fax: +1 616 395 7490.

http://dx.doi.org/10.1016/j.joep.2012.09.012
0167-4870/Published by Elsevier B.V.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 221

taxpayers is essential. Tax compliance decisions are of interest to researchers in multiple disciplines (e.g. accounting, eco-
nomics, psychology) that use varying experimental protocols. Thus, tax compliance experiments have been conducted using
diverse methods and have explored both economic and psychological variables as determinants of compliance (for over-
views, see Andreoni, Erard, & Feinstein, 1998; Kirchler, 2007; Kirchler, Muehlbacher, Kastlunger, & Wahl, 2007). In this study,
we examine two methodological factors in tax compliance studies, income source (earned versus endowed) and decision
context (tax versus nontax language), and their impact on compliance decisions. While these design choices are not part
of the economic models of tax evasion (see Allingham & Sandmo, 1972; Srinivasan, 1973; Yitzhaki, 1974), they may have
psychological effects that influence compliance choices. The individual impact of each of these factors has been addressed
in previous studies, but their joint effect on tax compliance decisions has not been carefully examined. Since any study of
tax compliance requires the specification of both factors (inevitably resulting in a joint effect), identifying their combined
impact is important in interpreting the results from these studies and in understanding compliance behavior.
We examine the impact of income source and decision context on tax compliance behavior using a 2  2 between-sub-
jects design. This results in four distinct treatment groups. In addition, we study the effect of these two factors on compliance
levels when subjects face varying income levels and make a series of tax compliance decisions. We find no significant joint
effect on compliance overall, but we find that income source and decision context jointly impact compliance behavior across
income levels. Income level is positively correlated to compliance in three of the treatment groups. The exception occurs for
the earned/tax group, in which we find a negative relationship between income level and compliance. This provides a pos-
sible contributing factor to the conflicting evidence on the relationship between income and compliance reported in previous
studies.
We find that subjects in the earned income and tax context treatment group not only behave differently than their coun-
terparts in response to income level variations, but across decision making periods as well. Subjects in this group demon-
strate relatively stable compliance behavior over time, while subjects in the other three treatment groups all show
decreasing compliance as the decision is repeated. Earned income interacts with a tax context to promote a stabilizing effect
on compliance behavior over time.
Kirchler et al. (2007) summarize the findings of tax compliance studies examining the economic factors thought to affect
compliance (e.g. income, tax rate, audit rate, fines). They report difficulty in drawing definitive conclusions about the effects
of these variables on compliance behavior because they are unstable or unclear across studies. The problem of tax compli-
ance is complex and not likely to be completely explained by a purely economic approach. The majority of taxpayers pay
their taxes honestly, and evasion is lower than neoclassical economic models would predict. Other factors not addressed
in these models must be playing a role. This study helps substantiate the notion that experimental design choices may
change these other factors and influence subject behavior. This is an important consideration in interpreting past conflicting
results from tax compliance experiments, and it signals the need for researchers to contemplate the impact of their design
choices when constructing future experiments.

2. Background and literature review

2.1. Income source

Even though economic theory suggests that resources are fungible, psychological research shows that the origin of wealth
affects individual decision making. Therefore, an important methodological issue in tax compliance studies is the origin of
income. Previous research exploring the relationship between object source and behavior (e.g., Cherry, Kroll, & Shogren,
2005; Clark, 2002; Gerxhani & Schram, 2006; Kroll, Cherry, & Shogren, 2007; Loewenstein & Issacharoff, 1994; Muehlbacher
& Kirchler, 2009; Muehlbacher et al., 2008) suggests that the source of income may affect its subjective valuation and there-
fore the choices made with regards to retaining it. In addition to theories about behavior in the presence of sunk costs, prop-
erty rights, and house money, this suggests that subjects who earn income may exhibit different compliance behavior than
those who are endowed with the same income.

2.1.1. Earned income and decreased compliance

Sunk cost effect. Even though economic theory asserts that decision-makers should ignore sunk costs, psychological re-
search shows that observed choices often violate this. The sunk cost effect (Arkes & Blumer, 1985; Thaler, 1980) refers to
the empirical finding that subjects consider costs incurred at an earlier time in a way that influences them to be more
risk-seeking in later decisions. There is a greater tendency to continue an endeavor once an investment in money, effort,
or time has been made.
Thaler (1980) describes the sunk cost effect in terms of prospect theory (Kahneman & Tversky, 1979). Prospect theory
indicates that choices are not evaluated in terms of the final assets the decision maker receives but rather in relation to a
reference point. The outcome is considered a gain if it is larger than the reference point and a loss if it is not. A sunk cost
may change the reference point from the status quo to an aspiration level instead. Equity theory (Walster, Walster, &
Berscheid, 1977) indicates that people learn that certain inputs are rewarded. So they anticipate an outcome that is
consistent with their inputs. Sunk costs may increase aspiration levels and induce evaluation in a loss frame. Because the
value function in prospect theory is concave in gains and convex in losses, decisions framed in the loss domain lead to more
222 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

risk-seeking behavior than those in the gain domain. Also, as the value function in prospect theory is steeper for losses than
for gains, a loss matters more than an equivalent foregone gain. The sunk cost effect suggests more risk seeking (less com-
pliant) behavior for subjects who incur a sunk cost from working for income than for those who are endowed with it.
Property rights and subjective value. Prior research from experimental economics in bargaining games also suggests that
varying the origin of income may produce differing behavior. Subjects who earn property rights or bargaining positions that
are reinforced in the instructions tend to exploit them more fully and act in a more self-interested manner than those who
are randomly assigned the same property rights (Hoffman & Spitzer, 1985; Hoffman et al.,1994; Sonnegard, 1996). Similarly,
psychological research shows that reaching an outcome by luck or skill can make a substantial difference in its valuation.
Loewenstein and Issacharoff (1994) report that subjects value a mug more highly when it is obtained by performance rather
than by chance. So the value function may be steeper for earned income than for endowed income, and the loss from paying
the same amount of taxes more severe when income is earned. Subjects who have invested more effort to earn income may
subjectively value it more or feel more of a property right to it, and so comply less in an attempt to retain it.

2.1.2. Earned income and increased compliance

Reverse sunk cost effect. Zeelenberg and van Dijk (1997) report a ‘‘reverse sunk cost effect,’’ observing that incurring behav-
ioral sunk costs (time and effort) increases risk averse choices rather than risk seeking choices as predicted by the sunk cost
effect.3 They argue that the essential component of previous studies finding a sunk cost effect is that the only way in which
subjects could reach their aspiration levels (or make up for prior losses) was through the risky alternative. The decisions in-
volved choosing to continue with a risky proposition that provides a chance of recovering the sunk cost or simply quitting
and accepting a sure loss. They argue that it is possible that an aspiration level may also be satisfied by risk averse choices,
and that in such cases, incurring a sunk cost promotes more risk averse behavior instead. Because of the sunk cost, the current
asset position is perceived as a loss, and since losses are more intense than gains, the safe option will be more appealing than the
risky one. Kirchler, Muehlbacher, and Hoelzl (2009) also find support for the reverse sunk cost effect. They observe that hard
earned money is more likely to be reported honestly to tax authorities.
House money effect. Past research has documented the presence of a ‘‘house money effect’’ for gains that come from little
or no effort (see Clark, 2002; Davis, Joyce, & Roelofs, 2010; Harrison, 2007; Keasey & Moon, 1996; Thaler & Johnson, 1990).
Individuals who experience gains exhibit risk seeking behavior for future bets whose payoffs are small relative to the amount
already gained. Neilson (1998) describes the house money effect in a sequence of decisions and explains that after an initial
gain, subsequent losses that are smaller than the original gain can be integrated with the prior gain. Until the previous win-
nings are completely depleted, losses are coded as reductions in a gain, and individuals are willing to bet with the house
money they have already accumulated. This effect is more likely with endowed income than with earned.
Windfall gains. In a similar vein, Arkes et al. (1994) find that windfall gains are spent more readily than other types of assets.
They find that, unlike house money, windfall gains are not based on the effort involved in the gain but rather on being unan-
ticipated. Whether or not windfall gains are a factor in the current study depends on the expectations of the subjects. At the time
they were recruited, subjects were not told how much they would be paid for their participation as we did not want to imply
what the expected or ‘‘correct’’ choices were. They were told they would receive $5 for showing up and whatever else they made
from their choices during the experiment. Therefore, we do not know what part, if any, of their earnings were unanticipated.
These various effects provide contradictory predictions for the relationship between income source and compliance. We
expect the relationship to depend on which of these effects are present and their relative sizes. Previous studies allow us to
anticipate when the sunk cost and reverse sunk cost effects may occur in the tax compliance setting. Which effect is present
depends in part upon the aspiration level. For an aspiration level at gross income, the sunk cost effect is predicted because
only the risky option (tax evasion) allows recovery of the sunk cost. If the aspiration level is at net income, the reverse sunk
cost effect is predicted because the safe option (compliance) allows recovery of the sunk cost. Zeelenberg and van Dijk (1997)
suggest that for compliance decisions it is plausible to assume that taxpayers who put different effort into earning their in-
come adapt to different aspiration levels. Kirchler et al., (2009) propose that it is more likely that high effort changes the
reference point to net income and that low effort decreases the likelihood that a taxpayer thinks about how much income
will remain after paying taxes and uses gross income as a reference point. They find that tax evasion is more pronounced
in low effort conditions, consistent with these hypothesized aspiration levels.

2.2. Context

Another important methodological issue in designing tax compliance experiments is specifying the context of the deci-
sion. Studies based in the tradition of experimental economics tend to adopt an economic approach to understanding tax
evasion and focus on the economic incentives involved. Because of the potential judgment and decision making effects aris-
ing from the use of contextual terms, these studies tend to minimize references to real-world institutions, roles and events in
order to maximize the reward dominance4 of the experiment and thereby maintain experimental control (e.g., Alm, Jackson, &
McKee, 1992a; Alm et al.,1990; Beck, Davis, & Jung, 1991; Cadsby, Maynes, & Trivedi, 2006).

3
Johnstone (2003) also describes a reverse sunk cost effect reflected in giving up on a failing venture ‘‘too easily’’ due to psychological forces such as being
upset by what is lost or because of self-imposed mental budgets.
4
Reward dominance is achieved when subjects are motivated by the monetary reward to the exclusion of other arguments in their utility functions.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 223

Tax Compliance Experiments Income Source Context Income Level Periods

(relationship
to compliance)a
Collins and Plumlee (1991) Earned Tax - 4
Collins, Murphy and Plumlee (1992) Earned Tax - 2
Becker, Buchner and Sleeking (1987) Earned Tax None/Negative 1
Anderhub, et al. (2001) Earned Tax Negative 4
Chung and Trivedi (2003) Earned Tax Negative 1
Alm and McKee (2006) Earned Tax Negative 30
Blumenthal, Christian, and Slemrod (1998) Earned Tax - 2
Cadsby, Maynes, and Trivedi (2006) Earned Nontax - 1

Alm, McKee and Beck (1990) Endowed Nontax Positive 25

Beck, Davis and Young (1991) Endowed Nontax - 30
Alm, Jackson and McKee (1992) Endowed Nontax Positive 25
Alm, McClelland, and Schulze (1992) Endowed Tax vs. Nontax - 45
Wartick, Madeo, and Vines (1999) Endowed Tax vs. Nontax - 6
Trivedi and Chung (2006) Endowed Tax vs. Nontax None/Negative 24
Mittone (2006) Endowed Tax vs. Nontax - 60
Baldry (1987) Endowed Tax Negative 6
Kim, Evans, and Moser (2005) Endowed Tax - 9
Worsham (1996) Endowed Tax - 7
Kim, Evans, and Moser (2005) Endowed Tax - 8
Trivedi, Shehata, and Lynn (2003) Endowed Tax - 20
Trivedi, Shehata, and Mestleman (2005) Endowed Tax - 10
Callihan and Spindle (1997) Endowed Tax - 1
Spicer and Becker (1980) Endowed Tax - 10
Gerxhani and Schram (2006) Endowed Tax - 8

Boylan and Sprinkle (2001) Earned vs. Endowed Tax - 1

Kirchler et al. (2009) Earned vs. Endowed Tax - 1
Boylan (2010) Earned vs. Endowed Nontax - 2

Current study Earned vs. Endowed Tax vs. Nontax Investigated 8

a
The compliance measure used differs across papers. Therefore, the relationships shown above cannot be directly
compared. The relationship indicated is based on the particular measure used and reported in the paper.

Exhibit 1. Design parameters from prior tax compliance experiments. (See above mentioned references for further information.).

In other disciplines, the psychological incentives are of primary interest and the use of context is considered vital. These
studies often involve hypothetical behavior or reactions to case situations. The use of tax terminology may influence com-
pliance decisions by allowing subject attitudes towards taxation, fairness, and norms to enter into their decision making. A
large body of research is dedicated to exploring how psychological factors affect tax compliance behavior (see Kirchler
(2007) for an overview).5
Context may also provide useful information to the decision maker for accomplishing the task at hand. Haynes and
Kachelmeier (1998) suggest that a richer context may facilitate understanding and aid in task optimization. They indicate
that context provides new or clarifying information, direct attention to important details or consequences, and/or interact
with the decision maker’s expertise through memory coding effects. A richer context provides an opportunity for individuals
to make better informed decisions.
Introducing context into the decision process allows for a closer parallel to naturally-occurring situations. Trivedi et al.
(2005) argue that both contextual and economic incentives affect decision making and ignoring either one necessarily over-
looks some of the factors affecting compliance decisions. In an effort to introduce this contextual richness into the economic
decision making process, some tax compliance experiments implement this method and include tax descriptions in the
materials. (e.g. Boylan & Sprinkle, 2001; Collins & Plumlee, 1991; Moser, Evans, & Kim, 1995; Trivedi, Shehata, & Lynn,
2003; Trivedi et al., 2005).

5
For example, social norms may affect tax compliance behavior because people tend to seek the respect of others (McAdams, 1997) and there is a social
stigma associated with tax evasion (Scott & Grasmick, 1981). Warneryd and Walerud (1982) find that attitudes towards tax crimes are important in explaining
self-reported tax evasion. Wilson and Sheffrin (2005) find that taxpayers who considered the tax system to be ‘‘very fair’’ were more likely to be honest when
compared to those who perceived the tax system as just ‘‘fair.’’ Trivedi, Shehata, and Mestelman (2005) report that personal characteristics are the most
significant predictors of actual compliance decisions in the laboratory. Coricelli and Joffily (2010) indicate that context may influence decision making by
triggering an emotion or trust issue. Muehlbacher, Kirchler, and Schwarzenberger (2011) find that voluntary compliance depends on trusting tax authorities,
supporting the slippery slope framework presented by Kirchler et al. (2008).
224 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

2.3. Tax compliance experiments

As previously discussed, there has been little consensus among tax researchers as to the appropriate choices for income
source and context. In order to illustrate the variety of design choices that have been implemented, Exhibit 1 presents a sum-
mary of the design parameters from a representative sample of experimental tax compliance studies. Previous experiments
conducted to explicitly test for the individual effects of income source and decision context in the laboratory are briefly
reviewed below.
Experimental tests of income source effect Experimental evidence regarding the effects of varying income source indicates
that the source of income is important in relation to changes in economic variables thought to influence tax compliance
behavior. The decision context in which the comparison is made varies across studies (see Exhibit 1). Boylan and Sprinkle
(2001) investigate the effects of endowed versus earned income and changes in the tax rate. They find that subjects with
earned income respond to an increase in the tax rate by reporting more income while those with endowed income respond
to an increase in the tax rate by reporting less. Boylan (2010) finds that compliance increases following an audit when in-
come is endowed, but that the reverse occurs when the income is earned.
Experimental tests of context effect Several studies examine the relationship of context and its effect on tax-related decision
making. The source of income varies across these studies (see Exhibit 1). They typically find either no difference in compli-
ance levels or strictly more compliance when a tax context is used. Some compare a tax context to a pure gambling context
while others compare it to the same decision using nontax terminology.
Baldry (1986) finds that when the compliance problem is presented as a pure gamble, all of the subjects choose some
level of evasion. In contrast, when the same decision was presented in a tax context, only approximately seventy percent
chose a positive level of evasion. Mittone (2006) confirms this and shows that the number subjects choosing evasion in a
tax context is typically lower than the number of evaders in a generic gambling context.
Cadsby et al. (2006) argue that taxpayers are not presented with a morally-neutral gambling game, but that tax author-
ities demand compliance. They show that even in a nontax context, explicitly demanding compliance produces very high
compliance rates and less sensitivity to change in economic variables than in earlier studies. In all but their invitation-
to- gamble setting, the majority of their participants were 100% compliant.
Alm, McClelland, and Schulze (1992b) and Swenson (1996) both find no difference in compliance behavior when tax (as
opposed to nontax) terminology is included and when it is not. Wartick, Madeo, and Vines (1999) show that subjects over
the age of 25 respond to contextual clues by reporting more income in a tax than nontax context, and subjects under the age
of 25 show no difference in compliance across contexts. Trivedi and Chung (2006) find that context and income level jointly
affect compliance. They find that compliance decreases as income increases from low to medium and then remains steady as
income increases to high in a nontax setting, but find no relationship between compliance and income level in the tax con-
text. Together, these findings indicate that a tax context, when compared to a pure gambling choice, tends to increase com-
pliance. However, when tax terminology is compared to nontax terminology, there is no clear evidence that the tax context
increases compliance.
Given the economic and psychological factors involved, the tax compliance decision is complex. Income source and
context affect compliance in laboratory markets, but their effects are not straightforward. Income source tends to matter
when altered in relation to other variables (namely tax rate and audit). This signals an important interaction effect be-
tween income source and other factors. Context impacts compliance behavior, especially when the tax context is com-
pared to a pure gambling context. Clearly, both of these factors affect tax compliance decisions, and we posit that their
interaction is important. No systematic relationship between the two factors and compliance has been determined. We
hypothesize that compliance decisions are affected by the choice of research design parameters in general, and by the
interaction of income source and decision context in particular. It is likely that the variety of design choices employed
in previous tax compliance studies contributes to their contradictory findings. This leads to the first testable hypothesis
presented below.

H1. Compliance is influenced jointly by income source and context.

2.4. Income level

The economic model typically used to evaluate the income tax reporting decision is that of Allingham and Sandmo (1972)
(see also Srinivasan, 1973; Yitzhaki, 1974). While the model predicts that the fraction of actual income declared will increase
with the penalty rate, audit rate, and tax rate, its prediction for the relationship between the compliance rate and income
depends on the individual’s risk preferences. The fraction of actual income declared is predicted to increase (decrease) with
income when relative risk aversion is also increasing (decreasing). Since it is not clear which assumption about relative risk
aversion is most realistic, the model provides no single prediction for how increases in income level will affect compliance
decisions.
Several studies, both experimental and archival–empirical, address the relationship between income level and compli-
ance with mixed results (see Andreoni et al. (1998), Kirchler (2007), Kirchler et al. (2007) for overviews.) For example,
Alm, McKee, and Beck (1990) and Alm, Jackson, et al. (1992a) find that the amount of income declared in their experiments
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 225

increases as income increases and report a positive relationship between income and compliance. Several other experimen-
tal studies report a negative relationship between income and compliance (e.g. Alm & McKee, 2006; Anderhub, Giese, Guth,
Hoffman, & Otto, 2001; Baldry, 1987; Becker, Buchner, & Sleeking, 1987; Trivedi & Chung, 2006). It is important to note that
because the compliance measures used differ across studies, some care must be taken in interpreting and comparing the re-
ported relationships.6 In addition, archival–empirical studies (which by their very nature mainly involve earned income and tax
terminology) often report a negative relationship between income level and compliance (Clotfelter, 1983; Forest & Sheffrin,
2002; Witte & Woodbury, 1985; Young, 1994).
We report income level and number of periods for the studies shown in Exhibit 1, as these are the variables manipulated
within-subjects in this study. While past research on the relationship between income level and compliance has produced
conflicting results, Exhibit 1 reveals somewhat of a pattern. Generally, when the experimental setting uses earned income
and a tax context, the relationship between income level and compliance is reported to be negative. On the other hand, when
other combinations of these factors are used, the results are more varied, and typically positive in the endowed/nontax mix.
We posit that the contradictory findings from studies on tax compliance may, in part, be a result of the choice of design
parameters. This leads us to our next testable hypotheses that the interaction of income source and context affects the rela-
tionship between income level and compliance.

H2a. The relationship between income level and compliance is influenced jointly by income source and context.

2.5. Period

The field of experimental economics abounds with evidence that subject behavior changes and evolves over time. Com-
petitive markets tend to converge to equilibrium over successive periods. Individual contributions to public goods often de-
cay over time. Laboratory monopolists search for the profit-maximizing price over successive periods. Assets often trade
above intrinsic value until the final periods of the experiment (see Davis & Holt, 1993 for a discussion of early experimental
work in these areas). These results lead us to predict that tax compliance decisions may also change over time and perhaps
more closely approach self-interested theoretical predictions in later periods.
The house money effect implies that prior gains and losses from earlier gambles may influence future behavior and there-
by influence compliance behavior over time. Additionally, Zeelenberg and van Dijk (1997) indicate that post-decision regret
(Bell, 1982) may occur if a decision maker learns the result of the risky option even if it is not chosen (as occurs in a tax com-
pliance setting). This makes the safe decision less attractive than when the information is not available because post-deci-
sion regret is more intense than post-decision joy. This has implications for behavior over time, as observations of past
outcomes may influence future choices. This leads us to our next testable hypotheses that the interaction of income source
and context affects compliance behavior over time.

H2b. The relationship between experimental period and compliance is influenced jointly by income source and
context.

3. Research design

The current study allows us to examine the relationships between compliance and both income level and period while
jointly considering income source and context. It utilizes a double auction market to allow subjects to earn income in the
earned treatment, and matches endowed income levels to those earned income levels. Income levels are inherently different
between subjects and can be further differentiated by changing the subject parameters between periods.

3.1. Participants

A total of 96 undergraduate students were recruited to participate in the experiment. Their average age was 22 years. Risk
questionnaires similar to those used in prior research (Boylan, 2010; Boylan & Sprinkle, 2001; Murnighan et al., 1988) were
given to the subjects in order to gain some basic insights into their risk preferences. Examination of the questionnaires
showed that the majority of the subjects were risk averse to varying degrees. No concentration of students with particular
risk preferences was found among the treatment groups. Therefore, behavioral differences across treatments are unlikely to
arise because of differing risk preferences.

6
For example, while Alm et al. (1992a) find that the amount of income declared increases as income increases and report a positive relationship between tax
compliance and income, the income elasticities of declared income that they report, while positive, are less than one, indicating a decrease in the compliance
rate. Becker et al. (1987) find that while the percentage of taxes evaded is not dependent on income level, the probability of evasion increases as income
increases. While Chung and Trivedi (2003) do not vary the income earned by the subjects in their experiments, they report a negative relationship between the
amount of income reported during the experiment and the actual income subjects indicated that they earned that year.
226 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

INCOME SOURCE
Endowed Earned
2 sessions 2 sessions
Nontax

CONTEXT
Income Levels: Income Levels:
Low and High Low and High
2 sessions 2 sessions
Tax
Income Levels: Income Levels:
Low and High Low and High

Exhibit 2. Experimental design.

3.2. Between-subjects and within-subject manipulations

The experimental design manipulated two factors at two levels each between subjects: income source (earned versus en-
dowed) and context (tax versus nontax). This created a 2  2 design with four treatment groups. Each of the four treatment
groups contained 24 subjects. Eight experimental sessions were conducted with 12 participants each. Thus two sessions had
identical between-subjects manipulations to create four treatment groups. Income level was manipulated within-subjects
across periods as described below. The research design is illustrated in Exhibit 2.

3.3. Procedures

3.3.1. Context, experimental phases, and audit

Twelve subjects participated in each of the eight experimental sessions. Context (tax versus nontax) was varied between
subjects.7 Each session consisted of eight periods, with two phases in each period. During the first phase, subjects either earned
profit by participating in a double auction market or were notified of the profit with which they had been endowed for that
period. When subjects earned profit, the instructions referred to the period profit they had ‘‘earned’’ rather than been ‘‘given.’’
During the second phase, subjects were asked to indicate their reported income (profit) and pay a 40% tax (market fee) on the
reported amount. The demand for compliance was the same across contexts and was not extremely high as was the relative
coercive power of the tax authority (Muehlbacher et al., 2011).8 Subjects were told that they had to decide how much of their
period profit they ‘‘wished’’ to report to the researchers and that they ‘‘would’’ pay an income tax (market fee) equal to 40% of
the amount they reported. They were told that they could choose to report all, some, or none of their period profit as their re-
ported income (profit). Therefore, any difference between behavior in the tax and nontax context is not likely to arise from a
difference in the explicit demand for compliance or the relative coercive power of the tax authority.
After the subjects submitted their reports, the auditing (verification) scheme was executed. The subjects to be audited
(verified) were chosen by drawing three balls, without replacement, from a bingo cage in which twelve bingo balls had been
placed. Therefore, a subject had a 25% chance of having his/her reporting sheet audited (verified) each period. A subject who
had underreported his/her income was required to pay the underreported tax (market fee) and a penalty (verification fee)
equivalent to the underreported tax.

3.3.2. Earned income treatment

In the earned income experiments, the first phase of each period consisted of the subjects participating in a computerized
double auction market which lasted 2½ min. Six of the subjects were assigned the role of buyers and six were assigned to be
sellers. Income levels were manipulated within subjects over the eight periods. The participants were assigned unit costs and
values such that, in equilibrium, three sellers and three buyers were predicted to earn a ‘‘high’’ income level (21.00 experi-
mental dollars) and three sellers and three buyers were predicted to earn a ‘‘low’’ income level (5.25 experimental dollars).
Each participant was given ‘‘high’’ income costs or values for four of the trading periods and ‘‘low’’ income costs or values for
the remaining four periods, alternating high and low every two periods. This means that there is likely to be little difference
in the effort required to earn high income and low income. In the earned income treatments, subjects were given two trial
periods in which to participate only in the market phase, one with high income costs or values and one with low. No report-
ing decisions were made during these trial periods. The predicted equilibrium price and quantity remained the same for all
periods.
Trading in a computerized double auction market was chosen as the earning task because of the strong convergence prop-
erties of the double auction institution. This provides some experimenter control over income levels while still allowing sub-
jects to influence their own payoffs. As the market reaches equilibrium, the unequal incomes earned by the subjects become

7
The main text of this section denotes language for the tax context treatment. Text in parentheses denotes language for the nontax context treatment.
8
Subject tax knowledge was not measured and thus cannot be incorporated into the power–trust–compliance slippery slope framework. Kirchler et al.
(2008) specifically recognize that tax knowledge can mitigate this dynamic relationship.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 227

Table 1
Correlation table of experimental variables.

Income source Context Income level Period Prior audit

Compliance rate .03 .02 .10a .17 .04
Income source .00 .0001 .01 .001
Context .01 .00 .001
Income level .01 .08
Period .12
a
Pearson correlation coefficients significant at the .05 level are shown in bold.

more the result of their position in the market (which the experimenters control) and less the result of their individual bar-
gaining abilities. This mitigates the possible problem of observing less compliance with higher income simply because more
aggressive subjects earn more income and also comply less.

3.3.3. Endowed income treatment

In the endowed income experiments, subjects were given slips of paper each period that indicated their endowment for
that period. The experiments were paired across treatments in order to keep the income levels consistent. Each subject in the
endowed income treatment was given the same income actually earned by his/her counterpart in the earned income exper-
iment. Subjects were assigned positions randomly.

3.3.4. Subject payments

In each of the treatments, subjects were paid $5 for showing up on time and then brought into a room and seated ran-
domly at a computer terminal. They were given a set of written instructions, and in the case of the earned income experi-
ments, were also asked to read a set of computerized instructions. After completing all of the instructions, the subjects were
given a brief quiz to test their understanding of the instructions.
Eight periods of actual data were collected for each experiment. No trial periods were run for the endowed income exper-
iments. At the conclusion of each experiment, subjects were asked to complete the risk questionnaire and a basic demo-
graphic questionnaire. They were then called up one at a time and paid their earnings in private. Subjects earned an
average of $29.82 for approximately 2 h of time.

4. Results

The compliance rate (reported income as a percentage of actual income) is computed for each of the periods per subject.
The total number of observations used in the analysis is 742.9 A correlation matrix including the compliance rate and the other
experimental variables is shown in Table 1. The key relationships to note, which we will discuss further below, are a significant
positive correlation between income level and compliance rate and a significant negative correlation between period and com-
pliance rate.

4.1. Repeated measures analysis of covariance

The experimental data set is analyzed using a repeated-measures analysis of covariance (ANCOVA). The dependent var-
iable for the analysis is the transformed compliance rate10 for the period. The repeated-measures estimation controls for un-
ique subject effects not captured in the measured variables. Income source and context are included as random effects and the
interaction of these between-subjects manipulations is added to test H1. Income level and period are included as continuous
variables. The interaction of both of these measures with the four treatment groups (income source  context) are added to test
H2a and H2b. Thus, two three-way interactions and the related two-way interactions are in the estimation. Prior audit11 is in-
cluded as a control for subject behavior in tax compliance experiments.
The results of this estimation are shown in Table 2. The between-subjects manipulations of income source and context are
not significantly related to compliance, and neither is the interaction of these two variables, initially indicating similar deci-
sion making across the four treatment cells. This result fails to support the first hypothesis that compliance is influenced by
the joint selection of income source and context.

9
The expected number of observations is 768 which represents 96 subjects participating in 8 periods each. Due to computer failure, one of the experimental
sessions in which subjects were trading to earn income only completed seven periods instead of eight. This was matched to the corresponding group for
endowed income. Therefore, the total observations are reduced by 24–744. Two additional observations were lost because in one period the subject failed to
earn any income in the experimental market. This was also matched to the corresponding group for endowed income.
10
In order to convert the dependent variable to a normal distribution and thus prevent negative predicted values, a log-log transform was applied to the
compliance rate for the repeated ANCOVA analysis.
11
An indicator variable is created set to 1 in the period following an audit for individual subjects.
228 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

Table 2
Repeated measures ANCOVA (Subjects = 96; Period observations = 742).

Dependent variable: Compliance rate, transformed Coefficient standard error (p-value)

Intercept 1.61a
.79
(.04)
Between-subjects
Income source 2.04
1.11
(.07)
Context .239
1.13
(.83)
Income source  Context 1.10
1.60
(.49)
Within-subjects
Income level .016
.03
(.00)
Period .015
.10
(.00)
Prior audit .293
.26
(.25)
Two-way interactions
Income level  Income source .057
.04
(.13)
Income level  Context .071
.04
(.06)
Period  Income source .587
.14
(.00)
Period  Context .195
.14
(.36)
Three-way interactions
Income level  Income source  Context .129
.05
(.02)
Period  Income source  Context .576
.20
(.00)
a
Coefficients significant at conventional levels are shown in bold.

The within-subjects manipulation of income is significantly related to compliance, with subjects choosing to comply
more (less) when income is higher (lower). Further, the period measure is significantly negative, showing a dynamic com-
ponent to compliance behavior. The prior audit variable is not related to compliance.12 Both three-way interaction terms and
one two-way interaction (period  income source) are all significantly related to compliance. The significant three-way inter-
actions provide support for H2a and H2b and are discussed below.

4.2. Behavior of treatment groups by income level

The three way interaction of income level  income source  context provides the empirical test for H2a. This coefficient
is significant and negatively related to compliance. To illustrate this interaction effect, the mean compliance rates for each of
the four treatment groups (income source  context) are computed for both high and low income levels. The results are
shown in Table 3 and graphically represented in Fig. 1. The main effect of income level is positively related to compliance
in the estimation. However, Fig. 1 shows that a positive relationship holds for only three of the four treatment groups.
The group that earned income and made decisions in a tax context demonstrates the opposite behavior. For these subjects,

12
Sensitivity analyses were also performed including subject age, major, class rank and risk preference score as additional controls. None of these controls
were significant at conventional levels.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 229

Table 3
Compliance rates by income level for four treatment groups (Income Source  Context).

Income source
Endowed Earned
Low High Low High
Context
Nontax
.37a .53 .40 .50
(.45)b (.42) (.44) (.44)
83c 97 83 97
Tax
.46 .54 .48 .43
(.46) (.42) (.45) (.45)
102 89 102 89
a
Mean.
b
Standard deviation.
c
Period observations.

Fig. 1. 3-Way interaction: Income Level  Income Source  Context.

compliance decreases as income increases. The interaction between income source and context clearly affects the relation-
ship between income level and compliance, providing support for hypothesis H2a.

4.3. Behavior of treatment groups by period

The three way interaction of period  income source  context provides the empirical test for H2b. This coefficient is sig-
nificant and positively related to compliance. The mean compliance rates for the four treatment groups by period are shown
in Table 4. The compliance rates by period were tested for differences for each of the four treatment groups separately,
adjusting for multiple comparisons. Significant differences across periods for a given treatment group are indicated by dif-
ferent letters.13 The primary effect of significantly decreasing compliance over time holds for only three of the four treatment
groups. Once again the same treatment group as discussed in Section 4.2 – earned income and tax context – exhibits behavior
distinct from the other three groups. Rather than decreasing, the compliance rates across periods for this group remain statis-
tically similar. Regression lines were estimated to fit the compliance rates by period for each of the four groups. Fig. 2 illustrates
these results. The line for the earned income and tax context group remains relatively flat while the others show significant
decreases over periods. Again we find that the interaction of income source and context exerts an important influence on com-
pliance decisions over experimental periods. This supports hypothesis H2b.

13
For example, the first treatment group shown in Table 4 is endowed income and nontax language. For this group, the compliance rate for period 1 is
significantly higher than period 7 while periods 2–6, and 8 are not significantly different from each other, period 1, or period 7.
230 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

Table 4
Compliance rates by period for four treatment groups (Income Source  Context).

Period Endowed/nontax (%) Endowed/tax (%) Earned/nontax (%) Earned/tax (%)

1 68 A 81 A 62 A 57 A
2 45 ABa 64 AB 55 AB 46 A
3 43 AB 51 B 40 AB 47 A
4 46 AB 57 B 41 AB 39 A
5 45 AB 36 BC 44 AB 37 A
6 48 AB 42 BC 42 AB 34 A
7 31 B 36 BC 38 AB 54 A
8 40 AB 30 C 31 B 51 A
a
Significant differences among compliance rates by period are noted by different letters.

Fig. 2. 3-Way interaction: Period  Income Source  Context.

5. Discussion and conclusions

We find that neither income source nor context has a significant impact on compliance overall. We also find that on aver-
age, subjects choose to comply more as income increases and to comply less over time. However, these relationships hold for
only three of the four treatment groups. The earned/tax treatment group exhibits behavior that is distinctly different, com-
plying less as income increases and displaying relatively stable compliance rates over time. The interaction between income
source and context significantly impacts the relationship between compliance and income level and the dynamics of the
compliance choice. This indicates that the specification of these factors is an important consideration when designing studies
of tax compliance behavior and when interpreting their results.

5.1. Joint effect of income source and context and income level

Unlike the other three treatment groups, a negative relationship between income and compliance occurs for those who
earn income in a tax context. This helps to explain the conflicting results from previous experimental studies examining this
relationship. The negative relationship is consistent with many previous experimental studies using earned income in a tax
context,14 while the positive relationship for the other groups is consistent with the mixed observations from other studies
employing different combinations of the two factors.
Because of the various possible effects that result from varying either income source or context, we do not attempt to
definitively determine the source of the joint effect on the relationship between income and compliance. However, we offer
one possible explanation that is consistent with the observed relationships for the different treatment groups based on the
possibility that income source and context interact to form differing aspiration levels.
Consistent with Kirchler et al. (2009), we suggest that earned income may lead to evaluation in terms of net income and
endowed income to evaluation in terms of gross income. Additionally, we propose that because context may evoke memory

14
It is also consistent with archival-empirical studies reporting a negative relationship between income level and compliance.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 231

cues and provide information about how to process the decision, the tax context may encourage subjects to form an aspi-
ration level of net as opposed to gross income. The nontax setting may not evoke these cues and is more likely to result
in a status quo reference point of gross income.
This suggests that two forces are at work when income source and context interact. Earned income and tax context rein-
force one another to encourage an aspiration level of net income. Endowed income and nontax context again reinforce one
another and encourage an aspiration level at the status quo of gross income. When the forces are opposing (earned/nontax
and endowed/tax), we posit an aspiration level somewhere between net and gross income.
Both sources of income required some level of effort in our experiment, though endowed is extremely low (reading
instructions, filling out record sheets, waiting for the experimenter to deliver materials, etc.) Within a particular income
source treatment, the same level of effort resulted in differing levels of income, based predominantly on the period. So if dif-
ferences in aspiration levels are connected to effort level as Zeelenberg and van Dijk (1997) suggest, this would argue that
within the earned and endowed treatments, aspiration levels may not change from low to high income situations. This may
cause the domain of the compliance decision to change across income levels. For our discussion, we suggest that aspiration
levels are associated with the gross and net incomes of the low income situation, as the earnings task is not onerous. These
different aspiration levels, in conjunction with house money effects, can explain our observations.
In the earned/tax treatment, the aspiration level is posited to be at the net of low income. In the low income periods, the
safe choice satisfies the aspiration level which promotes compliance (a reverse sunk cost effect). In the high income periods,
because the same effort leads to ‘‘extra’’ money, that extra money may be viewed as house money, creating a ‘‘partial’’ house
money effect, causing more risk seeking behavior at higher income levels. This supports the observed negative relationship
between income and compliance in this treatment.
In the earned/nontax and endowed/tax treatments, we suggest that the interaction of income source and context
leads to an aspiration level between the net of low income and the gross of low income. In the low income periods, the safe
option cannot satisfy the aspiration point. Therefore, more risk seeking (less compliant) behavior is predicted in the low
income periods (a sunk cost effect). In the high income periods, it is more likely that the aspiration level could be satisfied
by the safe option (the net income in high income periods might be larger than the gross income in low income periods), and
a reverse sunk cost effect occurs, resulting in a positive relationship between income and compliance.
In the endowed/nontax treatments, the aspiration level is predicted to be at the gross of low income. In low income peri-
ods, compliance falls in the loss domain, and the aspiration level cannot be satisfied by the safe option. Therefore, in low
income periods, risk seeking behavior (lower compliance) is predicted. In the high income periods, it is more probable that
the aspiration level could be satisfied by the safe choice, resulting in more risk averse behavior at the higher income level.
This is also consistent with the observed positive relationship for this treatment group.

5.2. Joint effect of income source and context and period

The interaction of income source and context also affects how compliance decisions change over time. Overall, compli-
ance rates are lower in later periods. There are several reasons why we might observe decreasing compliance over time.
The house money effect is likely to come into play in observing tax compliance behavior over a series of decision-making
periods, especially in an endowed income setting. A house money effect could even be argued for the earned income treat-
ment, although to a smaller degree, as the earning task was not arduous and could still be classified as low effort. Because
subjects always learn the result of the gamble even if it is not taken, post-decision regret may also come into play and lead to
more risk-taking behavior over time and with more experience with the results of the gamble.
Additionally, evidence from economics experiments leads us to believe that behavior changes over consecutive decision-
making periods and may become more self-interested, approaching the predictions of the economic models more closely
over time. It is also possible that the dynamics of compliance behavior also reflects some learning. In addition to accumu-
lating gains at a relatively low cost across periods, subjects are gaining experience with a complex compliance decision.
While these things can explain the compliance behavior overall, and specifically the behavior of the subjects in three of
the treatment groups, it does not explain behavior in the earned/tax treatment. Compliance rates remain relatively stable
across periods even in the presence of accumulated gains and experience when earned income is combined with a tax con-
text. The combination of the earning task in the auction market and the framing of the decision in tax-specific language may
be mitigating the other effects over time in this treatment.
Why does this specific interaction of the two factors result in distinctively different behavior over time? We suggest that
the earned/tax treatment is the most familiar environment for the subjects, as it is the most similar to what they face outside
of the laboratory. Thus, subjects have a better understanding of the task right away and do not need to adjust behavior much
over time. As Haynes and Kachelmeier (1998) discuss, the tax context may provide more information and cues for making
the optimal decision, and subjects’ experience with the environment better directs their attention to important details and
consequences of the decision. The house money effect is also likely to be smaller with earned income. So the combination of
contextual information and a smaller house money effect might lead to behavior that is more stable over time in the earned/
tax treatment. The familiar environment may foster quicker learning and promote more stable behavior over decision
making periods.
232 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233

5.3. Implications and limitations

The purpose of this study is to further explore the factors that determine tax compliance. Laboratory experiments offer
insight into tax compliance behavior, and being able to interpret the results from laboratory studies and what they might
imply for actual taxpayers is key to gaining this insight. While the outcomes from this current study may help us explain
the behavior of actual taxpayers, the primary purpose of the research is demonstrate that experimental design choices affect
behavior in the laboratory. This impacts interpretation of laboratory results and their implications and also demonstrates the
need to carefully consider experimental design choices.
We have discovered that the specification of income source and context influences compliance behavior. Consequently,
the results from previous and future experiments need to be examined with this in mind. Subjects who are endowed with
income and who make decisions in a tax context behave differently than those who do not. Which combination of these two
factors is the most appropriate choice for future studies is an open question. However, the simple fact that it makes a dif-
ference indicates that the specification of these two factors should be considered carefully.
Behavior in the laboratory can give us insights into behavior outside the laboratory, but there is no supposition that it
wholly reflects actual taxpayer behavior. The tax decision is much more complex than could ever be constructed in a labo-
ratory setting. The laboratory allows us to observe particular relationships in isolation, but necessitates choices that auto-
matically limit the generalizability of the results. For example,the average age of our subjects was 22. Certainly different
behavior might result from the use of older subjects. The experimental environment has a specific tax rate, audit rate,
and penalty, all variables that have all been found to affect tax compliance. Consequently, the specific relationships found
in this study may not be generalizable to other values of these determinants. Additionally, the nontax context used was
not completely neutral (a pure gamble or lottery), but only described the tax decision in nontax terms – market fees, veri-
fication fees, etc. So things like a demand for compliance and trust for authorities were still in play. Finally, income level was
not directly related to effort level in our earned treatment which may lead to concerns about interpreting income level var-
iation results.15 The fact that the earned/tax treatment group exhibits behavior different from the other three treatment groups
is significant. This is the treatment group that faces the environment most closely resembling the actual decision making envi-
ronment. One might argue that this group is more likely to exhibit behavior closer to that which occurs in the natural environ-
ment. This group exhibits relatively stable compliance behavior over time and a negative relationship between compliance and
income level. This gives insight into the conflicting results on income level and compliance found in previous studies. Although
the experimental evidence for the relationship between income and compliance is not conclusive, many of the studies support a
negative relationship between the two, and nearly all those that feature earned income within a tax context exhibit this neg-
ative relationship. Taken together, the evidence is mounting that the correlation between income and compliance is likely in-
verse. This has valuable and immediate implications for those involved in designing audit programs and compliance initiatives.
It suggests that tax authorities may wish to emphasize and pursue efforts targeted to high-income taxpayers.

Acknowledgements

The authors would like to thank James Alm, Don Finn, Geoffrey Sprinkle and Martha Wartick for their helpful comments
on the experiment materials and earlier drafts. We would like to thank workshop participants at the University of Arkansas,
Oklahoma State University, the Annual Meeting of the Western Economic Association and the Annual Meeting of the Amer-
ican Accounting Association for their helpful comments. We would like to thank the University of Arkansas Walton College of
Business for financial support.

References

Allingham, M., & Sandmo, A. (1972). Income tax evasion: A theoretical analysis. Journal of Public Economics, 1, 323–338.
Alm, J., Jackson, B., & McKee, M. (1992a). Estimating the determinants of taxpayer compliance with experimental data. National Tax Journal, 45, 107–114.
Alm, J., McClelland, G., & Schulze, W. (1992b). Why do people pay taxes? Journal of Public Economics, 48, 21–38.
Alm, J., & McKee, M. (2006). Audit certainty, audit productivity, and taxpayer compliance. National Tax Journal, 59, 801–816.
Alm, J., McKee, M., & Beck, W. (1990). Amazing grace: Tax amnesties and compliance. National Tax Journal, 43, 23–37.
Anderhub, V., Giese, S., Guth, W., Hoffman, A., & Otto, T. (2001). Tax evasion with earned income – An experimental study. FinanzArchiv, 58, 188–206.
Andreoni, J., Erard, B., & Feinstein, J. (1998). Tax compliance. Journal of Economic Literature, 36, 818–860.
Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35, 124–140.
Arkes, H. R., Joyner, C. A., Pezzo, M. V., Nash, J. G., Siegel-Jacobs, K., & Stone, E. (1994). The psychology of windfall gains. Organizational Behavior and Human
Decision Processes, 59, 331–347.
Baldry, J. C. (1986). Tax evasion is not a gamble: A report on two experiments. Economics Letters, 22, 333–335.
Baldry, J. C. (1987). Income tax evasion and the tax schedule: Some experimental results. Public Finance, 42, 357–383.
Beck, W., Davis, J., & Jung, W. (1991). Experimental evidence on taxpayer reporting under uncertainty. The Accounting Review, 66, 535–558.
Becker, W., Buchner, H., & Sleeking, S. (1987). The impact of public transfer expenditures on tax evasion. Journal of Public Economics, 34, 243–252.
Bell, D. E. (1982). Regret in decision making under uncertainty. Operations Research, 30, 961–981.
Blumenthal, M., Christian, C., & Slemrod, J. (2001). Do normative appeals affect tax compliance? Evidence from a controlled experiment in Minnesota.
National Tax Journal, 54, 125–136.
Boylan, S. (2010). Prior audits and taxpayer compliance: Experimental evidence on the effect of earned versus endowed income. Journal of the American
Taxation Association, 32, 73–88.

15
However, it could certainly be argued that effort is not always directly related to income level in the naturally-occurring economy either.
 Y. Durham et al. / Journal of Economic Psychology 40 (2014) 220–233 233

Boylan, S., & Sprinkle, G. (2001). Experimental evidence on the relation between tax rates and compliance: The effect of earned vs. endowed income. Journal
of the American Taxation Association, 23, 75–90.
Cadsby, C. B., Maynes, E., & Trivedi, V. U. (2006). Tax compliance and obedience to authority at home and in the lab: A new experimental approach.
Experimental Economics, 9, 343–359.
Callihan, D. S., & Spindle, R. M. (1997). An examination of contingent and noncontingent rewards in a tax compliance experiment. Advances in Taxation, 9,
1–23.
Cherry, T. L., Kroll, S., & Shogren, J. F. (2005). The impact of endowment heterogeneity and origin on public good contributions. Journal of Economic Behavior
and Organization, 57, 357–365.
Chung, J., & Trivedi, V. U. (2003). The effect of friendly persuasion and gender on tax compliance behavior. Journal of Business Ethics, 47, 133–145.
Clark, J. (2002). House money effects in public goods experiments. Experimental Economics, 5, 223–231.
Clotfelter, C. (1983). Tax evasion and tax rates: An analysis of individual returns. Review of Economics and Statistics, 65, 363–373.
Collins, J., Murphy, D., & Plumlee, R. (1992). The taxpayer’s dilemma: How hard to work and what to report? Advances in Taxation, 4, 31–53.
Collins, J., & Plumlee, R. (1991). The taxpayer’s labor and reporting decision: The effect of audit scheme. The Accounting Review, 66, 559–576.
Coricelli, G., & Joffily, M. (2010). Cheating, emotions, and rationality: An experiment on tax evasion. Experimental Economics, 13, 226–247.
Davis, D. D., & Holt, C. A. (1993). Experimental economics. Princeton, NJ: Princeton University Press.
Davis, L. R., Joyce, P., & Roelofs, M. R. (2010). My money or yours: House money payment effects. Experimental Economics, 13, 189–205.
Forest, A., & Sheffrin, S. (2002). Complexity and compliance: An empirical investigation. National Tax Journal, 55, 75–88.
Gerxhani, K., & Schram, A. (2006). Tax evasion and income source: A comparative experimental study. Journal of Economic Psychology, 27, 402–422.
Grassley, C. (2004). Remarks to the 2004 IRS research conference (June 2). <http://www.irs.gov/pub/irs-soi/keynote.pdf> (accessed 10.08.04).
Harrison, G. (2007). House money effects in public goods experiments: Comment. Experimental Economics, 10, 429–437.
Haynes, C., & Kachelmeier, S. J. (1998). The effects of accounting contexts on accounting decisions: A synthesis of cognitive and economic perspectives in
accounting experimentation. Journal of Accounting Literature, 17, 97–136.
Hoffman, E., McCabe, K., Shachat, K., & Smith, V. (1994). Preferences, property rights, and anonymity in bargaining games. Games and Economic Behavior, 7,
346–380.
Hoffman, E., & Spitzer, M. (1985). Entitlement, rights, and fairness: An experimental examination of subjects’ concepts of distributive justice. Journal of Legal
Studies, 14, 259–297.
Johnstone, D. (2003). Behavioral and prescriptive explanations of a reverse sunk cost effect. Theory and Decision, 53, 209–242.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291.
Keasey, K., & Moon, P. (1996). Gambling with the house money in capital expenditure decisions: An experimental analysis. Economics Letters, 50, 105–110.
Kim, C., Evans, J., III, & Moser, D. (2005). Economic and equity effects on tax reporting decisions. Accounting, Organizations and Society, 30, 609–625.
Kirchler, E. (2007). The economic psychology of tax behavior. Cambridge: Cambridge University Press.
Kirchler, E., Muehlbacher, S., Kastlunger, B., & Wahl, I. (2007). Why pay taxes? A review of tax compliance decisions. In: International studies program working
paper 07–30. Andrew Young School of Policy Studies, Georgia State University.
Kirchler, E., Hoelzl, E., & Wahl, I. (2008). Enforced versus voluntary tax compliance: The ‘‘slippery slope’’ framework. Journal of Economic Psychology, 29,
210–225.
Kirchler, E., Muehlbacher, S., & Hoelzl, E. (2009). Effort and aspirations in tax evasion: Experimental evidence. Applied Psychology: An International Review, 58,
488–507.
Kroll, S., Cherry, T. L., & Shogren, J. F. (2007). The impact of endowment heterogeneity and origin on contributions in best-shot public good games.
Experimental Economics, 10, 411–428.
Loewenstein, G., & Issacharoff, S. (1994). Source dependence in the valuation of objects. Journal of Behavioral Decision Making, 7, 157–168.
McAdams, R. (1997). The origin, development, and regulation of norms. Michigan Law Review, 96, 338–433.
Mittone, L. (2006). Dynamic behavior in tax evasion: An experimental approach. The Journal of Socio-Economics, 35, 813–835.
Moser, D., Evans, J., III, & Kim, C. (1995). The effects of horizontal and exchange inequity on tax reporting decisions. The Accounting Review, 70, 619–634.
Muehlbacher, S., & Kirchler, E. (2009). Origin of endowments in public good games: The impact of effort on contributions. Journal of Neuroscience, Psychology,
and Economics, 2, 59–67.
Muehlbacher, S., Kirchler, E., Hoelzl, D., Ashby, J., Berti, C., Job, J., et al (2008). Hard-earned income and tax compliance: A survey in eight nations. European
Psychologist, 13, 298–304.
Muehlbacher, S., Kirchler, E., & Schwarzenberger, F. (2011). Voluntary versus enforced tax compliance: Empirical evidence for the ‘‘slippery slope’’
framework. Europe Journal of Law and Economics, 32, 89–97.
Murnigan, J. K., Roth, A. E., & Schoumaker, F. (1988). Risk aversion in bargaining: An experimental study. Journal of Risk and Uncertainty, 1, 101–124.
Neilson, W. (1998). Reference wealth effects in sequential choice. Journal of Risk and Uncertainty, 17, 27–48.
Scott, W., & Grasmick, H. (1981). Deterrence and income tax cheating. Journal of Applied Behavioral Science, 17, 395–408.
Sonnegard, J. (1996). Determination of first movers in sequential bargaining games: An experimental study. Journal of Economic Psychology, 17, 359–386.
Spicer, M. W., & Becker, L. A. (1980). Fiscal inequity and tax evasion: An experimental approach. National Tax Journal, 33, 171–175.
Srinivasan, T. (1973). Tax evasion: A model. Journal of Public Economics, 2, 339–346.
Swenson, C. (1996). Experimental market evidence on implicit taxes. Working paper. University of Southern California.
Thaler, Richard. H. (1980). Towards a positive theory of consumer choice. Journal of Economic Behavior and Organization, 1, 39–60.
Thaler, R., & Johnson, E. (1990). Gambling with house money and trying to break even: The effects of prior outcomes on risky choice. Management Science,
36, 643–660.
Trivedi, V. U., & Chung, J. O. (2006). The impact compensation level and context on income reporting behavior in the laboratory. Behavioral Research in
Accounting, 18, 167–183.
Trivedi, V. U., Shehata, M., & Lynn, B. (2003). Impact of personal and situational factors on taxpayer compliance: An experimental analysis. Journal of Business
Ethics, 47, 175–197.
Trivedi, V. U., Shehata, M., & Mestelman, S. (2005). Attitudes, incentives, and tax compliance. Canadian Tax Journal, 53, 29–61.
Walster, E., Walster, W., & Berscheid, E. (1977). Equity: Theory and research. Newton, MA: Allyn & Bacon.
Warneryd, K., & Walerud, B. (1982). Taxes and economic behavior: Some interview data on tax evasion in Sweden. Journal of Economic Psychology, 2,
187–211.
Wartick, M., Madeo, S., & Vines, C. (1999). Reward dominance in tax-reporting experiments: The role of context. Journal of the American Taxation Association,
21, 20–31.
Wilson, J. L. F., & Sheffrin, S. (2005). Understanding surveys of taxpayer honesty. Finazarchiv, 61, 256.
Witte, A. D., & Woodbury, D. F. (1985). The effect of tax laws and tax administration on tax compliance: The case of the U.S. individual income tax. National
Tax Journal, 38, 1–13.
Worsham, R. (1996). The effect of tax authority behavior on taxpayer compliance: A procedural justice approach. Journal of the American Taxation
Association, 18, 19–39.
Yitzhaki, S. (1974). A note on ‘Income tax evasion: A theoretical analysis’. Journal of Public Economics, 3, 201–202.
Young, J. (1994). Factors associated with noncompliance: Evidence from the Michigan tax amnesty program. Journal of the American Taxation Association, 16,
82–105.
Zeelenberg, M., & van Dijk, E. (1997). A reverse sunk cost effect in risky decision making: Sometimes we have too much invested to gamble. Journal of
Economic Psychology, 18, 677–691.