cen01-2019
Constitutional
Economics
Network
Working Paper
Series
ISSN No. 2193-7214
CEN Paper
No. 01-2019
Testing preferences for basic income and its time
allocation effects in the German context:
A lab experiment
Ana Helena Palermo
Götz Werner Chair
of Economic Policy and Constitutional Economic Theory
University of Freiburg, Germany.
E-mail: ana.palermokuss@vwl.uni-freiburg.de
June 18th, 2019
University of Freiburg
Institute for Economic Research
Department of Economic Policy and Constitutional Economic Theory
Platz der Alten Synagoge / KG II D-79085 Freiburg
www.wipo.uni-freiburg.de
Testing preferences for basic income and its time allocation effects in
the German context: A lab experiment
Ana Helena Palermo1
University of Freiburg
Abstract
Inspired by Fröhlich and Oppenheimer (1990), an experimental survey in the lab was designed to
find out if preferences for three different redistribution schemes differ under a veil of ignorance.
The three schemes are a stylized version of the status quo German welfare state (A), a control
scheme without income taxation and redistribution (B) and one in which a flat tax-financed basic
income is paid to all (C). Furthermore, the study investigates whether the introduction of a basic
income induces a decrease in the time allocation to paid and unpaid work. The results point to no
significant difference in allocated working hours between A and C. Concerning preferences,
access to information on implications of schemes and self-interest played a central role in their
definition.
JEL classification: C91, I38, J22.
Keywords: lab experiment; basic income; welfare state; Germany; time allocation;
constitutional economics; labor supply.
1. INTRODUCTION
Many of the established social policy models have been designed for industrial societies and to
tackle social risks, which are related to industrial ones. Recently, we have entered a period in
which the digital and platform economies play an increasing role, and social security systems are
not designed for this new reality (Ollie Kangas in Reiners, 2018). In addition to the digitalization
1
This work was supported by the Konrad-Adenauer-Stiftung. Grant: PhD scholarship. Responsibility for the
information and views set out in this publication lies entirely with the author.
1
of the economy, other factors such as the emergence of new political forces, the persification of
household structures, migration movements and shifting dynamics in the labor markets have
contributed to the formation of new social and economic scenarios. In the face of these new
social dynamics, one has to recognize that a welfare state that is built on conditions of the past
cannot fulfill the demands of the coming generations (Straubhaar, 2017, p. 90).
As a way to cope with the increasing income and job insecurity that result from these
transformations in the world economy and to adapt social security systems to this new reality,
some suggest the introduction of an unconditional basic income. For the purposes of this paper,
unconditional basic income is defined as "an income (periodically) paid by a political community
to all its members on an inpidual basis, without means tested or work requirement" (van Parijs,
2004, p. 8). Basic income can be interpreted as a new concept for the social market economy
(Neumärker, 2018) or as a part of a broader set of reforms. Two critical points of such a change
are its potential effects on labor supply and whether such an initiative could win public approval.
This paper aims to analyze these two critical points in the German context and provide empirical
evidence for discussion.
An experimental survey in the lab was designed to find out if preferences concerning three
proposed income redistribution schemes may vary when peoples’ positions in society are
revealed. One scheme is a stylized representation of the German welfare state in its current form,
i.e., with a minimal means-tested income and relatively progressive income taxation system. The
second scheme functions as a control by representing a situation absent of taxation and
redistribution. The third scheme depicts a scenario where there is an unconditional basic income
financed by a flat income tax. The experiment was designed in order to focus on two issues. The
2
first concerns the possible effects of the introduction of a basic income on people’s paid and
unpaid work supply decisions. Moreover, the second concerns the participant’s preferences on
redistribution schemes and the influences of a constructed veil of ignorance on these preferences,
based on a constitutional economic approach. This veil of ignorance enables the construction of a
situation in which participants are not aware of their positions, which are represented by wages
and distributional schemes. When situated under a veil of ignorance, parties “do not know how
the various alternatives will affect their own particular case and are obligated to evaluate
principles solely on the basis of general considerations” (Rawls, 1971, pp. 136–137; Frohlich and
Oppenheimer, 1993, p. 15). Among the objectives of this experiment is the contribution to the
further development of experiments on basic income and the discussion on reforms for the
German welfare state.
The chosen methodology mixes elements of experiments and survey, as participants are both
introduced into hypothetical scenarios within a controlled laboratory setting and are asked how
they would react in such scenarios. There are a couple of reasons why an experimental survey
was chosen. Firstly, basic income as defined above has yet to be implemented in any political
system, and therefore there is no data available about how economic variables react to such an
institutional reform of the redistribution system. “Hence, any ex-ante analysis has to rely on some
kind of economic simulation” (Sommer, 2016, p. 108) or experiment. Secondly, this
methodology entails advantages, like the possibility to more easily control variables and enables
easy replication. The central drawback of this methodology concerns the external validity of
results. Also, when effects are statistically significant, it is not possible to extrapolate them
without reservations, as the sample selection was not random and is not fully representative of the
German population. However, one can argue that a fully representative data set is not always
3
required in experimental research and that external validity arises from the replication of an
experiment over different settings, using a variety of methods and measures (Runst, 2017;
McDermott, 2012). Thirdly, a survey experiment allows one to control the information which
participants receive, and the randomized treatments (exposure to information) allow for more
precise predictions (Page, 2018, p. 234). The experiment aims to observe whether the direction to
which the findings point to follows the stated hypotheses. In this sense, the replicability of the
experiment is an essential characteristic for evaluating external validity.
The remainder of this paper is organized in the following way. In the next section, a brief
literature review of the paper’s main topics is undertaken. In the third section the applied
methodology, experimental design, empirical estimation strategy, and the working hypotheses are
described in detail. In the fourth section, the stated hypotheses are contrasted with the experiment
results. In the fifth section, implications and improvement strategies are discussed.
4
2. ON BASIC INCOME, WELFARE STATE, AND WORK
The discussion on basic income is closely related to the debate on welfare institutions. When
concrete proposals for the introduction of a basic income are analyzed, basic income usually
constitutes one of many proposals in the reform package. Many of the recommendations include
changes in the tax and social security systems, sometimes putting into doubt if basic income is
indeed the central target of the project. These proposals can differ from each other considerably,
depending on the proposed level of basic income, the welfare state institutions it substitutes and
the suggested financing. In the German context, there are various basic income proposals, which
have been partially summarized according to their contents (Blaschke, 2017). Depending on the
proposal, the basic income can be interpreted as an extension, substitution or rearrangement of
the existing German welfare state.
The logic behind the type of basic income proposal that aims to extend the welfare state is the
universalization paradigm, which preconizes that social rights and access to social policies should
be decoupled from the accomplishment of a stipulated obligation (Monnerat et al., 2007). This
logic contrasts with the current German welfare system, which is strongly conditional. The
benefits in the ongoing order depend on the number of family members, level of earned and
unearned income, ability and willingness to work, among others (Gilroy et al., 2013, p. 45).
Further, this system lacks work incentives for inpiduals to leave unemployment, if one takes
into account the implicit taxes for the unemployed who start to work (implicit marginal tax rates
on the work income of single childless people who receive long-term unemployment benefits -
also called Hartz IV in Germany- can be up to 80%). Thus, work incentives constitute not only an
important issue within the basic income debate, but also in the discussion of the existent welfare
system.
5
Basic income advocates argue that labor supply would not necessarily decrease after such a
reform. Depending on the type of basic income, labor supply could even increase due to the
overcoming of poverty traps or be substituted by unpaid work. The skeptics draw instead a post-
basic income scenario where people would not be willing to work and welfare, as measured by
GDP, would decrease.
Some researchers have already discussed the possibility of interpreting data generation as a form
of labor (Arrieta-Ibarra et al., 2018) and argue that the notion of work will change so
fundamentally that the contemporary concept of unemployment will no longer exist (Daheim and
Wintermann, 2016, p. 11). Therefore, to rethink the meaning of unemployment and work is vital
to cope with the upcoming challenges. Along these lines, work can be defined not just as the
activities inpiduals develop in exchange for monetary payment but as all those social activities
that generate social effects for other inpiduals. Examples range from caregiving activities to
volunteering in different kinds of organizations.
3. EXPERIMENTAL DESIGN, ESTIMATION STRATEGY, AND HYPOTHESES
The experiment is the result of a mix of the methodologies found in Frohlich and Oppenheimer
(1990), Haigner et al. (2012) and Axelrad et al. (2016). The first paper builds the foundation of
the experimental design, more particularly of the sequence of events that constitutes each stage
(constitutional and post-constitutional) of the experiment (Frohlich and Oppenheimer, 1990). As
for the treatments that form the experimental strategy, they are inspired by those used in the
second paper (Haigner et al., 2012). Finally, the survey character of the experiment is rooted in
the innovative approach of the third paper (Axelrad et al. 2016), which constructed hypothetical
6
situations as a way to anticipate how people would potentially behave under specific
circumstances. Further, the experimental design is based on the manipulation of different rules
for taxation and redistribution, which differ among the tested treatments. The experiment was
programmed using z-tree (Fischbacher, 2007).
3.1. TREATMENTS (Schemes)
The experiment is composed of three treatments, A, B and C, which are also called redistribution
schemes throughout the experiment. Treatment A is a stylized representation of the taxation
status quo in Germany. Treatment B represents a scenario, in which there is neither taxation nor
income redistribution, and in treatment C there is a flat tax, and a basic income is paid to every
inpidual. Each participant played in just one treatment in order to become familiar with that one
specific treatment and to avoid confusion on rules specific to each treatment.
Treatment A
This treatment functions as a reference point for the participants and is the result of the
simplification of the actual German welfare state. Nevertheless, during the experiment, this
treatment is presented in a relatively neutral way, without mentioning its relation to the status quo
system in Germany. Here, the complexity of the income tax and social systems is reduced to a
chosen set of basic rules. It is assumed that participants have no other income source besides paid
work or social benefits in the form of Hartz IV (long-term unemployment benefit). A solidarity
tax, health insurance, old age insurance, nursing insurance, and other issues are not assumed for
the sake of simplification. The experiment abstracts from the German short-term unemployment
insurance (unemployment insurance I). Additionally, it is assumed all people are single without
children and in the same income tax class. The only withheld tax is income tax, which will vary
7
following the German rules for the year of 2016 (Bundesministerium der Finanzen, 2016, p. 31).
Besides, there is an unemployment insurance of 1.5% on wage from a gross income of €451,
which is called social contribution here and during the experiment. All this information is
decisive for the calculation of the net income. The social benefit is conditional and paid to
participants dependent on monthly net income. Every person with a net income up to
€1000/month receives a social benefit. However, the higher the net labor income, the smaller this
benefit will be. If net earnings of a benefit recipient are above €100, they are allowed to withhold
just 20% of the net labor income that exceeds €100. As a consequence of this rule, from a
monthly net labor income of €100/month on, the social benefit decreases linearly from €7222
until it reaches the value of zero.
Treatment B
This treatment works as a control. In B there is no taxation on income, no unemployment
insurance, and no social benefits. The objective of this treatment is to enable the comparison of
inpiduals' time allocation under taxation and no taxation.
Treatment C
The last one is treatment C, in which the income tax system is simplified, and a flat tax of 45%
substitutes the complicated income taxation of treatment A. This tax includes the unemployment
insurance. As for the social benefit, it is transformed into an unconditional one, i.e., every
inpidual receives it independent of income. The amount of payment is €722 to enable a direct
comparison with treatment A. In the experiment this payment is not communicated as a basic
2
The value of €722 was taken from the example calculation for the benefit of a single household in 2016, made by
the German Federal Ministry of Labour and Social Affairs/ Bundesministerium für Arbeit und Soziales (2017). This
calculation includes the base tariff (“Regelbedarf”) paid for single long-unemployed people in 2016 and estimated
additional subsidies for heating and housing costs.
8
income to avoid bias; it is just called Pauschalbetrag (lump sum payment). This flat tax of 45%
for the financing of a UBI of €722 is based on an adaptation of a basic income reform proposed
by Bergman (2014). For a basic income of €750 plus a health premium of €200 per month, he
proposed a 60% flat tax. As the health system was excluded, and the basic income is set at €722,
an approximation was made through a rule of three to find out a potentially feasible tax for the
case of treatment C. Two other proposals (Althaus, 2010 and “Flat Tax reform” found in Jessen
et al., 2015 ) also came into question, but the problem was that both also included the reform of
the pension system, from which was abstracted in the experiment for simplification. As the
experiment took place in Germany and aimed to simulate participants’ allocation of hours and
preference under different levels of information, it was decided to analyze a proposal that was
designed for the German system. Further, the decision to use a basic income financed by a flat tax
on income should not be interpreted as a normative preference for a particular type of basic
income funding. This decision was based on the comparability potential with the ongoing system
(treatment A). Therefore, to make the experiment more interesting and generate a trade-off
between treatments depending on the income, the introduction of a basic income accompanied by
an income tax reform was chosen.
Table 1 Summary of the treatments
Unemployment
Treatment Income tax system Redistribution rule
insurance
Stylized version of the German
A 1.5% Hartz IV system
income tax system in 2016
B None None None
Included in the
C 45% flat tax € 722 basic income
income tax
9
Comparing treatments A and C
One of the differences among these treatments concerns the issue of the poverty trap, which is
defined as the situation in which more impoverished inpiduals or households face higher
marginal rates because of the implicit taxes on their benefits (Barr, 2004, p. 225). The marginal
implicit tax rates for each treatment are represented in Figure 1, which shows that the rates faced
by those in treatment A with a gross income lower than €1000 are much higher than for those
with an income higher than €1000. In C this rate is constant among incomes, and therefore one
can say that the poverty trap in C is considerably less severe than in A, as the difference in the
implicit marginal tax rate among salaries disappears.
Figure 1 Implicit marginal tax rates by scheme (treatment)
Scheme A Scheme B Scheme C
90,00%
80,00%
IMPLICIT MARGINAL TAX RATE
70,00%
60,00%
50,00%
40,00%
30,00%
20,00%
10,00%
0,00%
0,00 €
200,00 €
400,00 €
600,00 €
800,00 €
1.400,00 €
4.400,00 €
1.000,00 €
1.200,00 €
1.600,00 €
1.800,00 €
2.000,00 €
2.200,00 €
2.400,00 €
2.600,00 €
2.800,00 €
3.000,00 €
3.200,00 €
3.400,00 €
3.600,00 €
3.800,00 €
4.000,00 €
4.200,00 €
GROS INCOME (MONTH)
3.2. EXPERIMENTAL DESIGN
The sequence of events (or steps) of the experiment, which is based on Frohlich and
Oppenheimer (1990), is pided into a constitutional and a post-constitutional stage. This
10
sequence is used as a method to identify preferred principles of justice that are represented by the
different forms of taxation and redistribution found in each treatment. On the constitutional stage,
participants chose their preferred treatment under a veil of ignorance, as they are not aware of
their future relative position in society, which is here represented by wages and the retribution
scheme (A, B or C) in use. The sequence of steps, which form this stage are the following:
Constitutional stage
1. Reading about the general rules of the game.
2. Reading about the three redistribution schemes, which were depicted in an
informative sheet I3.
3. Ranking I of schemes.
4. Reading about the implication of the schemes (gross-net-relation), which was
depicted in the informative sheet II4.
5. Ranking II of schemes.
In step 4, participants came to know about the consequences of the rules they were presented in
step 2. They learned about the net-gross relation of income for each of the three schemes they
were introduced to. This information was given both in the form of a table and in the form of a
graph. Also, in step 5 they were asked to rank the schemes a second time.
Post-constitutional stage
6. Random assignment of participants to one of the three schemes and an hourly
wage.
7. Ranking III of schemes.
8. First allocation period and report of both gross and net income.
9. Second allocation period and report of both gross and net income.
10. Ranking IV of schemes.
11. Questionnaire5
The random assignment in step 6 marked the transition from the constitutional stage to the post-
constitutional stage. The former is characterized by the revelation of the veil of ignorance, as
participants get to know their particular positions (wage and redistribution scheme) in the game.
3
Please check appendix A for the informative sheet I.
4
Please check appendix B for informative sheet II.
5
The questionnaire had questions on gender, age, semesters of study, program of study, if the participant worked or
not, how they financed their studies, number of kids, marital status and a question on the reason why they decided (if
at all) to invest part of the 48 hours in other activities besides paid work.
11
In step 6, participants were each randomly assigned to one of the three treatments- independent of
the declared preferences- and to one of five possible hourly gross wages (8.50 €, 13.50 €, 18.50
€, 23.50 €, and 28.50 €). The first hourly wage represents the German minimum wage for the
year 2016; different wages were chosen to test whether payment would affect participants'
choices. After participants learned about their position, they were asked a third time to rank the
schemes (step 7). Then, in both steps 8 and 9 they were asked the following question:
“You can use up to 486 hours of your disposable time in a week (Monday to Friday) to
work for the designated hourly wage. If you want, you can work less or do not work at
all and use these hours for other activities”.
“Based on the distribution scheme in which you are playing, how many of the 48
hours (if at all) would you effectively invest in paid work?”
The type of work was not included in the question. The only reference participants had
concerning their paid work was the hourly wage. After the first question was answered, a second
question was posed in both steps 8 and 9:
“If hours are left over, please distribute them among the following
activities. Enter a “0” where you would invest no hour.
Caring for household and non-household members, household
activities
Organizational and civic activities
Educational activities
Leisure, media use, social life
Hobbies
Other activities”
The categories of the second question are based on those used in the American Time Use Survey
(ATUS) (United States Department of Labor, 2017). The categories used enable classification of
the time use beyond a paid work/leisure dichotomy. As the first two categories mentioned above
generate direct social contributions to others, they were defined as unpaid work. At the end of
6
The 48 hours are based on the average weekly constraint of working hours in Germany according to the law.
12
both steps 8 and 9, participants’ net and gross wage were reported, taking both their scheme and
salary into account. A second allocation period (step 9) allowed for an examination of the relation
between experience with taxation on the one hand and attitudes towards treatments on the other
(Frohlich and Oppenheimer, 1990, p. 464).
3.3. DATA COLLECTION
The experiment was conducted in German with bachelor students and followed the instructions
displayed by a purpose build z-tree program (Fischbacher, 2007). Students received no other
information about the experiment besides those contained in the on-screen instructions and
appendices A and B. The experiment was run 12 times in groups of around 20 students, with 237
participants in total. Participants played inpidually, without discussions among each other.
Students’ study disciplines included economics, administration, political science, and other
subjects. Each participant was randomly assigned to one of the three treatments at the beginning
of the second (post-constitutional) stage of the game. Tables 2 to 4 summarize some
characteristics of the sample.
Table 2 Treatments
Treatment N Proportion
A 82 34.6%
B 82 34.6%
C 73 30.8%
Table 3 Descriptive statistics (means and standard deviations)
Variable Means SD Min Max
Age 23.53 2.97 20 42
Semesters of study 3.47 2.69 2 17
13
Table 4 Descriptive statistics (proportion of inpiduals in percent)
Variable Proportion
Woman 37.05%
Year of birth
until 1992 23.58%
1993-1994 18.11%
1995-1996 30.32%
1997-1998 28.00%
Work and study 70.32%
Receive Bafög7 14.00%
Marital status
married 1.68%
single 94.11%
other 4.21%
Course of study
Economics 64,98%
Administration 14,77%
Political science 7,59%
other 12,66%
3.4. HYPOTHESES:
3.4.1 Time allocation hypotheses
I. Effects on paid working hours: concerning the impact of distribution schemes on paid
working hours, the experiment has a more explorative approach. The intention is to
empirically investigate one of the central issues of the basic income debate, the effects of
basic income on paid work. The primary interest is in a potential difference between
treatments A and C concerning the hours invested in paid work. From the perspective of
the standard labor supply model, it is unclear how an increase in non-labor income may
affect labor supply. This will depend on whether leisure is seen as a normal or an inferior
good by an inpidual (Borjas, 2015, p. 36). Basic income critics often point to the
concern that basic income may lead to a withdrawal from the labor market. The objective
7
Public monetary support for students in need in Germany.
14
here is to analyze whether the experiment can provide some evidence on this line, or
instead in the opposite direction. As there is no redistribution in scenario B, it is expected
that the hours allocated to paid work by participants under treatment B will be statistically
significantly larger than those assigned by inpiduals under treatments A and C.
II. Effects on total working hours: here it is hypothesized that when receiving a basic income
(C) participants will not invest fewer hours in work than those playing under treatment A
when it is counted for both paid and unpaid work together. So, when comparing the hours
allocated to paid plus unpaid work in treatments A and C, it is not expected for them to be
significantly different from each other. As for the difference to treatment B, there is no
directional prediction. The central concern of this hypothesis is the analysis of the effects
of basic income on work understood in a broader sense when it is compared to a minimal
income scenario (A). In line with basic income literature and arguments put forth by many
advocates, it is not necessarily expected that basic income will lead to inpiduals
working less but rather it may lead to people having more freedom in choosing their time
allocation, thus being more prone to invest time in unpaid work since they would have a
guaranteed income source (van Parijs, 1997; Birnbaum, 2012; Blaschke, 2012;
Widerquist, 2013; Standing, 2015; Neumärker, 2018).
3.4.2 Constitutional hypotheses
III. It is expected that the provision of more information in step 4 has no significant effect on
the probability of an inpidual to choose scheme C and that the further explanation about
the implications of each scheme has a clarifying role and, therefore, will not have a
statistically significant effect on the probability to choose scheme C.
15
IV. The expectation is that the knowledge on one’s own hourly wage will have a statistically
significant effect on a participant’s probability to choose scheme C. It is hypothesized that
the higher the hourly wage one is assigned to, the lower the probability will be of this
participant selecting scheme C. This hypothesis is based on the expectation that
participants will behave as self-interested inpiduals once the veil is lifted. So, the veil is
used to convert self-interest into justice and, if lifted, it has the opposite effect; it
transforms justice into self-interest ” (Frohlich et al., 1987, p. 5).
V. The knowledge on one’s treatment is expected not to affect the dependent variable. So,
the assignment of a participant to a specific treatment should not influence the probability
of choosing scheme C. Therefore, the null hypothesis should not be rejected at a
statistically significant level for this variable.
3.5. ESTIMATION STRATEGY
Two types of regressions are used to test the hypotheses. The first type is used to test the time
allocation hypotheses (I and II) and is expressed by the equation below:
𝐻𝑜𝑢𝑟𝑠 𝑖,𝑗 = 𝛽0 + 𝛽1 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝐴 𝑖 + 𝛽2 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡𝐵 𝑖 + 𝛼𝑥 𝑖 + 𝜀 𝑖 (1)
In this OLS regression either the number of paid working hours (j = 1) or of paid plus unpaid
working hours (j = 2) is the dependent variable, which is denominated 𝐻𝑜𝑢𝑟𝑠. The explanatory
variables are dummies for each treatment, with treatment C being the base group. Further, it is
controlled for the five wage classes of the game, age, and gender, which are represented by 𝑥 𝑖 ,
the vector of the control variables. An interaction term between age and gender and age in the
quadratic form are also added. 𝛼 stands for the vector of the coefficients on the control variables,
16
and 𝜀 𝑖 is the error term. The second type is used to test the constitutional hypotheses and is
expressed by the two equations below.
𝐶𝑓𝑖𝑟𝑠𝑡 𝑖,𝑡 = 𝛽0 + 𝛽1 𝑀𝑜𝑟𝑒𝐼𝑛𝑓𝑜 𝑡 + 𝛽2 𝑉𝑒𝑖𝑙 𝑡 + 𝛽3 𝐸𝑐𝑜𝑛𝑒𝑥𝑝 𝑡 + 𝛼𝑥 𝑖 + 𝜀 𝑖,𝑡 (2)
𝐶𝑓𝑖𝑟𝑠𝑡 𝑖,𝑡 = 𝛽0 + 𝛽1 𝑀𝑜𝑟𝑒𝐼𝑛𝑓𝑜 𝑡 + 𝛽2 𝑤 𝑖 + 𝛽3 𝑤 𝑖 𝑉𝑒𝑖𝑙 𝑡 + 𝛽43 𝐸𝑐𝑜𝑛𝐸𝑥𝑝 𝑡 + 𝛼𝑥 𝑖 + 𝜀 𝑖,𝑡 (3)
Both equations represent a probit model, which is used to test the contributions of different
independent variables to the probability of an inpidual choosing scheme C as their priority
(𝐶𝑓𝑖𝑟𝑠𝑡). In the two versions of the model (equations 2 and 3), the choice of scheme C is the
dependent variable and is a binary variable. i refers to the participant and t to the rankings (I, II,
III or IV). The explanatory variables are variables that differentiate the moment in which each
ranking was made during the game. There are four rankings and therefore, three variables in the
model distinguish these rankings.
The first one is MoreInfo (more information), which tests if the information on the consequences
of the rules (step 4) influenced a person’s probability to select C as their first choice. The second
variable is related to the effects of the veil of ignorance on choice, i.e., captures the difference
between the constitutional (I and II) and post-constitutional (III and IV) rankings. The veil of
ignorance is captured by two variables, which are used in two variations of the estimation model.
One form is the dummy Veil (equation 2), which has the value of one for the first two rankings
and the value of zero for the two last ones. The second form (equation 3) is represented by
interaction terms between the variable Veil and wage classes (w), the base group is Veil with
wage class €8,50. The latter form considers that the effect of Veil may differ among the different
wage classes. The third variable is EconExp (economic experience), which is intended to capture
17
the possible effects of economic experience between rankings the first three rankings and the last
one and IV. An overview with the values of each explanatory variable in each of the rankings can
be found in Table 5. It is also controlled for age and gender, which are represented by the vector
𝑥 𝑖 and their coefficients by 𝛼.
Table 5 Description of rankings
Ranking MoreInfo Veil EconExp
I 0 1 0
II 1 1 0
III 1 0 0
IV 1 0 1
4. RESULTS
The results of the experiment are presented in the form of Mann-Whitney tests and regressions.
Further, descriptive statistics provide additional information on the collected data.
4.1. TIME ALLOCATION HYPOTHESES
4.1.1. HYPOTHESIS I: Effects on paid working hours
Descriptive statistics of paid working hours for each treatment are depicted in Table 6. The mean
was relatively similar among treatments, with treatment B having the highest mean. In Table 7 a
Mann-Whitney test was used to evaluate if the means of treatments differ significantly from each
other. The first glimpse at this table suggests that just the differences between A/B and B/C were
statistically significant. No relevant difference could be found between A/C.
18
Table 6 Descriptive statistics of paid working hours for each treatment (first allocation period)
Treatment N Mean Median SD Min Max
A 82 35.54 39 8.37 8 48
B 82 38.15 40 6.58 20 48
C 73 34.89 38 10.56 0 48
Table 7 Means and Mann-Whitney test results: paid working hours (first allocation period)
p-
Test Treatment Paid working hours z
value
A 35.54
1 2.52 0.01
B 38.15
A 35.54
2 0.09 0.93
C 34.89
B 38.15
3 2.33 0.02
C 34.89
Similar results can be interpreted from a regression (OLS) analysis of different explanatory
variables on paid working hours (see Table 8). Model 1 confirmed that participants in treatments
A and C invest on average respectively 2 and 2.5 hours less on paid work than those in treatment
B. Some controls were also statistically significant as Age_norm, Age_norm², and WAge_norm.
The variable Age_norm is a monotonic transformation of the variable Age. Age_norm grasps the
difference in years among the participants. The youngest participant, a 20-year old, was set as 0.
Therefore Age_norm= Age – 20. Age_norm revealed that older inpiduals tend to invest less
time in paid work. Age_norm² indicated that this effect is decreasing with the increase of
Age_norm. Finally, WAge_norm (interaction term between Woman and Age_norm) suggests that
the difference in paid working hours between man and woman increase with Age_norm. For each
year this difference increased by about one hour more for woman.
19
Table 8 Effects on paid working hours (OLS)
Paid working hours Ln (paid working hours)
(1) (2) (3) (4)
Base group Base group Base group Base group
Explanatory
(treatment B) (treatment A) (treatment B) (treatment A)
variables
EconExp 0.36 0.36 0.03** 0.03**
(0.30) (0.30) (0.02) (0.02)
TreatmentA -2.01* -0.08**
(1.10) (0.04)
TreatmentB 2.01* 0.08**
(1.10) (0.04)
TreatmentC -2.51* -0.50 -0.07* 0.00
(1.33) (1.51) (0.04) (0.05)
Wage
€ 13.50 2.05 2.05 0.06 0.06
(1.65) (1.65) (0.05) (0.05)
€ 18.50 -0.93 -0.93 -0.03 -0.03
(1.88) (1.88) (0.06) (0.06)
€ 23.50 0.65 0.65 0.01 0.01
(1.68) (1.68) (0.06) (0.06)
€ 28.50 0.35 0.35 0.00 0.00
(1.74) (1.74) (0.06) (0.06)
Woman -2.10 -2.10 -0.06 -0.06
(1.61) (1.61) (0.05) (0.05)
Age_norm -1.57*** -1.57*** -0.04*** -0.04***
(0.41) (0.41) (0.01) (0.01)
Age_norm² 0.08*** 0.08*** 0.00*** 0.00***
(0.02) (0.02) (0.00) (0.00)
WAge_norm 0.97*** 0.97*** 0.03** 0.03**
(0.36) (0.36) (0.01) (0.01)
constant 40.67*** 38.66*** 3.69*** 3.61***
(1.55) (1.58) (0.05) (0.05)
N 474.00 474.00 470.00 470.00
R² 0.09 0.09 0.07 0.07
F-statistic (F) 2.89 2.89 2.21 2.21
p-value (F) 0.00 0.00 0.01 0.01
Note: Standard errors are clustered at subject level and given in parentheses.
Significance levels: * p<0.10, ** p<0.05, *** p<0.01.
20
With the aim to account for possible non-linearity in the data, a log version of the OLS equation
was also run (see models 3 and 4 in Table 8). This log version produced similar results, except
for the control EconExp, which turned out to have a statistically significant positive effect on paid
hours. Another control was the variable Wage, which presented no statistically relevant results.
This finding evidences the complexity of the interactions between wage and time invested in paid
work. When a wage increases, the opportunity cost of a free hour increases, thus generating an
incentive to work longer. At the same time, when a wage increases people have an incentive to
work less because they will need to work fewer hours to achieve the same income as before. So,
from a purely monetary perspective, the effects of wage on paid working time are not evident.
They will depend on whether income or substitution effects are predominant.
4.1.2. HYPOTHESIS II: Effects on total working hours:
The effects on total working time (see Tables 9 to 11) are very similar to those presented in the
session above. Almost the same estimators were statistically significant. As for the difference
between treatments A and B, this ceased to be statistically significant, suggesting that when
unpaid working hours are accounted for, the difference between A and B tends to vanish. The
difference between B and C remained statistically significant, but with a lower coefficient (1.95
instead of 2.51). This result is partially in line with the hypothesis that if one accounts for both
paid and unpaid working hours, basic income will not necessarily lead to inpiduals working
less.
21
Table 9 Descriptive statistics of total working hours for each treatment (first allocation period)
Treatment N Mean Median SD Min Max
A 82 38.62 40 7.12 9 48
B 82 41.07 42 5.34 23 48
C 73 38.33 40 8.92 8 48
Table 10 Means and Mann-Whitney test results: total working hours (first allocation period)
Test Treatment Total working hours z p-value
A 38.62
1 2.36 0.02
B 41.07
A 38.62
2 0.07 0.94
C 38.32
B 41.07
3 2.18 0.03
C 38.32
22
Table 11 Effects on total working hours (OLS)
Total working hours Ln (total working hours)
(1) (2) (3) (4)
Base group Base group Base group Base group
Explanatory
(treatment B) (treatment A) (treatment B) (treatment A)
variables
EconExp 0.29 0.29 0.01 0.01
(0.24) (0.24) (0.01) (0.01)
TreatmentA -1.43 -0.05
(0.90) (0.03)
TreatmentB 1.43 0.05
(0.90) (0.03)
TreatmentC -1.95* -0.52 -0.08* -0.03
(1.09) (1.25) (0.04) (0.05)
Wage
€ 13.50 1.83 1.83 0.07 0.07
(1.35) (1.35) (0.04) (0.04)
€ 18.50 -0.40 -0.40 -0.01 -0.01
(1.54) (1.54) (0.06) (0.06)
€ 23.50 1.22 1.22 0.04 0.04
(1.43) (1.43) (0.06) (0.06)
€ 28.50 0.84 0.84 0.05 0.05
(1.35) (1.35) (0.05) (0.05)
Woman -0.98 -0.98 -0.02 -0.02
(1.32) (1.32) (0.04) (0.04)
Age_norm -1.02*** -1.02*** -0.03** -0.03**
(0.34) (0.34) (0.01) (0.01)
Age_norm² 0.05*** 0.05*** 0.00*** 0.00***
(0.02) (0.02) (0.00) (0.00)
WAge_norm 0.71** 0.71** 0.02** 0.02**
(0.29) (0.29) (0.01) (0.01)
constant 41.86*** 40.43*** 3.72*** 3.67***
(1.33) (1.25) (0.04) (0.04)
N 474.00 474.00 474.00 474.00
R² 0.08 0.08 0.07 0.07
F-statistic (F) 2.26 2.26 1.93 1.93
p-value (F) 0.01 0.01 0.04 0.04
Note: Standard errors are clustered at subject level and given in parentheses.
Significance levels: * p<0.10, ** p<0.05, *** p<0.01.
23
4.2. CONSTITUTIONAL HYPOTHESES
Here it is evaluated how different factors may interfere in the probability that an inpidual ranks
scheme C as the priority. To test the constitutional hypotheses probit, regressions (see Table 12)
were used. Model 1 includes the interaction term Veil x Wage. Model 2 includes neither the
variable wage nor the interaction Veil x Wage. The latter model focuses on the effects of each
ranking. Finally, in the third model, just half of the probit data was analyzed; the first and second
observations were excluded. This third model functions as a robustness check for the Wage
effects presented in model 1. In Table 13 there is an overview of the percent of participants who
ranked each scheme as first in each of the four rankings. Tables 14 and 16 show participants’ first
option for each hourly wage and Tables 15 and 17 indicate participants’ first option for each
treatment.
24
Table 12 Probability of choosing scheme C as the first option (probit)
Dependent variable: Cfirst
Explanatory Model 1 Model 2 Model 3
variables ME SE ME SE ME SE
MoreInfo 0.23*** (0.03) 0.24*** (0.03)
Veil x Wage
Veil x € 8.50 -0.06 (0.05)
Veil x € 13.50 0.00 (0.06)
Veil x € 18.50 0.13*** (0.04)
Veil x € 23.50 0.21*** (0.06)
Veil x € 28.50 0.26*** (0.06)
Veil 0.10*** (0.03)
Wage
€ 13.50 -0.08 (0.08) -0.08 (0.08)
€ 18.50 -0.13* (0.07) -0.12* (0.07)
€ 23.50 -0.25*** (0.07) -0.26*** (0.07)
€ 28.50 -0.34*** (0.07) -0.34*** (0.07)
EconExp -0.01 (0.02) -0.00 (0.02) -0.01 (0.02)
Age 0.01 (0.01) 0.00 (0.01) 0.01 (0.01)
Woman 0.38 (0.35) 0.22 (0.36) 0.09 (0.46)
Age x Woman -0.02 (0.02) -0.01 (0.02) -0.00 (0.02)
TreatmentA -0.12* (0.06)
TreatmentB -0.11* (0.06)
N 948.00 948.00 474.00
Notes: The values shown in the table are the marginal effects of each variable and not
the coefficients. Standard errors are clustered at subject level and given in parentheses.
Significance levels: * p<0.10, ** p<0.05, *** p<0.01.
Table 13 Participants’ first option by ranking (in percent)
Ranking Scheme A Scheme B Scheme C
I 65,4% 20,7% 13,9%
II 38,4% 24,1% 37,6%
III 31,2% 42,6% 26,2%
IV 30,4% 43,9% 25,7%
25
Table 14 Participants’ first option in ranking III by wage (in percent)
Hourly wage Scheme A Scheme B Scheme C
8,50 € 30% 28% 42%
13,50 € 40% 26% 34%
18,50 € 35% 35% 29%
23,50 € 25% 59% 16%
28,50 € 28% 61% 11%
Table 15 Participants’ first option in ranking III by treatment (in percent)
Treatment Scheme A Scheme B Scheme C
A 34% 44% 22%
B 30% 45% 24%
C 29% 38% 33%
Table 16 Participants’ first option in ranking IV by wage (in percent)
Hourly wage Scheme A Scheme B Scheme C
8,50 € 34% 21% 45%
13,50 € 37% 31% 31%
18,50 € 29% 43% 27%
23,50 € 30% 55% 16%
28,50 € 24% 67% 9%
Table 17 Participants’ first option in ranking IV by treatment (in percent)
Treatment Scheme A Scheme B Scheme C
A 34% 44% 22%
B 26% 52% 22%
C 32% 34% 34%
4.2.1. HYPOTHESIS III
Table 12 shows that the term MoreInfo is statistically highly significant (p< 0.01) in both models
1 and 2. Concerning the effect magnitude of MoreInfo, its marginal effect on the probability of
an inpidual to choose C as the first option is 0.23. That means that after inpiduals received
more information about the different schemes, the probability that they choose C as the first
26
option increased on average by about 23 percentage points. This result contradicts the hypothesis
that more information would have a clarifying role and could be attributed to a preference for
more progressive redistribution systems. The information provided in step 2 concerns solely the
general redistribution rules for each scheme. In this part, it is not clear what happens when
taxation and social payments are jointly analyzed. In step 4 this becomes clearer when the gross
and net income for different income levels are estimated for participants. Then, participants can
see that scheme C also entails progressivity, which was probably not clear for most of them in
stage 2. The access to this information can potentially explain why so many participants changed
their mind.
4.2.2. HYPOTHESIS IV
The probit regression confirmed that inpiduals with higher hourly wages are less prone to
choose scheme C as their priority. When one observes the marginal effects of wage dummies
(wage € 8.50 is the base groups), one perceives that they are consistently negative and increasing
in absolute terms with the hourly wage. Almost all wage dummies were statistically significant
(besides the first one).
The interaction term Veil x Wage aims to depict how participants behaved when they did not
know about their Wage, i.e., under the veil. The marginal effects of the interaction terms (see
Table 11) show that compared to the scenarios where participants did know about their Wage,
those with higher wages (€ 18.50, € 23.50, € 28.50) were more prone to choose C as their first
choice under the Veil. Moreover, these marginal effects are statistically highly significant
(p<0.01). These results point to a self-interest-oriented behavior of inpiduals, as participants
were less prone to choose C the higher their wage turned out to be.
27
4.2.3. HYPOTHESIS V
The regression results did not confirm this hypothesis. The results of model 3 (Table 11) point to
the fact that the treatment participants played in had a statistically relevant influence (p<0.10) on
their probability to choose C. Those participants who were assigned to treatments A and B were
less prone to pick C as their preferred scheme than those in treatment C (see also Tables 15 and
17), i.e. the experience one has within the treatment C tends to influence the choice of C as the
preferred scheme positively.
4.3. DISCUSSION
Inherent selection bias is part of this sample as a particular group solely constitutes it. So, any
extrapolation of results for the general population should be made with extreme care. On the
other hand, the fact that all participants are students and most do not have paid work as their
primary activity helps to avoid a bias in the experiment towards real earnings. People who have
paid work as their core activity would most probably tend to use their real wage as a reference
point, what would potentially bias the preferences stated in the game.
Another potential drawback of the experiment is related to the fact that stated and not revealed
preferences are measured, which are not necessarily the same (Hayo and Neumeier, 2017, p. 3).
Despite this fact, many survey experiments have been done in the last years (Hainmueller et al.,
2014, p. 27), which corroborates the important role played by this methodology. The use of such
surveys, which are based on stated preferences, is grounded on the assumption that people are
willing and able to report their attitudes (Holbrook, 2012; Runst, 2017). Lab experiments often
employ money as an incentive for participants to reveal their preferences. Nevertheless, there is
no evidence that such an incentive is necessary. Usually, people can put themselves in
28
hypothetical situations and "starting a question with imagine that …” achieves a focus similar to
one created by a small monetary incentive (Barbara et al., 2017, p. 599; Rubinstein, 2013).
Concerning the regressions of the time allocation part of the game (OLS), these contained
heteroscedasticity. Therefore, heteroscedasticity robust standard errors were used in these
regressions. Still, it is important to discuss where this bias may come from. One possible
explanation is a reference point bias. It seems that there is something like a "natural reference
point" for the allocation of hours to paid work. As people usually work around 40 hours a week,
participants may have tended to choose a number close to this. One possible explanation is the
existence of a hidden social norm that influences the participant's choice and that just a few
tended to deviate from it. Further, the reference point may also have affected the participants'
option in the first ranking, where they tended to prefer treatment A, which represents the welfare
state in which they live. As they are familiar with this system, it may be that they tended to prefer
this one in the beginning because it implicitly represented the status quo. In the experiment, no
explicit statement related A to the status quo system in Germany.
For further research, the experiment should be replicated with more heterogeneous groups. It
would be a crucial test for the external validity of results, to find out if other groups present
similar patterns of behavior and if not, how do they differ. The experiment was designed in a way
that enables uncomplicated replicability. Another improvement possibility concerns the
investigation of the influence of other variables on an inpidual's preferred redistribution system
and time allocation. Potentially other personal, cultural, social and economic characteristics may
help to explain inpiduals' preferences. It would also be interesting to test other basic income
proposals with the method presented in this paper.
29
5. SUMMARY AND POLICY IMPLICATIONS
The results for the constitutional part show, firstly, that the manner in which information is
framed and the amount of information one has access to can considerably influence the choice
and acceptance of reforms. How a reform is described and presented plays a central role in the
willingness of citizens to accept it, as can be derived from the effects of MoreInfo. Thus, “how a
UBI scheme is presented and framed in the political agenda, and how its implementation is
phased in, may well decide people's attitude toward the proposal" (Noguera and Wispelaere,
2006, p. 6). Consequently, a certain degree of complexity should not be ruled out of debates on
social security and taxation systems, otherwise many erroneous conclusions and impressions may
be taken. Secondly, self-interest has also been detected as a critical determinant. As the variable
Wage reveals in the constitutional part of the experiment, the higher the wage, the lower the
probability that a participant would select scheme C as the first choice. This finding expresses the
value of the constitutional economic approach in the research design, which enabled the
identification of self-interest's role. Thirdly, another interesting result was the fact that economic
experience within a redistribution scheme positively affected participants' preference towards this
same scheme, suggesting that familiarity with a scheme may influence inpiduals' propensity to
choose it.
As for the time allocation part, no significant difference in the paid working hours between
schemes A and C could be found, indicating that the tested basic income had no effect on the
hours allocated to paid work. For the social security debate, this finding provides evidence that a
basic income on the level tested would probably not structurally affect the labor market decisions
of people already employed. Additionally, evidence was provided that different kinds of
reference points affect preferences on time allocation. Among them are expectations that
30
participants may have on how many hours they would like to work. These expectations are
probably connected to what is usual in their social sphere and to what the surrounding social
norms describe as acceptable and desirable working hours. Consequently, the effects of
redistributions systems on paid work time may not be as significant as expected due to social
norms and related expectations.
REFERENCES
Althaus, D. (2010), Solidarisches Bürgergeld. Einkommenssteuer. http://www.solidarisches-
buergergeld.de/de/einkommensteuer.html. Accessed 5 June 2018.
Arrieta-Ibarra, I., L. Goff, D. Jiménez-Hernández, J. Lanier and E. G. Weyl (2018), ‘Should We
Treat Data as Labor? Moving beyond "Free"’, AEA Papers and Proceedings 108, 38–42.
Axelrad, H., I. Luski and M. Malul (2016), ‘Behavioral biases in the labor market, differences
between older and younger inpiduals’, Journal of behavioral and experimental economics
60, 23–28.
Barbara, L., G. Grolleau and N. Mzoughi (2017), ‘Do You Prefer Having More or More than
Others in the Workplace? A Quasi-experimental Survey in Algeria’, Managerial and
Decision Economics 38, 595–606.
Barr, N. A. (2004), The economics of the welfare state, 4th ed., Oxford University Press, Oxford
[etc.].
Bergmann, S. (2014), In zehn Stufen zum BGE. Über die Finanzierbarkeit und Realisierbarkeit
eines bedingungslosen Grundeinkommens in Deutschland, 1. Aufl., Books on Demand,
Norderstedt.
Birnbaum, S. (2012), Basic Income Reconsidered, Palgrave Macmillan US, New York.
Blaschke, R. (2012), ‘From the idea of a basic income to the political movement in Europe.
Development and questions’, Papers. Rosa Luxemburg Stiftung, 1–48.
Blaschke, R. (2017), Grundeinkommen und Grundsicherungen. Modelle und Ansätze in
Deutschland. Eine Auswahl. https://www.grundeinkommen.de/wp-
content/uploads/2017/12/17-10-%C3%9Cbersicht-Modelle.pdf. Accessed 4 June 2018.
Borjas, G. J. (2015), Labor economics, Seventh edition, international student edition, McGraw-
Hill Education, New York, NY.
Bundesministerium der Finanzen (2016), Datensammlung zur Steuerpolitik. Ausgabe 2015.
http://www.bundesfinanzministerium.de/Content/DE/Downloads/Broschueren_Bestellservice
/2016-04-04-datensammlung-zur-steuerpolitik-2015.html (Accessed 5 June 2018.
Bundesministerium für Arbeit und Soziales (2017), Arbeitslosengeld II / Sozialgeld.
http://www.bmas.de/DE/Themen/Arbeitsmarkt/Grundsicherung/Leistungen-zur-Sicherung-
des-Lebensunterhalts/2-teaser-artikelseite-arbeitslosengeld-2-sozialgeld.html#collapse98532.
Accessed 12 June 2017.
31
Daheim, C. and O. Wintermann (2016), 2050: Die Zukunft der Arbeit. Ergebnisse einer
internationalen Delphi-Studie des Millennium Project, 1st edn, 1–34.
Fischbacher, U. (2007), ‘z-Tree: Zurich toolbox for ready-made economic experiments’,
Experimental Economics 10, 171–178.
Frohlich, N. and J. A. Oppenheimer (1990), ‘Choosing Justice in Experimental Democracies with
Production’, The American Political Science Review 84, 461.
Frohlich, N. and J. A. Oppenheimer (1993), Choosing justice. An experimental approach to
ethical theory, University of California Press, Berkeley, CA.
Frohlich, N., J. A. Oppenheimer and C. L. Eavey (1987), ‘Laboratory Results on Rawls's
Distributive Justice’, British Journal of Political Science 17, 1.
Gilroy, B. M., A. Heimann and M. Schopf (2013), ‘Basic Income and Labour Supply: The
German Case’, Basic Income Studies 8.
Haigner, S., W. Höchtl, F. G. Schneider, F. Wakolbinger and S. Jenewein (2012), ‘Keep On
Working: Unconditional Basic Income in the Lab’, Basic Income Studies 7.
Hainmueller, J., D. J. Hopkins and T. Yamamoto (2014), ‘Causal Inference in Conjoint Analysis:
Understanding Multidimensional Choices via Stated Preference Experiments’, Political
Analysis 22, 1–30.
Hayo, B. and F. Neumeier (2017), ‘Public Preferences for Government Spending Priorities:
Survey Evidence from Germany’, German Economic Review 43, 487.
Holbrook, A. L. (2012), ‘Attitude Change Experiments in Political Science’, in: J. N. Druckman
(ed.), Cambridge handbook of experimental political science, 2. ed., Cambridge Univ. Press.
Cambridge [u.a.], pp. 141–154.
Jessen, R., D. RostammAfschar and V. Steiner (2015), ‘Getting the Poor to Work: Three Welfare
Increasing Reforms for a Busy Germany’, SOEPpapers on Multidisciplinary Panel Data
Research.
McDermott, R. (2012), ‘Internal and external validity’, in: J. N. Druckman (ed.), Cambridge
handbook of experimental political science, 2. ed., Cambridge Univ. Press. Cambridge [u.a.],
pp. 27–40.
Monnerat, G. L., M. d. C. M. Senna, V. Schottz, R. Magalhães and L. Burlandy (2007), ‘Do
direito incondicional à condicionalidade do direito: as contrapartidas do Programa Bolsa
Família’, Ciência & Saúde Coletiva 12, 1453–1462.
Neumärker, K. J. B. (2018), ‘Bedingungsloses Grundeinkommen aus ordnungspolitischer Sicht’,
WISU, 324–330.
Noguera, J. A. and J. de Wispelaere (2006), ‘A Plea for the Use of Laboratory Experiments in
Basic Income Research’, Basic Income Studies 1, 728.
Page, D. (2018), ‘How the Criteria for Joining the European Union Affect Public Opinion: The
Case of Equal Pay between Women and Men in Bosnia and Herzegovina’, JCMS: Journal of
Common Market Studies 56, 230–246.
Rawls, J. (1971), A Theory of Justice, Harvard University Press, Cambridge, MA.
32
Reiners, J. (2018), Finnisches Experiment: Fakten statt Fake-News.
https://www.grundeinkommen.de/17/05/2018/finnisches-experiment-fakten-statt-fake-
news.html. Accessed 4 June 2018.
Rubinstein, A. (2013), ‘Response time and decision making: An experimental study’, Judgment
and Decision Making 8, 540–551.
Runst, P. (2017), ‘Does Immigration Affect Demand for Redistribution? - An Experimental
Design’, German Economic Review 2, 187.
Sommer, M. (2016), A Feasible Basic Income Scheme for Germany, Springer International
Publishing, Cham.
Standing, G. (2015), ‘Why Basic Income’s Emancipatory Value Exceeds Its Monetary Value’,
Basic Income Studies 10, 495.
Straubhaar, T. (2017), Radikal gerecht. Wie das bedingungslose Grundeinkommen den
Sozialstaat revolutioniert, edition Körber-Stiftung, Hamburg.
United States Department of Labor (2017), Data Retrieval: American Time Use Survey (ATUS).
https://www.bls.gov/webapps/legacy/tusa_1tab1.htm. Accessed 6 June 2018.
van Parijs, P. (1997), Real Freedom for All, Oxford University Press.
van Parijs, P. (2004), ‘A Basic Income for All: A Brief Defence × To Secure Real Freedom,
Grant Everyone a Subsistence Income’, in: L. Groot (ed.), Basic Income, Unemployment and
Compensatory Justice, Springer US. Boston, MA, pp. 11–23.
Widerquist, K. (2013), Independence, Propertylessness, and Basic Income, Palgrave Macmillan
US, New York.
33
APPENDIX A
Presentation of three redistribution schemes
Scheme A
Income tax8: is dependent on gross income (see function below).
40,00%
35,00%
30,00%
Income tax
25,00%
20,00%
15,00%
10,00%
5,00%
0,00%
600,00 €
0,00 €
1.200,00 €
1.800,00 €
2.400,00 €
3.000,00 €
3.600,00 €
4.200,00 €
4.800,00 €
5.400,00 €
6.000,00 €
6.600,00 €
7.200,00 €
7.800,00 €
8.400,00 €
9.000,00 €
9.600,00 €
Gross income (month)
Social contribution: 1,5% on the gross income that exceeds 450€.
Redistribution rule: every person with a net income up to 1000€/month receives a lump sum
payment on top of their net income. However, the bigger the net income, the lower the lump sum
payment. Beginning with a net income of 100€/month the lump sum payment decreases
continuously from 722€/month until it reaches 0€/month.
800,00 €
Lump sum payment (month)
700,00 €
600,00 €
500,00 €
400,00 €
300,00 €
200,00 €
100,00 €
0,00 €
Net income (month)
8
Always when income tax in mentioned this text, we mean average income taxes.
34
Lump sum payment (month) Income tax
0,00 €
0,10 €
0,20 €
0,30 €
0,40 €
0,50 €
0,60 €
0,70 €
0,80 €
0,90 €
1,00 €
0,00 €
0,10 €
0,20 €
0,30 €
0,50 €
0,60 €
0,70 €
0,80 €
0,90 €
1,00 €
0,40 €
0,00 € 0,00 €
600,00 € 400,00 €
1.200,00 € 800,00 €
1.800,00 € 1.200,00 €
2.400,00 € 1.600,00 €
3.000,00 € 2.000,00 €
3.600,00 € 2.400,00 €
Income tax: no income taxation.
4.200,00 €
2.800,00 €
4.800,00 €
3.200,00 €
5.400,00 €
6.000,00 € 3.600,00 €
Social contribution: no social contribution.
4.000,00 €
Gross income (month)
Net income (month)
6.600,00 €
7.200,00 € 4.400,00 €
7.800,00 € 4.800,00 €
Scheme B
8.400,00 € 5.200,00 €
9.000,00 € 5.600,00 €
9.600,00 €
6.000,00 €
Redistribution rule: no redistribution through income taxation.
35
Scheme C
Income tax: every person (independent of the net and gross income) pays a 45% flat tax.
50,00%
45,00%
40,00%
35,00%
Income tax
30,00%
25,00%
20,00%
15,00%
10,00%
5,00%
0,00%
7.200,00 €
0,00 €
1.200,00 €
1.800,00 €
2.400,00 €
3.000,00 €
3.600,00 €
4.200,00 €
4.800,00 €
5.400,00 €
6.000,00 €
6.600,00 €
7.800,00 €
8.400,00 €
9.000,00 €
9.600,00 €
600,00 €
Gross income (month)
Social contribution: is included in the income tax.
Redistribution rule: every person (independent of the net and gross income) receives 722
€/month lump sum payment on top of their net income.
800,00 €
Lump sum payment (month)
700,00 €
600,00 €
500,00 €
400,00 €
300,00 €
200,00 €
100,00 €
0,00 €
3.500,00 €
7.000,00 €
0,00 €
1.000,00 €
1.500,00 €
2.000,00 €
2.500,00 €
3.000,00 €
4.000,00 €
4.500,00 €
5.000,00 €
5.500,00 €
6.000,00 €
6.500,00 €
7.500,00 €
8.000,00 €
8.500,00 €
9.000,00 €
9.500,00 €
10.000,00 €
500,00 €
Net income (month)
36
Net income after redistribution (month)
€-
€1.000,00
€2.000,00
€3.000,00
€4.000,00
€5.000,00
€6.000,00
€7.000,00
0,00 €
200,00 €
400,00 €
600,00 €
800,00 €
1.000,00 €
1.200,00 €
1.400,00 €
1.600,00 €
1.800,00 €
2.000,00 €
2.200,00 €
2.400,00 €
2.600,00 €
2.800,00 €
3.000,00 €
3.200,00 €
3.400,00 €
APPENDIX B
3.600,00 €
Gross income (month)
3.800,00 €
4.000,00 €
4.200,00 €
4.400,00 €
4.600,00 €
4.800,00 €
5.000,00 €
5.200,00 €
Net income after redistribution: schemes A, B and C (graph)
5.400,00 €
5.600,00 €
5.800,00 €
6.000,00 €
Scheme C
Scheme B
Scheme A
37
Net income after redistribution: schemes A, B and C (table)
Gross income Net income after redistribution (month)
Income class
(month)
Scheme A Scheme B Scheme C
1 0,00 € 722,00 € 0,00 € 722,00 €
2 200,00 € 842,00 € 200,00 € 832,00 €
3 400,00 € 882,00 € 400,00 € 942,00 €
4 700,00 € 939,90 € 700,00 € 1.107,00 €
5 1.000,00 € 989,33 € 1.000,00 € 1.272,00 €
6 1.500,00 € 1.308,09 € 1.500,00 € 1.547,00 €
7 2.000,00 € 1.664,22 € 2.000,00 € 1.822,00 €
8 3.000,00 € 2.335,90 € 3.000,00 € 2.372,00 €
9 4.000,00 € 2.953,48 € 4.000,00 € 2.922,00 €
10 6.000,00 € 4.089,51 € 6.000,00 € 4.022,00 €
11 8.000,00 € 5.246,51 € 8.000,00 € 5.122,00 €
12 10.000,00 € 6.406,51 € 10.000,00 € 6.222,00 €
Average income 2.364,95 € 3.066,67 € 2.408,67 €
Income distance between 1 and 12 5.684,51 € 10.000,00 € 5.500,00 €
38