Abstract
Using an artefactual field experiment, this paper tests the long-term implications of living in a specific economic system on individual dishonesty. By comparing cheating behaviour across individuals from the former socialist East of Germany with those of the capitalist West of Germany, we examine behavioural differences within a single country. We find long-term implications of living in a specific economic system for individual dishonesty when social interactions are possible: participants with an East German background cheated significantly more on an abstract die-rolling task than those with a West German background, but only when exposed to the enduring system of former West Germany. Moreover, our results indicate that the longer individuals had experienced socialist East Germany, the more likely they were to cheat on the behavioural task.
1. Introduction
Individual dishonesty is very costly for society. For example, the average annual tax gap in the US for the years 2008 to 2010 is estimated to be $458 billion (Internal Revenue Service, 2016). The costs of insurance fraud (excluding health insurance) is estimated to be more than $40 billion per year, which translates to costs in the form of increased premiums between $400 and $700 per year for an average US family (U.S. Department of Justice, 2018). Hence, understanding determinants of dishonest behaviour is a major concern for society.
In this paper, we investigate a context effect of dishonesty by asking the following question: What are the long-term implications of living in a specific economic system for individual dishonesty? We explore whether existing economic systems, socialism and capitalism, have a different effect on people’s dishonesty. To understand how exposure to economic systems influences individual behaviour, we use a historical event, the division of Germany into two different formerly existing economic regimes within a single country, socialist East and capitalist West Germany. Specifically, we compare cheating behaviour between East Germans, who were exposed to socialism for over 40 years, and West Germans, who were at the same time living in a market economy. Several studies have documented differences in individual behaviour between citizens of former East and West Germany, for example in national solidarity (Ockenfels and Weimann, 1999; Brosig-Koch et al., 2011) or preferences for redistribution and levels of social trust (Alesina and Fuchs-Schündeln, 2017; Heineck and Süssmuth, 2013). Heineck and Süssmuth (2013) even find that specific cultural traits are passed down through generations. We add to this strand of literature by investigating the persistence of behavioural differences in dishonesty. Moreover, we compare the impact of two economic systems within one single country to better understand the effect of extant socialistic versus capitalistic regimes on individual dishonesty.
It is important to note that social interactions—between friends as well as strangers—influence dishonest behaviour. People cheat more when they observe others behaving dishonestly (Gino et al., 2009). In this vein, Dieckmann et al. (2016) and Rauhut (2013) find support for social conformity to the observed behaviour in the form of contagiousness and spread of norm violations. Mann et al. (2014) also show a transmission through social networks and document that people’s tendency to lie is associated with the lying behaviour of their friends and family members. However, social interaction can also lead to anti-conformity in behaviour. An experimental study by Fortin et al. (2007) finds evidence of a social anti-conformity effect, which suggests that individuals deviate from the tax compliance behaviour of their reference group, using an income reporting task. In the specific case of former East and West Germany, fairness considerations may spur social interaction effects, since fervent debate erupted after the reunification of Germany about whether economic and social injustice befell parts of the country because of the reunion (Schmitt and Maes, 1998). Individuals who believe they were treated unfairly in an interaction with another person are more likely to cheat in a subsequent unrelated game (Houser et al., 2012). Moreover, when individuals are aware that they were poorly compensated relative to another group, they cheat more to increase their earnings (John et al., 2014). In this paper, we explore whether social interactions might explain individual differences in dishonesty between citizens from socialist versus capitalist systems.
We use an artefactual field experiment (Harrison and List, 2004) to investigate the impact of different economic systems within a single country on individual cheating behaviour. We compare cheating among people exposed to the two existing economic systems of former Germany. In particular, we compare Germans living in Berlin, where citizens with East and West German backgrounds co-exist, with Germans living outside of Berlin, where citizens typically live with peers sharing the same historical background. To measure cheating behaviour, we use a die task adapted from previous research where participants were paid based on the number of dots on reported die rolls (Fischbacher and Föllmi-Heusi, 2013; Jiang, 2013; Mann et al., 2016; see Garbarino et al., 2016 and Abeler et al., 2018 for two meta-analyses). It has been shown that even abstract cheating tasks predict behaviour in the field. For example, a widely used reporting task in the lab significantly predicts classroom misbehavior in middle and high school students (Cohn and Maréchal, 2018) and it has been shown that abstract as well as contextualized cheating tasks in the laboratory correlate with rule violations in real life (Dai et al., 2017).
Our results show that social interaction is an important mechanism underlying individual cheating: participants with an East German background cheated significantly more on an abstract die-rolling task than those with a West German background, but only when exposed to the enduring capitalist system of West Germany. Moreover, our results indicate that the longer individuals living in Berlin had experienced socialist East Germany, the more likely they were to cheat on the behavioural task. In contrast, we did not observe differences in cheating behaviour between East and West German individuals living in the respective cities of Leipzig (East Germany) and Dortmund (West Germany). Unlike in Berlin, individuals from Leipzig and Dortmund have less opportunity for comparison against the alternative economic system, due to being situated at some distance from the former inner German border.
The remainder of the paper is structured as follows. Section 2 outlines related literature on differences between former East and West Germany. Thereafter, section 3 presents our materials and methods. Section 4 lays out the empirical results of our study. Section 5 concludes.
2. Differences between former East and West Germany
From 1961 to 1989, the Berlin Wall divided one nation into two distinct economic and political regimes: socialism (East Germany) and capitalism (West Germany). Socialist systems have been characterized by extensive scarcity, which in the case of East Germany, ultimately led to the collapse of the German Democratic Republic (GDR). In many instances, socialism pressured or forced people to work around official laws. For example, in East Germany stealing a load of building materials in order to trade it for a television set might have been the only way for a person to acquire a valuable good and connect to the outside world (Hornuf and Rieger, 2017). Moreover, a high degree of infiltration by intelligence apparatuses is also considered as a key characteristic of socialist systems. In East Germany, the secret service (Staatssicherheit) kept records on more than one third of its citizens (Koehler, 1999). Moreover, freedom of speech was not a virtue upheld in socialist regimes and citizens had to often misrepresent their thoughts to avoid repression.
Earlier studies have shown differing degrees of national solidarity between East and West Germans. In a laboratory experiment with economics students, East Germans showed significantly less solidarity five years after the German reunification (Ockenfels and Weimann, 1999). When asked how much money they would be willing to hand over to anonymous future losers if they won 10 Deutsche Mark in a solidarity game, East Germans were willing to give up roughly half as much as West Germans. Interestingly, East Germans also expected to receive much less from potential winners. These results were recently confirmed by another study showing that there was no convergence in solidarity 20 years after the German reunification, which the authors attribute to slow changes in social behaviour due to the necessity of coordination on social norms in the society as well as complementarities involved in individual social behaviour (Brosig-Koch et al., 2011).
Based on data from the German Socioeconomic Panel (SOEP), Alesina and Fuchs-Schündeln (2007) provide evidence that East Germans have stronger preferences for public policies that involve redistribution. They find that economic and political regimes greatly shape individual preferences for state interventions and that these preferences change slowly. According to the authors’ analysis, one fourth of the effect that East Germans’ have stronger preferences for state intervention is because East Germans became poorer during the socialist epoch, while the remainder can be attributed to the impact of socialism on individual preferences itself. Yet, one limitation of the study is that people might distort their true preferences when responding to a survey like SOEP. For example, people might overstate their willingness to contribute to redistributive policies because they do not actually have to pay for them. Using a discrete choice experiment, another study shows that the stated preferences of East Germans towards redistribution indeed differ from their actual preferences (Pfarr et al., 2013). While East Germans indicate that they prefer higher degrees of redistribution, they are not actually willing to pay for such policies.
Relatedly, Heineck and Süssmuth (2013) investigate the effect of the economic regime on individuals’ trust, risk preferences, and cooperativeness using SOEP data. Relative to West Germans, East Germans showed persistently lower levels of social trust and were less inclined to see others as fair. This study also suggested that East Germans are more risk-loving. Most importantly, the authors find that these cultural traits appear to be passed down through generations. While this research provides valuable insights on differences in solidarity and individual social preferences, little is known about how the economic systems of former East and West Germany influenced individual dishonesty. Torgler (2003) indicates that at one point in time, East Germans were more likely to say that cheating on their taxes cannot be justified than West Germans, but this difference disappeared seven years after the German reunification. However, this finding is based on self-stated preferences in a survey and therefore might not reflect individuals’ actual behaviour when they are in a position where dishonesty financially pays off.
3. Materials and methods
3.1. Experimental setup
We conducted two experimental sessions in three German cities. The sessions differ in the cities of data collection only. Study materials and tasks were identical for both sessions. For session 1, we collected data in Berlin over the course of five days (from December 2nd, 2013 through December 6th, 2013). For session 2, we collected data over two days in Leipzig, East Germany (from March 13th to March 14th, 2015) and two days in Dortmund, West Germany (from September 9th to September 10th, 2015). The varying timeframes of data collection results from the obtainment of permission from the citizen centres and city administration, coordination of experimenters, and arrangement of the experimental setup and procedures at the citizen centres.
We vary the cities of data collection to explore the effect of social interaction between East and West German backgrounds on dishonest behaviour. We argue that individuals living in Berlin are closely exposed to the former other system (capitalism versus socialism), because Berlin itself was divided into a capitalist West and a socialist East, whereas those living in other average cities in Germany (Dortmund or Leipzig) are not so apt to exposure to the other economic system. Thus, we predict to find a significant difference in cheating behaviour when comparing former East versus West Germans in Berlin but not in Dortmund or Leipzig, i.e., we predicted an interaction between economic system exposed to (former East vs. former West) and location (Berlin vs. Dortmund/Leipzig).
In session 1, we collected data in six different administrative regions in Berlin, including regions directly at the former border of the Berlin Wall (Neukölln, Mitte and Pankow) and regions in former East and West Berlin hinterlands (Zehlendorf, Steglitz and Weissensee). The sample size of session 1 was 259 participants. In session 2, we collected data in four different administrative regions in Leipzig (Gohlis-Centre, Leipzig Mitte, Paunsdorf-Centre and Ratzelbogen), a city in former East Germany as well as two different administrative regions in Dortmund (Hörde and Innenstadt), a city in former West Germany (Fig. 1). Both cities were chosen because they are similar in terms of size and average income of the citizens.1 The sample size of session 2 was 275 participants (139 in Leipzig and 136 in Dortmund).
Fig. 1. Data collection in two economic systems. Red areas represent former East Germany and blue areas represent former West Germany. Dortmund, Leipzig, and Berlin (Zehlendorf, Steglitz, Neukölln, Mitte, Weissensee and Pankow) mark the regions where we conducted the experiment. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3.2. Participants
According to Article 1 of the German ID Card Law (Personalausweisgesetz), every German is required to possess and show upon request a passport or ID card from the age of sixteen. Passports and ID cards are valid for up to 10 years and thereafter need to be renewed. A fine of up to 5000 EUR can be imposed in cases of non-compliance. Hence, even Germans who do not need a passport or ID card for travel purposes have a strong incentive to possess at least one of the two documents. Furthermore, when a passport or ID card is ordered, the applicants must collect it themselves. A proxy person can collect the documents only in narrowly defined exceptional cases. This set of requirements gives us confidence that the people collecting their passport or ID card at the citizen centre represent a random sample of the population in the respective cities.2 The average age of our participants is 37.8 years with a standard deviation of 14.1 years; 48 percent are female.3 For further descriptive statistics, see Appendix A Table A.1.
3.3. Experimental protocol
The experimental design and procedures were identical for both sessions; additionally, the same two experimenters, one male and one female with East and West German family background respectively, collected the data in the three cities in Germany. Experimenters operated independently and approached participants as they came in the citizen centre waiting rooms and sat down (see Appendix B Figs. 1–2 for photographs of two locations of the experiment). An experimenter then briefly described the study and asked the participant whether they would be interested in participating. If a potential participant verbally consented to take part in the study, the experimenter continued with the protocol; otherwise the person was thanked and the experimenter moved on to the next person. The study in both sessions 1 and 2 included three consecutive parts in the following order: 1) a behavioural cheating task, 2) a questionnaire including measures of participants’ family background (East or West German) as well as demographics and control questions, and 3) an opportunity to anonymously donate the earnings.
3.4. The die task
To measure cheating behaviour, we used a behavioural task adapted from Jiang (2013) and Mann et al. (2016) where participants were paid based on the number of dots on reported die rolls (see Appendix C for die task instructions). Participants threw a six-sided die 40 times, and at the end of the task, their payoff was determined by the random selection of one of their rolls (the experimenter randomly drew a number from 1 to 40 out of an envelope). Before each roll, participants were instructed to either decide on the top or bottom side of the die in their mind, and to memorize their decision before rolling the die. They then threw a physical die, observed the outcome, and recorded their outcome on a piece of paper. Participants completed the task privately and earned 1 EUR for each dot on that particular roll.
In this task, participants could cheat in two possible ways every time they threw the die: first, by claiming that they chose the side of the die that leads to the higher payoff when in fact they chose the lower-paying side (done so by reporting the side of the die with more dots, which would lead to a higher payoff); and second, by simply making up a roll. If participants were completely honest, we would expect an average payoff of 3.50 EUR, equivalent to selecting the high-paying side 50 percent of the time. Thus, we can observe the first dimension of cheating by examining the percentage of high-paying rolls, which should be 50 percent in a completely honest population. In addition, we can identify the second dimension of cheating by examining the distribution of the combined numbers of rolls with 3 or 4 dots, rolls with 2 or 5 dots, and rolls with 1 or 6 dots, which should be equally distributed in a completely honest population. If, for example, the combined rolls of 6 and 1 are reported more often than the combined rolls of 5 and 2 or 4 and 3, this can be attributed to participants falsifying rolls (presumably making up rolls with 6 dots).
3.5. Questionnaire
In both sessions, we measure the family background of each participant using a survey (see Appendix D Supplementary data, Appendix E). To assess participants’ background, we collected five different measures in our questionnaire: participants’ family background (primary variable of interest) as well as place of birth, place of living in the 1980s, place of living in the 1990s, and participants’ self-appraised categorization of their social background as secondary background variables. Experimenters were blind with regard to the social background of the participants. Additionally, two participants refused to complete the questionnaire after they took part in the die-rolling task and were therefore excluded from the sample.
4. Results
In the analyses reported below, our independent variables are 1) the economic system participants were exposed to, which is operationalized by their family background (East German versus West German), and 2) participants’ location (session 1 versus session 2). In our survey, we included five measures to capture participants’ backgrounds, and report results using these five measures in Appendix A Table A.2. However, for our main results reported below, we consider family background as our primary measure of East or West background.
Our statistical approach combines differences in means tests and multivariate Probit regression models to examine the effect of the two economic systems on cheating behaviour. As a robustness check, we investigate participants’ length of exposure to a particular economic system. Besides using common control variables (age, gender, education, political orientation, living standard and marital status), we also included controls capturing trust and fairness considerations because prior work has shown that socialism in East Germany led to lower levels of trust (Heineck and Süssmuth, 2013). As participants’ trust might impact individual honesty behaviour (Dieckmann et al., 2016), we adapted a question from the World Value Survey (Inglehart and Welzel, 2012) and asked them how much they would trust East and West Germans respectively. We also included a dummy variable inquiring whether participants believed that East Germans have been betrayed by West Germans after the reunification of Germany, since Houser et al. (2012) showed that being treated unfairly by others influences individual cheating behaviour.
4.1. Cheating behaviour
Our results reveal that on average, both groups cheated on the first dimension by selecting the high-paying side of a die roll significantly more often than would be expected by chance. Across both sessions, participants with a West German family background reported 55 percent high rolls on average (mean roll = 3.68), and participants with an East German family background reported 58 percent high rolls on average (mean roll = 3.77). Both statistics significantly differ from the pure chance outcome of a fair die of 50 percent high rolls on average (mean roll = 3.50) on a simple t-test, hence reflecting some level of dishonesty in each group (p < 0.001). Beyond that, t-test as well as Wilcoxon-Mann-Whitney-test statistics reveal that Germans with an East German family background cheated more than those participants with a West German family background. The difference is, however, marginally significant (p = 0.069, Prob > |z| = 0.099). See Appendix A Table A.2. for detailed statistics.4
Comparing results from session 1 and 2 reveals that the behavioural difference between the two groups only holds for participants living in Berlin, as they chose the high-paying sides on 60 percent of rolls (East German family background) versus on 55 percent of rolls (West German family background) respectively (Fig. 2). The difference in cheating for the Berlin sample (session 1) is statistically significant at conventional levels (p = 0.013, Prob > |z| = 0.011). In contrast, participants from Leipzig and Dortmund (session 2) show no significant differences in levels of cheating between East and West German family backgrounds.
Fig. 2. Percentage of high rolls by East and West German family background. Percentage of high rolls by family background and region where the experiment was conducted. Error bars represent standard error of the mean (SEM). * indicate significance at the 5% level.
Next, we assessed whether participants cheated on the second dimension by simply making up rolls. The observed distribution of rolls in our sample suggests that participants invented rolls: rolls with 3 or 4 dots combined appeared less frequently than rolls with 1 or 6 dots combined. However, this effect is only marginally significant (p = 0.053) (see Appendix F Figure F.1.D). Finally, participants with East and West German backgrounds do not differ regarding this second dimension of cheating across the two sessions.
We ran Probit regressions with high or low-paying rolls as the dependent variable and used the random effects panel estimator to account for clustering of specific effects within individuals. In order to identify the effect of the participant’s background and location on cheating behaviour, we included a dummy variable for family background (dummy = 1 if the participant has an East German family background, 0 otherwise), a dummy variable for the location, i.e., session 1 versus session 2 (dummy = 1 if location is Berlin, 0 otherwise), and an interaction term of the two dummy variables (interaction term = 1 if the participant has an East German family background and the experiment was conducted in Berlin, 0 otherwise). Thus, the interaction term identifies the additional effects of having an East German background and being exposed to the citizens of former West Berlin – while controlling for the particular characteristics of participants who have an East German family background or those living in Berlin. Examining our model’s variables, we find that most of the explanatory variables are only weakly correlated. Consequently, multicollinearity does not appear to be a matter of concern for this study (see Appendix A Table A.3). Table 1 reports the results of our Probit regressions. Allowing for a more intuitive interpretation, we report average marginal effects of the explanatory variables. The baseline estimates reveal that, when controlling for other potential predictors of cheating, family background or living in Berlin per se does not change the probability of participants reporting the high side of the die. However, the interaction term of having an East German background and living in Berlin is significant (p = 0.020), indicating an increase in participants’ probability of cheating by 8.1 percent. This effect is stronger for participants with an East German family background that were also born in and currently live in East Germany (column 2). The probability of cheating increases by 9.5 percent (p = 0.018).
Table 1. Regression results for cheating behaviour.
| Dependent variable | High roll | High roll | High roll | High roll | High roll | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| (full sample) | (East background and birth and living) | (born before 1989) | (born before 1980) | (born before 1970) | ||||||
| Marginal effect | p-value | Marginal effect | p-value | Marginal effect | p-value | Marginal effect | p-value | Marginal effect | p-value | |
| East German | −0.017 | 0.508 | −0.033 | 0.235 | −0.055† | 0.068 | −0.084* | 0.043 | −0.083 | 0.120 |
| Location Berlin | −0.024 | 0.346 | −0.050† | 0.093 | −0.044 | 0.136 | −0.064 | 0.153 | −0.082 | 0.161 |
| East German * Location Berlin | 0.081* | 0.020 | 0.095* | 0.018 | 0.126** | 0.002 | 0.171** | 0.003 | 0.207** | 0.007 |
| Age | 0.002** | 0.001 | 0.002** | 0.002 | 0.002** | 0.002 | 0.003** | 0.009 | 0.002 | 0.356 |
| Female | 0.013 | 0.449 | 0.021 | 0.275 | 0.004 | 0.849 | −0.005 | 0.862 | 0.006 | 0.885 |
| Trust East | 0.002 | 0.721 | −0.003 | 0.649 | 0.001 | 0.849 | −0.003 | 0.665 | −0.007 | 0.463 |
| Trust West | −0.005 | 0.261 | −0.002 | 0.703 | −0.002 | 0.694 | 0.001 | 0.933 | 0.007 | 0.465 |
| West betrayed | −0.044* | 0.027 | −0.044† | 0.051 | −0.055* | 0.013 | −0.061* | 0.044 | −0.085* | 0.023 |
| Education | −0.016* | 0.011 | −0.016* | 0.031 | −0.019** | 0.006 | −0.027** | 0.004 | −0.011 | 0.394 |
| Political orientation | 0.002 | 0.627 | 0.003 | 0.502 | 0.005 | 0.335 | 0.000 | 0.992 | −0.007 | 0.448 |
| Living standard | −0.010 | 0.323 | −0.008 | 0.469 | −0.009 | 0.454 | −0.018 | 0.287 | −0.021 | 0.287 |
| Marital status | −0.036† | 0.056 | −0.034 | 0.102 | −0.041† | 0.051 | −0.069* | 0.025 | −0.069† | 0.097 |
| N (participants) | 341 | 266 | 276 | 167 | 100 | |||||
| N (rolls) | 13,640 | 10,640 | 11,040 | 6680 | 4000 | |||||
| χ ̅² | 30.522 | 21.860 | 32.429 | 27.734 | 19.301 | |||||
This table shows results of Probit regressions with participant random effects. The dependent variables captures whether participants report the high paying side of the die or not. For variable definitions, see Appendix E Table E.1. The table shows results for the baseline regression (1), a model analysing participants with an East German family background that were also born in and currently live in East Germany (2), as well as regressions for age cohorts (3)–(4). Regression models report average marginal effects and p-values. **,* and † indicate significance at the 1%, 5% and 10% level.
With regard to our control variables, we find that age has a significant impact on cheating, contributing to the mixed findings on the relationship between age and cheating in existing literature (Crown and Spiller, 1998). Being an additional year older increases the probability of participants reporting the high side by 0.2 percent (p < 0.003). In line with Hartshorne et al. (1929), we find higher levels of education reduce the probability of cheating. In addition, the results show that the belief that East Germans have been betrayed by West Germans after the reunification of Germany significantly reduces the probability of cheating (p = 0.027). However, West Germans, who believe that East Germans were betrayed after the reunification, drive this effect. Participants’ marital status has a marginally significant effect, which indicates that married participants tend to cheat less. None of the other explanatory variables have a statistically significant effect on cheating. Notably, our two control variables capturing participants’ trust in East and West Germans have no significant impact on cheating. The finding that there are no gender differences in cheating contrasts somewhat with Muehlheusser et al. (2015), but is in line with more recent research by Ezquerra et al. (2018).
In addition, we consider that participants might change their cheating behaviour over time as they roll the die over and over. Conceivably, the higher level of cheating observed for East German participants could be a consequence of adapting to the task and cheating more over time, even if their initial dishonesty was identical to participants with a West German background. Thus, we start by visually investigating the development of roll outcomes over time and across participants for each of the 40 rolls (see Appendix F Figure F.2.A – C). Testing for autocorrelation does not yield any significant results. When we split the sample in half, participants’ cheating levels in the first half (roll 1 to 20) and second half (roll 21 to 40) of the task are also not significantly different (p = 0.284).
4.2. Cheating behaviour by age cohorts
Next, we examine participants’ length of exposure to their respective economic system and assess whether this strengthens differences in cheating behaviour. If economic systems have an effect on individuals’ cheating, we would expect people who have had longer exposure to socialism in East Germany to cheat even more compared to participants with a West German background. In order to capture this degree of exposure, we analysed subsamples of Germans who were born before the dissolution of the GDR, who were at least 10 years old at the time of the reunification (born before 1980), and who were at least 20 years old when the Berlin Wall came down (born before 1970) (see Table 1, column 3–5). We investigated the exposure to socialism by examining these distinct age cohorts separately. The results show that differences in cheating levels for older age cohorts are greater between participants with an East and West German background living close to the former border (session 1). Participants born before the dissolution of the GDR were 12.6 percent more likely to cheat (p < 0.002), whereas participants who lived at least 10 years in a socialist society were 17.1 percent more likely (p < 0.003), and participants who lived for 20 years or more under socialism were 20.7 percent more likely (p < 0.007) to cheat than those participants of West Germany from the same age cohort from session 1 (i.e., location Berlin). Thus, the length of exposure to the economic system strengthens the difference in cheating behaviour. However, this is again only true for participants in Berlin who have the opportunity for directly experiencing both economic systems.5
5. Discussion
5.1. Berlin characteristics
The documented difference in cheating for the Berlin sample could be driven by specific characteristics of Berlin citizens. However, three arguments underline that the observed effect in our Berlin sample should not be related to specific unobserved characteristics of this group. First, Table 2 shows that almost half of the Berlin participants also lived, or were born, in other cities across Germany, ensuring that the effect is not only driven by specific factors related to being a native of Berlin. Second, our age cohort effect supports the notion that the observed behavioural difference is determined by variables related to living in East Germany over a long time and not being a native of Berlin per se. Third, we show that people with a West Berlin background have very similar cheating levels to those living in Dortmund or Leipzig, indicating that participants living in West Berlin are after all not that different from other participant cohorts.
Table 2. Participants’ living history in the Berlin sample.
| Berlin sample | |||
|---|---|---|---|
| Freq. | Percent | Cum. | |
| Place of birth | |||
| East Berlin | 39 | 15.29 | 15.29 |
| West Berlin | 69 | 27.06 | 42.35 |
| East Germany | 58 | 22.75 | 65.10 |
| West Germany | 65 | 25.49 | 90.59 |
| Others | 24 | 9.41 | 100.00 |
| Total | 255 | 100.00 | |
| Living in the 1980s | |||
| East Berlin | 48 | 22.97 | 22.97 |
| West Berlin | 60 | 28.71 | 51.67 |
| East Germany | 40 | 19.14 | 70.81 |
| West Germany | 48 | 22.97 | 93.78 |
| Others | 13 | 6.22 | 100.00 |
| Total | 209 | 100.00 | |
| Living in the 1990s | |||
| East Berlin | 64 | 25.40 | 25.40 |
| West Berlin | 82 | 32.54 | 57.94 |
| East Germany | 34 | 13.49 | 71.43 |
| West Germany | 58 | 23.02 | 94.44 |
| Others | 14 | 5.56 | 100.00 |
| Total | 252 | 100.00 | |
5.2. Selection effects
Given that almost half of the Berlin participants come from other regions in Germany, another alternative explanation of the observed effect might be that people who moved to Berlin are special. After the reunification, the former West German capital of Bonn moved to Berlin, resulting in many public officials relocating to Berlin in 1999. As of 2015, about 11,301 out of 17,997 government employees work in Berlin.6 However, assuming all currently employed officials moved from Bonn to Berlin, the probability of including such an official in our sample is only 0.0032 percent. Calculations are based on data of the German Federal Statistical Office. As of December 31st, 2015, the total population of Berlin amounts to 3,520,031. Our effect for the Berlin sample is clearly not driven by any one participant.7 Nevertheless, people moving to Berlin might generally be special and the observed effect could mainly be driven by those East German citizens who moved to Berlin. However, Fig. 3 reveals that the East Germans who moved to Berlin still cheat more, but not (yet) as much as natives of East Berlin, who faced a relatively longer period of exposure to both systems. Thus, we argue that our effect is not driven by the selection of a particular group of East Germans moving to East Berlin.
Fig. 3. Percentage of high rolls by place of birth for Berlin sample. Percentage of high rolls by place of birth for participants from Berlin sample. Error bars represent standard error of the mean (SEM). * indicate significance at the 5% level.
6. Conclusion
From 1961 to 1989, the Berlin Wall divided one nation into two distinct regimes. We exploited this fact to investigate whether the economic context impacts individual dishonesty by running an artefactual field experiment. Using an abstract die-rolling task, we find evidence that when social interaction is likely, East Germans exposed to socialism cheat more than West Germans exposed to capitalism. In contrast, when the other economic system is less salient, East Germans and West Germans do not differ in terms of cheating behaviour. Furthermore, we find that the observed difference in cheating behaviour is stronger for older age cohorts, indicating that the longer participants were exposed to socialism in East Germany, the greater is their dishonesty. Thus, we provide evidence for long-term implications of living in a specific economic system for individual dishonesty when social interactions are innate.
Our findings add a new perspective to dishonesty. Previous work provided 658 Germans with an opportunity to earn 15 EUR by misreporting the side of a coin toss on the telephone (Abeler et al., 2014). While the results of this study did not reveal differences in cheating behaviour, we show that differences in dishonesty exist between different groups within the country. This is in line with the finding that morality differs more between social groups within one country than between different countries (Mann et al., 2016; Haidt et al., 1993; for differences in unethical behaviour between countries see Gächter and Schulz, 2016 as well as Shalvi, 2016). More generally, considering the findings by Falk and Szech (2013), which indicate that markets decay morals, our results potentially provide a first hint that other economic systems, such as socialism, may have an even more detrimental effect on individuals’ morality. Future studies might investigate the specific factors that might affect morality in capitalist versus socialist as well as democratic versus authoritarian societies.
Behavioural variation across different populations or groups can be driven by many different factors, such as markets, institutions, or norms. In contrast to much existing cross-cultural research, we investigate the impact of economic systems on social behaviour within one single country, which makes our results more robust with regard to unobservable heterogeneity across the sample. Moreover, we use a non-student sample, taking into account that behavioural differences among student populations are not representative and are rather small compared to the range of existing economic environments. For future work, it would be interesting to examine the observed effect of social interaction, isolating what drives the variation in behavioural differences across different economic systems and whether it can be found in other types of prosocial and antisocial behaviour. Laboratory experiments validating this finding from the field seem to be promising.
Appendix A. Sample statistics
Table A.1. Descriptive statistics for the full sample and subsamples regarding family background. See variables description in Appendix E.
| Full sample | Family background | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| East German | West German | German | Others | N/A | ||||||
| mean | median | SD | min | max | mean | mean | mean | mean | mean | |
| Average roll | 3.74 | 3.68 | 0.47 | 2.28 | 5.85 | 3.77 | 3.68 | 3.73 | 3.82 | 3.64 |
| High rolls (%) | 0.57 | 0.55 | 0.14 | 0.28 | 1.00 | 0.58 | 0.55 | 0.57 | 0.59 | 0.55 |
| Earnings | 3.67 | 4.00 | 1.71 | 1 | 6 | 3.64 | 3.61 | 3.61 | 4.04 | 3.00 |
| Age | 37.77 | 34.00 | 14.11 | 12 | 87 | 41.14 | 36.63 | 35.81 | 33.17 | 58.00 |
| Female | 0.48 | 0.00 | 0.50 | 0 | 1 | 0.49 | 0.42 | 0.47 | 0.61 | 0.33 |
| Education | 3.39 | 3.00 | 1.42 | 0 | 6 | 3.41 | 3.58 | 2.94 | 3.24 | 2.00 |
| Living standard | 2.74 | 2.00 | 1.03 | 1 | 7 | 2.77 | 2.66 | 2.97 | 2.59 | 3.00 |
| Political orientation | 3.89 | 4.00 | 2.03 | 0 | 10 | 3.87 | 3.79 | 4.40 | 3.73 | 3.00 |
| Trust East | 2.61 | 2.00 | 2.42 | 0 | 10 | 2.70 | 2.24 | 2.43 | 3.67 | 2.00 |
| Trust West | 3.10 | 2.50 | 2.55 | 0 | 10 | 3.80 | 2.34 | 2.78 | 3.69 | 2.00 |
| West betrayed | 0.30 | 0.00 | 0.46 | 0 | 1 | 0.26 | 0.36 | 0.25 | 0.27 | 0.00 |
| N | 534 | 189 | 195 | 75 | 70 | 5 | ||||
Table A.2. Participants’ reported share of high rolls by background. Reported share of high rolls for different measures of East and West German background. **,* and † indicate significance at the 1%, 5% and 10% level.
| N | High rolls (%) | SD | Differences between East and West background | ||
|---|---|---|---|---|---|
| t-test | Wilcoxon-Mann-Whitney-test | ||||
| Panel A: Full sample | |||||
| Family background | |||||
| East German | 189 | 0.58 | 0.15 | 0.03† | |
| West German | 195 | 0.55 | 0.13 | (p = 0.069) | Prob > |z| = 0.099 |
| German | 75 | 0.57 | 0.14 | ||
| Others | 70 | 0.59 | 0.16 | ||
| N/A | 5 | 0.55 | 0.12 | ||
| Consider yourself | |||||
| East German | 73 | 0.59 | 0.14 | 0.01 | |
| West German | 70 | 0.58 | 0.15 | (p = 0.656) | Prob > |z| = 0.416 |
| German | 298 | 0.56 | 0.15 | ||
| Others | 85 | 0.58 | 0.15 | ||
| N/A | 8 | 0.50 | 0.11 | ||
| Place of birth | |||||
| East Germany | 215 | 0.58 | 0.15 | 0.02 | |
| West Germany | 265 | 0.56 | 0.14 | (p = 0.248) | Prob > |z| = 0.222 |
| N/A | 54 | 0.55 | 0.13 | ||
| Living in the 1980s | |||||
| East Germany | 191 | 0.58 | 0.16 | 0.02 | |
| West Germany | 202 | 0.56 | 0.14 | (p = 0.115) | Prob > |z| = 0.291 |
| Others | 25 | 0.55 | 0.12 | ||
| N/A | 116 | 0.57 | 0.14 | ||
| Living in the 1990s | |||||
| East Germany | 211 | 0.58 | 0.16 | 0.03* | |
| West Germany | 279 | 0.55 | 0.13 | (p = 0.044) | Prob > |z| = 0.152 |
| N/A | 44 | 0.58 | 0.15 | ||
| Berlin sample (N = 259) | Leipzig & Dortmund samples (N = 275) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | High rolls (%) | SD | Differences between East and West background | N | High rolls (%) | SD | Differences between East and West background | |||
| t-test | Wilcoxon-Mann-Whitney-test | t-test | Wilcoxon-Mann-Whitney-test | |||||||
| Panel B: East and West Berlin, Leipzig and Dortmund samples | ||||||||||
| Family background | ||||||||||
| East German | 90 | 0.60 | 0.15 | 0.05* | 99 | 0.56 | 0.15 | 0.00 | ||
| West German | 98 | 0.55 | 0.13 | (p = 0.013) | Prob > |z| = 0.011 | 97 | 0.55 | 0.14 | (p = 0.850) | Prob > |z| = 0.881 |
| German | 37 | 0.52 | 0.13 | 38 | 0.61 | 0.13 | ||||
| Others | 30 | 0.60 | 0.16 | 40 | 0.58 | 0.15 | ||||
| N/A | 4 | 0.53 | 0.12 | 1 | 0.63 | – | ||||
| Consider yourself | ||||||||||
| East German | 33 | 0.60 | 0.12 | 0.02 | 40 | 0.58 | 0.15 | 0.00 | ||
| West German | 40 | 0.58 | 0.14 | (p = 0.510) | Prob > |z| = 0.229 | 30 | 0.57 | 0.16 | (p = 0.902) | Prob > |z| = 0.877 |
| German | 141 | 0.56 | 0.15 | 157 | 0.56 | 0.14 | ||||
| Others | 40 | 0.56 | 0.13 | 45 | 0.60 | 0.16 | ||||
| N/A | 5 | 0.52 | 0.11 | 3 | 0.48 | 0.13 | ||||
| Place of birth | ||||||||||
| East Germany | 97 | 0.59 | 0.15 | 0.03 | 118 | 0.57 | 0.15 | 0.00 | ||
| West Germany | 134 | 0.56 | 0.14 | (p = 0.118) | Prob > |z| = 0.060 | 131 | 0.57 | 0.15 | (p = 0.889) | Prob > |z| = 0.934 |
| N/A | 28 | 0.54 | 0.14 | 26 | 0.56 | 0.12 | ||||
| Living in the 1980s | ||||||||||
| East Germany | 88 | 0.59 | 0.16 | 0.03† | 103 | 0.57 | 0.16 | 0.01 | ||
| West Germany | 108 | 0.56 | 0.14 | (p = 0.095) | Prob > |z| = 0.132 | 94 | 0.56 | 0.14 | (p = 0.528) | Prob > |z| = 0.953 |
| Others | 13 | 0.52 | 0.10 | 12 | 0.58 | 0.13 | ||||
| N/A | 50 | 0.56 | 0.15 | 66 | 0.57 | 0.13 | ||||
| Living in the 1990s | ||||||||||
| East Germany | 98 | 0.59 | 0.16 | 0.04† | Prob > |z| = 0.130 | 113 | 0.57 | 0.16 | 0.02 | |
| West Germany | 140 | 0.55 | 0.13 | (p = 0.058) | 139 | 0.56 | 0.13 | (p = 0.329) | Prob > |z| = 0.578 | |
| N/A | 21 | 0.56 | 0.14 | 23 | 0.60 | 0.16 | ||||
Reported share of high rolls for different measures of East and West German background and for East and West Berlin as well as Leipzig and Dortmund subsamples. **,* and † indicate significance at the 1%, 5% and 10% level.
Table A.3. Correlation matrix showing pairwise correlations between variables. For variable definitions, see Appendix E.
| [1] | [2] | [3] | [4] | [5] | [6] | [7] | [8] | [9] | [10] | [11] | [12] | [13] | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| [1] | High rolls % | 1.0000 | ||||||||||||
| [2] | Donations % | −0.1708 | 1.0000 | |||||||||||
| [3] | East German | 0.0928 | 0.0232 | 1.0000 | ||||||||||
| [4] | Location Berlin | 0.0004 | 0.0166 | −0.0264 | 1.0000 | |||||||||
| [5] | Age | 0.0701 | 0.0577 | 0.1612 | 0.0029 | 1.0000 | ||||||||
| [6] | Female | 0.0768 | −0.0033 | 0.0718 | −0.0767 | −0.0578 | 1.0000 | |||||||
| [7] | Trust East | 0.0524 | −0.0095 | 0.0976 | 0.0123 | 0.0447 | −0.0046 | 1.0000 | ||||||
| [8] | Trust West | 0.0778 | −0.0379 | 0.2862 | −0.0128 | 0.1189 | −0.0388 | 0.6604 | 1.0000 | |||||
| [9] | West betrayed | −0.0518 | −0.0132 | −0.1018 | −0.2186 | 0.2240 | −0.1354 | −0.0391 | 0.0051 | 1.0000 | ||||
| [10] | Education | −0.1072 | −0.0062 | −0.0618 | 0.0397 | 0.0950 | −0.0399 | −0.1653 | −0.1735 | 0.0407 | 1.0000 | |||
| [11] | Political orientation | −0.0272 | −0.0764 | 0.0205 | −0.0899 | 0.0358 | −0.1576 | 0.1139 | 0.1571 | 0.0138 | −0.0408 | 1.0000 | ||
| [12] | Living standard | 0.0370 | −0.1102 | 0.0564 | 0.0916 | 0.0606 | 0.0252 | 0.0696 | 0.1564 | 0.0183 | −0.1972 | −0.0734 | 1.0000 | |
| [13] | Marital status | 0.0443 | 0.0591 | 0.1696 | −0.0523 | 0.4738 | 0.0486 | −0.0076 | 0.0435 | 0.1749 | 0.0078 | 0.0473 | −0.0506 | 1.0000 |
Appendix B. Exemplary locations field experiment
Fig. 1. Citizen centre East Berlin.
Fig. 2. Citizen centre West Berlin.
Appendix C. Cheating Die Task Instructions
In this task, you are asked to throw a die 40 times. Every time before you roll the die, you will be asked to choose one side of the die in your mind: top or bottom. Be sure to make your choice before you roll the die. Then, after rolling the die, please enter the outcome of your roll in this sheet, i.e. the number of dots shown on the chosen side of the die (top or bottom).
Bear in mind that for each roll there are two possible sides: top or bottom. Here, the figure shows the different outcome combinations.
Please return your record sheet to the experimenter after you threw the die 40 times. The experimenter will then randomly draw a number from 1 to 40. This number determines which of your rolls is relevant for your payment.
You will get paid 1€ for each dot of this randomly drawn roll. For example, assume the experimenter draws number 14, then your 14th roll determines your payment. The experimenter will check your record sheet and pay you 1€ for each dot of your 14th roll.
Please discuss any questions with the experimenter before starting the task!
Instructions (donation).
The hospital of Hannover [Leipzig, Rostock] located in West Germany [East Germany] is a public institution that stands for local medicine and care, high quality treatment, as well as sophisticated diagnostics and therapy. Given its traditional duty, the hospital regards itself as an institution that constitutes a medical-social centre. In addition to offering medical care, being close to people is also at the core of the hospital’s work. The clinic therefore also administers social services especially for West German [East German] patients and relatives and is currently expanding this service. To continue with that work, the hospital of Hannover [Leipzig, Rostock] depends not only on public funding but also on public donations. Please donate for the patients and relatives in West Germany [East Germany].
Appendix D. Supplementary data
The following is the supplementary data to this article:
Multimedia component 1.
Appendix E.
Table E.1. Variable descriptions.
| Variable | Description |
|---|---|
| Earnings | Participants’ earnings from the die task in Euro (1 EUR, 2 EUR, 3 EUR, 4 EUR, 5 EUR, or 6 EUR). |
| High roll | 0 = Participant reported that they had rolled a 1, 2, or 3. 1 = Participant reported that they had rolled a 4, 5, or 6. |
| High rolls in % | Participants’ share of reported high rolls over the 40 rolls. |
| East German | 0 = Participant indicated to have a West German family background 1 = Participant indicated to have an East German family background. |
| Location Berlin | 0 = Experiment was run in Leipzig (East Germany) or Dortmund (West Germany). 1 = Experiment was run in East or West Berlin. |
| Age | Participants age as of January 1, 2015. |
| Female | What is your gender? 0 = male, 1 = female. |
| Education | What is the highest level of education you have completed? Scale: 0 = none, 1 = ‘Hauptschule’ (lower level high school), 2 = ‘Realschule’ (high school), 3 = ‘Abitur/Fachabitur’ (some college), 4 = ‘Bachelor/Fachhochschulabschluss’ (3–4 years of university), 5 = ‘Maste/Diplom’ (4–5 years of university) and 6 = ‘Promotion/Aufbaustudium’ (doctoral degree, post-graduate degree). |
| Living standard | What describes your standard of living? Scale from 1 = very well off to 6 = poor. |
| Marital status | 0 = Participant reported to be single, separated, divorced, widowed or indicated ‘other’. 1 = Participant reported to be in a relationship or married. |
| Political orientation | In political matters, people talk of “the left” and “the right”. How would you place your views on this scale, generally speaking? Scale from 0 = left to 10 = right. |
| Trust East | Generally speaking, would you say that East Germans can be trusted? Scale from 0 = most East Germans can be trusted to 10 = you cannot be too careful in dealing with East Germans. |
| Trust West | Generally speaking, would you say that West Germans can be trusted? Scale from 0 = most West Germans can be trusted to 10 = you cannot be too careful in dealing with West Germans. |
| West betrayed | Would you agree that West Germans cheated East Germans after the fall of the Berlin Wall? 0 = no, 1 = yes. |
Appendix F. Reported rolls
Figure F.1. Reported roles by East and West German family background.
Percentages of reported rolls of 1, 2, 3, 4, 5 and 6. The red line refers to the fair outcome of 16.6 percent and 33.3 percent, respectively. All rolls for East and West Germans (as well as both groups combined) are statistically different from the 16.6 percent baseline for outcomes 1, 2, 3, 4, 5 and 6 at the p < 0.05 level.
Fig. F.1.A. Full sample.
Fig. F.1.B. Berlin sample.
Fig. F.1.C. Leipzig and Dortmund sample.
Percentages of reported rolls of 3–4, 2–5 and 1–6 combinations. None of the two-side combinations for East and West Germans (as well as both groups combined) are statistically different from the 33.3 percent baseline at the p < 0.05 level.
Fig. F.1.D. Full sample.
Fig. F.1.E. Berlin sample.
Fig. F.1.F. Leipzig and Dortmund sample
Figure F.2. Percentage of high rolls by stage of experiment, family background and location where experiment took place.
Fig. F.2.A. Full sample.
Fig. F.2.B. Berlin sample.
Fig. F.2.C. Leipzig and Dortmund sample.
References
- Abeler et al., 2014
-
J. Abeler, A. Becker, A. Falk. Representative evidence on lying costsJ. Public Econ., 113 (2014), pp. 96-104
- Abeler et al., 2018
-
J. Abeler, D. Nosenzo, C. Raymond Preferences for truth-telling, Econometrica(2018)forthcoming
- Alesina and Fuchs-Schündeln, 2007
-
A. Alesina, N. Fuchs-Schündeln Goodbye Lenin (or not?): the effect of communism on people’s preferencesAm. Econ. Rev., 97 (4) (2007), pp. 1507-1528
- Brosig-Koch et al., 2011
-
J. Brosig-Koch, C. Helbach, A. Ockenfels, J. Weimann Still different after all these years: solidarity behavior in East and West GermanyJ. Public Econ., 95 (11–12) (2011), pp. 1373-1376
- Cohn and Maréchal, 2018
-
A. Cohn, M.A. Maréchal Laboratory measure of cheating predicts school misconductEcon. J., 128 (6) (2018), pp. 2743-2754
- Crown and Spiller, 1998
-
D.F. Crown, M.S. Spiller Learning from the literature on collegiate cheating: a review of empirical researchJ. Bus. Ethics, 17 (6) (1998), pp. 683-700
- Dai et al., 2017
-
Z. Dai, F. Galeotti, M.C. Villeval Cheating in the lab predicts fraud in the field: an experiment in public transportationManag. Sci., 64 (3) (2017), pp. 1081-1100
- Dieckmann et al., 2016
-
A. Dieckmann, V. Grimm, M. Unfried, V. Utikal, L. Valmasoni On trust in honesty and volunteering among Europeans: cross-country evidence on perceptions and behaviorEur. Econ. Rev., 90 (2016), pp. 225-253
- Ezquerra et al., 2018
-
L. Ezquerra, G.I. Kolev, I. Rodriguez-Lara Gender differences in cheating: loss vs. gain framingEcon. Lett., 163 (2018), pp. 46-49
- Falk and Szech, 2013
-
A. Falk, N. Szech Morals and marketsScience, 340 (2013), pp. 707-711
- Fischbacher and Föllmi-Heusi, 2013
-
U. Fischbacher, F. Föllmi-Heusi Lies in disguise—an experimental study on cheatingJ. Eur. Econ. Assoc., 11 (3) (2013), pp. 525-547
- Fortin et al., 2007
-
B. Fortin, G. Lacroix, M.C. Villeval Tax evasion and social interactionsJ. Public Econ., 91 (11–12) (2007), pp. 2089-2112
- Gächter and Schulz, 2016
-
S. Gächter, J.F. Schulz Intrinsic honesty and the prevalence of rule violations across societiesNature, 531 (2016), pp. 499-696
- Garbarino et al., 2016
-
E. Garbarino, R. Slonimb, M.C. Villeval Loss Aversion and Lying Behavior: Theory, Estimation and Empirical Evidence(2016)IZA Discussion Paper No. 10395
- Gino et al., 2009
-
F. Gino, S. Ayal, D. Ariely Contagion and differentiation in unethical behavior: the effect of one bad apple on the barrelPsychol. Sci., 20 (3) (2009), pp. 393-398
- Haidt et al., 1993
-
J. Haidt, S.H. Koller, M.G. Dias Affect, culture, and morality, or is it wrong to eat your dog?J. Personal. Soc. Psychol., 65 (4) (1993), pp. 613-628
- Harrison and List, 2004
-
G.W. Harrison, J.A. List Field experimentsJ. Econ. Lit., 42 (4) (2004), pp. 1009-1055
- Hartshorne et al., 1929
-
H. Hartshorne, M.A. May, J.B. Maller Studies in the Nature of Character II: Studies in Aervice and Self-ControlThe Macmillian Company, New York (1929)
- Heineck and Süssmuth, 2013
-
G. Heineck, B. Süssmuth A different look at Lenin’s legacy: social capital and risk taking in the two GermaniesJ. Comp. Econ., 41 (3) (2013), pp. 789-803
- Hornuf and Rieger, 2017
-
L. Hornuf, M.O. Rieger Can Television Reduce Xenophobia? The Case of East Germany(2017)Research Papers in Economics No. 1/17
- Houser et al., 2012
-
D. Houser, S. Vetter, J. Winter Fairness and cheatingEur. Econ. Rev., 56 (8) (2012), pp. 1645-1655
- Inglehart and Welzel, 2012
-
R. Inglehart, C. Welzel Nation-level Mean Scores on Traditional/Secular-Rational and Survival/Self-Expression Values Dimensions, 1981–2007(2012)(World Values Survey) Available at:
- Internal Revenue Service, 2016
-
Internal Revenue Service Tax Gap Estimates for Tax Years 2008-2010(2016)
- Jiang, 2013
-
T. Jiang Cheating in mind games: the subtlety of rules mattersJ. Econ. Behav. Organ., 93 (2013), pp. 328-336
- John et al., 2014
-
L.K. John, G. Loewenstein, S.I. Rick Cheating more for less: upward social comparisons motivate the poorly compensated to cheatOrgan. Behav. Hum. Decis. Process., 123 (2) (2014), pp. 101-109
- Koehler, 1999
-
J.O. Koehler STASI: The Untold Story of the East German Secret PoliceWestview Press, Boulder (1999)
- Mann et al., 2016
-
H. Mann, X. Garcia-Rada, L. Hornuf, J. Tafurt, D. Ariely Cut from the same cloth: similarly dishonest individuals across countriesJ. Cross Cult. Psychol., 47 (6) (2016), pp. 858-874
- Mann et al., 2014
-
H. Mann, X. Garcia-Rada, D. Houser, D. Ariely Everybody else is doing it: exploring social transmission of lying behaviorPLoS One, 9 (10) (2014), Article e109591
- Muehlheusser et al., 2015
-
G. Muehlheusser, A. Roider, N. Wallmeier Gender differences in honesty: groups versus individualsEcon. Lett., 128 (2015), pp. 25-29
- Ockenfels and Weimann, 1999
-
A. Ockenfels, J. Weimann Types and patterns: an experimental East-West-German comparison of cooperation and solidarityJ. Public Econ., 71 (2) (1999), pp. 275-287
- Pfarr et al., 2013
-
C. Pfarr, A. Schmid, V. Ulrich You Can’t Always Get what You Want: East and West Germans’ Attitudes and Preferences Regarding the WelfareState(2013)MPRA Working Paper No. 47240
- Rauhut, 2013
-
H. Rauhut Beliefs about lying and spreading of dishonesty: undetected lies and their constructive and destructive social dynamics in dice experimentsPLoS One, 8 (11) (2013), Article e77878
- Schmitt and Maes, 1998
-
M. Schmitt, J. Maes Perceived injustice in unified Germany and mental healthSoc. Justice Res., 11 (1) (1998), pp. 59-78
- Shalvi, 2016
-
S. Shalvi Behavioural economics: Corruption corruptsNature, 531 (2016), p. 456
- Torgler, 2003
-
B. Torgler Does culture matter? Tax morale in an East-West-German comparisonFinanzArchiv Public Financ Anal: Public Finance Anal., 59 (4) (2003), pp. 504-528
- U.S. Department of Justice, 2018
-
U.S. Department of Justice Insurance Fraud(2018)Available at:
Footnotes
1. Information published in the German Statistical Yearbook 2015 and the National Accounts of the German Federal Statistical Office: 551,871 inhabitants in Leipzig and 589,283 inhabitants in Dortmund as of December 2014; disposable household income in 2013 of 8801€ in Leipzig and 10,325€ in Dortmund.
2. The acceptance rate to take part in the study was between 25 percent to 50 percent mostly depending on the time of the day and how long people had to wait in the citizen centre.
3. The age of our participants ranged from 12 to 87 years. Two participants at the age of 12 conducted the experiment jointly with their parents.
4. When we treat the rolls of participants as statistically independent and run a χ2 test, we find East Germans to cheat more for all measures except “How do you consider yourself?”.
5. The results in Table 1 are robust in terms of effect size and statistical significance when estimating a linear model.
6. See BT-printed matter 18/7274 of the German parliament, page 9–10.
7. 0.0032 percent x 252 participants = 0.81 participant ∼1 participant.
