Unethical behavior such as dishonesty, cheating and corruption occurs frequently in organizations or groups.Recent experimental evidence suggests that there is a stronger inclination to behave immorally in groups than individually. We ask if this is the case, and if so, why. Using a parsimonious laboratory setup, we study how individual behavior changes when deciding as a group member. We observe a strong dishonesty shift. This shift is mainly driven by communication within groups and turns out to be independent of whether group members face payoff commonality or not (i.e. whether other group members benefit from one's lie). Group members come up with and exchange more arguments for being dishonest than for complying with the norm of honesty.Thereby, group membership shifts the perception of the validity of the honesty norm and of its distribution in the population. IntroductionGroups and organizations sometimes fail to comply with a moral norm. They lie, they cheat, they are dishonest, they are corrupt, and they commit fraud. However, it is not organizations that take those decisions; it is individuals that are part of the organization. Can we thus explain undesired behavior in organizations simply by aggregating individual failures to comply with the norm? Or, are there other elements inherent to the organization or to its structure that can help us better understand how undesired behavior emerges? And, how does undesired behavior of individuals differ from behavior within or by an entire organization? While these are relevant questions, surprisingly little empirical evidence exists (Conrads et al., 2013; Sutter, 2009). This paper addresses these questions in a parsimonious setup that allows us to identify some of the potential reasons for collective failure to follow a moral norm or to comply with desired behavior. This study provides a twofold contribution. First, we implement a parsimonious laboratory setup to investigate whether groups (as our proxy for small organizations) are indeed more inclined to engage in dishonest or unethical behaviors than individuals, as casual observation and some previous results in the literature suggest (e.g., Chytilova and Korbel, 2014; Conrads et al., 2013; Gino et al., 2013; Muehlheusser et al., 2015; Sutter, 2009; Weisel and Shalvi, 2015). We find that the answer is affirmative. Individuals lie less frequently when deciding alone as compared to groups. Second, we offer and disentangle explanations for this "dishonesty shift". There are several candidate explanations: (i) a simple aggregation of individual inclinations as a consequence of aggregation rules (i.e. decision making procedures) within the group; (ii) the incentive structure inherent to many group decisions (oftentimes, all members share group payoffs equally and an individual deviation from either of the strategies -behaving dishonest or honest -can sometimes reduce payoffs for everyone dramatically); (iii) the decreased observability of one's actions within a group, potentially making the individual...

In many cases individuals benefit differently from the provision of a public good. We study in a laboratory experiment how heterogeneity in returns and uncertainty about the own return affects unconditional and conditional contribution behavior in a linear public goods game. The elicitation of conditional contributions in combination with a within subject design allows us to investigate belief-independent and type-specific reactions to heterogeneity. We find that, on average, heterogeneity in returns decreases unconditional contributions but affects contributions only weakly. Uncertainty in addition to heterogeneity reduces conditional contributions slightly. Individual reactions to heterogeneity differ systematically. Selfish subjects and one third of conditional cooperators do not react to heterogeneity whereas the reactions of the remaining conditional cooperators vary. A substantial part of heterogeneity in reactions can be explained by inequity aversion with respect to different reference groups.

We investigate whether risk, time, environmental, and social preferences affect single family homeowners' investments in energy efficient renovations and energy quality of their house using established experimental measures and questionnaires. We find that homeowners who report to be more risk taking are more likely to have renovated their house. Pro-environmental and futureoriented renovators, i.e. renovators with lower discount factors, live in homes with higher energy efficiency. Controlling for the energy efficiency of houses, we further find that energy consumption as measured by heating and energy costs are lower for future-oriented and proenvironmental individuals. Social preferences measured in a dictator and a generosity game play a mixed role for investments in energy efficiency and energy consumption.

Although understanding preferences for privacy is of great importance to economists, businesses and politicians little is known about the factors that shape the individual willingness to share personal data. This article provides three experimental studies with a total of 470 participants that help characterizing individual preferences for sharing personal data varying the characteristics of potential recipients. We find that participants' willingness to share personal data with anonymous recipients decreases with the number of recipients. However, social distance to the recipients and the extent of personal data a single recipient receives do not decrease the willingness to share personal data. Further, we provide a methodological insight by showing that verification of personal data is essential when eliciting privacy preferences. JEL Classification: C90, C91, D80, D82

We investigate whether automatic selling devices causally reduce investors' disposition effect (DE) in a laboratory experiment. Investors can actively buy and sell assets. Investors in the treatment group use stop-loss and take-gain options to automatically sell assets. In addition, we introduce a reminder condition that reminds investors about their selling plan if a limit is hit. Results show that the automatic selling device treatment significantly reduces the DEs, but the reminder treatment does not. Thus, the opportunity to ex ante commit to automatically selling at a loss causally reduces the disposition effect. (JEL C91, G02, G11) data can be found on The Review of Financial Studies web site.