Experimental Design
In our survey experiment, we illustrate a decision context of a hypothetical individual (person X) who decides about taking up a job offer or not. Participants of the survey are representative of the German population in several dimensions (such as education) and have to assess the social appropriateness of the decision. The context provides information about the income situation, so that we are able to identify the social norm against dependency on public assistance. For this reason, we clarify that the implications of taking up a job are negligible moneywise but being employed allows for compliance with this particular social norm. Such a scenario is plausible given the institutional background in Germany and hence realistic for survey participants who are asked to assess the social appropriateness of X’s decision to earn money in the labor market or not.
To identify the social pressure to earn a living, we compare the two different decisions as our main experimental treatments. First, X picks up the job, which would be expected as being perceived socially appropriate, despite that this decision is not rewarded by any relevant increase in household income. We test the hypothesis that society expects people to earn a living (vs. relying on society to help via public assistance) empirically by comparing the perceived appropriateness of taking up a job, compared to the alternative decision by X to stay non-employed, which is the second main treatment. Randomness of this manipulation in our experiment allows us to identify the causal effect of (non )employment on the social appropriateness of individual (mis)behavior in the eyes of the public, which indicates the level of social pressure to do what society expects you to do.
To ensure that survey responses of participants inform us about the social norm to earn a living, we ask them first to think about their belief of what the broader population perceives as socially appropriate, before we ask them about their personal views on the matter in a second step. As a well-established tool to foster successful identification of social norm effects, we incentive survey responses by allowing participants to receive a monetary reward if their belief matches the most common answer in all the survey participants (in a given treatment condition). This idea follows a concept by Krupka and Weber (2013) to measure social norms regarding appropriate behavior via an incentivized experimental approach. In our application of this approach, we allow participants to take part in a lottery. A prize of 50 Euro is awarded to each randomly drawn winner (from all the participants in a given treatment condition), provided that this participant reports the ‘correct’ response in the way described above.
In addition to our main manipulation regarding the job decision (take-up of the job offer vs. staying in non-employment), we further manipulate the scenario by providing information about time-use in unemployment. Accordingly, X can either be productive at home (e.g. by taking care of the children) or X enjoys leisure time (e.g. by pursuing hobbies). Furthermore, we manipulate the gender of X by using gender-specific terms like Mr. X vs. Ms. X. Thereby, we have a 2x2x2 experimental design with three different dimensions (job decision / time-use / gender). To complement our dataset, we collect information on the survey respondents including their attitudes towards work as a means to contribute to society (among other reasons why people go to work), which could play a role when assessing behavioral choices in the labor market.
By inspecting varying scenarios, we can learn more about the mechanisms underlying our earlier findings on labor supply decisions in our first study (based on data from the German Socio-Economic Panel study). First and foremost, we empirically test the idea of social pressure to earn a living, which could explain our behavioral evidence on people’s labor supply. Furthermore, our design answers the question whether this pressure could be alleviated by being productive at home (as opposed to leisure enjoyment). This is ex-ante unclear, given that there are reasons why society might not reward unemployed people for being productive at home. A possible explanation could be that doing useful things at home, such as childcare, does not receive the recognition by others, as doing market work does. Finally, our design allows us to study the role of potential gender differences in social pressure and hence labor market behavior. Previous research suggests that traditional gender roles could affect labor market behavior, with men being expected more than women to fulfill a breadwinner role by supplying market work. Due to societal trends and significant changes over time in beliefs regarding the role of women and men in the labor market, however, the existence of gender-specific effects on the assessment of socially appropriate behavior is an open question.