Mind the framing when studying social preferences in the domain of losses

There has been an increasing interest in altruistic behaviour in the domain of losses recently. Nevertheless, there is no consensus in whether the monetary losses make individuals more generous or more selfish. Although almost all relevant studies rely on a dictator game to study altruistic behaviour, the experimental designs of these studies differ in how the losses are framed, which may explain the diverging findings. Utilizing a dictator game, this paper studies the impact of loss framing on altruism. The main methodological result is that the dictators' prosocial behaviour is sensitive to the loss frame they are embedded in. More specifically, in a dictator game in which the dictators have to share a loss between themselves and a recipient, the monetary allocations of the dictators are more benevolent than in a standard setting without a loss and in a dictator game in which the dictators have to share what remains of their endowments after a loss. These differences are explained by the different social norms that the respective loss frames invoke.