We provide a reason for the wider economics profession to take social preferences, a concern for the outcomes achieved by other reference agents, seriously. Although we show that student measures of social preference elicited in an experiment have little external validity when compared to measures obtained from a field experiment with a population of participants who face a social dilemma in their daily lives (i.e., team production), we do find strong links between the social preferences of our field participants and their productivity at work. We also find that the stock of social preferences evolves endogeously with respect to how widely team production is utilized.