We study the endowment effect and expectation-based reference points in the field leveraging the setup of the Socio-Economic Panel. Households receive a small item for taking part in the panel, and we randomly assign respondents either a towel or a notebook, which they can exchange at the end of the interview. We observe a trading rate of 32 percent, consistent with an endowment effect, but no relationship with loss aversion. Manipulating expectations of the exchange opportunity, we find no support for expectation-based reference points. However, trading predicts residential mobility and is related to stock-market participation, i.e., economic decisions that entail parting with existing resources.