Robustness of Inferences in Risk and Time Experiments to Lifecycle Asset Integration

Résumé En

Participants in an experiment can engage in unobservable asset integration, mentally incorporating their own non-experimental “field” resources into an otherwise controlled scenario. This paper extends asset integration to include intertemporal tradeoffs like consumption smoothing. A model of “lifecycle asset integration” shows that exogenous and endogenous field resources cause different interference patterns. Exogenous resources cannot be affected by the experiment, and so their interference can be controlled by accounting for their level. Endogenous resources, by contrast, are highly substitutable with the experiment, and their interference can be controlled only by modeling the entire experiment-field interaction. The model's practical implications are investigated in the context of three classic laboratory experiments on risk and time: one static (Holt and Laury, 2002) and two dynamic (Andersen et al., 2008; Andreoni and Sprenger, 2012). As interference worsens, decisions in these tasks tend to exhibit a kind of attenuation bias toward less risk aversion and more patience. Interference occurs reliably when field resources are on household scales, but amounts on the scale of pocket change can also cause problems.