A pillar of behavioral research is that preferences are constructed during the process of choice. A prominent finding is that uninformative numerical "anchors" influence judgment and valuation. It remains unclear whether such processes influence market equilibria. We conduct two experiments that extend the study of anchoring to field settings. The first experiment produces evidence that some consumers' valuations can be anchored in novel situations; there is no evidence that experienced agents are influenced by anchors. The second experiment finds that anchors have only transient effects on market outcomes that converge to equilibrium predictions after a few market periods.