We propose a multi-period model with agents who differ in their risk sharing motives and participation costs. Predictions regarding return, order flow, and return-flow dynamics are empirically tested using a state-space model and proprietary NYSE trading data. We show monthly transitory price pressure is more that 28% the magnitude of efficient price variation for the average stock. A one standard deviation change in market makers’ positions (or individuals’ net trades) is associated with transitory volatility of 1.55% (or 1.66%). The results are larger for smaller stocks (2.48% or 1.97%). Together, our trading variables account for 37.81% of transitory variance (59.73% for small stocks.) Joint Work with Terrence Hendershott (U.C. Berkeley), Albert J. Menkveld (VU) and Mark S. Seasholes (HKUST).