In this study, I analyze Bitcoin transaction data and build an economic model on Bitcoin traders incentives to decompose the Bitcoin price into a utility-driven component, a speculative component, and a friction component. The model I build extends the LDA (Latent-Dirichlet-Allocation) model, and I perform a paralleled collapsed Gibbs Sampling method to estimate the realized incentives of each individual trader at each time point. For post-estimation analysis, I look into major headline news to see which how information or rumor affects the different components of the Bitcoin price. The preliminary results show interesting patterns of trading and pricing in the Bitcoin market for the first time.