This thesis investigates how the information dispersed among market participants dynamically aggregates in asset prices, the extent to which prices reflect available information, and how such information affects investors’ decisions. The main model considers a population of investors with different absolute risk aversions and time-varying, diverse signals on the growth rate of an asset’s dividends. Each investor bids the asset based on the information in his private signal and in the asset price itself, which is determined in equilibrium by the market-clearing condition and partially reflects the signals of other market participants. The dividend stream is driven by a latent variable, which investors strive to estimate based on their individual, private information, and on the common knowledge revealed by prices. We find in closed form equilibrium prices and the optimal behaviour of the agents. Price volatility depends on the volatility of dividends and on the volatility of the estimate of the latent variable, which is revealed to all agents through prices. Equilibrium prices do not reveal all the private signals of market participants, but the same estimate of the state of the economy that an agent with all private signals would be able to obtain. Put differently, prices reveal not all information but all relevant information. The first chapter presents a baseline model, where the only noise in the market is on the stochastic dividend process. In the second chapter dividends become mean reverting to a state variable observed by all agents - the state of the economy - which fluctuates over time. The state of the economy is unobservable in the last chapter, but market participants have individual information, which jointly with asset prices, helps them to estimate the latent variable.