I develop a dynamic asset pricing model in which subject beliefs about stock price behavior are heterogeneous and susceptible to peer effects. Two types of traders optimally learn from past price realizations and share beliefs in a social network. I show that, at each period, the equilibrium price is a function of traders' beliefs and the network structure. As a result, booms and busts of the price-dividend ratio emerge. The most (least) speculative trader is the most influential during booms (busts). More connected networks exhibit less volatile price dividend ratio, booms and busts episodes last longer, and the average price realization is higher. Also, there is less disagreement in beliefs. However, if traders of the same type are highly interconnected stock market volatility is higher and booms and busts are shorter. The model captures relevant empirical features of stock prices and returns, and it is also consistent with the survey evidence on investor expectations.