Traditional asset pricing models predict that covariance between prices of different assets should be lower than what we observe in the data. This paper introduces markets for information that generate high price covariance within a rational expectations framework. When information is costly, rational investors only buy information about a subset of the assets. Because information production has high fixed costs, competitive producers charge more for low-demand information than for high-demand information. The low price of high-demand information makes investors want to purchase the same information that others are purchasing. When investors price assets using a common subset of information, news about one asset affects the other assets' prices; asset prices comove. The cross-sectional and time-series properties of comovement are consistent with this explanation.

Emerging equity markets witness occasional surges in prices (frenzies) and cross-market price dispersion (herds), accompanied by abundant media coverage. An information market complementarity can explain these anomalies. Because information has high fixed costs, high volume makes it inexpensive. Low prices induce investors to buy information that others buy. Given two identical assets, investors learn about one; abundant information reduces its payoff risk and raises its price. Transitions between low-information/low-asset-price and high-information/high-asset-price equilibria resemble frenzies. Equity data and new panel data on news coverage support the model's predictions: Asset market movements generate news and news raises prices and price dispersion.