Bakos and Brynjolfsson analyze the strategy of bundling information goods, such as those available on the Internet, and selling them at a fixed price. They find that bundling a large number of unrelated information goods can be highly profitable due to the "predictive value of bundling," which reduces uncertainty in consumer valuations. This allows for greater sales, efficiency, and profits compared to selling goods separately. The key insight is that as the number of goods in a bundle increases, the variance of consumer valuations decreases, making it easier to predict demand and set prices. However, this strategy is less effective for physical goods due to higher marginal costs. The authors use statistical techniques to derive asymptotic results and bounds on profits for bundles of any size. They show that bundling can be more profitable than selling goods separately when consumer valuations are negatively correlated, or even independent. However, when valuations are positively correlated, bundling may not be optimal. The study also highlights that bundling can be more profitable when consumer valuations are drawn from different distributions, as it allows for self-selection by consumers. The authors also discuss the implications of bundling for different market structures, including the potential for increased profits through third-degree price discrimination. They argue that bundling can create new opportunities for price discrimination by reducing the importance of idiosyncratic factors in consumer valuations. The study concludes that bundling can be a powerful strategy for information goods with low marginal costs, and that it can lead to greater efficiency and profits compared to selling goods separately. The findings are supported by empirical evidence from the markets for Internet and online content, cable television programming, and copyrighted music.Bakos and Brynjolfsson analyze the strategy of bundling information goods, such as those available on the Internet, and selling them at a fixed price. They find that bundling a large number of unrelated information goods can be highly profitable due to the "predictive value of bundling," which reduces uncertainty in consumer valuations. This allows for greater sales, efficiency, and profits compared to selling goods separately. The key insight is that as the number of goods in a bundle increases, the variance of consumer valuations decreases, making it easier to predict demand and set prices. However, this strategy is less effective for physical goods due to higher marginal costs. The authors use statistical techniques to derive asymptotic results and bounds on profits for bundles of any size. They show that bundling can be more profitable than selling goods separately when consumer valuations are negatively correlated, or even independent. However, when valuations are positively correlated, bundling may not be optimal. The study also highlights that bundling can be more profitable when consumer valuations are drawn from different distributions, as it allows for self-selection by consumers. The authors also discuss the implications of bundling for different market structures, including the potential for increased profits through third-degree price discrimination. They argue that bundling can create new opportunities for price discrimination by reducing the importance of idiosyncratic factors in consumer valuations. The study concludes that bundling can be a powerful strategy for information goods with low marginal costs, and that it can lead to greater efficiency and profits compared to selling goods separately. The findings are supported by empirical evidence from the markets for Internet and online content, cable television programming, and copyrighted music.