Data Technology and Market Efficiency
Abstract
How will advances in data technology affect market equilibrium and informational efficiency when adoption by investors is uneven? I show that data technology development and dissemination improve asset managers' stock-picking ability, especially for those with larger shares of data-science labor. Using U.S. institutional holdings and employment records, I estimate information production across investors and a demand system with heterogeneity in private information and learning from prices. I estimate investors' private information from holdings' forecasting power for future payoffs beyond public information and prices. Data technology developments are captured by increases in data scientist productivity. Estimates show that better-informed investors have more elastic demand, while learning from prices makes demand 61 percent less elastic at the median informative stock. A counterfactual advance calibrated to more than a decade of data scientist productivity growth raises price informativeness by 26 percent, more for small stocks. New technology adopters crowd out other investors, increasing market segmentation as stocks' investor bases diverge. Aggregate demand becomes more inelastic for 47 percent of stocks.