Title: Endogenous Growth Under Multiple Uses of Data

Abstract: We model a dynamic data economy with fully endogenous growth where agents generate data from consumption and share them with innovation and production firms. Different from other productive factors such as labor or capital, data are nonrival in their uses across sectors which affect both the level and growth of economic outputs. Despite the vertical nonrivalry, the innovation sector dominates the production sector in data usage and contribution to growth because (i) data are dynamically nonrival and add to knowledge accumulation, and (ii) innovations "desensitize" raw data and enter production as knowledge, which allays consumers' privacy concerns. Data uses in both sectors interact to generate spillover of allocative distortion and exhibit an apparent substitutability due to labor's rivalry and complementarity with data. Consequently, growth rates under a social planner and a decentralized equilibrium differ, which is novel in the literature and has policy implications. Specifically, consumers' failure to fully internalize knowledge spillover when bearing privacy costs, combined with firms' market power, underprice data and inefficiently limit their supply, leading to underemployment in the innovation sector and a suboptimal long-run growth. Improving data usage efficiency is ineffective in mitigating the underutilization of data, but interventions in the data market and direct subsidies hold promises.