We thank Ach Adhvaryu, Amanda Agan, Naoki Aizawa, Heski Bar-Isaac, Panle Barwick, Alan Benson, Gordon Dahl, Chiara Farronato, Kory Kroft, Peter Kuhn, Rafael Lalive, Rosario Macera, Imke Reimers, Kathryn Shaw, Joe Vecci, and especially David Autor for helpful comments. We also thank seminar and conference participants at NBER Labor Studies, NBER SI Personnel, ASSA, Berkeley-Padova-Paris Workshop, Bocconi, Cornell, CUHK, EIEF, Fudan, Gothenburg, Lausanne, Peking, Renmin, Rochester, Shanghai Jiao Tong, SUFE, SUNY Buffalo, Toronto, Toulouse, UCR, UCSB, Vrije Universiteit, Wisconsin, and Zurich for useful suggestions. We thank Xiaonan Zhu, Weiyue Zhang, Ruibo Sun, Yaqi Du, Yu Xu, Yibing Xia, and Sunyu Ye for excellent research assistance. We thank Schmidt Sciences, the National Natural Science Foundation of China [Grants 72673168, 72192803], and the National Social Science Fund of China [Grant 26ZDA030] for support. This project has been supported by the Gates Foundation (investment ID INV- 044136), through a subgrant from IDinsight’s Digital Economy Research Impact Initiative (DERII). We received ethics approval from the University of Toronto. One author has previously received compensation from the Platform for work unrelated to the research reported in this paper. The paper was reviewed to ensure that proprietary or confidential information is not revealed. We thank the Platform and its employees for their collaboration and for providing data. We thank delivery workers for participating in the qualitative data collection. Any opinions, findings, conclusions and recommendations expressed are those of the authors and do not necessarily represent the views of the Platform. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.