The Effects of Service Integration on Platform Competition and Welfare:
Evidence from Fulfillment by Amazon

Extended abstract: Digital platforms are extending their role beyond marketplace intermediation by integrating complementary services. Google and Apple, for instance, operate their own payment systems for app stores, and Google offers ad tech tools alongside its ad exchange. Amazon and Walmart also run logistics services that store and deliver sellers’ products, and Fulfillment by Amazon (FBA) has been central to Amazon becoming the largest logistics operator in the U.S. Such integration by large digital platforms raises concerns about competition and welfare, prompting global regulatory scrutiny (Epic v. Google, 2020; Epic v. Apple, 2020; EC 2022; CMA 2023; FTC v. Amazon, 2023; United States v. Google, 2023). In this paper, I investigate the welfare effects of a platform’s vertical integration of complementary services, focusing on e-commerce competition between Amazon and Walmart in the U.S.

To what extent does a platform’s integration benefit or harm consumers and sellers? While integration eliminates double marginalization arising from marketplace commissions and logistics fees, it also allows a platform to make its service exclusive to its own marketplace or more costly to use on a rival platform. Such foreclosure weakens cross-platform competition, with the magnitude of this effect depending on how much sellers’ adoption of integrated logistics raises their costs of multihoming (i.e., selling on both platforms). Integration therefore creates a welfare trade-off, and its net effect is theoretically ambiguous.

I build an equilibrium model of competition between Amazon and Walmart to measure the welfare effects of integrated logistics. The model captures the welfare trade-off: platforms set commission rates and logistics fees jointly, accounting for consumer choice and sellers’ multi-platform entry, logistics adoption, and pricing. I construct a unique dataset linking products and sellers across the two platforms, estimate the model, and document evidence of both pro- and anti-competitive effects of integration. The platforms hold logistics fees down substantially to increase commission profits, yet sellers who use FBA are much less likely to enter Walmart. The latter finding is further supported by a regression discontinuity design: Amazon classifies products into logistics fee tiers, and sellers facing discontinuously higher fees are more likely to drop FBA and sell on Walmart.

On net, counterfactual separation of FBA from Amazon would reduce consumer welfare by 19%. Although separation removes exclusivity and increases sellers’ multihoming by 92%, it introduces double marginalization and raises logistics fees and commission rates, increasing consumer prices by 25%. Aggregate seller profits change little (+3%), yet separation substantially reduces the profits of small and medium-sized sellers and increases earnings inequality among sellers. A counterfactual policy that removes exclusivity and discriminatory fees while preserving integration, such as making integrated logistics interoperable, would target the anti-competitive effects of integration directly and raise consumer welfare.