The welfare effects of trade policy are shaped by the outcomes of imports re- allocation and price changes. In this paper, I show that these outcomes crucially depend on whether importing firms are matched with multinational suppliers or with single-country producers. I study an antidumping duty imposed by Colom- bia on the imports of Chinese truck tires. In the data I observe the full network of Colombian importers and global exporters, some of which are multi-origin and some of which are single-origin. Due to the policy, approximately 75% of imports of tires originated in China were reallocated into other origins, and the bulk of this realloca- tion was undertaken by importers matched with multi-origin exporters. I estimate a quantitative trade framework to match the reallocation and price changes in the data. I find that under a counterfactual network without multinational suppliers, the increase in the price index from the policy would have been almost twice the size of the actual change. The analysis suggests that ignoring the network structure could lead to sizable biases for the prediction of the welfare effects of tariffs.