Research question The world trading system has changed considerably over the past few years. The United States has imposed several protectionist measures that have prompted threats of retaliation. These developments have spurred new research into the aggregate effects of trade protectionism. Yet surprisingly little is known about the macroeconomic effects of U.S. trade protection on the United States’ major trading partners. This paper seeks to fill that gap with new empirical evidence on the international transmission mechanism of U.S. trade policy. Contribution The paper contributes to a growing literature that bridges the gap between data at the microeconomic and macroeconomic level. Isolating the surprise component of protectionist policy actions from confounding factors poses several challenges. First, protectionist trade policies may be endogenous to macroeconomic factors, making it difficult to draw causal inference. Second, substantial time delays occur between the announcement of trade-protective measures and their actual implementation, which gives rise to anticipation effects. I overcome these identification issues by developing a new measure of U.S. trade policy “announcement shocks”, using micro-level data on U.S. anti-dumping, countervailing duties, and safeguards for the period 1988-2015. The measure is cleansed of macroeconomic sources of endogeneity. Results Estimates using the new measure indicate that U.S. trade policy announcement shocks give rise to contractions in major trading partners’ output and investment. Economic expectations play an important part in the transmission of an announcement shock to the real economy. Evidence shows that the effects of the shock are substantially weaker when endogenous movements in expectations following the shock are not accounted for. An analysis that quantifies the extent of newspaper coverage of U.S. trade protection suggests that media attention facilitates the propagation of protectionist shocks, consistent with an expectations channel. Note finally that the results should be interpreted with caution when considering the post-2015 events because the sample does not include U.S. trade protection measures of the Trump administration.