We evaluate, empirically, the effect of changes in trade policy during the 2018-19 trade war on U.S. economic activity. We begin by documenting that sectors and states across the United States are heterogeneous in their exposure to international trade. To do that, we construct a measure of exposure that combines the share of a sector's gross output that is accounted for by trade with the pattern of comparative advantage of each state in that sector. We then exploit cross-state heterogeneity in exposure to international trade and correlate it with measures of economic activity across U.S. states. Our findings suggest that states that were very exposed to trade at the onset of the trade war experienced worse outcomes in terms of employment and output growth. Our analysis is not aimed at concluding any causality effects, but instead focuses on correlations.
The U.S. trade policy that started in 2017 has spiraled into a return to protectionism. The trade war began with recommendations from the U.S. administration to implement tariffs on specific goods (e.g., solar panels, washing machines, and steel and aluminum) in an attempt to protect national security and mitigate negative impacts faced by domestic producers of such products. Tariffs on these goods were enacted in 2018 and were followed by retaliatory tariffs from China, along with retaliatory tariffs from the European Union (EU) and Canada on steel, aluminum, and agricultural products. The United States has continued to levy tariffs onto major trading partners such as Canada, Mexico, the EU, and China. Perhaps the most noteworthy battle in the tariff front exists between the United States and China, given the growing tension between the two countries as the trade war progresses.
In this article, we exploit variation in sector-state exposure to international trade and evaluate, empirically, the effect of trade policy on U.S. output growth and employment. Recent papers have explored the effect of the trade war on the United States through the lens of a quantitative model (see Caldara et al., 2019; Fajgelbaum et al., 2019; and Santacreu, Sposi, and Zhang, 2019). However, empirical studies have been rather limited. We try to fill that gap by providing an empirical analysis of the effects of a trade war on the United States.
We start by constructing a measure of initial sectoral exposure to international trade, from the perspective of both imports and exports. We focus on trade in intermediate goods and abstract from final goods, as tariff announcements have fallen mainly onto intermediate goods. We use data for 19 sectors from the World Input-Output Database (WIOD) 2016 Release for 2014—the latest year available in the WIOD, prior to the trade war. We then compute, for each sector, a measure of trade exposure as the ratio of U.S. total trade of intermediate products with the world to U.S. gross output in that sector. To the extent that U.S. states specialize in the production of different sectors, exposure to trade policy will likely be heterogeneous across states. To capture this idea, we combine our measure of sectoral trade exposure with the production composition of each U.S. state—calculated as the share of value added of each sector in that state—and compute a measure of trade exposure at the state level. It is important to note that this measure of trade exposure does not rely on actual tariff data. We interpret it as a sector's overall trade exposure that depends only on data of traded intermediate goods by sector, regardless of actual trade policy.
We find that the sectors most exposed to imports—those that rely more heavily on imported intermediate products—are Coke and petroleum and Motor vehicles and trailers. Mining and quarrying and Forestry, fishing, and logging are the sectors least exposed. With respect to exports, the sectors most exposed—those that export more intermediate goods to the world—are Other transport equipment and Computer, electronic, and optical, and the sectors least exposed are Mining and quarrying and Other non-metallic mineral products. Louisiana and Michigan are very exposed to imports from the world. Louisiana specializes in Coke and petroleum, and Michigan specializes in Motor vehicles and trailers. Alaska and North Dakota are the states least exposed to imports, as they both specialize in Mining and quarrying. In the case of exports, Washington and California are very exposed, whereas Alaska and Wyoming are among the states least exposed.
We then follow an approach similar to Mian and Sufi (2009) to exploit the observed cross-state heterogeneity in exposure to international trade and correlate it with measures of economic activity across U.S. states. We focus on quarterly growth rates of employment and output between 2018:Q1 and 2019:Q1. We find the following: (i) There is a negative and significant correlation between the initial exposure to trade and economic activity; (ii) the negative correlation is stronger with employment growth than it is with output growth; and (iii) the negative correlation is stronger with import exposure than it is with export exposure. That is, those states more exposed to trade experienced lower increases or even decreases in output growth and employment growth between 2018 and 2019. These findings reflect that firms operating in states very exposed to trade adjusted their employment and production decisions after announcements of tariff increases. The adjustments were stronger in terms of employment than in terms of output, and they were stronger in states very exposed to U.S. tariffs than in states very exposed to retaliatory tariffs. Obviously, there are economic forces other than trade exposure that could be driving heterogeneous adjustments in employment and production across U.S. states. Our analysis is not aimed at concluding any causality effects. However, the strong negative correlation found in the data suggests that initial trade exposure could have played a role in those adjustments.
Finally, we analyze the effect of the 2018-19 U.S.-China trade war on U.S. economic activity. We construct exposure measures that use actual U.S. tariffs imposed on China and subsequent retaliatory Chinese tariffs. The measure of import exposure exploits the input-output structure of the economy, as it captures that those sectors that depend more on imports from sectors with larger increases of U.S. tariffs are more exposed to imports from China. These are Motor vehicles and trailers and Machinery and equipment, n.e.c. In the case of export exposure, sectors that the United States exports most or that are subject to higher retaliatory Chinese tariffs will be more exposed to trade with China. These are Other transport equipment and Agriculture. Heterogeneity in sector exposure translates into heterogeneity in state exposure, as states specialize in the production of different sectors. Michigan is very exposed to imports from China, whereas Washington is very exposed to exports. As before, we find a negative correlation—albeit a weaker one than in the case of exposure to the world—between our measures of trade exposure to China and both output growth and employment growth. In contrast to our previous results, the negative correlation is stronger with output growth than it is with employment growth, and it is stronger in the case of retaliatory tariffs.
Our results suggest that cross-state heterogeneity in trade exposure correlates negatively with U.S. economic activity. Although we cannot claim any causality effects, these findings are an indication that the trade war initiated by the United States may have had a stronger impact on U.S. employment and production than what is found through the lenses of standard models of trade. Accounting for this heterogeneity is thus key in capturing the negative impacts.
Our article is related to a very recent strand of literature on the effects of trade policy on the United States. Actual tariff increases, followed by the threat of future raises, have increased uncertainty in the tradable sector. According to several studies, higher uncertainty has already had a negative impact on the United States in terms of investment and production (see Caldara et al., 2019; Handley and Limão, 2015 and 2017; and Bloom, Bond, and Van Reenen, 2007). Recent studies using quantitative models of trade have found small aggregate effects of tariffs on the U.S. economy. These studies, however, do not take into account the effect of uncertainty and focus their analysis on actual tariff increases. Despite small aggregate effects, these models find highly heterogeneous effects of tariffs across both sectors and states (see Fajgelbaum et al., 2019; Amiti, Redding, and Weinstein, 2019; Auer, Bonadio, and Levchenko, 2018; and Santacreu, Sposi, and Zhang, 2019), which could potentially have heterogeneous effects on economic activity across U.S. states. Our article differs from those studies in that we explore empirically—rather than through the lens of a quantitative model—whether heterogeneous exposure to trade had an impact on economic outcomes.
Read the full article: https://files.stlouisfed.org/f...