Trade creates winners and losers. In the context of the US-China trade war – the unprecedented tit-for-tat increase in tariffs by the US and China – this maxim suggests that changes in US and Chinese trade policy will redistribute the gains from trade. I argue that empirical evidence supports the theory – the trade war is inducing concentrated losses in consumption and employment for communities in the US that are the most exposed to Chinese retaliatory tariffs.
The costs of the trade war
The trade war affects welfare through two mechanisms. Trade benefits consumers through lower prices and increased variety. In the context of the US-China trade war, the expectation is that US tariffs on Chinese goods will impose hardship on all US consumers through higher prices and a reduction in variety. This prediction is becoming apparent as several studies (Fajgelbaum et al. 2019, Amiti et al. 2019, Cavallo et al. 2019) find that the US tariffs are leading to higher prices and a reduction in welfare for all consumers.
From a US worker’s perspective, there is a second hardship of the trade war. Retaliatory Chinese tariffs on US exports affect labour income and/or production opportunities for those directly impacted, such as the farmers or workers engaged in agriculture and manufacturing production that China targeted with retaliatory tariffs. Unlike price effects – which are spread widely across the US population – this ‘labour income channel’ is concentrated: those who had a position of comparative advantage for the Chinese market and lost it due to tariffs bear this burden of the trade war alone.
Measuring the consumption response to trade shocks
In a recent paper (Waugh 2019), I focus on this second hardship and measure the effect of retaliatory Chinese tariffs on county-level consumption and labour market outcomes. The research design is simple: I exploit variation in a county’s exposure to Chinese retaliatory tariffs between 2017 and 2018 and correlate it with changes in consumption and employment at the county level. The focus on Chinese retaliatory tariffs stems from a desire to measure a trade-induced change in labour income/production opportunities (e.g. soybean farmers in Iowa lose the ability to sell their product due to Chinese retaliation).
The focus on consumption is a unique aspect of my work. Most research focuses on trade-induced labour market outcomes (for example, the seminal work of Autor et al. 2013). From a welfare perspective, however, labour market outcomes may not reflect how economic welfare is allocated across those who are differentially impacted by trade. Moreover, abrupt changes in trade policy will have different welfare consequences, depending on the opportunities households have to adjust to these shocks.
Measuring consumption at the microeconomic level is difficult, which is one reason for the focus on labour market outcomes. My approach to measuring consumption is to use a unique dataset with the universe of new auto sales at the US county level at a monthly frequency. This dataset allows me to proxy consumption behaviour with purchases of an easily defined object, and variation at both a narrow geographic dimension and at high frequency. High frequency is important in this context due to the rapidly changing nature of trade policy during 2018.
The unequal burden of the trade war
Figure 1 illustrates that the burden of Chinese retaliation is concentrated. It plots the change in a county’s tariff (which is an employment weighted average of the tariff at the sectoral level) between December 2017 and December 2018. In this map, a county is coloured according to its position within the distribution across counties: red indicates that a county’s tariff increased significantly; blue indicates that a county’s tariff did not increase significantly.
Figure 1 Tariff exposure by county in the continental US (the ‘lower 48’)
Consistent with the notion that much of the Chinese tariff retaliation targeted agriculture commodities, much of the US Midwest and agriculture intensive areas of California, Oregon, and Washington are heavily exposed to Chinese retaliation. Rural counties also experienced larger tariff increases.1 Overall, I find that a county’s share of employment in goods-producing industries and a county’s rural population share accounts for 17% of the variation in the change in tariffs.
Those bearing the burden of the trade war suffered
Both visually and through formal econometric specifications, I find that changes in trade policy had large effects on consumption, with high-tariff counties experiencing between a 2 and 5.5 percentage point decline in new auto sales growth relative to low-tariff counties.
This simplest way to arrive at these conclusions is to compare auto sales growth in counties that had large increases in tariffs to those that had small increases in tariffs. Here, high versus low is a comparison of counties in the upper quartile of the change-in-tariff distribution to those in the lower quartile, as of December 2018 (i.e. the dark red areas of Figure 1 versus the blue areas).
Figure 2 US county-level auto sales and Chinese retaliatory tariffs
Figure 2 plots this comparison between January 2018 and January 2019. Dashed vertical lines (with annotation) indicate important events during the trade war. Units on the vertical axis are in log points, so an interpretation of the value of 0.01 is a 1 percentage point difference in annual growth rates. Prior to the implementation of tariffs in July 2018, Figure 2 shows no difference in auto sales growth between high- and low-tariff counties. A difference immediately emerges after the implementation of the first round of tariffs in July 2018. For the second half of 2018, high-tariff counties grew more slowly than low-tariff counties. The magnitude is large, with a 2 percentage point difference. In other words, consumption in high-tariff counties grew two percentage points slower after the implementation of the tariffs.
In formal econometric specifications, I find even larger results. Depending on controls and inclusion of various fixed effects, I find that the elasticity of consumption growth to tariffs can be as large as −1.4 (i.e. a 1 percentage point increase in a county’s exposure to Chinese retaliatory tariffs leads to a 1.4 percentage point decrease in auto sales growth). This translates to a 5.5 percentage point decline in auto sales growth for counties in the upper quartile of the tariff distribution.
From an aggregate perspective, some simple tabulations suggest that the impact from auto sales alone are in the same ballpark relative to estimates of other effects from the trade war. My calculations show that the trade war caused $9.3 billion in lost auto sales. This number is as large as the aggregate effects ($7.8 billion) found by Fajgelbaum et al. (2019) with an important distinction: the loss that I measure is concentrated. It is the few communities who are exposed to retaliatory tariffs who are bearing this burden.
These consumption losses are connected with negative labour market outcomes. Using the monthly data from the BLS’s Quarterly Census of Employment and Wages, I find that total employment growth declines by one percentage point for counties in the upper quartile of the tariff distribution. More intriguing is the fact that non-tradable employment (e.g. employment in restaurants, retail, and services) declines by a similar magnitude. The decline in non-tradable employment suggests that through county-level equilibrium effects, consumption and local demand conditions are softening for the communities most affected by Chinese tariffs. Because non-tradable employment and, in turn, non-tradable consumption is declining, this finding also suggests that the aggregate consumption response is substantially larger than the effects discussed above.
Why? What now?
Overall, the employment effects connect well with the reductions in consumption. That is, counties that were more exposed to Chinese retaliatory tariffs experienced a reduced ability to export, which fed into the labour market (both tradable and non-tradable), and this mechanism reduced consumption.
Questions remain as to why this is so. In my results, there is suggestive evidence that something beyond the labour market is behind the fall in consumption. The role of expectations and uncertainty are primary culprits. To tease this out, a formal economic model is probably needed, one that takes into account the durable nature of consumption in the data I am using and can examine the idea that expectations play an important role.
A more pressing question is: What is going on now? The escalating US–China trade war makes understanding its effects particularly important in the context of the current economic environment in the US, which includes slowing economic growth and recessionary concerns. These results have policy implications for short-run demand management policy in the US and the appropriate response to the trade war. Conventional wisdom sees the trade war as a negative aggregate supply shock with declines in output and inflationary pressure. In contrast, the trade-war-induced declines in consumption that I am finding suggest that important demand-side effects from the trade war warrant consideration in the formulation of US policy.
Amiti, M, S J Redding, and D Weinstein (2019), “The Impact of the 2018 Trade War on U.S. Prices and Welfare”, NBER Working Paper 25672.
Autor, D, D Dorn, and G H Hanson (2013), “The China Syndrome: Local Labor Market Effects of Import Competition in the United States”, The American Economic Review 103: 2121–2168.
Cavallo, A, G Gopinath, B Neiman, and J Tang (2019), “Tariff Passthrough at the Border and at the Store: Evidence from US Trade Policy”, NBER Working Paper 26396.
Fajgelbaum, P D, P K Goldberg, P J Kennedy, and A K Khandelwal (2019), “The Return to Protectionism”, NBER Working Paper 25638.
Waugh, M E (2019), “The Consumption Response to Trade Shocks: Evidence from the US-China Trade War”, NBER Working Paper 26353.
 Casual observation also suggests that counties facing the most retaliation are the counties that voted in larger numbers for President Trump in the 2016 election. Though a formal regression analysis confirms that relationship, most of the variation in tariff retaliation is simply explained by the export exposure of a county to China in the past.