The Effects of U.S. Trade War Tariff Exemptions on U.S. Trade with China and Asia

Following a U.S. Section 301 investigation into China's acts, policies, and practices related to technology transfer, intellectual property, and innovation, the Trump administration acted against China in April 2018 by announcing its first round of punitive tariffs on Chinese imports. Because China responded to this move, and all subsequent U.S. tariffs, with immediate retaliatory actions of its own, U.S. and Chinese tariffs quickly rose in lock-step with each other until a phase 1 agreement signed on 15 January 2020 paused these tit-for-tat trade actions.¹ Bown's (2023) chronicle of the tariff escalation process documents the dramatic tariff increases that the United States and China levied against each other, as well as the broad coverage of imports that were caught in the fray. When tariff rises ceased, China's new tariffs hit 58.3 percent of U.S. exports to China, while the trade war tariffs introduced by the United States affected 66.4 percent of the value of U.S. imports from China.² Beyond tariffs, a further effect of the trade war was a dramatic increase in trade policy uncertainty. Notably, Handley and Limão's (2022) measure of U.S. trade policy uncertainty reveals that the spike in trade policy uncertainty during the trade war overwhelmingly exceeded the uncertainty effects of any other trade shocks or events going back to the measure's inception in 1960.

Within this setting, the U.S. Trade Representative (USTR) provided some firms with an avenue to seek relief from trade war duties via exemption petitions. If successful, importers of the products that were granted exemptions received one-year relief from the trade war duties. In addition, as these exemptions expired, beneficiary firms were given the opportunity to petition for continued relief. In practice, only a small share of petitions was granted relief in the initial round of exemption requests, and an even smaller group of firms received ongoing relief via further extensions.³ Nonetheless, the operation of this exemption process provides a unique opportunity to study the economic effects of tariffs and trade policy uncertainty. Further, because the operation of this exemption application process provides insight into the characteristics of products that received relief, this paper provides information on the operation of this trade war tariff exemption policy.

The first analysis in this paper focuses on the application for and granting of U.S. trade war tariff exemptions at the product level. This analysis uncovers two key regularities. First, the receipt of relief appears to be consistent with the stated exemption criteria, which sought to limit the granting of exemptions to firms that did not have good sourcing replacement alternatives in the United States or in third countries. While trade data do not definitively pin down the presence or absence of import sourcing alternatives, the existence of prior import flows ASEAN countries and Mexico provide an indication of availability and capability of third-country providers. Indeed, more widespread availability of ASEAN or Mexican options were associated with diminished success of exemption petitions. The second feature that emerges from this analysis is the heterogeneity of responses by product type (intermediate, capital and final goods). Here too, the differential granting of exemptions was aligned with expectations related to the relative sourcing flexibility for each of these product categories.

While tariff exemptions weren't ubiquitous, they noticeably reduced the level of U.S. effective tariffs on Chinese imports. Flaaen, Langemeier, and Pierce (2021) concluded that exemptions led to a 1.8 percentage point gap between effective and statutory tariff rates in the United States.⁴ Thus, the second major question of this paper is: How did exemptions affect subsequent U.S. import patterns from China, both in value and relative to other countries? On this question the results show that sourcing from China declined dramatically not only due to trade war tariffs, but due to the trade policy uncertainty caused by the handling of exemptions, and the effects of being included on trade war product lists. Here too, the magnitude of tariff and policy sensitivity differed by goods type.

In March 2018 the USTR released its report on “China's Acts, Policies, and Practices Related to Technology Transfer, Intellectual Property, and Innovation.”⁵ As noted by Swenson and Woo (2019), prior U.S. support of freer trade and increased international integration were facing headwinds created by a shifting political environment that was much less supportive of globalization and its associated policies. Rather than taking its concerns and evidence to the WTO, on 6 July 2018, the United States launched its first round of 25 percent tariffs on US$ 34 billion of its imports from China. On the same day, China skipped the WTO-prescribed channels for complaint, and launched its own set of 25 percent tariffs on US$ 34 billion of products it imported from the United States. A similar expansion of the trade war followed on 23 August 2018, as the United States set 25 percent tariffs on a second list of US$ 16 billion its imports from China, and was matched on the same day as China imposed 25 percent tariffs on an additional US$ 16 billion of its U.S.-origin imports. Tit-for-tat trade war trade escalation continued until a phase 1 agreement went into effect in 2020.⁶

All Road wheel products encompassed in HS8 8708.90.45 were included on the U.S. third round Section 301 tariff list. If the firms did nothing to contest their inclusion on this list, these products faced 10 percent U.S. import tariffs starting in September 2018, which were raised to 25 percent in May 2019 as the United States sought to increase pressure on China. However, individual firms could request exemptions for products within these listings at the HS10 product level. For example, any firm importing Road wheels of aluminum, or subcategory HS10 8708.90.4560, could petition for relief. If that firm was successful in its petition, the tariff exemption was made available to any other firms who imported the same product as the petitioner.

Ultimately, 52,746 exemption petitions were filed in response to the U.S. trade war tariff waves. As illustrated by a sample of exemption filings displayed in  Appendix A, individual firms could make several filings for exemptions on a range of imported items that they viewed as important to their business. Notably, although trade war tariffs affected 10,694 different HS8 product groups, exemption request filings emerged in only 3,058 of these HS8 groups. Within HS8 product groups that involved at least one exemption request, the average number of cases was 17.25, while the median number of filings in each HS8 group that had one or more filings was three cases.

Firms were required to submit individual applications for each product on which they were seeking a trade war tariff exemption. The guidelines for Section 301 tariff exemptions set three criteria for exemption: (1) whether the product was only available from China, and not from the United States or third countries, (2) whether duties would significantly harm the claimant or the United States, and (3) whether the product in question was part of China's China 2025 program.⁸ In making their requests, firms were instructed to document the quantity and value of their previous three years’ purchases of the product from China.

A summary of firm-product exemption requests is provided in Table 1. Notably, firm filing decisions varied dramatically across HS10 product groups. Overall, panel (A) shows that, of the 52,746 firm petitions, only 12.9 percent were granted relief.

Panel (B) in Table 1 reveals that product-level heterogeneity was a major feature of exemption filing behavior. Among the ten HS10 codes that attracted the largest number of firm petition filings, the number of individual firm exemption requests ranged from 447 to 3,656 filings. Notably, the number of filers does not appear to be strongly related to the case approval rate. Among the products with the largest number of filing firms, the approval rate was highest for HS10 8413.91.9080 (Parts of Air Pumps, Others), with a success rate of 32.8 percent. The next highest approval rate in the top ten most intensively filed HS10 products involved HS10, 7326.90.8688, or Articles of Iron or Steel, Others, where firms had a 4.3 percent success rate. The third highest rate of 2.1 percent applied to HS10 8408.99.6890, titled “Parts of Motor Vehicles, Others, Of Headings 8701 To 8705: Other Parts for Power Trains.” The remainder of the ten most-intensively filed HS10 products had zero to very low approval rates.

Firms that received trade war tariff exemptions also had the opportunity to request an extension of the protection they received. This culminated in a one-year extension that was granted to 549 products. This decision was met with disappointment, as expressed by the U.S.–China Business Council's encouragement of the “USTR to articulate a clear process and criteria to evaluate requests for tariff exclusions.”⁹ This statement decried the USTR's opaqueness, in light of its failure to provide the rationale for its decisions, and for the absence of opportunities for other products to seek tariff redress. Later yet, in 2023, firm expectations were still uncertain, in part due to conflicting statements from different branches of government about the desirability of ongoing exemptions and the U.S. trade stance toward China more generally.¹⁰ Although the four-year review of trade war tariffs was to be concluded by the end of 2023, In January 2024, Ambassador Katherine Tai communicated to China that she expected the four-year review to be completed within the “next few months.”¹¹

Given the criteria for exemptions, which included the availability of alternative sourcing options outside of China, Table 2 provides information about the sourcing of products that were and were not on the original Section 301 tariff lists. Further, for products that were included on the trade war Section 301 tariff lists, Table 2 documents the prior HS10 product sourcing composition, as it related to exemption filing propensities, and the success rates for cases that were filed. Although the USITC did not publish information about the reasoning and assessments they applied in their determination of exemption filings, the stated criteria noted that successful filers would need to explain why they needed to source from China, as well as their efforts to source from the United States or third countries.

To provide information about the relative availability of alternative sourcing option at the HS10 product level, this paper takes the view that the availability of alternative sourcing opportunities is revealed, at least in part, by information on aggregate U.S. sourcing decisions prior to the start of the trade war. For example, if a large share of imports in a particular HS10 product group were already purchased by U.S. buyers from ASEAN sources, this suggests that ASEAN suppliers could provide an alternative source of supply for U.S. buyers. As the USTR determined whether suitable alternatives were available, they presumably considered comparable trade options in manufacturing-oriented countries with relatively low cost of labor. For this purpose, this paper views prior HS10 product imports from ASEAN and Mexico as a strong indicator of the feasibility of such a shift. To implement this idea, the data analysis includes variables on the ASEAN-share and Mexico-share of individual HS10 products, which measures the share of U.S. imports within an HS10 product group that were purchased from ASEAN member countries or Mexico, respectively. Unless otherwise noted, this share refers to the import share in 2017, which preceded the imposition of Section 301 tariffs, which were first introduced in 2018.

The presence or absence of domestic, U.S.-based, alternatives to the Chinese supply of specific HS10 products is harder to evaluate since U.S. production data are not reported with product or industry classifiers that correspond to the HS10 product identifiers that are used to record international trade transactions. However, from an economic perspective, the availability of U.S. options prior to the imposition of Section 301 tariffs would be based on the presence or absence of U.S. comparative advantage. Thus, given the economic similarities between the United States and the European Union in terms of GDP, population, and development, this project uses the EU share, or EU28-share (which represents the country membership of the EU at the time of the trade war, and prior to the U.K.’s Brexit departure in January 2020), of HS10 product-level trade as a proxy for the potential strength or capability of high-income and high-skill U.S. exports of the HS10 products that were subject to potential trade war tariffs. As with the measures of the ASEAN- and Mexico-shares, this variable is also constructed based on data from 2017.

Table 2 provides summary evidence on firm exemption filing behavior and on the characteristics of products that received tariff relief. Notably, there were strong differences between the products that were included on U.S. trade war tariff lists versus those that were not. While products that were named on the Section 301 lists involved HS10 products where the average China-share was 28.7 percent, the China-share of products that did not appear on the lists was 58.2 percent. More generally, the relative availability of alternatives also appears to be correlated with the inclusion of products on the Section 301 trade war tariff lists since the availability of alternatives was lower for those products that were not on the trade war tariff lists, as those product groups had lower ASEAN, Mexico, or EU28 import shares. For example, products that were included on a Section 301 tariff list had an average 2017 ASEAN-share of 10.5 percent, while those that were off the list involved a 6.9 percent ASEAN-share.

Table 2 also reveals that the filing of exemption requests was related to prior sourcing patterns. First, while all named 301 tariff products had an average 28.7 percent reliance on China, firms that took the effort to file an exemption claim were more reliant on China, averaging a 35.9 percent China-share. Similarly, while the average product on the Section 301 product tariff lists involved products where the ASEAN-share was 10.5 percent, the average filer made claims in HS10 products that were typified by lower levels of 2017 sourcing, as represented by an 8.9 percent ASEAN-share. This characterization of sourcing in 2017 also applies to sourcing from Mexico and the EU28, as the HS10 products that attracted exemption request cases had lower 2017 levels of Mexico-share and EU28-shares of U.S. sourcing than the average HS10 product that was placed on the Section 301 tariff lists.

Although the USITC did not disclose the basis for its exemption decisions, the data on Table 2 suggest that they were broadly consistent with the named criteria. First, compared with the average product on the Section 301 lists, the HS10 products that were granted exemptions involved higher China-shares in U.S. 2017 sourcing. This appears to indicate that these were cases where China was more central to firm profitability. Similarly, the data show that the levels of ASEAN- and EU28-shares were lower in the granted list than on the entire list of named products in the Section 301 trade war tariff plans. The fact that ASEAN and EU28 sources of supply were smaller in the successful exemption cases suggests that these firms used this information to support their claim that switching away from China would not be feasible, or easily feasible in the short run. The one aberration in these summary statistics is relative value of Mexico-share, which was higher in the case of successful exemptions than it was on the named Section 301 lists.

To provide a more detailed view on the sourcing of products at the time of the trade war, Table 3 summarizes the sourcing differences for goods that were subject to trade war tariffs disaggregated by product type: capital, intermediate and final goods. Comparison of the top two data blocks in this table shows that the selection of Section 301 tariff items was strongly related to sourcing reliance on China. In each product category products named to the Section 301 tariff lists were only half as reliant on China as an import source. Comparing named, with unnamed, products, the sourcing reliance was 37.5 percent versus 72.3 percent in the case of final goods, 22 percent versus 49 percent in capital goods, and 24.7 percent versus 43.9 percent in the case of intermediate goods imports. Similarly, for all three product categories, the inclusion on the trade war tariff lists was higher in HS10 products where the Mexico-share was higher, presumably due to the availability of alternative sources that would mitigate the impact of tariffs. However, in the case of ASEAN and EU28 shares, the evidence is mixed, as the heavier presence of alternatives in trade war HS10 products is only apparent in three of the six comparisons.

Table 3 also uncovers differences in the sourcing characteristics of HS10 products that were on Section 301 trade war tariff lists, related to the application for tariff exemptions, and in the success rate of the applications. First, by comparing the second and third blocks of the table, it is possible to view sourcing differences for products that were included on the tariff lists, and the products that sought to gain exemptions. Similar with the results for all products, for capital, intermediate and final goods the table shows that HS10 products that sought exemptions, involved imports where the China-share was notably higher. Further, in terms of alternatives, the fact that applicant HS10 products were all in sectors where the EU28- and Mexico-shares were lower suggests that firms were reserving their efforts to cases where replacement of Chinese import sources would be more difficult.

The bottom data block in Table 3 shows the sourcing characteristics of HS10 products that received exemption relief. For all goods—final, intermediate and capital—the rates of success involved applicant products that had higher China-shares and lower ASEAN-shares, both signifying the relative challenges in re-orienting this sourcing away from China.

The first regression in Table 4, displayed in panel (A), seeks to understand filing decisions at the HS10 product level. For this purpose, the data sample is restricted to all products that were contained on HS8 trade war tariff lists, which were the basis for Section 301 trade war actions. This is done since any products that were not covered by one of the trade war lists had no need to seek an exemption as a means of avoiding the new tariffs. In this panel, the dependent variable Product Exemption Challenges is set to 1 for all HS10 products that were on the trade war tariff lists, and had at least one firm filer, and 0 for the HS10 products that were on the trade war tariff lists but did not attract any exemption request filings. The results in panel (A) provide three messages. First, the application for a tariff exemption was higher for HS10 products that had a higher reliance on China as a source of imports, as shown by the positive and significant coefficients on the China-share variables. Second, economic factors also affected the rate of challenges. On this dimension, the results in the second and third columns show that filings were more likely for HS10 products for which the overall U.S. import value from the world, in 2017, was higher. In addition, the coefficient on Imp Growth Rate, which measures the percentage change in U.S. imports of the product between 2017 and 2023, is also positive and significant. While firms in 2018 and 2019 would not have known the realization of this variable, they may have had information about the products that were likely to become more or less important over time. If so, this positive coefficient on this variable suggests that firms were more likely to seek an exemption if they viewed the HS10 product as one that would become increasingly important in the future. Finally, panel (A) indicates that filing behavior was somewhat related to current sourcing decisions, though these results decline in significance as the trade measures are added. Considering the first column of panel (A), it is interesting to see that filing rates were depressed if the EU28-share was larger in value. However, if sourcing options in the United States and EU28 are similar, it is possible that larger EU28 shares indicated that the HS10 product had sources of comparative advantage in EU28, that will also similarly shape U.S. production capabilities and opportunities. If so, since the presence of U.S. domestic sourcing options would reduce the likelihood of gaining an exemption, it may have deterred firms from applying in these cases.

To evaluate robustness of this initial result, the regressions in panel (A) were re-estimated using probit. In this setting, the coefficient signs and significance are very similar to the results shown in panel (A), though the coefficient magnitudes differ. However, though this is a comforting result, the paper does not use probit for the analysis, since re-estimation of equation (1) leads to the loss of almost one-third of the original sample due to the already-discussed absence of any challenges in many product groups.

Panels (B), (C), and (D) of Table 4 relate the outcomes of exemption approval cases to the same set of regressors and specification given by equation (1). In panel (B), this begins with exploration of Some-Success(es) as the new dependent variable. In this setting, the set of HS10 products used in this regression is limited to products that had at least one exemption filer. Within this group, the dependent variable $yi$ is set to 0 if all the cases were unsuccessful, and to 1 if at least one of the cases in the group received a tariff exemption. Panel (B) has four consistent results. First, the frequency of some-success was positively and significantly related to China's share of U.S. imports in a HS10 product category, consistent with the USTR's intent to evaluate the importance of China as a source for individual firm importers. Second, the frequency of some-success was negatively and mostly significantly related to ASEAN's share of U.S. imports in a HS10 product category. This, too, was consistent with the USTR's stated metrics for approval; if ASEAN was already a large supplier of a product, it would be difficult for firms to claim they needed an exemption due to the absence of third country alternatives. Similarly, the coefficients on the Mexico and EU28 variables were negative, but not statistically significant. Third, economic size also affected the presence of approvals, as approvals were more likely in product groups for which the overall value of U.S. imports from the world were larger. Finally, the likelihood of approval was also related to the number of applicants for relief. The larger the number of HS10 cases filed (# Cases Filed), the larger the odds of approval. However, since this coefficient is small, it appears that this statistically significant result was not an economically significant determinant of filing success.

In panel (C) of Table 4, the new dependent variable, HS10 Product Exemption, reports the rate of approval for exemption requests within an HS10 product code. The key message of this set of results is that the rate of relief was strongly related to the USTR stated interest in the availability of alternatives. In all three specifications, the approval rate is shown to be negatively associated with import shares from all three sources, and statistically so in the case of 2017 ASEAN and EU28-shares. As before, this result suggests that more cases were approved in cases where there were most likely fewer alternatives in third countries, or taking EU28 capability as a proxy for U.S. capability, as an indication that case success was lower if domestic options were available.

For a final view of the approval of exemption requests, panel (D) of Table 4 looks at the approval in the first round of exemption extension requests, which led to an additional year of relief. Here, the sample is limited to products that received prior exemptions. The main takeaway is that extended exemptions were most likely to be granted to products that had relatively high China-shares in 2017 and had larger world import values.

Given the wide differences by product type in filing behavior, the original data are broken into three groups based on product type (intermediate, capital, and final goods) to search for differential responses. Table 5 contains the results from repeating the application of estimating specification (1) individually for each of these subgroups. Viewing panel (A) makes it clear that filing decisions are similar for each of the three product groups. Filing rates are higher for HS10 products where the value of world imports is higher, and where the share sourced from China is larger. Notably, the effect of China-share is largest for capital goods and followed by intermediate goods and final goods. Such an order is consistent with firm choices based on the degree of sourcing flexibility. If capital goods are tailored to purchasers, substitution toward non-Chinese suppliers will be particularly difficult. In contrast, many final goods may be produced by multiple suppliers located in many countries, which would explain the smaller coefficient on China-share in the Table 5 exemption request estimates.¹³

Next, panel (B) of Table 5 evaluates product exemption success(es) for each of the product types. Within each of these subgroups the value of U.S. HS10 imports from the world exerts a positive effect, though this effect is more than twice as large for intermediate and capital goods as it is for final goods. In addition, while the measures of alternative sourcing options all have negative coefficients, the only case that is statistically significant is the estimate of the negative relationship of ASEAN-share in final good subgroup. Taken together, these regression results suggest that the USTR was making evaluations that considered the presence of alternative sourcing options. Further, if final goods sourcing has the greatest sourcing flexibility, this could explain why ASEAN-share options had a statically identified effect on depressing exemption approvals in the case of final goods, while the negative coefficients in the other sectors were weaker. For example, the case of capital goods, none of the share variables is statistically identified. However, if capital goods have limited sourcing flexibility, this would show why broad measures are inadequate to distinguish capital goods sectors with reasonable options as compared with those with the least flexibility.

The important variables are the policy treatment variables, $Zi$’s, for the HS10 products. These include a dummy variable for products that were included on the Section 301 trade war tariff lists, and whether the HS10 product managed to receive any tariff exceptions, and exception extensions for HS10 products that received exemptions. The estimating specification also includes a measure of the final trade war tariff, $Ln(1+TradeWarTariff)$⁠.¹⁵

As before, the estimation regressions include fixed effects for HS04 product groups, or $γg$⁠. These variables are intended to capture differences in China's comparative advantage at the industry/sector level that would affect U.S. firm's affinity for sourcing from China. Finally, the specification includes a normal independent and identically distributed error term, $ɛi$⁠.

Several regularities emerge in the first two columns of Table 6. First, the degree of persistence, as measured by the coefficient on China-share2017 is roughly 0.5. More notably, the coefficient on $Ln(ImportGrowth)$ is negative, which indicates that U.S. sourcing was moving away from China over the six-year interval.

However, the more striking effects are evident in the policy treatment variables. Here, inclusion on the trade war tariff lists was associated with a 13.8 percent decline in product China-shares by 2023 compared with other U.S. imports that were not placed on any of the trade war lists. Products that managed to secure inclusion on a first-year exemption list only managed to reverse a mere 2.2 percent of that decline, while products that made it to the subsequent exemption extension list made no statistically identifiable gains. Thus, exemptions had a limited effect on U.S. sourcing of imports. When Table 6 regressions are repeated for robustness over the shorter window of 2017 to 2021, the estimated offsetting effects of exemptions are no larger than those reported over the full estimation window.

While the U.S.-China trade war had significant impacts on U.S. and global trade, assessment of trade impacts from 2020 and on, must also grapple with similarly disruptive effects of the COVID-19 crisis. To this end, two key sets of robustness checks were run. First, while trade policy uncertainty effects are likely to unfold over a longer time horizon, the regressions in Table 6 were rerun using the 2021 value of China-share, rather than the 2023 value (China-share2023). This avoids the effects of changes in China's supply capabilities, among other effects. If the shorter interval is used, the only major change in coefficients involves the coefficient on China-share2017, which is generally around 0.60, rather than 0.49 as in Table 6. Otherwise, the policy variables are almost identical. This suggests that the shift away from China was already under way in 2021, though the changes were not as strong as they would be by 2023.

For a second robustness check related to COVID-19, the Table 6 regressions were also run as before, but with exclusion of medical items. This exclusion has two motivations. First, some Section 301 tariff items received later exemptions during COVID-19 pandemic due to their necessity for combatting the crisis. Second, it is reasonable to believe that U.S. importers of these products may have increased their import volumes, based on elevated expectations of future tariff relief. To implement this, the 101 HS10 products that were in HS2 30 (pharmaceutical products), the 42 HS10 products that were in HS4 9018 (instruments and appliances used in medical, surgical, dental or veterinary sciences), the 33 HS10 products in HS4 6307 (textile product area that included surgical drapes and items and face masks), the 10 HS10 products that were in HS4 9019 (range of items including respiration apparatus), and the three products that were in HS4 9020 (range of items that including breathing apparatus) are dropped from the sample. In this case, re-estimation of the Table 6 specification while excluding medical imports has almost no effect on the estimated responses to import growth, or the policy implementation of the trade war.

Since the results in Table 6 suggest that the damaging effects of trade policy action were not limited to tariff effects but were also affected by the treatment of products via inclusion on the lists or through exemptions, a second approach is taken to consider the effects of trade policy uncertainty. As shown by Dixit and Pindyck (1994), increases in uncertainty hinder investment and action. Based on the trade literature, new trade connections are believed to involve fixed costs, whether it is the fixed costs of distributing the product in a new country location, or the fixed costs of identifying a new trade partner. Since these fixed costs are related to the expansion of sourcing from China, it is important to ask whether there were differential effects related to positive U.S. import growth versus negative U.S. import growth, as the fixed costs would only apply to the former.

To put these results in perspective, consider the first column in Table 7, which provides estimates based on the subsample of HS10 Intermediate Goods imports. In this setting, the coefficient on U.S. import growth (⁠$-$⁠.011) is much smaller than the coefficient on U.S. import decreases (⁠$-$⁠.043). This implies that in the case of HS10 products that experienced U.S. import expansion between 2017 and 2023, China's share declined. Relative to other import sources, firms were less likely to invest the resources to maintain or grow China's share in these sectors. In contrast, in HS10 products that experienced declining U.S. imports between 2017 and 2023, the negative coefficient estimate suggests that the share of imports purchased from China grew.¹⁷ This positive effect on China-share is consistent with circumstances where firms continue past relationships, in cases where the fixed costs were already sunk in the past but forgo costly expansions due to the combined effects of sunk costs and uncertainty. Similar results are attained in all other subsets in Table 7 (final goods, capital goods, and non-ITC products), except for ITC goods.

A second message of Table 7 is that the effects of tariffs and exemption policy have differentially powerful effects on products, whether they are disaggregated by good type (intermediate, capital, final), or by ITC versus non-ITC. Nonetheless, the signs and statistical strength of almost all coefficients are comparable to those from the previous table. To the extent that the ITC subgroup is the one outlier, the overall estimates for this segment suggest that ITC imports shifted away from China in all cases, aside from HS10 products had secured one of the longer U.S. tariff exemption exceptions.

When this new specification is taken to the data, Table 8 uncovers a distinct asymmetry in strength of response to positive versus negative changes in level of U.S. HS10 product imports. Though the level of Chinese exports grew when the level of U.S. import demand rose, this positive response was smaller in magnitude than the rate of decline in China's exports when U.S. import demand fell. China's exports grew relatively weakly when U.S. import opportunities emerged. In contrast, China's exports experienced much larger declines when U.S. imports shrank in value.

Considering declines in China's share of U.S. imports following the trade war, it is interesting to ask whether these changes affected Asian trade with the United States and China's immediate neighbors in ASEAN. To this end, Table 9 provides a comparison of the effects on China with those in ASEAN, using trade changes between 2017 and 2021. To begin, similar regressions for China's and ASEAN's share of U.S. imports are displayed in the first two columns of Table 9. A number of results suggest the changes in China's fortunes accrued to ASEAN countries through three statistically identified and meaningful differences that explained declines in China's share and growth in ASEAN's share: (1) Inclusion on the United States Section 301 tariff lists interacted with the original China share, (2) the size of trade war tariffs, and (3) the reaction to increases in U.S. product-level imports. On this third point, another difference is apparent. While the Chinese data suggest that Chinese trade reacted asymmetrically to growth and decline in U.S. product-level import, this form of asymmetry is not identifiable in regressions that search for such differential effects. This suggests that China's differential response was related to trade war trade policy uncertainty, rather than a general asymmetry in responses to positive versus negative developments that applied to all ASEAN exporters.

The remainder of Table 9 searches for evidence of differential reactions in the U.S. share of imports from ASEAN, as differences are related to policy and growth. The coefficients on the indicator variable, Onlist, and the interaction between Onlist and the original China share show that the primary effect of 301 tariffs on U.S. final goods imports from ASEAN was related to the tariffs charged as well as the mere listing in the group of Section 301 products that were part of the trade war. In contrast, the effects on being Onlist in the case of intermediate and final goods were positively mediated by the original level of China share. This result also emerges for non-ITC goods, when the sample is divided into ITC versus non-ITC products.

A second dimension of ASEAN response was entry into export of HS10 products that ASEAN had not exported to the United States in 2017. To focus on entry of this form that was related to U.S. treatment of Chinese imports, Table 10 considers the sample of 5,142 products that the United States imported from China in 2017 but did not import from ASEAN in 2017. By 2021, the United States imported 1,667 of those products from ASEAN. Thus, it is possible to create a dependent variable, ASEAN entry, that is set to one if the United States started import from ASEAN by 2021, and zero if no ASEAN import had commenced.

Viewing Table 10 reveals that the policy variable that had the most impact generally was the inclusion of products on the U.S. Section 301 tariff lists. ASEAN entry is positively associated with Onlist, in the case of the full sample, non-ITC goods, and most strongly for final goods. Presumably, these products were the easiest to relocate, and given the uncertainty associated with the trade war, firms took advantage of extensive margin sourcing changes to help reduce the risks. The other policy item that has a positive effect on the shift is the presence of some approvals of the exemption petitions. Here, it would suggest that the USTR managed to grant its relief to importers that were actively able to change their sourcing patterns and needed time to accomplish the shift. Lastly, in all cases, the entry of ASEAN into new export product lines was fostered by product-level U.S. import growth.

This paper studies the effects of the U.S. tariff exemption process that accompanied the widespread jump in U.S. tariffs during the U.S.–China trade war. Although the USTR did not disclose its grounds for decisions, analysis of trade data at the fine product level shows that the approval of these exemptions was consistent with the expressed criteria, which sought to limit exemptions if firms had alternative sourcing options in the United States or third countries. Most notably, petitions that had higher odds of success involved products where sourcing from ASEAN countries and Mexico was more limited, and where reliance on Chinese supply was heavier.

Following the documentation of stylized facts regarding the HS10 products that sought exemptions, and the characteristics of the products that were granted exemption relief, this paper examines how the combined effects of trade war tariffs, and the presence or absence of protectionist threat and relief shaped changes in U.S. imports from China and ASEAN by 2023. The results show that both tariffs levied and trade war procedures and risks affected trade outcomes. For example, the inclusion on tariff lists led to import sourcing shifts toward ASEAN sources of supply—an effect that was not reversed, even for the products that received tariff exemptions. This suggests that firms were wary of future return of tariffs in these product segments. Importantly, the asymmetric relationship of these effects to overall U.S. product-level import growth, suggests that trade war impacts had enduring effects related to ongoing trade policy uncertainty. This asymmetric response suggests that trade connections may be deterred by potential threats in other settings, such as Donald Trump's suggestion that he may initiate tariffs as economic leverage against multiple trade partners, including Canada and Mexico.

Due to these impacts, the long-term effects of the trade war should be studied as ongoing experience provides a longer time series of data. To the extent that the trade war affected firm supply chains, the potential for permanent scarring is suggested by Handley, Kamal, and Monarch (2020) and shows how U.S. import tariffs damaged exports by U.S. firms that relied on imported inputs, while Benguria et al. (2022) show the damage to the profitability and investment activity of Chinese firms who faced U.S. trade war tariffs. The damage to firms, and the new ASEAN exporter entry documented in this paper, suggest that tariffs and uncertainty will both continue to reshape trade. Although Boer and Rieth's (2024) study of macroeconomic data from U.S. participation in NAFTA and the WTO suggests that tariffs may ultimately have a larger effect in the long run, the long-run impacts of the trade war will ultimately be revealed.

This paper combines information on exemption applications with U.S. trade data. The collection of the exemption data was sponsored by the Mercatus Center at George Mason University and is available for download at https://www.quantgov.org/tariffs. Information on these data is provided by Brunk, McDaniel, and Parks (2019). The U.S. import data are available on the U.S. International Trade Commission Web site, https://dataweb.usitc.gov/. This project uses the 10-digit HS10 data, from the “Imports for Consumption” data series.