Disentangling the Effects of the 2018-2019 Tariffs on a Globally Connected U.S. Manufacturing Sector

Finance and Economics Discussion Series

Divisions of Research & Statistics and Monetary Aﬀairs

Federal Reserve Board, Washington, D.C.

Disentangling the Eﬀects of the 2018-2019 Tariﬀs on a Globally

Connected U.S. Manufacturing Sector

Aaron Flaaen and Justin Pierce

2019-086

Please cite this paper as:

Flaaen, Aaron, and Justin Pierce (2019). “Disentangling the Eﬀects of the 2018-2019 Tar-

iﬀs on a Globally Connected U.S. Manufacturing Sector,” Finance and Economics Discus-

sion Series 2019-086. Washington: Board of Governors of the Federal Reserve System,

https://doi.org/10.17016/FEDS.2019.086.

NOTE: Staﬀ working papers in the Finance and Economics Discussion Series (FEDS) are preliminary

materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth

are those of the authors and do not indicate concurrence by other members of the research staﬀ or the

Board of Governors. References in publications to the Finance and Economics Discussion Series (other than

acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

Disentangling the Eﬀects of the 2018-2019 Tariﬀs on a

Globally Connected U.S. Manufacturing Sector

∗

Aaron Flaaen

Federal Reserve Board

Justin Pierce

Federal Reserve Board

December 23, 2019

Abstract

Since the beginning of 2018, the United States has undertaken unprecedented tariﬀ

increases, with one goal of these actions being to boost the manufacturing sector. In

this paper, we estimate the eﬀect of the tariﬀs—including retaliatory tariﬀs by U.S.

trading partners—on manufacturing employment, output, and producer prices. A key

feature of our analysis is accounting for the multiple ways that tariﬀs might aﬀect the

manufacturing sector, including providing protection for domestic industries, raising

costs for imported inputs, and harming competitiveness in overseas markets due to

retaliatory tariﬀs. We ﬁnd that U.S. manufacturing industries more exposed to tariﬀ

increases experience relative reductions in employment as a positive eﬀect from import

protection is oﬀset by larger negative eﬀects from rising input costs and retaliatory

tariﬀs. Higher tariﬀs are also associated with relative increases in producer prices via

rising input costs.

∗

We are grateful to Vivi Gregorich for superb research assistance. We thank Ryan Monarch and partici-

pants at the 2018 Mid-Atlantic Trade Workshop at Duke for useful comments. Any opinions and conclusions

expressed herein are those of the authors and do not necessarily represent the views of the Board of Governors

or its research staﬀ.

1 Introduction

The unprecedented increase in tariﬀs imposed by the United States against its major trading

partners since early 2018 has brought renewed attention to the economic eﬀects of tariﬀs.

While there are already vast theoretical and empirical literatures documenting the eﬀects of

changes in trade policy, it is not clear how prior estimates apply to the present day when

there are virtually no modern episodes of a large, advanced economy raising tariﬀs in a way

comparable to the U.S. in 2018-2019. Further complicating the process of estimating the

eﬀects of tariﬀs is the rapid expansion of globally interconnected supply chains, in which

tariﬀs can have impacts through channels beyond their traditional eﬀect of limiting import

competition.

Another important feature of these tariﬀs is that they were imposed, in part, to boost the

U.S. manufacturing sector by protecting against what were deemed to be the unfair trade

practices of trading partners, principally China. Thus, while existing research has mostly

documented negative consequences of the tariﬀ increases on the broad economy–including

higher prices, lower consumption, reduced business investment, and drops in the valuations

of aﬀected ﬁrms–some might view these eﬀects as an acceptable cost for achieving the policy

aim of ensuring more robust manufacturing activity in the United States.

This paper provides the ﬁrst comprehensive estimates of the eﬀect of recent tariﬀs on

the U.S. manufacturing sector. A key feature of this analysis is accounting for the diﬀerent

ways that tariﬀs could aﬀect manufacturers in the presence of global trade and supply chain

linkages. On the one hand, U.S. import tariﬀs may protect some U.S.-based manufacturers

from import competition in the domestic market, allowing them to gain market share at

the expense of foreign competitors. On the other hand, U.S. tariﬀs have also been imposed

on intermediate inputs, and the associated increase in costs may hurt U.S. manufacturers’

competitiveness in producing for both the export and domestic markets. Moreover, U.S.

trade partners have imposed retaliatory tariﬀs on U.S. exports of certain goods, which could

again put U.S. ﬁrms at a disadvantage in those markets, relative to their foreign competitors.

Disentangling the eﬀects of these three channels and determining which eﬀect dominates is

an empirical question of critical importance.

Toward this end, we construct straightforward industry-level measures of exposure to each

of these three channels. We measure the import protection channel as the share of domestic

absorption aﬀected by newly imposed tariﬀs. We account for declines in competitiveness

associated with increased input costs as the share of industry costs subject to new tariﬀs.

Finally, we measure an industry’s potential exposure to retaliatory tariﬀs by U.S. trading

partners as the share of industry-level exports subject to new retaliatory tariﬀs. We construct

these measures using detailed data on each industry’s input-output structure, trade ﬂows,

and shipments, as well as information on the set of products covered by both U.S. and foreign

retaliatory tariﬀs. We then relate the measures for these three channels of tariﬀ exposure to

monthly data on manufacturing employment, output, and producer prices.

We employ a diﬀerences in diﬀerences approach that compares outcomes in industries

that are more subject to each of the three tariﬀ channels to those that are less aﬀected.

We begin by regressing the industry-level outcomes on interactions of measures of the three

channels with a set of month dummies, which allows us to observe the timing of any eﬀects

of tariﬀs, while also determining whether more-exposed industries were on diﬀerential trends

prior to the start of trade tensions. We adopt two approaches to control for potential pre-

trends, including using detrended measures of the dependent variables, and diﬀerencing out

the eﬀect of the pre-trends as in Finkelstein (2007). Industry and month ﬁxed eﬀects in the

regressions control for time-invariant characteristics of industries and aggregate shocks. In

addition, we control for industries’ import share of domestic absorption and export share

of shipments to allow for the possibility, for example, that more internationally exposed

industries may perform diﬀerently at diﬀerent stages of the business cycle.

We ﬁnd that tariﬀ increases enacted in 2018 are associated with relative reductions in

manufacturing employment and relative increases in producer prices. In terms of manufac-

turing employment, rising input costs and retaliatory tariﬀs each contribute to the negative

relationship, and the contribution from these channels more than oﬀsets a small positive

eﬀect from import protection. For producer prices, the relative increases associated with

tariﬀs are due solely to the rising input cost channel. We ﬁnd little evidence for a relation-

ship between industrial production and any of the three tariﬀ channels considered.

In terms of economic signiﬁcance, we ﬁnd that shifting an industry from the 25th per-

centile to the 75th percentile in terms of exposure to each of these channels of tariﬀs is

associated with a reduction in manufacturing employment of 1.4 percent, with the positive

contribution from the import protection eﬀects of tariﬀs (0.3 percent) more than oﬀset by

the negative eﬀects associated with rising input costs (-1.1 percent) and retaliatory tariﬀs

(-0.7 percent). For producer prices, we ﬁnd that an interquartile shift in exposure to rising

input costs is associated with a 4.1 percent increase in factory-gate prices. These estimates

provide information about the responses of more-exposed relative to less-exposed industries,

but do not reﬂect general equilibrium eﬀects of the tariﬀ increases.

The results point to the importance of increased costs from tariﬀs on inputs as a mecha-

nism through which tariﬀs aﬀect the manufacturing sector, with this channel yielding highly

statistically signiﬁcant eﬀects for both employment and producer prices. The importance

of this channel in our estimates is consistent with the purposeful targeting of intermediate

inputs for early rounds of U.S. tariﬀs in order to avoid tariﬀs on consumer goods, as discussed

in section 2. Furthermore, the results point to the relevance of global supply chains when

evaluating the eﬀects of tariﬀs.

These results necessarily represent short-term eﬀects of tariﬀs, and the longer-term impli-

cations of trade tensions may diﬀer from those estimated here. For example, some adjustment

to the imposition of tariﬀs may take time as ﬁrms complete previously agreed-upon contracts

with customers and suppliers. Furthermore, some eﬀects may dissipate if tariﬀs were to re-

turn to their previous levels, while others may be longer-lived. We also note that this paper

focuses speciﬁcally on realized changes in tariﬀs, but does not explicitly consider the eﬀects

of increased uncertainty about future trade policy. Indeed, the increase in uncertainty gen-

erated by recent rounds of tariﬀ increases may be one of the ways in which the eﬀects of past

policy persist.

This paper builds on the evolving literature examining the eﬀects of recent global trade

tensions on the U.S. economy. Early work in this literature includes Amiti, Redding and

Weinstein (2019) and Fajgelbaum et al. (2019) who ﬁnd near-complete pass-through of U.S.

tariﬀ increases to domestic prices, implying welfare losses, though of a relatively small mag-

nitude. Cavallo et al. (2019) show that product composition appears to be a key determinant

in the diﬀerences in tariﬀ pass-through between U.S. imports and U.S. exports during the

2018-2019 tariﬀ escalation, while also showing that the majority of U.S. tariﬀ increases are

being absorbed by U.S. retailers. Flaaen, Horta¸csu and Tintelnot (2019) examine the case

of U.S. tariﬀs imposed on washing machines, showing that tariﬀs on individual countries can

lead to the relocation of production across borders, while tariﬀs on a broader sets of coun-

tries lead to substantial retail price increases for both targeted products and complementary

goods. Handley, Kamal and Monarch (2019) ﬁnd that those ﬁrms impacted by U.S. import

tariﬀs respond by reducing U.S. exports, consistent with our ﬁnding of the importance of

the intermediate input channel. Huang et al. (2019) ﬁnd that the eﬀects of tariﬀs also carry

through to ﬁrms’ ﬁnancial performance, with ﬁrms more engaged in trade with China expe-

riencing lower stock returns and higher default risk after the announcement of new rounds

of tariﬀs targeting China.

Focusing on the eﬀects of retaliatory tariﬀs on consumption, Waugh (2019) ﬁnds that

counties specializing in industries subject to Chinese retaliatory tariﬀs experienced reductions

in new auto sales. Blanchard, Bown and Chor (2019) also ﬁnd that these retaliatory tariﬀs

can explain a shift in voting away from Republican House candidates in the 2018 election.

In research focusing on uncertainty regarding tariﬀ rates, Caldara et al. (2019) ﬁnd that

increases in measured trade policy uncertainty reduce investment in ﬁrm-level and aggregate

data. Other research (Reyes-Heroles, Traiberman and Van Leemput 2019) notes that the

eﬀects of tariﬀ actions by major trading countries can also have implications for the trade

patterns of emerging market economies.

Although we highlight the recent and rapidly expanding literature on the 2018-2019

tariﬀs, the ideas of accounting for retaliatory tariﬀs and supply chain eﬀects of tariﬀs go back

decades. Early examinations of optimal tariﬀs given the potential for retaliation can be found

in Kaldor (1940) and Johnson (1953). The counteracting eﬀect of tariﬀs on intermediate

inputs used in further domestic production–the rising input cost channel described above–

was highlighted in Balassa (1965) and also Corden (1966). The scale of the 2018-2019

tariﬀs, the increased availability of data, and the immensely expanded network of global

supply chains permits a quantitative examination of these channels that was not possible

before.

Our paper makes several contributions to the existing literatures. First, we explicitly

measure and estimate the eﬀects of several channels through which tariﬀs could aﬀect man-

ufacturers, which we ﬁnd to be important given that tariﬀs can simultaneously protect an

industry’s output, while raising prices for its inputs and subjecting it to retaliation in its

export markets. Second, we focus speciﬁcally on the manufacturing sector, the sector whose

output and employment were targeted to be boosted by tariﬀs. We believe this focus is

important, because our results indicate that tariﬀs have been a drag on employment and

have failed to increase output. Third, we provide the ﬁrst simultaneous examination of the

output, employment, and price eﬀects of the 2018-2019 tariﬀs in a particular sector.

The remainder of the paper proceeds as follows. Section 2 describes the timing of recent

trade actions by the U.S. and its trading partners, lists the data sources used in the analysis,

and details the calculation of the three measures of exposure to tariﬀs. Section 3 presents

our empirical strategy and results and Section 4 concludes.

2 Background, Data, and Industry-Level Measurement

2.1 Recent Trends in U.S. Manufacturing

We begin by providing some brief background on recent trends in manufacturing activity

in the period leading up to and during the imposition of tariﬀs. Toward that end, Figure

1 displays manufacturing employment and production from January 2017 to August 2019,

with each data series converted to an index that equals 100 in January 2018, just before

the imposition of the ﬁrst round of new tariﬀs. As indicated in the Figure, manufacturing

employment and output increased at a robust pace in 2017 and, indeed, through much of

2018. Since the end of 2018, however, manufacturing output has declined noticeably and

manufacturing employment growth has stalled. Given this inﬂection point, which came

after the imposition of substantial tariﬀs by the U.S. and its trading partners, it seems

reasonable to ask whether the tariﬀs implemented or planned in 2018 played some role in

this manufacturing slowdown. Given the patterns evident in Figure 1, it is now helpful to

provide additional detail on the timing of the implemented tariﬀs we study.

Figure 1: Measures of Manufacturing Activity: 2017m1 to 2019m8

98 99 100 101 102 103 104

Jan. 2018 = 100

2017 2018 2019

Jan Jul Jan Jul Jan Jul

Employment

Sources: Federal Reserve Board (FRB) for industrial production; U.S. Department of Labor, Bureau of

Labor Statistics for employment.

2.2 Timing of U.S. and Retaliatory Tariﬀs

The escalation in tariﬀs beginning in 2018 can be classiﬁed under three separate actions,

which occurred on distinct timelines and were subject to distinct rounds of retaliatory tariﬀs.

Figures 2 and 3 display the scope and timing of these trade actions.

The ﬁrst tariﬀ action was enacted in early February 2018 against large residential washing

machines and solar panels/modules. Applying the rarely used Section 201 of the Trade Act

of 1974, these tariﬀs aﬀected all import partners based on a determination of “whether an

article is being imported in such increased quantities that it is a substantial cause of injury,

or threat thereof, to the U.S. industry producing an article like or directly competitive with

the imported article” (U.S. Code: Title 19, Code 2252). These tariﬀs are scheduled to last

for three years before automatically expiring. The value of 2017 U.S. imports aﬀected by

these tariﬀs is shown by the green area in Figure 2.

The second major tariﬀ action aﬀected steel and aluminum imports beginning in March

2018. In only the third instance of its use by the United States, the U.S. administration

self-initiated an investigation under Section 232 of the Trade Expansion Act of 1962, which

determines whether an imported product is being imported “in certain quantities or under

such circumstances as to impair or threaten to impair U.S. national security” (U.S. Code:

Title 19, Code 1862). The announced tariﬀ rates were applied at 25 percent on steel and

certain steel products and 10 percent on aluminum. The administration initially exempted

the imports of several countries, including Canada, Mexico, Korea, and the countries com-

prising the European Union; however, in June 2018 these exemptions were removed.

The

value of imports aﬀected by these tariﬀs is shown by the red area in Figure 2.

In response to this second action, China announced retaliatory tariﬀs on U.S. exports

in April of 2018, while the European Union, Canada, and Mexico followed with their own

retaliatory tariﬀs in June and July of 2018. These tariﬀs focused on aluminum waste, scrap,

pork and various agricultural products for the case of China, and steel, aluminum and other

agricultural goods for the cases of the E.U., Canada, and Mexico. The value of 2017 U.S.

exports subject to these retaliatory tariﬀs is shown by the red portion of Figure 3. The

equal scales of the axes in Figures 3 and 2 makes clear that the magnitude of retaliation on

U.S. exports has been signiﬁcantly lower than the level of tariﬀs applied by the U.S. on its

imports.

The third and most signiﬁcant tariﬀ action imposed tariﬀs on U.S. imports from China

based on Section 301 of the Trade Act of 1974. This provision provides enforcement mecha-

nisms against a wide range of trade practices deemed to be unjustiﬁable, unreasonable, dis-

criminatory, or that violate trade agreements. The initial section 301 investigation against

Chinese trade practices focused on intellectual property and technology transfer. The re-

sults of the investigation were released in March of 2018, and tariﬀs on Chinese goods were

announced in April. These tariﬀs were actually applied in several phases as both the United

States and China retaliated against import tariﬀs imposed on one another. Phase 1 of U.S.

tariﬀs occurred in July 2018, targeting $34 billion of Chinese imports, with additional Phase

2 tariﬀs on $16 billion following in August. Each of these initial phases were set at a rate of

25 percent. The largest set of tariﬀs, which covered nearly $200 billion of Chinese imports,

went into eﬀect in late September 2018 at a rate of 10 percent. These initial phases of the

section 301 tariﬀs can be seen as the blue areas of Figure 2.

Each of these phases of U.S. tariﬀs against Chinese imports was accompanied by Chinese

retaliatory tariﬀs on U.S. exports. Retaliation on Phases 1 and 2 covered equivalent dollar

values of U.S. exports at identical rates, while the Phase 3 covered a smaller amount — $60

billion of U.S. exports — at rates ranging from 5-10 percent. These values are shown in the

blue areas of Figure 3.

There was an extended lull in the exchanges of U.S.-China tariﬀs between October 2018

and April 2019 as negotiations between the two sides took place. In May of 2019, citing a re-

treat from Chinese commitments in previous rounds of the negotiations, the U.S. announced

that the Phase 3 tariﬀs would be increased to 25 percent beginning May 10th. China re-

taliated by raising tariﬀ rates on a little over half of its Phase 3 retaliatory tariﬀs. Further

U.S. tariﬀs were announced in August 2019, but fall outside our period of analysis: a Phase

4 of U.S. tariﬀs against roughly $110 billion in Chinese imports to take eﬀect in September

The Republic of Korea managed to secure a permanent exemption after agreeing to stringent import

quotas to the U.S. market.

Figure 2: Timeline of New U.S. Import Tariﬀs: 2018-2019

Solar Panels,

Wash. Mach.

Sec.201: $7B

Steel, Aluminum

Sec. 232: $40B

Chinese Imports

Sec. 301: $506B

0 200 400 600

Billions of USD (Based on 2017 Imports)

2018m1 2018m7 2019m1 2019m7 2020m1

Sources: USITC for 2017 import values. See Table A1 for details on the set of relevant products and trade

values.

Notes: Our analysis below covers only the 2018 rounds of tariﬀs.

2019, and a Phase 5 of tariﬀs against roughly $160 billion in December 2019. China once

again retaliated with a plan for additional tariﬀs on U.S. exports to take eﬀect with a similar

timing.

As we highlight below, the eﬀect of these tariﬀs on the U.S. manufacturing sector depends

at least in part on how they impact global supply chains. U.S. manufacturers competing

with Chinese imports would fare quite diﬀerently than manufacturers that rely on Chinese

inputs in their U.S. production. For a rough guide for how these tariﬀs were split along this

dimension, we apply the United Nations Broad Economic Categories (BEC) classiﬁcation

to these tariﬀs (see also Bown, Jung and Lu (2019b) for a similar breakdown).

As shown

in Appendix Figure B1, the early U.S. tariﬀs predominantly covered intermediate goods,

represented by the blue areas of the section 232 and initial section 301 phases of U.S. tariﬀs.

There were some reports that this focus on intermediate goods over consumer goods was a

purposeful eﬀort on the part of the United States to shield U.S. consumers from any eﬀects

of tariﬀs.

Of course, the BEC classiﬁcation does not substitute for analysis using input-output tables, as one U.S.

industry’s ﬁnal good output can be another industry’s intermediate input.

Figure 3: Timeline of Retaliatory Tariﬀs on U.S. Exports: 2018-2019

Mexico, Turkey, Russia

China, EU, Canada

& Aluminum Tariffs

Retaliation on Steel

Retaliation on

Sec. 301 Tariffs

China

0 200 400 600

Billions of USD (Based on 2017 Exports)

2018m1 2018m7 2019m1 2019m7 2020m1

Sources: USITC for 2017 export values. See Table A2 for details on the set of relevant products and trade

values.

2.3 Data and Measurement

This section describes the data sources and measurement for the empirical analysis presented

in Section 3. We use publicly available data on the sets of products covered by U.S. import

tariﬀs and foreign retaliatory tariﬀs. For U.S. tariﬀs, product lists are from the United States

Trade Representative and as published in the U.S. Federal Register. For retaliatory tariﬀs

by U.S. trade partners, data are drawn from the relevant government agencies including

the Canadian Department of Finance, the European Commission, as well as the World

Trade Organization. These lists of aﬀected products have been helpfully collected by other

researchers who have made them available on their websites.

Table A1 provides a list of

links to all lists of aﬀected products.

We map the Harmonized System (HS) codes covered by tariﬀs described above to the

North American Industry Classiﬁcation System (NAICS) using the concordance developed

by Pierce and Schott (2012). For U.S. import tariﬀs, this requires a simple application of

the concordance. For tariﬀs imposed by U.S. trade partners, this process is complicated by

the fact that the import product codes published by foreign governments cannot be matched

to the Schedule B system used for U.S. exports below the six-digit HS level. Therefore, for

foreign retaliatory tariﬀs, we treat an entire six-digit HS code as being covered by tariﬀs if any

product with that six-digit HS preﬁx is covered by a tariﬀ, following Blanchard, Bown and

See, for example, the website, maintained by the Crowell-Moring International Trade law ﬁrm and Bown

and Kolb (2019).

Chor (2019), Waugh (2019) and Bown, Jung and Lu (2019a). We ﬁnd that this assumption

is justiﬁed because the value of U.S. exports that we classify as being covered by retaliatory

tariﬀs lines up well with those calculated by other researchers as well as those announced by

U.S. trade partners, as reported in Table A2.

Our measures of exposure to the various rounds of tariﬀs imposed by the U.S. and its

trading partners also require data on the value of overall imports, exports and shipments.

We collect data on the dollar value of U.S. imports and exports from the United States

International Trade Commission. For annual levels of industry shipments, we use the most

recent year of the Annual Survey of Manufactures (ASM), 2016. Data on the input usage

of each industry are drawn from the BEA’s input-output tables for 2012, the most recent

vintage available.

Lastly, we draw monthly values of the dependent variables for our analysis—industry

output, employment, and producer prices—from three sources. Our measures of monthly

industry output come from the Federal Reserve’s G.17 Release on Industrial Production

and Capacity Utilization. For monthly estimates of employment at the industry-level, we

utilize data from the Current Employment Statistics (CES) program of the Bureau of Labor

Statistics. Finally, we use the producer price index, also from the Bureau of Labor Statistics,

to measure monthly changes in prices across industries.

2.4 Level of Aggregation

We conduct the analysis at the level of four-digit NAICS industries, which is the most de-

tailed level at which comprehensive data for industrial production, producer prices, and

employment are available at a consistent level of aggregation, with several exceptions. First,

we separate aluminum manufacturing (NAICS 3313) into three sub-industries to reﬂect an

important distinction within the industry that is impossible to ignore given the set of tariﬀs

we study. Primary aluminum production (NAICS 331313) consists of establishments largely

making aluminum from raw materials like alumina and bauxite. The remaining components

of NAICS 3313 (NAICS 331314: Secondary Smelting and Alloying of Aluminum, NAICS

331315: Aluminum Sheet, Plate, and Foil Manufacturing, and NAICS 331318: Other Alu-

minum Rolling, Drawing, and Extruding) manufacture aluminum from scrap, dross, or other

existing aluminum. As is clear from this deﬁnition, primary aluminum manufacturers would

be protected by the U.S. tariﬀs on aluminum, while the same tariﬀs would apply to both the

output and inputs of the second group of producers. Therefore, we split NAICS 3313 into

primary aluminum production (NAICS 331313), secondary aluminum production (NAICS

331314), and the combination of NAICS 331315 and 331318, consistent with the industry

splits used in the BEA input-output tables.

Second, there are exceptions to the use of the four-digit NAICS classiﬁcation due to data

limitations. In particular, there are some industries (NAICS 315: Apparel Manufacturing,

and NAICS 316: Leather and Allied Product Manufacturing) where declining domestic pro-

duction has made 4-digit NAICS detail increasingly uncommon in U.S. federal statistics. We

have therefore used the 3-digit classiﬁcation for these two industries. Data on manufacturing

employment and producer prices also lack suﬃcient detail on the aluminum splits detailed

above, and so we use the aggregate NAICS 3313. After factoring in the necessary aggrega-

tions, our baseline samples contain 76 industries for employment, 84 industries for industrial

production, and 81 industries for producer prices.

2.5 Industry-Level Measures of Trade Policy Impact

This section describes the measures we construct to quantify the industry-level eﬀects of the

trade policies enacted by the U.S. and its trading partners in 2018. Our focus in construct-

ing these measures is capturing the eﬀect of realized changes in tariﬀs on forces likely to

impact outcomes in the manufacturing sector, including the amount of import competition

in the U.S. market, foreign demand for U.S. exports, and input costs. In particular, we

construct three industry-level measures capturing each of these channels of potential trade

policy impact. Uncertainty regarding trade policy may have led to additional eﬀects; indeed,

uncertainty has been a prominent focus of industry anecdotes as well as academic research

(see, for example Caldara et al. (2019)).

New Tariﬀ Import Share of Domestic Absorption

One of the most salient ways in which tariﬀs could aﬀect an industry’s economic activity is by

restricting foreign competition. To measure the extent of this potential protection, we relate

the scale of imports aﬀected by new tariﬀs to the level of domestic absorption. Formally,

let Ω

be the list of U.S. imported product-country pairs (pc) subject to new tariﬀs. The

variables imp

and exp

identify total industry i imports and exports, and Q

equals domestic

production. Then, the new tariﬀ import share of domestic absorption is given by:

Import Protection

pc∈Ω

imp

ipc

+ imp

− exp

. (1)

Throughout the remainder of the draft, we typically refer to this measure as an industry’s

degree of “import protection” from the 2018 tariﬀs.

In our baseline results, we calculate equation (1) for the cumulative set of tariﬀs from

2018 as a whole.

Table 1 lists the top ten industries from this cumulative measure of

Our measures of the channels of tariﬀ eﬀects are based on country-product-level trade ﬂows, but do not

incorporate information on the ad valorem tariﬀ rate assessed. In the case of U.S. tariﬀs, the overwhelming

majority of import value was covered by ad valorem rates that were either initially set at or eventually

reached 25 percent.

In Appendix B.3 we describe additional results in which we calculate separate measures of import

new import protection. Despite the focus on China-speciﬁc tariﬀs in media coverage of

recent trade actions, the global tariﬀs – Section 232 tariﬀs on steel and aluminum and the

Section 201 tariﬀs on washing machines and solar panels – play a prominent role in this

list of protected industries.

Other industries such as electric lighting equipment (NAICS

3351), household and institutional furniture and kitchen cabinets (NAICS 3371), and other

electrical equipment and component (NAICS 3359) are primarily aﬀected by the U.S. tariﬀs

on Chinese imports.

Table 1: Top Ten Industries by New Import Protection

New Tariﬀ

Import Share

Rank NAICS Industry Description of Absorption

1 331313 Primary Aluminum Production 67%

2 3351 Electric Lighting Equipment 30%

3 3371 Household and Institutional Furniture

and Kitchen Cabinet

24%

4 3344 Semiconductor and Other Electronic

Component

21%

5 3311 Iron and Steel Mills and Ferroalloy Mfg 21%

6 33131B Aluminum Sheet/Plate/Foil and

Rolling/Drawing/Extruding

19%

7 3352 Household Appliance Manufacturing 18%

8 3359 Other Electrical Equipment & Component 16%

9 3160 Leather and Allied Product 15%

10 3332 Industrial Machinery 14%

Sources: Authors’ calculations based on equation (1) in the text.

Notes: This measures corresponds to the cumulative eﬀect of all new tariﬀs

from January through September 2018.

New Tariﬀ Export Share of Output

While U.S. tariﬀs may reduce competition for some industries in the domestic market, U.S.

trading partners responded to these tariﬀs by imposing retaliatory tariﬀs. These retaliatory

tariﬀs may harm U.S. manufacturers by decreasing their competitiveness in foreign markets.

We measure this potential eﬀect as the share of U.S. output that has been impacted by

new retaliatory tariﬀs. Deﬁning Ω

to be the list of U.S. exported product-country pairs

(pc) subject to retaliatory tariﬀs, we calculate the new tariﬀ export share of output – which

we refer to as an industry’s exposure to “foreign retaliation” – as the following:

protection for each individual wave of tariﬀs.

Solar panels and modules are included in NAICS industry 334413; NAICS industry 3344 is listed as

number 4 in Table 1 though this industry comprises a wide range of chips, microprocessors, and other

electrical components.

Foreign Retaliation

pc∈Ω

exp

ipc

. (2)

The ten industries most aﬀected by new foreign retaliatory tariﬀs (for 2018 as a whole)

are shown in Table 2. As is clear from the table, even the most aﬀected industries experience

increases in tariﬀs on less than 10 percent of their overall shipments.

Table 2: Top Ten Industries by New Tariﬀ Export Share of Output

New Tariﬀ

Export Share

Rank NAICS Industry Description of Output

1 3346 Manufacturing and Reproducing

Magnetic & Optical Media

8.6%

2 3160 Leather and Allied Product 7.7%

3 33131B Aluminum Sheet/Plate/Foil &

Rolling/Drawing/Extruding

7.7%

4 3311 Iron and Steel Mills and Ferroalloy Mfg 6.9%

5 3361 Motor Vehicle Manufacturing 4.9%

6 3352 Household Appliance 4.7%

7 3211 Sawmills and Wood Preservation 4.5%

8 3343 Audio and Video Equipment 4.3%

9 3253 Pesticide, Fertilizer, and Other

Agricultural Chemical

4.1%

10 3341 Computer and Peripheral Equipment 3.9%

Sources: Authors’ calculations based on equation (2) in the text.

Notes: This measures corresponds to the cumulative eﬀect of all new tariﬀs

from January through September 2018.

New Tariﬀ Share of Costs

The ﬁnal channel we study traces the impact of U.S. tariﬀs on input costs via supply chain

linkages with foreign countries. The principal data on an industry’s sources of inputs used in

U.S. production are the input-output tables produced by the Bureau of Economic Analysis.

At the core of these tables are the “make” and “use” tables. The “use” table provides a

matrix use

–the dollar value of commodity j used in industry i production, and similarly

the “make” table provides a matrix make

–the dollar value of commodity j produced in

industry i production. It is helpful to ﬁrst consider the overall share of an industry’s costs

coming from foreign sources. Deﬁne SC

as the share of input costs of commodity j in

industry i:

use

+ Comp

, (3)

This measure of retaliatory tariﬀs includes retaliatory tariﬀs by China on U.S. exports of motor vehicles

(NAICS 3361), which were imposed in July of 2018, but then suspended in January 1, 2019.

where the variable M

equals total intermediate uses of industry i, and Comp

equals

compensation of employees of industry i. Then, deﬁne IS

as the import share of domestic

supply of commodity j:

imp

+ imp

, (4)

where here the variables imp

and Q

, are imports and output of commodity j, respec-

tively. By multiplying the terms from equations (3) and (4) we arrive at the (implied) import

share of costs in industry i from commodity j. Without additional detail on the sources of

inputs across countries, here we must use the “proportionality assumption”: that the dis-

tribution of the uses of imported commodities in an industry is proportional to domestic

commodity usage. Summing across commodities j yields the total import share of costs for

industry i. This implied import share of costs is given by:

ISC

use

+ Comp

| {z }

product share

of costs

imp

+ imp

| {z }

product import

(5)

We use data from the BEA’s 2012 input-output tables, the most recent year available,

updated with 2016 information on values of imports and shipments. To incorporate the

more recent data, we separate our measure of input usage into two components. The ﬁrst

component, Equation (3) uses only data from the 2012 detailed I-O tables and measures the

share of total costs of each commodity j for an industry i. The second component, Equation

(4), uses the more recent data on output and imports. It is worth pointing out that we are

only able to update equation (4) for manufactured goods, as annual output measures for

non-goods are unavailable. For non-goods commodities, we will use the 2012 values that

come from the IO-tables.

Finally, we construct our measure of an industry’s share of costs subject to new tariﬀs

to be the share of equation (5) that is covered by new U.S. import tariﬀs. We refer to this

measure below as the degree of “rising input costs” for a given industry, and calculate it as

follows:

Rising Input Costs

NT SC

use

+ Comp

pc∈Ω

imp

jpc

+ imp

, (6)

where as before the term Ω

denotes the list of U.S. imported product-country pairs (pc)

subject to new tariﬀs.

Table 3 lists the top domestic industries most aﬀected by increased costs from recent

import tariﬀs, where once again we show the cumulative eﬀects from all new 2018 tariﬀs.

As is apparent in the table, all of these industries are heavily dependent on various met-

als for domestic production. In some cases, such as the most aﬀected industry, aluminum

sheet/plate/foil (NAICS 33131B), our measures indicate that any beneﬁts of increased im-

port protection are oﬀset by increases in costs.

Table 3: Top Ten Industries by New Tariﬀ Import Share of Costs

New Tariﬀ

Rank NAICS Industry Description of Costs

1 33131B Aluminum Sheet/Plate/Foil &

Rolling/Drawing/Extruding

17.6

2 3312 Steel Product Mfg from Purchased Steel 8.4

3 3324 Boiler, Tank, and Shipping Container 8.3

4 3321 Forging and Stamping 7.4

5 331313 Primary Aluminum Production 7.2

6 331314 Secondary Smelting and Alloying of Aluminum 7.1

7 3323 Architectural and Structural Metals 5.3

8 3369 Other Transportation Equipment 5.0

9 3339 Other General Purpose Machinery 4.5

10 3352 Household Appliance 4.4

Sources: Authors’ calculations based on equation (6) in the text.

Notes: This measures corresponds to the cumulative eﬀect of all new tariﬀs

from January through September 2018.

3 Short-Run Impacts of Tariﬀs on Manufacturing

This section presents the diﬀerences-in-diﬀerences empirical strategy we use to estimate the

relationship between recent tariﬀs and outcomes in the manufacturing sector and presents

our baseline results.

This is a result of a signiﬁcant degree of input requirements “on the diagonal” of the use tables published

by the BEA.

3.1 Empirical Strategy

The distribution of our measures of tariﬀ impacts across industries, shown in part in Tables 1,

2, and 3 demonstrates the need for a systematic approach to decompose the impacts of tariﬀs

on the manufacturing sector. Any bivariate relationship between an outcome measure and

one of the channels identiﬁed above could end up conﬂating multiple, potentially oﬀsetting

eﬀects on an industry. Hence, we will control for all channels of exposure to tariﬀs in our

baseline speciﬁcation, allowing us to calculate estimates of the eﬀect of each channel holding

the others constant, and determining which tariﬀ channel dominates.

In addition, rather than designating a set period of time over which to analyze the eﬀects

of tariﬀs on the outcome variables, we adopt a ﬂexible setup that allows the eﬀects of each

of the channels to vary over time. In particular, we interact the industry-level measures

for each of the tariﬀ channels with a full set of month dummies. This allows us determine

the exact timing of any estimated tariﬀ eﬀects in high-frequency data, while also providing

information on the presence of any diﬀerential pre-existing trends in outcome variables across

industries.

Finally, recognizing that international trade will have industry-level eﬀects even in the

absence of changes in trade policy, we include additional controls that account for a baseline

level of export and import exposure for each industry.

These controls should account for

general exposure to international conditions such as changes in the value of the dollar and

foreign GDP growth. In addition, these controls may also serve as a coarse proxy for exposure

to trade policy uncertainty.

Our estimating equation is given by:

= α +

1(M

= t)(Import Protection

) +

1(M

= t)(Input Cost

) ... (7)

1(M

= t)(Foreign Retaliation

) + +

1(M

= t)(Import Share

) ...

1(M

= t)(Export Share

) + δ

+ δ

+ ε

where 1(M

= t) indicates a monthly categorical variable, and the outcome of interest,

, is either log employment, log output, or the log of the producer price index of industry

i in time t.

One concern inherent in this type of diﬀerences-in-diﬀerences analysis is the issue of

diﬀering trends across industries prior to the implementation of new tariﬀs. We utilize two

approaches to account for pre-trends. First, we replace the measure y

with the equivalent

measure after removing an industry-speciﬁc trend in 2017 (denoted ey

), the last full year

Our export exposure measure is simply the export share of output (from 2016), and the import share of

domestic absorption (also from 2016).

before the implementation of new tariﬀs. In a second approach, we estimate equation (7)

and then follow Finkelstein (2007) by diﬀerencing out the pre-trend path for each coeﬃcient,

thereby arriving at a point estimate of the particular channel for the period of our study.

Speciﬁcally, for a given coeﬃcient (say, the γ

coeﬃcient above) we calculate the following:

∆y

= (

Jun-Aug19

− γ

Mar18-May18

) − (γ

Mar18-May18

− γ

Feb17-Apr17

) (8)

where we utilize the average of March-May, 2018, as the approximate period (t = 0) of the

cumulative 2018 tariﬀs, with the end period being the average of the last three months of

our sample (June-August, 2019) and the pre-period being the average of three months from

an equivalent number of months before (February-April, 2017).

3.2 Results

Figure 4 provides a visual representation of the results of equation (7) when using the

industry-detrended (ey

) outcomes of interest. Speciﬁcally, the three panels of the ﬁgure

display the 90 percent conﬁdence intervals of the coeﬃcient estimates for the interactions of

the tariﬀ channel measures with the month dummies. The dashed vertical lines indicate the

timing of the ﬁrst (February) and last (September) trade actions imposed by the U.S. and

its trading partners in 2018.

Panel A displays estimated eﬀects for detrended employment. The left column of panel

A shows a negative and statistically signiﬁcant relationship between the rising input cost

eﬀect of tariﬀs and manufacturing employment. An interesting feature of this relationship

is that it materializes with a lag following the application of tariﬀs aﬀecting U.S. imported

intermediates, suggesting that the eﬀects of this channel take some time to manifest itself

in manufacturing employment. This result, therefore, provides additional context for the

aggregate trends shown in Figure 1, in which manufacturing activity did not slow noticeably

until the end of 2018. The middle column of panel A indicates a modest positive eﬀect on

employment associated with import protection, but these estimates are somewhat imprecise.

Finally, the right column of panel A exhibits a negative eﬀect of foreign retaliatory tariﬀs on

U.S. manufacturing employment that is apparent immediately after retaliatory tariﬀs begin

to be imposed. For all three sets of estimates shown in panel A of Figure 4, coeﬃcient

estimates during the period prior to the implementation of tariﬀs are reasonably ﬂat, though

not always centered exactly around zero.

The ﬁrst column of Table 4 provides alternative estimates of the relationship between the

tariﬀ channels and employment based on the de-trending approach from Finkelstein (2007)

described above. These estimates are consistent with those reported in Figure 4, and also

Figure 4: Eﬀects of Cumulative Tariﬀs (Detrended)

(a) Employment

-.8 -.6 -.4 -.2 0 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.05 0 .05 .1 .15 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-1.5 -1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

(b) Industrial Production (Output)

-1.5 -1 -.5 0 .5 1

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-2 -1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

-1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.2 -.1 0 .1 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-1 0 1 2

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

Sources: Federal Reserve Board (FRB), U.S. Department of Labor, Bureau of Labor Statistics; authors’

calculations.

Notes: Each panel represents separate regressions, with each row of plots corresponding to the ﬁrst three

components of equation (7), after removing an industry-speciﬁc trend in the dependent variable. The shaded

areas represent 90 percent conﬁdence intervals, and the two dashed lines represent February 2018 and

September 2018, the ﬁrst and last wave of new 2018 tariﬀs.

provide a more straightforward way of characterizing their economic signiﬁcance. We ﬁnd

that shifting an industry from the 25th percentile to the 75th percentile in terms of exposure

to each of the three channels of tariﬀs is associated with a reduction in manufacturing

employment of 1.4 percent, with the positive contribution from the import protection eﬀect

of tariﬀs (0.3 percent) more than oﬀset by the negative eﬀects associated with rising input

costs (-1.1 percent) and retaliatory tariﬀs (-0.7 percent).

Panel B of Figure 4 presents the coeﬃcient estimates pertaining to the relationship be-

tween tariﬀs and industrial production. Here, we see little evidence of signiﬁcant impacts

from the 2018 tariﬀs, on net, though there is a brief period of a negative relationship with the

input cost channel in mid-2018 and a noticeable shift down in estimates toward the end of

our sample. The estimates pertaining to the other channels – import protection and foreign

retaliation – do not exhibit statistically signiﬁcant patterns. Point estimates in column 2 of

Table 4 similarly lack statistical signiﬁcance.

Finally, panel C of Figure 4 indicates that these new tariﬀs had signiﬁcant eﬀects on

producer prices, primarily through the pass-through of rising input costs (the left column).

There is also suggestive evidence that producer prices rose modestly through the import

protection channel (the middle column); however, the estimates lack precision. Although one

might expect negative eﬀects on producer prices impacted by foreign retaliatory tariﬀs, we

see no evidence for this in our analysis (the right column). In terms of economic signiﬁcance,

the estimates in column 3 of Table 4 indicate that an interquartile shift in exposure to rising

input costs is associated with a 4.1 percent increase in factory-gate prices.

Table 4: Point Estimates of Cumulative Eﬀect by Channel:

Industrial Producer

Variable Employment Production Prices

Import Protection 0.080* -0.042 -0.053

(0.049) (0.165) (0.069)

Rising Input Costs -0.463*** -0.357 1.780***

(0.122) (0.330) (0.486)

Foreign Retaliation -0.506** 0.180 0.341

(0.249) (0.657) (0.538)

Industry Fixed Eﬀects yes yes yes

Number of Industries 76 84 81

Observations 2,440 2,688 2,586

Sources: Federal Reserve Board (FRB), U.S. Department of Labor, Bureau of

Labor Statistics; authors’ calculations.

Notes: See equations (7) and (8) in the text. Robust standard errors in paren-

theses. ∗ p < 0.10, ∗∗ p < 0.05, ∗ ∗ ∗ p < 0.01.

An alternative way of calculating the economic signiﬁcance of these estimates is to consider the eﬀect of

shifting the average industry—in terms of exposure to the three channels—to an alternative scenario with

zero tariﬀ exposure. This change in exposure to tariﬀs is also associated with a 1.4 percent relative reduction

in employment. We caution, however, that this estimate does not account for additional general equilibrium

eﬀects that might be associated with the recent tariﬀs, which might be the result of either changes to applied

tariﬀs or uncertainty regarding future tariﬀs.

3.3 Discussion

The results presented in Figure 4 and Table 4 pertain to the set of tariﬀs enacted in 2018,

with estimated eﬀects through August 2019. As additional data become available, it may

be possible to estimate eﬀects of the Phase 4 tariﬀs against Chinese imports enacted in

September 2019. Our results suggest that the eﬀects of this later round of tariﬀs will be

highly dependent on the composition of aﬀected products. As shown in Appendix Figure

B1, the September 2019 U.S. tariﬀs focused to a much greater extent on consumer goods

rather than the intermediate inputs that were heavily targeted in the 2018 tariﬀ actions.

Therefore, the input cost channel could play a smaller role in the eﬀects of the 2019 tariﬀs

on the manufacturing sector.

We also note that the longer-run eﬀects of these tariﬀs could be qualitatively diﬀerent

than the short-run eﬀects that we estimate here. On the one hand, there may be more

substantial expansion of U.S. manufacturing activity in the longer-term as ﬁrms fully adjust

their supply chains to avoid U.S. import tariﬀs, highlighting a potential longer lag structure

for the import protection channel. That said, there is suggestive evidence that the United

States is not typically the immediate destination for production relocation from China due

to increased tariﬀs. In the washing machine case studied in Flaaen, Horta¸csu and Tintelnot

(2019), ﬁrms ﬁrst moved production to other East Asian countries (Thailand and Vietnam)

following China-speciﬁc antidumping duties imposed in mid-2017. After the later Section

201 tariﬀs against worldwide imports of washing machines (discussed above), these same

ﬁrms did indeed shift some sizable production to the United States, though this occurred

with substantial costs to consumers via rising prices.

4 Conclusion

This paper examines the eﬀect of the tariﬀ increases imposed by the United States and its

trading partners in 2018 on outcomes in the U.S. manufacturing sector. We calculate mea-

sures of each industry’s exposure to tariﬀ changes via three channels: the import protection

that comes when an industry’s output is subject to U.S. tariﬀs, the increase in production

costs resulting from tariﬀs on imported inputs, and the reduction in foreign competitiveness

due to retaliatory tariﬀs in U.S. export markets. We then estimate the relationship between

these measures of exposure to tariﬀs and manufacturing employment, output, and producer

prices.

We ﬁnd that the 2018 tariﬀs are associated with relative reductions in manufacturing

employment and relative increases in producer prices. For manufacturing employment, a

small boost from the import protection eﬀect of tariﬀs is more than oﬀset by larger drags

from the eﬀects of rising input costs and retaliatory tariﬀs. For producer prices, the eﬀect

of tariﬀs is mediated solely through rising input costs.

These results have implications for evaluating the eﬀects of recent U.S. trade policy.

While one may view the negative welfare eﬀects of tariﬀs found by other researchers to be an

acceptable cost for a more robust manufacturing sector, our results suggest that the tariﬀs

have not boosted manufacturing employment or output, even as they increased producer

prices. While the longer-term eﬀects of the tariﬀs may diﬀer from those that we estimate

here, the results indicate that the tariﬀs, thus far, have not led to increased activity in the

U.S. manufacturing sector.

In addition, our results suggest that the traditional use of trade policy as a tool for

the protection and promotion of domestic manufacturing is complicated by the presence of

globally interconnnected supply chains. While the potential for both tit-for-tat retaliation

on import protection and input-output eﬀects on the domestic economy have long been rec-

ognized by trade economists, empirical evidence documenting these channels in the context

of an advanced economy has been limited. We ﬁnd the impact from the traditional import

protection channel is completely oﬀset in the short-run by reduced competitiveness from

retaliation and higher costs in downstream industries.

References

Amiti, Mary, Stephen J. Redding, and David Weinstein. 2019. “The Impact of the

2018 Trade War on U.S. Price and Welfare.” Journal of Economic Perspectives, 33(4): 187–

210.

Balassa, Bela. 1965. “Tariﬀ Protection in Industrial Countries.” Journal of Political Econ-

omy, 73(6): 573–94.

Blanchard, Emily J., Chad P. Bown, and Davin Chor. 2019. “Did Trump’s Trade

War Impact the 2018 Election?” National Bureau of Economic Research Working Paper

26434.

Bown, Chad P., and Melina Kolb. 2019. “Trump’s Trade War Timeline: An Up-to-Date

Guide.” Peterson Institute for International Economics.

Bown, Chad P., Euijin Jung, and Zhiyao Lu. 2019a. “China’s Retaliation to Trump’s

Tariﬀs.” Peterson Institute for International Economics.

Bown, Chad P., Euijin Jung, and Zhiyao Lu. 2019b. “Trump, China, and Tariﬀs: From

Soybeans to Semiconductors.” Peterson Institute for International Economics.

Caldara, Dario, Matteo Iacoviello, Patrick Molligo, Andrea Prestipino, and An-

drea Raﬀo. 2019. “The Economic Eﬀects of Trade Policy Uncertainty.” Journal of Mon-

etary Economics.

Cavallo, Alberto, Gita Gopinath, Brent Neiman, and Jenny Tang. 2019. “Tariﬀ

Passthrough at the Border and at the Store: Evidence from U.S. Trade Policy.” Becker-

Friedman Institute Working Paper 2019-124.

Corden, W.M. 1966. “The Structure of a Tariﬀ System and the Eﬀective Protective Rate.”

Journal of Political Economy, 74(3): 221–237.

Fajgelbaum, Pablo D., Pinelopi K. Goldberg, Patrick J. Kennedy, and Amit K.

Khandelwal. 2019. “The Return to Protectionism.” National Bureau of Economic Re-

search Working Paper 25638.

Finkelstein, Amy. 2007. “The Aggregate Eﬀects of Health Insurance: Evidence from the

Introduction of Medicare.” Quarterly Journal of Economics, 122(1): 1–37.

Flaaen, Aaron, Ali Horta¸csu, and Felix Tintelnot. 2019. “The Production Relocation

and Price Eﬀects of U.S. Trade Policy: The Case of Washing Machines.” National Bureau

of Economic Research Working Paper 25767.

Handley, Kyle, Fariha Kamal, and Ryan Monarch. 2019. “Rising Import Tariﬀs,

Falling Export Growth: When Modern Supply Chains Meet Old-Style Protectionism.”

Working Paper.

Huang, Yi, Chen Lin, Sibo Liu, and Heiwei Tang. 2019. “Trade Networks and Firm

Value: Evidence from the U.S.-China Trade War.” Centre for Economic Policy Research

Working Paper DP14173.

Johnson, Harry G. 1953. “Optimum Tariﬀs and Retaliation.” The Review of Economic

Studies, 21(2): 142–53.

Kaldor, N. 1940. “A Note on Tariﬀs and the Terms of Trade.” Economica, 7(28): 377–80.

Pierce, Justin, and Peter Schott. 2012. “A Concordance Between U.S. Harmonized

System Codes and SIC/NAICS Product Classes and Industries.” Journal of Economic

and Social Measurement, 37(1-2): 61–96.

Reyes-Heroles, Ricardo, Sharon Traiberman, and Eva Van Leemput. 2019.

“Emerging Markets and the New Geography of Trade: The Eﬀects of Rising Trade Barri-

ers.” Working Paper Working Paper.

Waugh, Michael E. 2019. “The Consumption Response to Trade Shocks: Evidence from

the US-China Trade War.” National Bureau of Economic Research Working Paper 26353.

A Expanded Detail on Implemented Tariﬀs

Tables A1 and A2 provide additional information regarding the data on products covered by

tariﬀs. Speciﬁcally, the tables report the value of trade–based on 2017 annual data from the

U.S. Census Bureau–that we calculate was subject to new tariﬀs, along with comparisons to

values of trade publicly announced by governments and those calculated by other researchers.

In addition, we provide links to sources of the lists of HS codes covered by new tariﬀs.

Table A1: New U.S. Import Tariﬀs by Trade Action and Wave

2017 Reported

Reference Import by Other Source

for Aﬀected Volume Government Estimates for Other

Import Tariﬀ Products Billions of U.S. Dollars Estimates

Sec. 201: Solar Panels 7 8.5

Sec. 201: Washing Machines 1.85 1.8

Sec. 232: Steel Link 27.7 10.2 29 Source

Sec. 232: Aluminum Link 17.4 7.7 17 Source

Sec. 301 Part 1 Link 32.3 34

Sec. 301 Part 2 Link 13.7 16

Sec. 301 Part 1+2 46.0 50 45.7 Source

Section 301 Part 3 Link 189 200 177 Source

Table A2: New Retaliatory Tariﬀs on U.S. Exports by Trade Action and Wave

2017 Reported

Reference Export by Other Source

for Aﬀected Volume Government Estimates for Other

Retaliatory Tariﬀ Products Billions of U.S. Dollars Estimates

China on US – Apr. 2018 Link 2.44 2.4 2.39 Source

EU on US – Jun. 2018 Link 4.23 3.2 3.24 Source

Canada on US – Jul. 2018 Link 17.8 12.8 12.76 Source

China on US – Jul. 2018 Link 29.2 34

China on US – Aug. 2018 Link 21.9 16

China on US – Jul.+Aug. 51.1 50 49.8 Source

China on US – Sep. 2018 Link 52 60 53.4 Source

Mexico on US – Jun. 2018 Link 4.51 3.8

India on US – Jan. 2019 Link 0.89 1.3 1.3 Source

Turkey on US – Jun. 2018 Link 1.56 1.8

Russia on US – Aug. 2018 Link 0.27 0.43

China on US – Sep. 2019 Link 112

China on US – Dec. 2019 Link 160

B Additional Results

Figure B1 breaks down Figure 2 in the main text based on major classiﬁcations on the types

of goods subject to new U.S. import tariﬀs. As indicated in the ﬁgure, the tariﬀs imposed

by the U.S. in 2018 were targeted mostly against intermediate goods (shown in blue) and

capital goods (red), which are inputs to production for U.S.-located manufacturers. Later

rounds of tariﬀs imposed in 2019, which fall outside of our period of analysis and are shown

by light shading, were focused relatively more on consumer goods.

Figure B1: Composition of New U.S. Import Tariﬀs: 2018-2019

Solar Panels,

Wash. Mach.

Sec.201: $7B

Steel, Aluminum

Sec. 232: $40B

Chinese Imports

Sec. 301: $506B

0 200 400 600

Billions of USD (Based on 2017 Imports)

2018m1 2018m7 2019m1 2019m7 2020m1

Consumer Goods

Capital Goods

Intermediate Goods

Source: U.S. International Trade Commission for 2017 Import Values. See Table A1 for details on the set of

relevant products and trade values. Classiﬁcation comes from the Broad Economic Categories (BEC) from

the United Nations. Further details can be found here. Lightly shaded areas are used for tariﬀs imposed in

late 2019, which fall outside the main period of analysis for this paper.

We also report coeﬃcient estimates for the control variables used in equation (7). These

variables—industry export share of output and industry import share of domestic absorption—

are intended to capture any features of industry exposure to things like exchange rate move-

ments or overall foreign growth; these eﬀects may also capture some of the potential impact

from increased uncertainty on international markets more generally. Figure B2 reports these

results pertaining to employment, industrial production, and PPIs.

Figure B2: Coeﬃcient Estimates from General International Conditions

(a) Employment

-.06-.04-.020.02.04

90% CI

2017m1 2018m1 2019m1 2020m1

Import Share of Consumption

-.05 0 .05 .1 .15

90% CI

2017m1 2018m1 2019m1 2020m1

Export Share of Shipments

(b) Industrial Production (Output)

-.2 0 .2 .4

90% CI

2017m1 2018m1 2019m1 2020m1

Import Share of Consumption

-.3 -.2 -.1 0 .1 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Export Share of Shipments

-.1 0 .1 .2 .3

90% CI

2017m1 2018m1 2019m1 2020m1

Import Share of Consumption

-.3 -.2 -.1 0 .1

90% CI

2017m1 2018m1 2019m1 2020m1

Export Share of Shipments

Sources: Federal Reserve Board (FRB), U.S. Department of Labor, Bureau of Labor Statistics; authors’

calculations.

Notes: Each panel represents results from separate regressions, where we plot the coeﬃcient estimates of

and ϕ

from the detrended version of equation (7). The shaded areas represent 90 percent conﬁdence

intervals, and the two dashed lines represent February 2018 and September 2018, the ﬁrst and last wave of

new 2018 tariﬀs.

B.1 Alternative Speciﬁcations

Figure B3 presents the results from estimating equation (7) without ﬁrst removing an

industry-speciﬁc trend from each dependent variable. As indicated in the ﬁgure, there is

evidence that some industries were on diﬀerent pre-trends in terms of the outcome variables

prior to the imposition of tariﬀs. For example, the estimates for employment, shown in Panel

a of Figure B3 appear to be on upward trends with respect to rising input costs and foreign

retaliation and a downward trend for import protection.

As discussed in the main text, we control for these diﬀering pre-trends by either detrend-

ing the dependent variable based on its 2017 linear trend or subtracting out the pre-trend

following the approach in Finkelstein (2007). Our approach, therefore, treats these time

trends as resulting from independent random shocks that aﬀect outcomes in each industry

over the short term, rather than long-term trends associated with particular industry char-

acteristics. This assumption seems reasonable given our use of high-frequency data focusing

on a relatively short period of time. It is also supported by our ﬁnding, shown in Figure

B3 that the pre-trends often move in diﬀerent directions for diﬀerent tariﬀ channels, even

though measures of each of the tariﬀ channel exposure variables are positively correlated

with one another.

B.2 Univariate Results

Figure B4 presents results of regressions of the three outcome variables on individual tariﬀ

channel measures, one at a time, as opposed to including the three channels together in the

same regression. Thus, Figure B4 shows the results of nine separate regressions, rather than

three separate regressions as in Figure 4 in the main text.

There are some similarities between these “univariate” regression results and the main

results shown in Figure 4. Figure B4 still ﬁnds negative relationships between employment

and both the rising input cost and foreign retaliation channels. There are also important

diﬀerences, however. While the middle column of Panel C in Figure B4 ﬁnds a positive

relationship between the eﬀects of import protection and producer prices, the equivalent

panel in Figure 4 reveals that this result is not present once the measure for rising input

costs is included in the regression.

B.3 Results by Tariﬀ Wave

The main results presented in Figure 4 calculate exposure to three tariﬀ channels based

on cumulative values of aﬀected trade, covering all tariﬀs imposed over the course of 2018.

Figures B5, B6, and B7, on the other hand, show the results of regressions that include

separate measures for each of the individual waves of tariﬀs. Therefore, Figures B5, B6,

and B7 each show the results of a single regression. The vertical lines in each of the ﬁgures

represent the time that each individual tariﬀ wave took eﬀect.

These ﬁgures yield several ﬁndings on the eﬀects of each individual tariﬀ wave. First,

Figure B3: Eﬀects of Cumulative Tariﬀs (Non-Detrended)

(a) Employment

-.5 0 .5 1

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.2 -.1 0 .1 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-1.5 -1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

(b) Industrial Production (Output)

-1 -.5 0 .5 1 1.5

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.4 -.2 0 .2 .4

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-2 -1 0 1 2

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

-2 -1 0 1 2

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.3 -.2 -.1 0 .1 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-2 -1 0 1

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

Sources: Federal Reserve Board (FRB), U.S. Department of Labor, Bureau of Labor Statistics; authors’

calculations.

Notes: Each panel represents separate regressions, with each column of the panel corresponding to the ﬁrst

three components of equation (7). The shaded areas represent 90 percent conﬁdence intervals, and the two

dashed lines represent February 2018 and September 2018, the ﬁrst and last wave of new 2018 tariﬀs.

we typically ﬁnd the most pronounced relationships with outcome variables coming from

the March (steel and aluminum) and September (largest China 301 wave) waves of U.S.

tariﬀs. This is apparent, for example, in the eﬀects of the rising input cost tariﬀ waves on

employment (Panel a of Figure B5) and producer prices (Panel a of Figure B7). Second, we

Figure B4: Univariate Eﬀects of Cumulative Tariﬀs (Detrended)

(a) Employment

-.8 -.6 -.4 -.2 0 .2

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.1 -.05 0 .05 .1

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-1.5 -1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

(b) Industrial Production (Output)

-1 -.5 0 .5 1

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.4 -.2 0 .2 .4

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-2 -1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

-1 0 1 2 3 4

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs

-.2 0 .2 .4 .6 .8

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection

-1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation

Sources: Federal Reserve Board (FRB), U.S. Department of Labor, Bureau of Labor Statistics; authors’

calculations.

Notes: Each chart represents separate regressions, with only the particular channel present. The shaded areas

represent 90 percent conﬁdence intervals, and the two dashed lines represent February 2018 and September

2018, the ﬁrst and last wave of new 2018 tariﬀs.

ﬁnd that producer prices respond quickly to tariﬀs. Panel a of Figure B7 shows producer

prices rising immediately in response to the March and September rounds of tariﬀs.

Figure B5: Employment Eﬀects by Tariﬀ Wave (Detrended)

(a) Rising Input Costs

-1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Mar.)

-5 0 5 1015

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Jul.)

-20 -10 0 10

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Aug.)

-6 -4 -2 0 2

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Sep.)

(b) Import Protection

-.2 0 .2 .4 .6

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Mar.)

-1-.50 .5 11.5

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Jul.)

-3 -2 -1 0 1

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Aug.)

-.10.1.2.3.4

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Sep.)

-60-40-20 0 20

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Apr.)

-5 0 5 10

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jun.)

-4 -3 -2 -1 0 1

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jul.)

-3 -2 -1 0 1 2

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Aug.)

-1.5 -1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Sep.)

Sources: U.S. Department of Labor, Bureau of Labor Statistics; authors’ calculations.

Notes: The shaded areas represent 90 percent conﬁdence intervals.

Figure B6: I.P. Eﬀects by Tariﬀ Wave (Detrended)

(a) Rising Input Costs

-1.5-1-.50 .5 1

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Mar.)

-20 0 20 40

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Jul.)

-40-20 0 2040

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Aug.)

-20 -10 0 10

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Sep.)

(b) Import Protection

-.5 0 .5 1 1.5

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Mar.)

-10 -5 0 5

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Jul.)

-5 0 5 10

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Aug.)

-1 -.5 0 .5 1

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Sep.)

-40-20 0 20 40

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Apr.)

-5 0 5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jun.)

-6 -4 -2 0 2 4

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jul.)

-10 0 10 20

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Aug.)

-2 -1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Sep.)

Sources: Federal Reserve Board; authors’ calculations.

Notes: The shaded areas represent 90 percent conﬁdence intervals.

Figure B7: P.P.I. Eﬀects by Tariﬀ Wave (Detrended)

(a) Rising Input Costs

-1 0 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Mar.)

-10-5 0 5 10

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Jul.)

-20-10 0 10

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Aug.)

-5 0 5 1015

90% CI

2017m1 2018m1 2019m1 2020m1

Rising Input Costs (Sep.)

(b) Import Protection

-1 0 1 2

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Mar.)

-2-10 1 2 3

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Jul.)

-2 0 2 4

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Aug.)

-1.5-1 -.5 0 .5

90% CI

2017m1 2018m1 2019m1 2020m1

Import Protection (Sep.)

-50 0 50 100

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Apr.)

-6 -4 -2 0 2 4

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jun.)

-2 0 2 4 6

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Jul.)

-15-10 -5 0 5

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Aug.)

-4 -2 0 2 4

90% CI

2017m1 2018m1 2019m1 2020m1

Foreign Retaliation (Sep.)

Sources: U.S. Department of Labor, Bureau of Labor Statistics ; authors’ calculations.

Notes: The shaded areas represent 90 percent conﬁdence intervals.