VoxEU Column International trade

Global value chains have made countries remove some trade protection

The decades preceding the Trump era saw a significant decline in trade barriers and a concurrent rise in global value chains. Evidence on the direction of causality between the two is still lacking. Using an exogenously timed WTO requirement for countries to re-evaluate previously imposed tariffs, this column argues that increased activity through global value chains had an important role to play in the countries’ choice to reduce trade protection during this period. 

The decline in trade barriers and the rise in global value chains are two of the most important trade-related developments in the decades preceding the Trump era. Countries opened up to trade in a variety of ways during this period: multilaterally, reciprocally, preferentially, and sometimes even unilaterally. Additionally, this period also witnessed an increase in fragmentation of firms’ production across borders. This was brought about through foreign direct investment in some cases and arms-length transactions in others, generating the spider or snake supply chains which explain the boom in international commerce in parts and components during the period.

The concurrence of events has raised obvious questions: Did trade liberalisation cause supply chain growth? Or did these newly evolving supply chains change the political economy of protection? Surely it was some of both, and elements of each have been exploited in different empirical settings, as described in Johnson and Noguera (2017).1 Yet, examining causality remains a challenge, given that natural experiments do not often lend themselves to trade policy.

In new research, we investigate this question by taking advantage of an exogenously timed WTO requirement for countries to re-evaluate some of their previously imposed tariffs (Bown et al. 2020). We use the setting of this quasi-natural experiment to document evidence that countries’ choice to reduce their trade protection is caused by increased activity arising through global value chains.

The policy change creating the testing environment

Antidumping protection quickly became a highly utilized form of protection across the major economies during the WTO period (Prusa 2001). Prior to 1995, antidumping use was mostly confined to four high-income economies: the US, the European Community, Australia, and Canada. Yet even before the WTO, exporting countries such as Japan and South Korea increasingly found their industries hurt by imposition of such duties. Consequently, the Uruguay Round featured negotiations from countries concerned that such temporary protection would remain in place for decades and thus become quasi-permanent, eroding the market access benefits of broader tariff reduction commitments. 

The 1995 WTO agreement introduced a new legal requirement in response to these fears. Countries that imposed antidumping duties would be forced to undertake a mandatory ‘Sunset Review’ investigation at the five year mark and consider removing the protection. Additionally, countries with large stocks of antidumping duties already in place in 1995, like the US, were also told to review all of those pre-existing duties.

Our empirical setting relies on this newly agreed WTO legal requirement as a quasi-natural experiment to examine the influence of changes in global value chain activity on levels of antidumping protection. Our key measure of global value chains follows the theoretical approach of Blanchard et al. (2017) and involves the domestic value added (DVA) content in foreign production. The basic theoretical prediction of their model is that more DVA makes tariff protection less desirable. This relationship is arrived at through a terms-of-trade channel. The intuition is that an imported product with higher DVA is more exposed to losses feeding back through the supply chain – i.e. reduced factor returns on domestic inputs – when a tariff drives down the world (exporter-received) price. 

The key variable for each country’s protection removal decision is the change in DVA over the four years prior to protection renewal decision. Thus, imported products with changed economic circumstances in the form of additional value added ascribed to the country that originally imposed the protection, should be more likely to have that original protection removed.

Anecdotes illustrate the point

In 2000, the new WTO legal requirement forced the US to make two decisions involving previously imposed antidumping duties on the same product – brass sheet and strip – from two different countries: Japan and the Netherlands. Between 1995 and 1999, the US value added in Japanese basic metals production declined by 64%, whereas in Dutch metals it grew by 3%. The US decided to keep the tariffs on Japan but removed those on the Netherlands.

Our econometric framework allows us to ensure such decisions are not simply a trading-partner specific phenomenon. In 2005, the WTO requirement forced the EU to reconsider three different previously imposed duties on India. The EU removed tariffs on non-alloy steel hot rolled flat products and cathode-ray colour TV picture tubes where its DVA in Indian basic metals and electronics production had accelerated by 71% and 64%, respectively. At the same time, the EU did not remove its tariffs on steel wire rope where its DVA growth in Indian fabricated metals was only 32%.

Our formal econometric setting exploits these, and other, margins of variation in the data, and introduces a host of other determinants of protection.

The formal approach and results

Our empirical setting involves the universe of antidumping review decisions for 10 countries, 41 trading partners, and 18 industries over 1995-2013. The 10 policy-imposing countries include four high-income economies (Australia, Canada, the EU, and the US) and six emerging economies (Argentina, Brazil, China, Mexico, India, and Turkey) that together accounted for 71% of world GDP in 2013. The OECD’s Trade in Value Added (TiVA) data are used to link these removal decisions to industry-level estimates of bilateral value added content in production. In addition to exploiting the quasi-natural setting of the WTO’s new rules beginning in 1995, the econometric approach is to instrument DVA growth by exogenous export supply shocks, since DVA growth in production might be related to an existing trade barrier.

Our results indicate that the decisions of high-income countries to remove trade barriers are (statistically and economically) affected by the extent of value chain integration. This result is not driven by imports from China, which is by far the top export target of virtually all countries’ antidumping duties during this period. However, the results do not hold for the antidumping removal decisions of emerging countries, the major new users of such policies after onset of the WTO. 

Antidumping protection is the most important temporary trade barrier in use during 1995-2013, though it is, of course, not the most important trade policy empirically. That would be the first line of protection – most favoured nation (MFN) status and preferential tariffs. The benefit of the antidumping removal decision is that it is a relatively clean policy environment to test the theory.

Additional policy implications

These results may contribute to a separate debate, involving what really led to the decline in use of antidumping (and other trade remedies) in a country like the US during the 2000s. This debate is critical right now, as Trump administration officials are blaming the WTO’s Appellate Body for the American downsizing of such protection. In their view, WTO judges made rulings that forced the US to excessively remove antidumping tariffs. For that and other reasons, the US administration has disabled the function of the WTO’s dispute settlement system (Bown and Keynes 2020, USTR 2020).

The results from this research, however, suggest that at least some of that blame is misplaced. Reduction in antidumping – through the channel of increased removal of antidumping duties previously in place – was partially due to the changing nature of American firms’ own economic interests arising through supply chains. As their exposure to the adverse consequences of tariffs through global value chains increased, in many instances, they simply no longer wanted the protection.


Blanchard, E, C P Bown, and R C Johnson (2017), “Global Value Chains and Trade Policy”, CEPR Discussion Paper No. 11044 (revised).

Blanchard, E and X Matschke (2015), “U.S. Multinationals and Preferential Market Access”, The Review of Economics and Statistics 97(4): 839-854.

Bown, C P, A Erbahar, and M Zanardi (2020), “Global Value Chains and the Removal of Trade Protection”, CEPR Discussion Paper No. 14451.

Bown, C P and S Keynes (2020), “Why Trump Shot the Sheriffs: The End of WTO Dispute Settlement 1.0”, CEPR Discussion Paper No. 14477.

Johnson, R C and G Noguera (2017), “A Portrait of Trade in Value-Added over Four Decades”, The Review of Economics and Statistics 99(5): 896-911.

Ludema, R D, A M Mayda, M Yu, and Z Yu (2018), “The Political Economy of Protection in GVCs: Evidence from Chinese Micro Data”, mimeo.

Prusa, T J (2001), “On the Spread and Impact of Antidumping”, Canadian Journal of Economics 34(3): 591-611.

USTR (2020), Report on the Appellate Body of the World Trade Organization, Washington, DC.


1 Johnson and Noguera’s (2017) empirical contribution is to examine this question over four decades by focusing on regional trade agreements. Other approaches include tariff preferences offered to poor countries under GSP and US outbound foreign direct investment (Blanchard and Matschke 2015), a cross-country study of tariffs and value-added measures of production and trade (Blanchard et al. 2017), and a study that focuses on China using transaction-level data (Ludema et al. 2018).

2,310 Reads