VoxEU Column International trade

Tilting the playing field: FTAs and the changing pattern of protection

While countries rush to enact more and more free-trade agreements, not enough is known about their impact. This column presents evidence suggesting that free-trade agreements are more discriminatory than their preferential tariffs suggest. It finds a stark increase in contingent protection as free-trade agreements cause a 10%-30% increase in the number of antidumping disputes against non-member countries.

Two of the most popular changes in trade policy in recent years are free-trade agreements and using antidumping duties to restrict trade from low-wage countries.

  • The number of countries using antidumping has increased five-fold and the annual number of antidumping disputes has more than doubled (Prusa 2005 and Bown 2010).
  • The growth in free-trade agreements (FTAs) is no less astounding. Some 474 FTAs have been notified to the WTO to date, counting goods and services notifications separately. There is no reason to think the proliferation of FTAs will end soon.

Both of these trade policies can be viewed as double-edged swords. Preferential liberalisation introduces new distortions (the preferences) while removing others (the liberalisation). Antidumping duties are protection pure and simple, but are frequently used as the "escape valve" necessary to grease the wheels of prior trade liberalisation.

While often analysed in isolation, one long-standing hypothesis is that the two interact in pernicious ways. Bhagwati (1992, 1993) and Bhagwati and Panagariya (1996), for example, argue that the selective use of antidumping can increase the risk that preferential trade agreements will end up stressing the preferences and minimising the liberalisation.

In recent research (Prusa and Teh 2010), we find support for the interaction conjecture. We find that FTAs discriminate both directly – through preferential tariffs – and indirectly – through the altered pattern of antidumping activity. Our research establishes a protection analogue to the well known “trade creation-trade diversion” impact of FTAs. FTA members are spared antidumping actions (“protection reduction”) but non-FTA members face even greater antidumping scrutiny (“protection diversion”).

Data used in the study

Our antidumping data consists of all such cases filed since 1980 using official country reports to the WTO. Since 1980, 5,006 antidumping filings were initiated. Of this total, only 201 were initiated by countries that are not part of any FTA. This leaves a sample of 4,805 antidumping filings, 745 of which were taken against FTA members while 4,060 were taken against non-FTA members.

The mapping of antidumping provisions in 80 FTAs is based on work done by Teh et al. (2009). The FTAs cover almost 50% of worldwide exports; include the two largest FTAs, the EU and NAFTA, and are geographically diverse with North-North, North-South and South-South FTAs. We divide the antidumping provisions in FTAs into three mutually-exclusive groups:

(i) those with no antidumping provisions (17 FTAs);

(ii) those that prohibit the use of antidumping (9 FTAs); and

(iii) those that include some rules governing antidumping use (54 FTAs).

Because antidumping use is governed by the WTO Antidumping Agreement, we expect that if FTAs rules have any impact, they will serve to make antidumping duties more difficult to impose on FTA members.

Figure 1 provides compelling support for the protection reduction hypothesis. It shows the number of antidumping disputes drop sharply in the year the FTA is introduced (t=0) and remains much lower in subsequent years as compared to the years prior to enactment. On average, during the 10 years prior to enactment there were 29.5 antidumping cases per year and during the 10 years following enactment there were just 23.6 cases per year.

Figure 1. Intra-FTA antidumping filings

Estimating the impact of FTAs on antidumping actions

We then apply a widely used statistical approach – the difference-in-difference estimator. This compares antidumping activity against two groups of countries (FTA members and non-FTA members) for two time periods (pre- and post-FTA establishment) to isolate the impact of FTA membership from the impact of other factors over time. Table 1 depicts antidumping filings when we categorise antidumping filings in this way. As seen, countries file about 58% of antidumping cases against non-FTA countries prior to FTA enactment but a remarkable 90% following enactment. Again, this strongly suggests that FTAs are changing the pattern of protection.

Table 1. Antidumping activity by FTA status

Target Country
Non-FTA Country
FTA Country
Protection reduction

Antidumping filings are driven by a variety of economic factors (Knetter and Prusa 2003, Blonigen 2005) and a statistical analysis must control for the effect of being an FTA member as well as other exogenous factors. The formal regression analysis implies that FTAs cause as much as a 60% reduction in intra-FTA antidumping disputes. Importantly, we find this result is not solely driven by those FTAs that have abolished antidumping (for whom intra-FTA antidumping activity is essentially eliminated). When we just look at those FTAs that have adopted antidumping rules we find a 33-55% reduction in intra-FTA antidumping disputes. We find no significant change in antidumping activity for FTAs without antidumping rules.

Protection diversion

We next turn to the Bhagwati-Panagariya hypothesis. According to their explanation, contingent protection is driven by import volume. Who is targeted in the antidumping petition is entirely up to the discretion of the domestic industry. If antidumping provisions make FTA members more difficult to sanction, then the domestic industry will simply target other sources. As a result, we might see an increase in antidumping protection directed toward non-FTA members when in fact the injury mostly stems from intra-FTA imports. Once we control for the changing pattern of bilateral imports our results imply FTAs cause a 10%-30% increase in the number of antidumping disputes against non-FTA members.

Net effect

Taking the protection reduction and diversion results together, we find the reduction in intra-FTA activity is more than offset by the increase in activity against the far larger set of non-FTA members. Our bottom-line assessment is that FTAs increase the overall number of antidumping disputes by perhaps as much as 10%. Therefore, the main impact of FTAs is on the pattern of protection rather than the level of protection.

Confirming that the results stand up

How robust are these results? It turns out that they are extremely stable, and the estimates remain economically large and statistically significant under a large number of alternative specifications. We provide several robustness checks including controls for the possibility that other FTA provisions may be driving our results, We also drop a number of big users and targets such as the EU and China to ensure our findings are not distorted. Finally, we control for possible inconsistency in our standard errors (see Prusa and Teh 2010 for more details).


Our research presents new evidence that FTAs significantly alter the pattern of protection. To the extent that FTAs adopt special or additional rules on antidumping actions against members’ trade, they can effectively increase the level of discrimination against non-members. This increase in discrimination occurs when FTA members abolish antidumping actions against FTA members but not against non-members’ trade. It also occurs when FTA members adopt rules that strengthen disciplines on antidumping actions against FTA members but not against non-members.

At first sight, FTA efforts moves to strengthen disciplines on antidumping appear good for trade. Yet, because FTAs thrust us into the world of the second best, actions that look like they will lead to an increase in economic efficiency may achieve the exact opposite effect. Our results reaffirm the importance of multilateral trade reform and tougher scrutiny of FTAs in the WTO.

The opinions expressed in this paper should be attributed to the authors. They are not meant to represent the positions or opinions of the WTO and its Members and are without prejudice to Members' rights and obligations under the WTO.


Bertrand, M, E Duflo and S Mullainathan (2004), “How Much Should We Trust Differences-in-Differences Estimates?” Quarterly Journal of Economics, 119(1): 249-75.

Bhagwati, J (1992), “Regionalism versus multilateralism”, World Economy 15(5): 535-56.

Bhagwati, J (1993), “Regionalism and Multilateralism: An Overview,” in J de Melo and A Panagariya (eds.), New Dimensions in Regional Integration, Cambridge University Press.

Bhagwati, J and A Panagariya (eds.) (1996), The Economics of Preferential Trade Agreements. Washington, DC: AEI Press.

Blonigen, BA (2005), “The Effects of NAFTA on Antidumping and Countervailing Duty Activity,” World Bank Economic Review, 19(3): 407-424.

Bown, Chad P (2010), “The WTO dispute settlement system would survive without Doha”, VoxEU.org, 19 June.

Dee, P (2008) “Multinational corporations and Pacific regionalism,” in J Palacios (ed.), Multinational Corporations and the Emerging Network Economy in Asia and the Pacific, Routledge, London and New York, 232-266.

Knetter, MM and TJ Prusa (2003). “Macroeconomic Factors and Antidumping Filings: Evidence from Four Countries”, Journal of International Economics 61(1): 1-17.

Prusa, TJ (2005), “Antidumping: A Growing Problem in International Trade’, The World Economy, 28(5): 683–700.

Prusa, TJ and R Teh (2010), “Protection Reduction and Diversion: FTAs and the Incidence of Antidumping Disputes,” NBER Working Paper 16276.

Teh, R, M Budetta, and TJ Prusa (2009) “Trade remedy provisions in regional trade agreements,” in A Estevadeordal, K Suominen, and R Teh (eds.), Regional Rules in the Global Trading System, (Cambridge University Press), Chapter 4, 166-249.

1,575 Reads