引言

在一般和特定情況下的的保護措施的非關稅貿易壁壘的重要性的增長，已成為全球貿易政策行為的一個顯著特征，無論最近是在發達國家還是在發展中國家（Prusa2001和2005；Zanardi2004；Bown2008）減少關稅保護和保護措施的范圍正日益被視為替代貿易政策工具來保護國內生產商和工業（Blonigen，Prusa，2003；Konings 和Vandenbusche,，2005）。

最初的“平等”貿易措施反傾銷特點突出存在于有機保護措施中。給出了反傾銷政策制定和實施的方法以及為了產生政治經濟影響力實現相當大的空間，因此為保護政策的游說提供了另一種渠道（Nelson，2006）。


Introduction

The growth in importance of non-tariff trade barriers in general and contingent protection measures in particular has become a remarkable feature of the conduct of global trade policy,both in the developed and more recently also in the developing world(Prusa,2001 and 2005;Zanardi,2004;Bown,2008).With reduced scope for tariff protection contingent protection measures are increasingly seen as alternative trade policy instruments to protect domestic producers and industries(Blonigen and Prusa,2003;Konings and Vandenbusche,2005).

Originally devised as‘fair’trade measures antidumping features prominently amongst the forms of contingent protection.Given the way in which antidumping policies are set up and implemented considerable room for political-economy influences tends to be created,thus generating an alternative channel for protection lobbying(Nelson,2006).

Support for the hypothesis of declining tariff protection being replaced by an enhanced use of antidumping investigations may be found in early descriptive studies identifying anti-dumping as“a major loophole in the free-trading disciplines of the world trading system”(Lindsay and Ikenson,2001:5).2 Thorough empirical evidence on the subject matter is however still scarce and tends to be industry analysis and characterized by mixed results.3 Focusing on the impact of the Uruguay Round tariff concessions,Feinberg and Reynolds(2007)analyse subsequent antidumping investigations in 19 different industries for several countries between 1996 to 2003.They find evidence for trade policy substitution mostly in developing countries.4 Traditional users of anti-dumping measures(i.e.Australia,Canada,New Zealand,the EU and US)are not found to show a positive correlation between tariff protection and anti-dumping proceedings.

1 Moore and Zanardi(2011)further add to these findings by examining the relationship between the probability of AD investigations and applied(rather than bound)tariffs between 1991 and 2002.Analysing AD fillings in 291 Nelson(2006:554)reviews of the literature on the political economy character of antidumping measures highlighting that“[..]antidumping is a much worse a problem than its small coverage and marginal contribution to aggregate protection would imply.

2 Vandenbussche and Zanardi(2010),moreover,find in this context that antidumping measures considerably affect trade in industries which are not directly involved in the investigation thereby characterising antidumping investigations as a potentially very powerful tool of alternative import protection.A view which is also held by Blonigen and Prusa(2003:253)who state that most people“agree that AD has nothing to do with keeping trade‘fair’[…]It is simply another form of protection”.

3 There is a related literature that focuses on the political choice between tariffs and other forms of(non-ADrelated)import protection.Hillman(1990),Hillman and Ursprung(1988)and Feenstra and Lewis(1991)analyse the use of tariffs and Voluntary Export Restraints(VERs)showing that the latter may,under certainassumptions,be preferred to tariffs.Limão and Tovar(2011)provide theoretical and empirical evidence for a substitution scenario of tariffs for non-tariff-barriers(NTBs)in general.

4 Feinberg and Reynolds(2007)focus on AD petitions in HS 1-digit industries.ISIC 3-digit manufacturing industries and numerous countries,the authors are not able to confirm the findings of a positive correlation between tariffs and antidumping,with the exception of a small group of developing economies.

5 By contrast,the sole study conducted at a detailed product level,Bown and Tovar(2011)provides support for the hypothesis of tariffs being substituted by more frequent AD investigations when analysing India’s antidumping proceedings in the face of a major tariff reform programme.

6 We seek to contribute to the existing literature by examining a potential product-level link between(bound)mfn tariff cuts conceded by one of the world’s largest traders–the EU–and the latter’s subsequent antidumping investigations.Our study contrasts to much of the previous empirical evidence by focusing on detailed and country-specific HS 8-digit productlevel AD investigations,and(bound)MFN tariff concessions for a large and developed economy.

7 As pointed out by Feinberg and Reynolds(2007)the fact that industry classifications usually include several hundreds of individual product lines,industry and country level studies may lead to biased results,since sectors with a large variation in product level tariff cuts and possibly very small aggregate tariff reductions are be more likely to attract subsequent AD investigations than industries with a large aggregated degree of tariff liberalization but no extreme product level tariff reductions.Our research is motivated by Anderson and Schmitt’s(2003)theoretical contribution
which analyses the effect of binding tariff reductions on the use of quantitative import restrictions and anti-dumping measures.

8 Based on Brander and Krugman’s(1983)reciprocal dumping model,these authors derive a theoretical framework of preference progression for different forms of trade policy protection.They show that in an unrestricted trade policy.These include Argentina,Brazil,China,India,Mexico,Peru and South Africa(Moore and Zanardi,2011

As a result,when constraining the use of tariffs by coordinated negotiations,policy-makers are likely to resort to the use of quantitative trade policy instruments which are again superseded by the use of antidumping actions in the presences of additional agreements on‘quota tariffication’.Restrictions on the use of tariffs and quotas will thus result in an enhanced use of antidumping protection.This trade policy preference progression tends to be in line with some stylized facts regarding the historical use of trade policy instruments.Coinciding with the end of the Kennedy Round(1964-1967),the 1960’s witnessed an upsurge of quantitative import barriers which was followed by an increasing trend towards antidumping measures since the 1980s(Renner,1971;Finger and Olechowski,1987).The Uruguay Round(1986-1994)finally established a guideline for the‘tariffication’of quantitative import restrictions for all GATT-signatory countries and additionally required them to restrict the use of quotas in the future,whereas the use of AD measures remains largely WTO-unconstrained.Import protection following the Uruguay Round(UR)tariff commitments therefore represents an interesting testing environment for a potential substitution effect of greater use of antidumping measures in response to falling tariffs.

Focusing on the UR trade policy outcome,our findings show a highly significant, albeit small,positive impact of bound MFN tariff concessions on the probability of subsequent antidumping investigations;having controlled for other influences.Employing a variety of different econometric techniques,including random-effects and a Chamberlain- Mundlak approach to control for unobserved heterogeneity,this finding is robust to a series of sensitivity tests.

Legal framework
The EU’s trade policy is governed by the European Council and the European Commission. While the Commission proposes and enforces trade policy actions, the Council, consisting of Member States’ representatives, decides about approval or rejection of the Commission’s propositions. Antidumping measures represent a major component of the EU’s trade policy mix (Rovegno and Vandenbusche, 2011). Guided by Article 207 of the Treaty on the Functioning of the European Union as well as Council regulation 1225/2009, the EU’s antidumping legislation is embedded in the WTO’s antidumping policy framework allowing GATT signatory countries to impose discriminatory trade protection measures if foreign exporters sell their goods at a price lower than their ‘normal value’,and if the latter results or threatens to result in ‘material injury’ for the domestic industry。

The initiation of an antidumping investigation on part of the EU’s antidumping authorities requires an officially lodged complaint by a Community industry which needs to provide evidence of dumping and the resulting causal material injury. Additionally, any antidumping complaint must be supported by enough EU producers responsible for at least 25% of the EU’s product-specific production. EU regulations further specify a timeframe of 45 days for the Commission to decide whether to open an investigation or not. Preliminary measures may be imposed after an initial investigation period of 9 months, during which (mostly questionnaire-based) consultations are held with EU producers and importers as well as the investigated exporters. The time span from the opening of an investigation to the publication of the final decision may therefore take up to 15 months. The ‘normal value’ of a product is in general defined as the country of origin’s production costs plus reasonable profit margins and additional costs for selling and administration. In calculating the normal value the European Commission distinguishes between whether the investigated country is a market economy or not. If it is not, an analogue country, often already proposed by the complaining industry, serves as a proxy (Liu and Vandenbusche, 2002). In light of the difficulties of estimating production costs, the European Commission often uses domestic sales prices in the exporting country to calculate the normal value. Price information of the analogue country is also used if domestic sales in the exporting, or analogue, country are too small to be representative. For more detailed information on the determination or ‘construction’ of the normal value see Macrory et al. (1991

The EU can initiate anti-dumping investigations against all non-EU member countries, with an almost complete exception of goods stemming from Iceland, Lichtenstein and Norway (i.e. the EEA countries). Comparing foreign suppliers’ export prices with ‘normal values’, the European Commission first investigates whether there is enough proof for the existence of dumping following a complaint of a Community industry. While the investigated export price refers to the ex-factory price - i.e. the price for goods sold to the EU net of rebates, discounts, taxes, etc. (Macrory et al., 1991), the normal value of a product is most often calculated on the basis of domestic sales prices of the like product in the exporting country. The difference between the latter two – i.e. the dumping margin – is then calculated according to one of three alternative measures specified in the WTO’s Antidumping Agreement (ADA).

The determination of causal material injury to the domestic industry, or a threat thereof, includes an economic analysis of various domestic industry factors such as, output, productivity, profits, utilisation capacity, stocks, sales, market share, cash flow return on investment and employment, and also compares the foreign producers’ export prices to the prices charged by the domestic industry (i.e. the injury margin).12 If the Commission considers the evidence for dumping and material injury to be sufficient as well as potential trade defence actions to be in line with the general interest of the Community, the former finally proposes antidumping measures which may either take the form of price-undertakings or additional duties to offset the injury caused by the dumped products.

Despite the fact that antidumping investigations directly target exporting firms and tend to impose firm-specific trade remedy duties, not investigated firms originating from the same country are most often also subjected to additional duties even if the latter were not involved in dumping activities.

The difference between a calculated normal value and the foreign firm’s export price determines the dumping margin. The WTO’s Antidumping Agreement (Article II) specifies three alternative approaches for contrasting the latter two prices: (i) comparing weighted averages of both price indices, (ii) comparing both price indices for each (product-level) transaction averaging the latter to compute the overall dumping margin, or (iii) contrasting weighted normal values with individual transaction based foreign producers’ export prices if the latter vary substantially across purchasers, time periods or regions. The latter method is also followed by the averaging of all transaction-to-transaction based dumping margins. Closely associated with the calculation of dumping margins is the methodology of ‘zeroing’. ‘Zeroing’ denotes the replacement of negative dumping margins by zeros which may finally results in larger average dumping margins. For a recent discussion regarding the different approaches of zeroing and associated WTO litigations see Prusa and Vermulst (2010

When calculating the material injury of alleged dumping activities, the EU, like many other users of AD actions, often applies the principle of cumulation, which allows considering the combined impact of all imports from the investigated exporting countries on the domestic industry. Hansen and Prusa (1996) as well as Tharakan et al. (1998) find that cumulation significantly increases the probability of finding evidence for material injury.