3. Winning Coalition Size and the Use of Anti-dumping

According to Aggarwal (2004), AD cases account for almost 90 % of contingent protection cases with countervailing duties and safeguards coming a distant second and third.

Bown (2009a) classifies disputes from 1995 to 2008 by alleged policy causes that led to market access in order to gain insights into different sources of dispute and the impact of visibility. Note that this categorization counts cases that “directly” concern AD measures, whereas cases in which the general legality of rules and practices is challenged are treated separately but can also concern AD.

For the formal explanation see Bueno de Mesquita et al. (2003).

In the model, resources are raised through taxation and then get reallocated. The amount is endogenous with the level of taxation being determined in the model.

For a similar narrative centred on incentives see for instance Acemoglu and Robinson (2012) who argue that ultimately development (failure) comes down to institutions.

The exact size of the winning coalition can vary depending on characteristics of the electoral systems though. See Bueno de Mesquita et al. pp. 54–55.

Note that the operationalisation of w and the possibility to distinguish the effect empirically from other aspects of democracy will be further addressed in the empirical section of this chapter.

Members of the existing winning coalition are partly selected on affinity.

Affinities only break ties in the model but they do not dominate decision-making. In the basic model, the incumbent always survives because affinities are known and he can match the challenger’s proposal.

To be clear, Bueno de Mesquita et al. do address international interactions and propose an argument how institutions shape leaders incentives to engage in armed conflict (also see Bueno de Mesquita et al. 1999). However, their theory does not address “conflicts” among good policies. For instance compliance with international treaties and free trade would typically be considered a public good. So would high public health and safety standards. Similarly, leaders in the EU and the US may hold diverging opinions on issues such as imports of hormone-treated beef and may even end up in a trade dispute. Selectorate theory does not prescribe a hierarchy between public health, free-trade and compliance with international rules, other than that leaders would give priority to the policy goal which is most likely to keep them in office.

However, for a critique of Sachs and Warner see Rodriguez and Rodrik (2001).

See Bueno de Mesquita et al. (2003, p. 197): “Leaders in need of a large coalition must look beyond this small group, which means that they must attract support from voters who do not gain from cutting the economy off to trade.”

See Bueno de Mesquita et al. (2003, p. 74).

Aidt and Gassebner (2010) propose a model with a similar argument, with possible complementarities between institutions and the quality of bureaucracy.

For example, Prusa (2005, p. 334) notes: “If AD is not about making markets competitive, what is it? For all intents and purposes, AD is simply protectionism dressed up in a nice suit. In many ways, AD is an almost ideal instrument of modern protection.”

“Predatory pricing” occurs if exporters deliberately try to put domestic producers out of business by charging unfairly low prices first in order to get rid of competition, later raise prices again and then capture monopoly rents.

There is no specific legal notion of predation in the WTO AD code but typically domestic antitrust rules provide a definition of predation. Also recall that predatory pricing as a strategy would only work with a small number of domestic producers and high domestic entry barriers. Otherwise, once the firm that applied predatory pricing strategies dominates the market and raises prices, it would again attract new competitors. In addition, if duties are levied on similar products exported by different countries this seems difficult to match with the idea of clearly targeted predatory pricing strategies.

The development of two largely separate theoretical literatures focusing on dumping and anti-dumping separately in a way also provides additional indication that often anti-dumping is not necessarily about dumping (Zanardi and Vandenbussche 2008, p. 105).

Note that a recent scholarly debate has evolved asking to what extent the countercyclical pattern applies to the recent financial crisis. As this is not the main focus of this work, results may be sensitive to the methodological approaches taken and the jury is still out on mid- to longer term effects, it will not be reviewed in this context. See however Rose (2012a, b) and Bown (2011) for recent contributions and Evenett (2013) for the latest figures on trade restrictions imposed during the crisis.

Note that Takacs (1981) uses data on safeguards, and that results are based on data that is already a couple of decades old and was before major administrative reforms of the AD mechanism took place. Nevertheless, further analyses also suggest that administered protection can be influenced by domestic conditions.

The analysis includes AD, safeguards and CVD duties.

However, this could be the case if there is a global systemic shock.

The US scores in the 75–90th percentile for government effectiveness and in the 90–100th percentile on control of corruption (values for 2008). See for instance the World Bank Governance indicator (compiled by Kaufmann et al. http://​info.​worldbank.​org/​governance/​wgi/​worldmap.​asp)

The analysis includes safeguards and CVD’s with AD accounting for the majority of observations.

Some evidence (Hansen and Park 1995) suggests a similar effect for Committee representation and financial contributions on the dumping decision and the margin but this should be interpreted with caution due to the skewed distribution of the dependent variable.

Estimates based on a sample of US public companies between 1980 and 1992. Recent anecdotal evidence also suggests that AD investigations impact on firms’ stock value, but this is an area warranting further systematic assessment. See for instance Börsenzeitung (Sept. 26th 2012): “Beschwerde in Brüssel verleiht Solarworld Auftrieb.”

Countervailing duties are meant to punish the use of subsidies that are inconsistent with GATT/WTO provisions. AD measures are by far used more frequently than CVD’s and therefore it seems fair to attribute most of the effects to them accordingly.

Note that these estimates are based on values for 1993 and are only for one country. Hence they do not even take into account the further spread of AD in subsequent years.

Analyses found duties to be on average 10–20 times higher than the scheduled levels of protection.

Calculations based on data from US AD cases.

See Kucik and Reinhardt (2008, p. 485): “We must emphasize one point before proceeding: our hypotheses speak to purely behavioural aspects of the flexibility argument. In contrast, the theory’s welfare claim—that flexibility provisions can make outcomes more efficient—is far more difficult to substantiate.”

See Kucik and Reinhardt (2008, ibid): (…) “Whether breach of contract is on net ‘efficient’ also depends on the possibility for abuse of the flexibility provision, shaped by a number of factors: for example, how clearly its standards are defined; how likely enforcement is in the wake of abuse; and how possible it is for parties to agree on, and for the regime to estimate, the compensation due in the event of an invocation of the flexibility provision.”

See Bown (2005, p.1): “… WTO dispute panels have held and the Appelate Body has confirmed WTO-inconsistencies with at least one element of almost every trade remedy action that they have ruled on”.

Dispute cases do suggest that AD is a source of conflict among members and that there are frequent cases of misuse. Not all of them may be challenged which might partly be due to the fact that targets consider themselves in a position too weak to effectively induce compliance. In that sense, it seems reasonable to expect that part of misuse is unnoticed. At the same time, the high proportion of positive rulings for actual cases can also reflect selection effects to a certain extent. However, Bown (2009a) notes that defendants in AD cases are often proficient AD users, which rules out “inexperience” of new users or lack of administrative capacity to administer AD correctly as a reason for disputes and rather suggests a deliberately wide interpretation on behalf of investigating countries.

Note that AD measures can be imposed on imports from both WTO members and non-members. Therefore information on actions by WTO members matters for non-members as well.

However, they do not find evidence that democracy impacts AD law adoption when controlling for GATT/WTO membership.

In addition, it could be argued that a more heterogeneous group of members after 1995, participation in WTO meetings, negotiation, access to staff and training etc. can help to mitigate capacity constraints at least to some extent.

The main source on countries’ AD laws are Zanardi (2004), Zanardi and Vandenbussche (2008) that is complemented with information from the WTO (regularly reports and notes publicly available on the organisation’s website).

Similarly, Hong Kong and Macau are considered as independent units before they become part of China. Cross sectional averages are adjusted accordingly, i.e. they only consider independent use.

Part-time but not full-time: Bahrain, Cameroon, Costa Rica, Czech Republic, Dominican Republic, Egypt, Estonia, Fiji, Guatemala, Jordan, Kenya, Latvia, Macedonia, Macao, Morocco, Oman, Paraguay, Poland, Slovak Republic, United Arab Emirates.

Note that only cases initiated since WTO membership are counted. This excludes, e.g. cases which China filed previously to joining the WTO as well as Latvia, which imposed several measures in 2001 but had initiated them prior to joining the WTO.

The variable can take on four values 1 = civilian, 2 = military-civilian, 3 = military, 4 = “other”.

In particular, I checked the case of Thailand, which received a score of 1 until 2005 but was coded military-civilian (2) in 2006. This change is most likely due to the military coup, taking place in September 2006. Following the coup, the military proposed a new constitution, which was adopted in summer 2007. In between (for approx. 15 months), Thailand was ruled by a government installed by the military (hence it is coded “2” in 2007). Elections took place in December 2007, which the Thaksin-party won. When ex-prime minister Thaksin came back in 2008, he was first arrested and later freed. A new prime minister took office later in autumn (and because this was in September, most of the year can be considered a kind of “interregnum” but with the military still being important. Hence it is coded “2” for 2008).

Annual values are adjusted to account for changes in EU membership. Averages exclude Cyprus, Malta and Luxembourg because Polity IV does not provide data for them.

The temporary trade barriers database contains material formerly available via the Global Anti-dumping Database also collected by Bown. All data files were downloaded from the Global AD Database as of mid-2010.

This concerns the following countries and years: Costa Rica: 1995, Ecuador: 1996, 1997, Indonesia: 1995, Jamaica: 1995–1999, Pakistan: 1995–2001, Paraguay: 1995–1998, Thailand: 1995, Trinidad & Tobago: 1995, 1996, Uruguay; 1995, 1996.

There is one case listed for Trinidad and Tobago in 1996 but this is included in Bown’s data.

Cumulative counts are weighted by WTO membership years to get comparable averages. For the count model specifications in robustness checks, the information based on WTO membership years is used to extrapolate AD counts for the full time period (also see robustness checks for further explanation). This yields adjusted figures countries that joined the WTO after 1995 or became members of the EU. The hypothetical count figures should be interpreted with caution, as these are rough approximations rather than ‘real’ values. Nevertheless the count models provide a useful additional robustness check to the averages used in the Tobit specifications.

Imports are in USD based on data from the World Development Indicators (World Bank).

Note that slight differences to WTO reporting can arise due to time periods considered (reporting from mid to mid-year vs. Jan. to Dec. and investigations initiated in 2008 but concluded in 2009 for this analysis).

Note that this is also consistent with the usage patterns by country presented in Sect. 3.4. Standard deviations are consistently larger than means across all three categories and samples. The largest variation can be observed for the sample, which includes only countries with AD law during the entire period.

For example, one pattern that is clearly visible in bivariate plots and descriptive analyses are “size effects”, i.e. there seems to be a strong correlation between the number of AD cases and log population/GDP.

Data is based on the Worldbank’s WDI dataset. Values are averaged across the respective observation years.

These characteristics include regional dummies and the religious composition of the respective countries which are used to predict the level of per capita GDP. The batchfile to generate the controls can be obtained here: http://​politics.​as.​nyu.​edu/​docs/​IO/​8146/​BatchFileFinal.​do

Used for w: XRCOMP, XROPEN and PARCOMP. All three of them are also used to construct the Polity-scale which is used to control for democracy. The Polity measure additionally includes XTCONST and PARREG. See Bueno de Mesquita et al. (2003, p. 395).

Democracy is measured using the Polity variable.

Note that when using the instrumental variable, the EU is excluded from estimation (problems with aggregating/lack of values for religion).

Scholars suggest a substitution effects to account for the increased use of AD and the greater diffusion across countries. Zanardi (2004) notes that while the GATT and the WTO have been successful in terms of reducing tariff barriers, countries have increasingly resorted to non-tariff protection with AD playing a major role. Sometimes this was even facilitated by institutional adjustments, such as during the Tokyo Round (Hoekman and Kostecki 2001) which introduced tariff reductions and at the same time relaxed AD provisions and made them easier to use. Note also that developing countries’ increasing use of AD coincides with their greater involvement in the global trading system during the end of the 1980s and subsequently in the WTO.

For a similar rationale see e.g. Busch and Reinhardt (1999) who examine NTBs and control for tariff protection or Mansfield and Busch (1995).

For alternative model specifications also see robustness checks in the appendix.

Note that real GDP p.c.—not the instrument—is used when estimating model 5. The reason is that there is no theoretical rationale to assume a relationship between executive constraints and income per capita similar to w and income levels. Hence, the results of model 5 cannot be compared directly to specifications 1–4.

Alternative specifications using GDP and import volumes as controls show a similar effect.

Values for the entire sample. Marginal effects calculated using CLARIFY. All other variables are set at their means. Despite the significant increase, it should be noted that the confidence intervals also indicate some variation and uncertainty which reflects the heterogeneity of the sample.

Respective adjustments are made to account for countries joining the WTO after 1995 or joining the EU during the observation period. The number of measures is divided by years of WTO membership to obtain comparable averages. Similarly, other variables are adjusted to reflect the respective period of membership.

See Kono (2008) for use of tobit specification in a similar context.