Veto players and the design of preferential trade agreements

ABSTRACT

The concept of domestic veto players has become a popular explanation for foreign policy rigidity. We argue that veto players can be amenable to new policy initiatives – in our case preferential trade agreements (PTAs) – but then choose to exert a strong influence on their contents. Drawing upon more than a dozen PTA-design variables for an expanded collection of postwar trade agreements, our quantitative tests reveal that veto players systematically shape what is included in, and excluded from, PTAs. Most notably, agreements concluded under greater veto-player constraints contain fewer liberalization commitments, weaker dispute settlement, and more opt-outs in the form of trade remedies. Our findings highlight domestic political actors as important determinants of international institutional design and show that veto players exert additional, and perhaps stronger, influence on what is proscribed in agreements, not just whether they are enacted in the first place.

KEYWORDS:

• international trade
• international negotiation
• preferential trade agreements
• institutional design
• veto players
• ratification

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. “EU and Canada sign deal amid fears about future of trade policy.” Financial Times, 30 October 2016; accessed at https://www.ft.com/content/450bea68-9d9f-11e6-86d5-4e36b35c3550.

2. ‘Die heikle Frage der Menschenrechte im China-Abkommen’, Neue Zürcher Zeitung, 15 May 2013; accessed at https://www.nzz.ch/schweiz/die-heikle-frage-der-menschenrechte-im-china-abkommen-1.18081364.

3. They split PTAs into three general types: (1) preferential agreements; (2) free trade areas; and (3) customs unions/common markets/economic unions (Mansfield and Milner, 2012, 137–145; also Mansfield et al., 2008).

4. We do not assume that executives are unitary actors, particular in coalition governments. However, we posit that when governments decide to launch trade negotiations, some degree of internal consensus has been achieved.

5. Executives are argued to support free-trade because they are in charge of foreign economic policy and represent the entire country (e.g. Baldwin 1985; Lohmann and O'Halloran 1994; Rogowski 1987), which also makes them more concerned about enhancing general welfare (Mansfield and Milner 2012).

6. The various logics are as follows. A complainant will select the venue that it believes allows for the swiftest resolution and/or most sympathetic considerations of its claims. The lack of exemptions allows any issue to be addressed via the PTA's DSM. Also, the process will be faster when there are time limits and apolitical procedures for selecting a panel chair. Moreover, an unbiased chair not only prevents delay, but should result in a more comprehensive consideration of the complainant's claims.

7. In cases involving the EU, we use values for Germany (West Germany before 1991), whose support is crucial for agreement success and whose veto-player score is among the top third among EU member states (see Elsig 2016). To check for robustness we also substitute the EU-wide average as well as the scores for France (another pivotal EU member on trade) and Belgium (consistently the highest veto-player score among EU members). Our overall findings are robust to these different measurement logics for the EU.

8. For consistency, for country-level variables, we once again take the average score across the dyad, although we also consider alternate measurement approaches in our sensitivity checks discussed later.

9. GDP data are taken from the World Development Indicators (and expressed in billions of dollars) and GDP pc data are taken from Gleditsch (2002) and expressed in thousands of dollars.

10. These and all other calculations of substantive effects are conducted by varying the veto-player variable from its value at the 10^th percentile (.033) to its 90th percentile value (.513) while setting the WTO membership variable equal to 1 and holding all other variables at their median values. Predictions are calculated using the prvalue command in SPost 9 (Long and Freese 2005).

11. As we did for the calculations in Table 2, in this case we examine the change in democracy from the 10th percentile value (Polity score of −4) in our sample to the 90th percentile value (polity score of 9) for WTO members, holding all other variables at median values. The substantive effects for veto players are typically 1.5–2 times greater than those for democracy.

12. In only one case, for the likelihood of a PTA allowing post-award sanctions, does the estimated finding for veto players exhibit reduced confidence? Even then, the coefficient is positive but falls just short of conventional levels of statistical significance. By contrast, in one other case, for the effect of veto players on AD provisions, the estimated level of confidence moves from the 95% level to the 99% level.

13. For these purposes, we classify PTAs as being among countries solely in Africa, the Americas, Asia, Europe, or Oceania, as well as inter-region agreements. We also insert regional time-trend variables and our findings remain unchanged. None of these results are sensitive to the decision of which region to treat as the omitted category.

14. Across various models, they typically are correlated at right around .7.

15. We choose this exclusion restriction for both theoretical and empirical reasons. We believe that political ties play an important role in determining with whom to sign PTAs, as ongoing political debates about current and possible trade agreements illustrate. These ties, along with distance, are consistent predictors of PTA signing (e.g. Mansfield and Milner 2012). However, once PTA partners are selected, we believe that governments pursue the same elements in all of their PTAs. They have a type of template for what they want (e.g. Allee and Elsig 2014; Allee and Lugg 2016) and do not cave or adapt depending on their relationship to the partner. In our estimations here, alliance ties (positive) and distance (negative) are indeed strong, individual predictors of PTA signing, and the three variables jointly are statistically significant predictors of PTA signing. None of the three are consistent predictors of PTA design.

16. Findings are consistent across all methods for selecting non-PTA dyads. We weakly prefer to present these estimations because we have fewer concerns about measurement and missing data, given the smaller number of cases, and because the estimated rho in these models tends to be larger, thus presenting somewhat of a tougher test for possible selection bias.

Additional information

Funding

This work was financially supported by the Swiss National Science Foundation's NCCR Trade Regulation.

Notes on contributors

Todd Allee

Todd Allee is an associate professor of government and politics at the University of Maryland. He has published extensively on issues related to international trade and foreign direct investment, including on the settlement of investment, trade, and territorial disputes.

Email: [email protected]

Manfred Elsig

Manfred Elsig is a professor of international relations at the University of Bern and a deputy managing director of the World Trade Institute. He holds a PhD in political science from the University of Zurich. He has published extensively on political economy, international institutions, and trade policy.