Rules of origin of the European Union’s Preferential Trade Agreements with special reference to the EU–ACP Economic Partnership Agreements
Quantification of the restrictiveness of PTA RoO was initiated by Estevadeordal (2000) in relation to NAFTA. Estevadeordal developed a rating system whereby specific RoO were attributed a score between 1 (least restrictive) and 7 (most restrictive). In allocating scores to rules, two basic principles were followed: (i) for CTC rules, change at the level of chapter was considered more restrictive than at the level of heading, change at the level of heading more restrictive than at the level of sub-heading, and so on; (ii) VA and TR conditions added to a CTC were considered to be more restrictive than a CTC alone. In a comparison of NAFTA and contemporary EU PTA rules, Estevadeordal and Suominen (2004) concluded that, while both sets of rules were restrictive, NAFTA RoO had a higher overall level of restrictiveness than the EU’s. This result was based on a comparison of averages of restrictiveness ratings across 20 HS chapter headings. The EU’s ratings were highest for fruit and vegetable products (6.6), textiles and clothing (6.1) and food products (5.0), while NAFTA’s were highest for textiles and clothing (6.9), fruit and vegetable products (6.0) and leather goods (5.6). The NAFTA average for all 20 sections was 5.1 and the EU PTA was 4.5. At the same time, the NAFTA RoO showed a higher level of variance in their restrictiveness than the EU’s, which were consistently ‘moderately restrictive’ or ‘restrictive’. A critical point in the quantitative literature is that a correlation was established at product level between high levels of RoO restrictiveness and high margins of preference. In other words, where incentives for preference utilisation were highest, the highest barriers to their utilisation were found. This is demonstrated by Cadot et al. (2005a), using a modified version of Estevadeordal’s method. Cadot et al. (2005a) compare levels of restrictiveness of RoO for tariff lines where there are MFN tariff peaks and those for tariff lines on which low tariffs were applied.3 The NAFTA average scores for high and low tariffs are 6.2 and 4.8, respectively, while the EU PTA averages are 4.2 and 3.4, respectively. Cadot et al. (2005a) go on to demonstrate an association between RoO restrictiveness at product level and preference under-utilisation. Cadot et al. (2005b), examining this relation for the Eastern and Southern Africa (ESA) grouping in
the EPA negotiations, show that the main losers have been non-LDCs, since these are the countries most capable of exporting the value-added products to which restrictive rules most typically apply. Annual losses for the four ESA nonLDCs, represented by their export of goods under EU MFN RoO rather than CPA ones, are computed at €201 million (or equivalent to 16 per cent of the value of the preferences for the tariff lines concerned). Cadot et al. (2006) find similar levels of losses for all EU PTA beneficiary countries. Estevadeordal’s method has been subject to criticism for the arbitrariness of its assumptions by Erasmus et al. (2004). According to these authors, because gradations in the HS were not mainly designed to reflect differences in levels of processing at product level, inputs and outputs of particular processes are often found at the same sub-heading level. Likewise, TRs are not restrictive per se, only specific TRs are. Finally, a combination of less restrictive conditions may be less restrictive than a single more restrictive one. On the other hand, Rivas (2006), reporting the use of an alternative and somewhat more nuanced method for measuring restrictiveness, comes to conclusions broadly consistent with those of Cadot et al.