The effects of EU preferential treatment

by Samuel Herzfeld (Author) Laurent Hauwaert (Author) Benjamin Levon (Author)

Seminar Paper 2011 31 Pages

Business economics - Economic Policy


Table of contents:

1) Introduction
1.a. What is a preferential treatment and historical background
1.b. Name of the different kinds of trade partnerships
1.c. Main objectives of our work
1.d. The structure: what our article will be about and how will it be decomposed

2) Main body
2.1 Literature review
2.1.a. Description of the different kinds of preferential treatments
2.1.b. Previous studies’ results
2.1.c. Why do we need to go further
2.2 Description of the data
2.2.a. Econometrical evidence: which model to use
2.2.b. Persson and Wilhelmsson: Assessing the Effects of EU Trade Preferences for Developing Countries
2.2. c. Aiello and Demaria: Do Trade Preferential Agreements enhance Exports of Developing Countris? Evidence from EU GSP.
2.2. d. Further considerations
2.3) Results in both studies
2.3.a. Persson and Wilhelmsson’s results
2.3.b. Aiello and Demaria’s results
2.3.c. Divergences between these two articles and comparison with the former literature
2.3.d. The future of GSP and its derived schemes

3) Conclusion
3.a. Main results: evidences and disparities
3.b. Opening on the future

4) References

5) Appendix

I) Introduction

For the last 60 years, the main goal of the World Trade Organization (WTO) -and its predecessor the General Agreement on Tariffs and Trade (GATT) - has been to supervise and promote trade between nations.

Throughout its multiple rounds, the organization implemented various measures to reach this objective. Some of them have been the systems of preferences offered by the European Union (or the EEC or EC in the past) to developing countries.

These preferential treatments consist mainly of the Generalised System of Preferences (GSP), which the WTO defines as “programs established by developed countries granting preferential tariffs to imports from developing countries2 ”. In addition to this frame of rules, other types of partnerships between the EU and developing countries exist, as for instance ACP, Euromed or other schemes derived from the ordinary GSP, such as Everything But Arms (EBA) or the GSP-Drug which we will examine more in details in our study.

Thus, the main purpose of this paper is to analyse the effects of these treatments and partnerships on trade, or more specifically on the impact of exports to the preference-given countries. By studying thoroughly the literature and comparing the different studies led in the past, we will try to find out which models are the most reliable to evaluate trade growth. This deep investigation will bring us to the conclusion that on the one hand, it’s nowadays accepted that there are some factors which either enhance or limit trade between developing and developed countries, such as language or distance. On the other hand, contradictory results between the different statistical models or data used will lead us to a less successful conclusion that the outcomes of some comparisons are puzzling and therefore, we won’t be able to affirm beyond any doubt the assumptions made.

The structure of our work will be decomposed as follows: firstly, we will need to summarize the literature on the subject that has been conducted in the past, and explain to which extent the two core studies on which we deepened our interest try to solve the various problems of the previous analyses. Secondly, we will describe the data used in both studies, the factors taken into account and the ones voluntarily omitted. We will carefully examine the similarities and differences between the two research papers, and explain how and why do the hypotheses and variables taken into consideration impact the model. Thirdly, we will make a thorough comparison between the different results and present how these results have been verified until now, before concluding on the challenges for tomorrow’s world.

II. Main Body

2.1. Literature review

2.1. a. Description of the different kinds of preferential treatments

Throughout the years, the targets of these preferential treatments offered by European countries to developing countries have been not only to “increase export volumes, [but also to] offer export diversification and the possibility for exporters to charge higher prices”, as pointed out by Persson and Fredrik Wilhelmsson (2006). But have these objectives been reached? Before going into details, it is useful to describe the different kinds and degrees of trade partnerships that have been established in these efforts of enhancing trade with developing countries.

The General System of Preferences (GSP) is the widest arrangement, in terms of both number of countries and product coverage. It has been granted unilaterally by the EU since 1971 to almost all developing countries. It is necessary to underline the fact that these partnerships are non-reciprocal but result of the eagerness of the European Union to help the developing countries to thrive by promoting trade. But these general arrangements have been at the bottom of the pyramid of privilege: together with this GSP regime, some other schemes of trade partnerships have been established, allocating the beneficiaries better preferences. The ones that will be addressed in our study are mainly the following:

- The ACP preferences refer to the preferential trade agreements offered by the EU to the African, Caribbean and Pacific Group of States (ACP). As mentioned by Persson and Wilhelmsson (2006), the main differences with the GSP are firstly, “the wider preference margin and commodity coverage for ACP countries”, secondly, the fact that “ACP preferences are contractual, which makes market access more certain”, and thirdly the “more generous rules of origin [and] less restrictive safeguard clauses”.3

- The Mediterranean preferences designate “countries around the Mediterranean Sea [which] have been involved in different trading arrangements with the EU since the late 1960s and early 1970s”.4
- The GSP- Plus is, as pointed out by Aiello and Demaria (2010), granted to “countries [which] (…) adopt recommendations (…) on human and labour rights, environmental protection, good governance and the fight against drugs”.5
- The Everything But Arms (EBA), which “provides duty-free and quota-free treatment for all products originating in [Least Developed Countries] (LDCs), except for arms and ammunition.

2.1. b. Previous studies’ results

Having gone through the different trade preferences for developing countries that have been used by the EU since the early 1960s, we will now report the empirical results of the effects they had and comment whether the objectives have been reached. Using yearly cross-sectional OLS regressions of a gravity model based on data for the period 1967-1978, Sapir (1981) found a positive effect of as much as 48% of gross trade creation for the GSP regime. As the reference of his paper is north-north trade, he did not examine the other schemes of preferential treatment. Using the same statistical method, Oguledo and MacPhee (1994) find a statistically significant and positive effect for the three kinds of arrangements studied, namely GSP, Mediterranean preferences and ACP preferences. Their findings summarize well the so-called “pyramid of privilege”, a notion commonly used in the literature (cf appendix B): the ACP effect is superior to the Mediterranean one, which in turn overtakes that of the GSP. In a similar model-based study conducted in 2002, Nilsson also finds statistical significant results for ACP and GSP preferences, but mostly insignificancy for the Mediterranean ones. However, in a more recent study that Nilsson led with Matsson in 2009, the results stick to the previous results of Oguledo and MacPhee, thus giving still more credit to the pyramid of privilege.

To go into details and hence to evaluate these results more precisely, we will draw our attention to studies which went further and decomposed the effects per sector, or even per product. Therefore, we might need to qualify the previous results by deepening our interest into the disintegrated structure of the arrangements. In this respect, Subramanian and Wei found in 2007 that despite the “general positive impact of EU GSP on total trade, (…) the effect is negative for the agro-food sector”. Agostino (2008) came to a similar conclusion:

“the ordinary GSP only has a positive effect in the meat sector; (…) its impact is negative and significant in the livestock and sugar sectors”. Other authors such as Brenton (2003), Hoekman (2001) or Tangermann (2002) observe that “the preferential treatment of GSP is only generous with regards to a few products”. To illustrate this point, an example of these restrictions can be found in the EBA initiative which strongly limited “access to the EU market of three particularly important products for LDCs: rice, sugar and bananas”. To sum up that point, it would be false to say that every type of product has been eligible for preferential treatments. Even though vegetable products and fisheries have been widely exported, the agricultural sectors of developing countries suffered from the highly protected European market.

In addition to those barriers, we would like to highlight that the EU GSP preferences are under-utilised for a few reasons. First of all, as stated by Aiello and Demaria (2010), “the GSP often competes with other preferential arrangements”. In other words, for instance, as the ACP preferences are more generous than the regular GSP, an ACP country would rather export under its most favourable arrangement, in this case the Cotonou Agreement (which is the treaty the ACP countries benefit from). Secondly, the constraints of RoO (Rules of Origin) may explain this under-utilisation: as enunciated by Candau and Jean (2005), “the utilisation of preferential schemes is often costly” for technical and administrative reasons. Indeed, we can easily imagine that the beneficiary countries are not always able to meet the requirements necessary to implement these schemes.

2.1.c. Why do we need to go further

Having gone through the past literature, it is now necessary to assert the limits of these studies, or more specifically to which extent we can and can’t rely on them. The repeatedly inappropriate method is the main point we need to emphasize here. In fact, as supported by Persson and Wilhelmsson (2006), “cross-sectional regressions of the gravity model [the most commonly used in the past] do not fully control for country heterogeneity, which leads to biased estimates due to omitted variables”. In other words, the new literature argues that these models are incorrect insofar as they don’t take into account the bias due to the fact that the countries are not identical. The results are consequently distorted since it is likely that the divergences within the countries’ (or group of countries’) characteristics benefiting from preferential treatments included in their analysis are not taken into consideration. These discrepancies could be due for instance to factors such as distance, or language. In order to tackle heterogeneity due to these omitted variables, the best strategy would have been to conduct a randomized trial, specifically by “taking a group of volunteers and randomly assigning them to either a “treatment group” that receives the intervention, either a “control group” that does not receive [it]”.6 It seems unfortunately merely not possible to carry out such an experiment in practice, notably because it takes a long time and it appears obvious that no country would be willing to be part of the “control group”. Moreover, it is necessary to bear in mind that the successive EU enlargements may have had an impact on the exportations from developing countries. In this respect, as the European customs union became larger, as stated by Persson and Wilhelmsson (2006), trade diversion might have occurred -a phenomenon arising when “a free trade area (FTA) diverts trade, away from a more efficient supplier outside the FTA, towards a less efficient supplier within the FTA.”7 Hence, these unintended consequences could involve negative effects on the developing countries economic results (as well as the country’s national welfare in some cases -cf Appendix C). Therefore, the EU enlargements component also needs to be taken into account in the assessment of the effects of EU preferential treatments. Thus, we will examine in the second part of our paper how the more recent studies of Persson and Wilhelmsson (2006) and Aiello and Demaria (2010) could overcome these difficulties and, in this manner, get to better conclusions.

2.2 Description of the data

2.2. a. Econometrical evidence: which model to use

As the gravity model is used in the couple of papers on which we’ve deepened our interest, we need to explain shortly what it refers to. In fact the model’s name is assigned to its origin: it relies on the concept of gravitational force and applies it to some other cases. For example, the gravity model can be used in order to explain the volume of trade between countries. Most of the time, when it is applied to trade volumes, the determinants of the model are GDP (gross domestic product), size of the population and distance between the countries of interest. In some other cases, it may also try to clarify capital flows and migration between countries.6 But as explained by Kenneth A. Reinert (School of Public Policy, George Mason University), “the effect of policies on trade flows can then be assessed by adding the policy variables to the equation and estimating deviations from the baseline flows. In many instances, gravity models have significant explanatory power, leading Deardorff (1998) to refer to them as a “ fact of life ” .8 In this regard, this model seems to be the best for assessing whether the preferential treatments were effective or not. This is notably one of the reasons why we chose to focus on these papers in particular.

2.2. b. Persson and Wilhelmsson: Assessing the Effects of EU Trade Preferences for Developing Countries.


1 First page, short abstract, table of contents, references and appendix excluded. Footnotes included.

2 WTO | Glossary, GSP, Retrieved on november 6th 2011 from http://www.wto.org/english/thewto_e/glossary_e/gsp_e.htm

3 We voluntarily won’t talk about the Yaoundé and Lomé Conventions or the Cotonou Agreement as these trade agreements refer to the ACP preferences, despite the fact that these systems could have different effects, as mentioned by Persson and Wilhelmsson (2006), “due to (…) the time period when they were granted and the number of developing countries involved”. Therefore, “it should be valid to compare other preferential systems with the ACP system as a whole”

4 The Mediterranean preferences might evoke other kinds of Cooperation Agreements signed with the Maghreb or the Mashreq or the more recent Euro-Mediterranean Association Agreements, but as they are akin, we will refer to the Mediterranean preferences for facility purposes.

5 Even though GSP-Drug is less wide with regard to the number of eligible countries and the requirements that have to be fulfilled (more focused on South America countries and their anti-drug policy), we will use in this paper GSP-Drug and GSP-Plus as synonyms, the latter incorporating the former.

6 Definition from the course of Introduction to Public Economics, R. Foucart, September 2011

7 International Trade Theory and Policy, S. Suranovic. Available on http://internationalecon.com/Trade/Tch110/T110-2A.php

8 Reinert, K., “World economy gravity models”, http://mason.gmu.edu/~kreinert/paperspdf/gravmod.pdf


ISBN (eBook)
ISBN (Book)
File size
1.2 MB
Catalog Number
Institution / College
Solvay Brussels School of Economics and Management
EU preferential treatment GSP preferential treatment GSP ACP preferences system of preferences trade flows preferential trade agreements



Title: The effects of EU preferential treatment