Elsevier

Journal of Policy Modeling

Volume 38, Issue 5, September–October 2016, Pages 837-854
Journal of Policy Modeling

Macroeconomic effects of the Barcelona Initiative

https://doi.org/10.1016/j.jpolmod.2016.07.001Get rights and content

Abstract

This paper compares ex post and ex ante assessments of the macroeconomic effects of trade liberalization in the Mediterranean. Using implications from a standard Ramsey growth model augmented for anticipation and implementation effects, we pool cross section and time series data to estimate ex post the effects of trade liberalization on a set of Arabic Southern Mediterranean Partner countries (SMPCs). We find significant and robust evidence for positive effects on major macro variables and discuss the appropriate policies. Second, we review a number of computable general equilibrium (CGE) studies which aimed at assessing the macroeconomic impacts for the same countries ex ante. CGE projections are very much at odds with the econometric findings and the biases seem to be systematic for all macro variables. Third, we use ANOVA techniques to identify possible shortcomings both with respect to design and target country of the CGE study. We find that even well-designed CGE studies targeted to an average type of country do not seem to yield reliable results. Overall, our analysis suggests that there is no sound statistical evidence to believe that CGE analysis has been useful in assessing the macroeconomic effects of trade liberalization in the Mediterranean. But we find considerable econometric evidence to support the view that free trade policies have enhanced growth in the MENA region.

Introduction

Recent unrest and political instability in arabic states has re-alerted European policy makers to the threat of uncontrolled immigration from its southern neighboring region. But the European Union (EU) is not as unprepared as it may seem, for it has, in the last 15 years, spent considerable effort on promoting economic integration and development in the Middle-East and North-Africa (MENA). Nevertheless, there is a widespread perception that this region is slow in responding and adjusting to globalization, cf. World Bank (2003).

The MENA region is a large developing market with more than 400 million customers (about the size of the EU27). Therefore, the EU initiated the Euro-Mediterranean Partnership, aimed at strengthening economic and political ties between the Common Market and MENA. A cornerstone of this so-called Barcelona Initiative (BI) was the gradual creation of a free trade area for industrial products between the EU and its Mediterranean Partners. In the following years, the EU negotiated bilateral Association Agreements (AA) with each partner country, typically allowing for a twelve-year transition phase of tariff and non-tariff barrier dismantling.

The effects of such preferential trade liberalization need not be mutually benefitial due to the possibility of harmful trade diversion effects. In the late 1990s and early 2000s a wave of applied economic research in Computable General Equilibrium (CGE) modeling was directed at assessing this issue from an ex ante perspective. A more recent CGE approach is Elshennawy (2012).2

By 2011, the Barcelona Initiative has been pushed on further into a “Union for the Mediterranean”. But few academic studies exist which provide quantitative assessments of what has been achieved so far.3 Nor has much attention been devoted to how well today's experience is in line with the ex-ante projection of CGE studies.

Of the few retrospective studies which exist, the most notable ones estimate gravity equations to test for a significant impact of trade liberalization on exports or imports. Peridy (2005), Peridy (2006) finds beneficial effects of lower EU tariffs on Mediterranean countries’ exports. However, most of this tariff dismantling took place in the 1970s prior to the BI. Söderling (2005) explicitly studies the effects of the first AAs and finds that some MENA countries’ exports seem to have benefited while others have not. Hagemejer and Cieslik (2009) conclude that imports of MENA countries have clearly increased while there is no significant effect of BI-induced trade liberalization on exports.

Unfortunately, the welfare implications of these results are far from clear. The economic well-being of MENA populations does not primarily hinge on foreign trade but on income (GDP), consumption, and investment (as a proxy for future consumption possibilities). Thus, 15 years after the launch of the BI little is known about whether its key element, unilateral trade liberalization, has made the MENA countries better or worse off. This question is of immediate policy relevance and it is the first to be addressed in this paper.

Econometric methods evaluate the macro impacts ex-post while the CGE exercises of the 1990s provided ex-ante evaluations of the BI's macroeconomic effects. Our study seems to be the first which aims at a serious comparison between these results.4 But CGE-based claims should, according to good scientific practice, be falsifiable. Surprisingly, however, CGE-analyses have flourished for decades without much econometric review. Filling this gap is the second issue addressed in this paper.

Since some countries have not yet completed tariff dismantling, we focus our analysis on the semi-elasticity of macro variables with respect to tariff rates. This magnitude can be estimated econometrically while trade liberalization is still under way, and it is also readily computed from CGE studies. Comparing projected and realized effects of trade liberalization then permits inference on the reliability of CGE models.5

Section snippets

The sequel of the paper is structured as follows:

In section II we introduce a theoretical model which explicitly distinguishes between anticipation and implementation effects of trade liberalization. We use this to derive the appropriate specification of our regression analysis. We discuss the data in section III. In section IV we apply dynamic panel estimators to our data set and find significant and robust evidence for positive effects on major macro variables. In section V, we review CGE studies and compute semi-elasticities for ready

Deriving the regression equation

Trade liberalization did not come unexpected for consumers and investors in SMPCs. Tariff dismantling was announced long before it was implemented. We start by specifying a theoretical model which takes this kind of informational structure into account. Unlike the related literature which emphasizes the importance of news in DSGE models (cf. Beaudry and Portier (2006), Beaudry and Lucke (2010)), we focus exclusively on a deterministic setting. Future tariff rates are known with certainty from a

Data

For southern Mediterranean countries, national accounts data is available only at the annual frequency. To obtain a reasonably sized sample which may allow for valid inference even in the face of noisy data we pool cross section and time series data of seven Arabic countries, Algeria, Egypt, Jordan, Lebanon, Morocco, Syria, and Tunesia. We generally use all available data for the analysis, i.e. we typically work with unbalanced panels.9

Econometric analysis

Eq. (2) is a dynamic equation – the lagged dependent variable is among the set of regressors. As is well known, standard fixed or random effects estimators are inconsistent in this context. The widely used Arellano–Bond (1991) estimator (AB), however, is consistent, irrespective of whether individual effects are fixed or random. This estimator uses a dynamic set of intruments applied to the first difference of the regression equation. It eliminates the individual effects which cause the

CGE based semi-elasticities and their econometric counterparts

This section examines CGE studies which simulate trade liberalization measures for SMPCs. We use data and information reported in each paper to calculate what was, at the time of writing, an ex-ante perspective on the semi-elasticities of main macroeconomic variables20 with respect to quantitative import barriers in the partner countries. In the following section, these figures will be compared with the

Conclusions

Summing up, we find strong and robust econometric evidence for positive effects of BI-induced trade liberalization. GDP, consumption, investment and imports respond positively to lower tariff rates and they do so not only for implemented tariff changes but also in response to credible announcements of future trade liberalization.

These results are highly relevant from a policy perspective. Not only do lower applicable tariff rates stimulate growth of GDP, consumption and investment, but the mere

References (41)

  • A.H. Bayar

    Fiscal challenges of the Euro-Mediterranean agreements

    (2001)
  • A. Bhargava

    On the theory of testing for unit roots in observed time series

    Review of Economic Studies

    (1986)
  • D.K. Brown et al.

    Some economic effects of the free trade agreement beween Tunisia and the European Union

    (1996)
  • R. Chatti

    A CGE assessment of FTA between Tunisia and the EU under oligopolistic market structures

    Review of Middle East Economics and Finance

    (2003)
  • M.A. Chemingui et al.

    Assessing non-tariff barriers in Syria

    (2004)
  • M.A. Chemingui et al.

    Agricultural trade liberalization and poverty in Tunisia: Microsimulation in general equilibrium framework

    (2008)
  • J. Cockburn et al.

    Les leçons du mariage entre les modèles d’équilibre general calculable et la nouvelle théorie du commerce international: Application à la Tunisie

    (1998)
  • S. Dessus et al.

    Trade and competition policies for growth in Lebanon: A general equilibrium analysis

    Review of Middle East Economics and Finance

    (2008)
  • S. Dessus et al.

    Trade integration with Europe, export diversification and economic growth in Egypt

    (1998)
  • S. Dessus et al.

    Trade integration with Europe and labour reallocation in the Southern Mediterranean Countries: The case of Egypt

    (1998)
  • Cited by (0)

    1

    This is an updated and extended version of previous research by Lucke and Nathanson (2007).

    View full text