Non-stationarity and the import demand for virgin olive oil in the European Union J. M. GIL,* B. DHEHIBI y, M. BEN KAABIA z and A. M. ANGULO z Departamento de Ingenierı´a Agroalimentaria y Biotecnologı´a, UPC, Comte d’Urgell, 187, 08036-Barcelona, Spain, y INRAT, Tunisia and zDepartamento de Ana ´lisis Econo ´mico. Facultad de C.C.E.E y E.E de Zaragoza The aim of this study is to analyse the importation of virgin olive oil to European Union countries, paying special attention to the Spanish export contribution. The method used is based on the estimation of an imports demand system. The novelty of the paper lies not in the modelling approach but in the explicit consideration of the univariate characteristics of series that is included in the analysis. Since prices are non-stationary, cointegration among them has been tested. Results indicate that they are cointegrated and that homogeneity holds. As a result, relative prices are included in the imports demand system. Structural change is also considered so as to account for the entrance of both Spain and Greece into the EU during the period studied. Results demonstrate the leadership of Spain within the EU virgin olive oil market as well as the increasing competitiveness of Greek oil. I. INTRODUCTION In the last two decades many papers have dealt with the estimation of import demand elasticities for a large number of products and geographical locations. Most of these papers have specified models consistent with economic theory, such as the Armington’s model (Sarris, 1983; Abbot and Paarlberg, 1986; Figueroa and Webb, 1986; Suryana, 1986; Babula, 1987; among others), the AIDS model (Hadden, 1990; Heien and Pick, 1991; Yang and Koo, 1994; Andayani and Tilley, 1997), or the Rotterdam model (Seale et al., 1992). Among them, the AIDS system, applied for the first time by Winters (1984) to import demand analyses, has been one of the most frequently used due to its easy estimation as well as its flexibility in testing all theoretical restrictions (homo- geneity, symmetry and negativity). However, results from the specification and estimation of AIDS systems when dealing with time series data have been criticized from the theoretical point of view due to the frequent rejection of homogeneity, symmetry and negativ- ity restrictions (although it must be noted that no attention has been paid up to now to the last restriction in import demand models). The rejection of theoretical restrictions has been related to misspecification problems. To solve this problem, dynamic specifications have been adopted following Anderson and Blundell (1983a, 1983b) who showed that the theoretical restrictions were compatible with the data, at least in the long run. Assuming that changes in endogenous variables are generated by unex- pected changes in prices and the total imported quantity to ensure the maintenance of the long-run equilibrium, Arnade et al. (1994) used the just-mentioned approach to specify an AIDS system in error correction form to analyse the import demand for cotton in various countries. However, recent studies on non-stationarity and cointe- gration have opened a new approach for introducing dynamics in demand systems, making the specification of AIDS models in error correction form quite popular. While the stochastic properties of series have been considered in some studies dealing with traditional demand analyses (Denbaly and Vroomen, 1993; McAvinchey, 1996; and Chambers and Nowman, 1997; among others), this tech- nique has yet to be applied to import demand systems. If all *Corresponding author. E-mail: jmgil@posta.unizar.es Applied Economics ISSN 0003–6846 print/ISSN 1466–4283 online # 2004 Taylor & Francis Ltd 1859 http://www.tandf.co.uk/journals DOI: 10.1080/0003684042000227877 Applied Economics, 2004, 36, 1859–1869