Évaluation de l’hypothèse de la moyenne-variance : une application au portefeuille des banques canadiennes

Le but de cet article est d’évaluer la pertinence du modèle de la moyenne-variance comme schéma explicatif du portefeuille domestique des banques canadiennes. Le modèle théorique entraîne certaines restrictions que nous tentons de vérifier à l’aide d’un test de chi-carré dans le cas de sept banques à charte. Les résultats indiquent que le modèle postulé ne peut être accepté.In this paper we evaluate the mean-variance hypothesis (MVH) in the case ofthe domestic portfolios of seven Canadians banks. Following previous studies byParkin, Gray, Barrett and Courakis we test a particular version of the MVH byrestricting the objective function of the banks (presented as a negative exponentialutility function). The banks select their optimal portfolios by maximizing theirexpected utility of profit over a one period horizon. Optimal portfolios are revisedin each period. The MVH yields a number of properties which characterize theasset demands: homogeneity, additivity, symmetry and non-negativity of ownexpected rate effects.These properties are tested under the joint hypothesis that expected rates ofreturn are evaluated either by a rational expectations process or an ARIMAprocess. Chi-square tests of the overall restrictions as implied by the MVH areapplied to the data set which comprises the seven largest Canadian charteredbanks. The results indicate a strong rejection by the data of the conjectured model