Cette thèse s’attèle à déterminer les conséquences théoriques et empiriques de la considération d’indicateurs socialement responsables dans la sélection de portefeuille traditionnelle. Le premier chapitre étudie la significativité de la perte d’efficience moyenne-variance d’un portefeuille d’obligations souveraines lorsque l’on introduit une contrainte sur la notation socialement responsable moyenne des Etats. En utilisant un échantillon d’obligations d’Etats développés sur la période 1995-2008, nous montrons qu’il est possible d’augmenter sensiblement la notation socialement responsable moyenne sans perdre significativement en termes de diversification. Le second chapitre propose une analyse théorique de l’effet sur la frontière efficiente d’une contrainte sur la notation socialement responsable du portefeuille. Nous mettons en évidence les différents cas de figure pouvant se produire en fonction de la corrélation entre les rendements attendus et les notations socialement responsables et de l’aversion au risque de l’investisseur. Enfin, puisque la question de l’efficience des portefeuilles investis en fonction de critères socialement responsables fait débat dans la littérature financière, un dernier chapitre propose un nouveau test d’efficience moyenne-variance dans le cas réaliste où aucun actif sans risque n’est disponible. / This thesis aims at determining the theoretical and empirical consequences of the consideration of socially responsible indicators in the traditional portfolio selection. The first chapter studies the significance of the mean-variance efficiency loss of a sovereign bond portfolio when introducing a constraint on the average socially responsible ratings of the governments. By using a sample of developed sovereign bonds on the period 1995-2008, we show that it is possible to increase sensibly the average socially responsible rating without significantly losing in terms of diversification. The second chapter proposes a theoretical analysis of the impact on the efficient frontier of a constraint on the socially responsible ratings of the portfolio. We highlight that different cases may arise depending on the correlation between the expected returns and the socially responsible ratings and on the investor’s risk aversion. Lastly, as the issue of the efficiency of socially responsible portfolios is a central point in the financial literature, the last chapter proposes a new mean-variance efficiency test in the realistic case where there is no available risk-free asset.

This thesis aims at determining the theoretical and empirical consequences of the consideration of socially responsible indicators in the traditional portfolio selection. The first chapter studies the significance of the mean-variance efficiency loss of a sovereign bond portfolio when introducing a constraint on the average socially responsible ratings of the governments. By using a sample of developed sovereign bonds on the period 1995-2008, we show that it is possible to increase sensibly the average socially responsible rating without significantly losing in terms of diversification. The second chapter proposes a theoretical analysis of the impact on the efficient frontier of a constraint on the socially responsible ratings of the portfolio. We highlight that different cases may arise depending on the correlation between the expected returns and the socially responsible ratings and on the investor’s risk aversion. Lastly, as the issue of the efficiency of socially responsible portfolios is a central point in the financial literature, the last chapter proposes a new mean-variance efficiency test in the realistic case where there is no available risk-free asset. / Doctorat en Sciences économiques et de gestion / info:eu-repo/semantics/nonPublished