Le vin comme outil de diversification de portefeuille

Abstract

As correlations between traditional asset classes increased and diversification effects disappeared during the 2007-2008 financial crisis, the search for untraditional and potentially uncorrelated asset classes gains importance in the scientific literature. Consequently, several alternative asset classes have been analyzed by academics and practitioners to determine their risk-return properties. This dissertation is intended to investigate the diversification potential of investment-grade wines. Therefore, this study attempts to determine the extent to which the inclusion of fine wines in a well-diversified portfolio can be considered relevant. First, this study suggests that the historical returns of investment-grade wines are highly serially correlated. Indeed, markets for infrequently traded assets tend to exhibit smoothed returns. As a result, the presence of high first-order autocorrelation can severely bias the estimation of the volatility. Therefore, this dissertation highlights the importance of eliminating autocorrelation from the return time series through the Okunev-White procedure. Second, this dissertation addresses the issue of non-Gaussian distributions. Indeed, markets for financial assets tend to exhibit returns that are not normally distributed. As a result, the ignorance of skewness and kurtosis may lead to an inaccurate estimation of risk-adjusted performance measures. Hence, this study emphasizes the necessity to take into account higher moments through the modified value-at-risk. Finally, the results of this study indicate that investment-grade wines have positive weights in the optimal portfolios constructed using mean-variance-liquidity-skewness-kurtosis analysis. In other words, the consideration of fine wines leads to upward and/or leftward shifts of the efficient frontiers. However, using the spanning test proposed by Gibbons, Ross and Shanken, the enhancements of the Sharpe ratios are not found to be statistically significant.