Minimum Variance Portfolios in the Swedish Equity Market

Abstract: Recent research in finance has suggested an investment strategy aimed at reducing volatility without sacrificing returns. This is the concept of minimum variance trading strategies, which has gained popularity in both academia and the industry in recent years. According to the supporters of the theory, a portfolio of stocks constructed for the sole purpose of minimizing risk will on average generate the same returns as a capitalization stock market index with substantially lower risk. In this paper we focus on the Swedish equity market and construct minimum variance portfolios of OMXS30 members from 1991-2010, and compare the properties of this strategy to the OMXS30 value weighted index and an equally weighted index constructed from the same OMXS30 members. The study finds that a covariance estimation using twelve months of data is the optimal strategy and that the performance is improved with shorter holding periods. Adjustments to the covariance matrix estimation through shrinkage or exponentially weighted moving average (with any decay factor) ads no further value to the portfolio construction in the analysis. Over the last 20 years, a minimum variance portfolio with a 3 month holding period dominates portfolios with 1 or 6 month within the 0.106 percent to 0.237 percent rebalancing cost range. In addition, the minimum variance portfolios with leverage to equal the volatility of the benchmark (44 % for portfolios with 1 month holding period, 41 % for 3 month and 36 % for 6 month) are only resilient at a 0.28 percent cost for a one month holding period compared to 0.56 percent for a 3 month window in order to beat the index on a risk adjusted basis. We find that the minimum variance strategy with a 3 month holding period and a 12 month estimation window generates persistent positive alphas over the last decade, but that the excess returns are strongly correlated with a value factor. The study finds that in absolute terms the equally weighted index produces greater returns (15.5 %) than the OMXS30 (14.9 %) and the minimum variance strategy (13.5 %) for the full period. However, the minimum variance portfolio exhibits the highest risk adjusted returns with consistently and significantly lower volatility.