Study on portfolio selection with skewness at Oslo Stock Exchange
Abstract
In this paper we investigate the statistical measure of skewness in a portfolio management
setting at Oslo Stock Exchange (OSE). Our analysis follows earlier research on the topic
of non-normal investor preferences which prices skewness as a relevant factor. We analyze
distributional properties of monthly returns in individual assets and nd on OSE that 1)
skewness is pervasive, 2) positive skewness has a moderate level of persistence in the long
term and can reasonably be predicted, and 3) diversi cation and skewness are negatively
correlated. As a second major focal point, we form strategies which include a preference for
skewness using Polynomial Goal Programming. We compare them to traditional portfolios
using a traditional nancial performance measure (Sharpe Ratio) and skewness. With two
model speci cations we nd mixed results regarding skewness - the strategies are only able to
produce higher portfolio skewness than the classical mean-variance portfolio in one scenario.
A second nding is that we can not reject a null hypothesis of equal Sharpe Ratio between
the skewness-strategies and the mean-variance portfolio. We also nd that the two skewnessstrategies
are 1) less diversi ed than the other portfolios and 2) more risky as a consequence.
Skewness and variance seem to be opposing goals for an individual with non-normal investor
preferences.
1