This paper shows that performance attribution considered alone can be misleading. Indeed, portfolio managers who know perfectly the distribution of an asset's returns and who perform a relative portfolio optimisation according to that information may be penalised in some of their choices by the performance attribution process. In order to solve this apparent paradox, this paper proposes taking risk into account. It is established that the appropriate definition of risk in this management context is relative risk, as measured by the standard deviation of the tracking error. This measure makes it possible to justify the choices which were previously penalised. Moreover, it is proved that the information ratio of each decision (asset allocation and security selection) is the same. This means that some equilibrium between expected (or ex post mean) return and relative risk has been reached.

This article proposes a transactional analysis of the performance of international joint ventures formed in mainland China and Taiwan. This analysis is carried out for a sample of 104 Chinese partners who, between January 2000 and December 2003, announced the formation of an international joint venture. An event study methodology is used to measure Chinese partners' stock market reactions to the formation of these joint ventures. The different empirical results found concur to envisage an analysis of Chinese partners' performance that is distinct from that of foreign partners. These results raise doubts concerning the appropriateness of transaction cost theory as a framework for analyzing Chinese partners' performance.

The theory of portfolio insurance is important theory since some well-known past …nancial crisis. The objective of the portfolio insurance is to give the in- vestor the ability to limit downside risk while allowing some participation in upside markets. This return pattern has seemed attractive to many investors who have poured up to billions of dollars into various portfolio insurance prod- ucts. Here, we are interested in a widely used one : the Constant Proportion Portfolio Insurance (CPPI). The CPPI method uses a simpli…ed strategy to allocate assets dynamically over time. It requires that two assets are exchanged on the …nancial market : the riskless asset, B, with a constant interest rate r ( =5 %) (usually Treasury bills or other liquid money market instruments) and the risky one, S (usually a market index or a basket of market indexes). This method is rather simple and consists mainly on …xing a deterministic ‡oor such that the portfolio value will never be below this level. The exposure which is the amount invested on the risky asset is such that it is always equal to a …xed proportion of the total portfolio value. This is done by choosing an appropriated constant, called the multiple. Nevertheless, this multiple must not ne too high to keep the guarantee. In this paper, we apply the extreme value theory to the study of this CPPI method, in particular to determine the multiple. A quantile hedging approach is introduced, which provides an upper bound on the standard multiple m. This bound is statistically estimated from the behavior of extreme variations in rates of asset returns. Moreover, we introduce the distributions of interarrival times of these extreme movements and show their impact on the portfolio insurance. We illustrate these results on S&P 500 data.