Quantitative forecast model for the application of the Black-Litterman approach

The estimation of expected security returns is one of the major tasks for the practical implementation of the Markowitz portfolio optimization. Against this background, in 1992 Black and Litterman developed an approach based on (theoretically established) expected equili-brium returns which accounts for subjective investors' views as well. In contrast to historical estimated returns, which lead to extreme asset weights within the Markowitz optimization, the Black-Litterman model generally results in balanced portfolio weights. However, the existence of investors' views is crucial for the Black-Litterman model and with absent views no active portfolio management is possible. Moreover, problems with the implementation of the model arise, as analysts' forecasts are typically not available in the way they are needed for the Black-Litterman-approach. In this context we present how analysts' dividend forecasts can be used to determine an a-priori-estimation of the expected returns and how they can be integrated into the Black-Litterman model. For this purpose, confidences of the investors' views are determined from the number of analysts' forecasts as well as from a Monte-Carlo simulation. After introducing our two methods of view generation, we examine the effects of the Black-Litterman approach on portfolio weights in an empirical study. Finally, the perfor-mance of the Black-Litterman model is compared to alternative portfolio allocation strategies in an out-of-sample study. -- analysts' earnings forecasts ; discount rate effect ; equity premium puzzle ; implied rate of return