Through the eyes of a Danish pension fund, this thesis seeks to implement economic forecasts to improve asset allocation and earn an excess return over a benchmark. By doing so, the model diverts from the classical sole use of historical data to calculate expected returns. The benchmark asset weights are constructed based on typical investment strategies in Danish pension funds. The Black-Litterman model uses the benchmark as entry point and adds subjective information to improve portfolio returns. The composite leading indicators (CLI) are published monthly by the OECD, and used as input in the allocation model - using the indicator for the entire OECD membership nations. As a result the asset indices are as well extracted on a monthly basis ultimo, to prevent asset allocation prior to release of forecasts. Expected returns and covariance are calculated on rolling 12-month historic data. The asset allocation model used in this thesis is the Idzorek’s version of the Black-Litterman model incorporating a bayesian aspect to the certainty provided to the individual views. Four views are constructed to represent each of the four economic phases. A Black-Litterman view is a view on two or more assets relative performance expressed in excess returns over the risk free rate. The thesis has four different paradigms based on the two dimensions: How the forecasts provide confidence to the views – categorical vs. dynamic approach. Access to data in/out of sample. The thesis analyzes the expected returns and risk of the four paradigms, due to the impact of the incorporated views. Each paradigm has a benchmark (the same for all four paradigms), an unconstrained Black-Litterman-portfolio and two constrained portfolios; the tangency and minimum-variance. The last part of the analysis focuses on the ex-post returns of all the paradigms’ portfolios. Despite the dynamic paradigms ability to be more precautious in the view shifts, the portfolios that generate the highest returns over the entire period of analysis are the Black-Litterman and tangency porfolios from the optimized categorical paradigm (“analyseparadigme 3”), both beating the benchmark with a large margin of excess return. Although the correlation between ex-ante and ex-post risk premiums is doubled for the best performing portfolio, the actual correlation figure is low hence the model is not able to perfectly predict future returns. However the correlation of excess returns between the benchmark and the portfolio is high and the excess return of the portfolio accelerates with ex-post performance of the benchmark.
|Educations||MSc in Finance and Accounting, (Graduate Programme) Final Thesis|
|Number of pages||87|