Use this URL to cite or link to this record in EThOS:
Title: Fixed income portfolio construction : a Bayesian approach for the allocation of risk factors
Author: Vamvakas, Orestis Georgios
ISNI:       0000 0004 5918 238X
Awarding Body: City University London
Current Institution: City, University of London
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
Active portfolio management is driven by the trade-off between the expected return and the associated risks. In light of the most recent extensions of Black-Litterman model, we stick to a Bayesian approach for the construction of active fixed income portfolios. Within the investment grade universe, the equilibrium returns are approximated by the yield levels implied by the market prices and these are blended together with investment views. In parallel, risk factors are preferred over asset class risk modelling. Affinity towards risk factors rather than asset classes is primarily linked with two elements; the reduction of the dimensionality of the risk estimation problem and the intuitive way in which portfolio exposures per risk factor can be expressed as performance drivers. The first empirical part of the thesis deals with the optimisation of a relative to an index portfolio where the centre of gravity is the chosen benchmark. The first ingredient of the optimisation is the blend of the yield advantage over the index and the expectations for excess returns over the index emanating from the investment views. The second ingredient is the risk estimated by a multifactor risk model. Then, a set of relative to the index investment grade portfolios is constructed. The second empirical part investigates whether there is scope to blend the multifactor risk framework with more sophisticated risk estimation techniques such as resampling. Tail risk estimated by block bootstrapping on the risk exposures of real actively managed portfolio exposures vs. the Barclays Capital US Aggregate index is compared with the parametric and exponentially weighted moving average risk model findings. The multifactor risk estimate using block bootstrapping exhibits better performance than the alternatives tested but struggles to capture the out of sample extremes. Finally, the third empirical part aims to enhance the allocation model by taking advantage of the findings of the second empirical part. The blending mechanism of equilibrium returns and investment views, which are expressed as optimisation constraints, is performed with the aid of a numerically approximated returns’ distribution. The resampled distribution deviates from the normality assumption imposed initially in the Black-Litterman model and forms a more realistic basis for the evaluation of investment views and for the portfolio construction against tail risk measures such as value at risk and conditional value at risk.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: HG Finance