Perspectives on Wealth Management for Foundations: Asset Allocation Strategies at Novo Holdings

Lorenzo Tracanna

Student thesis: Master thesis

Abstract

Nowadays, foundations benefit people and society by supporting projects that would not have been realized without their generous grants and donations. Because the wealth of foundations is invested, it is exposed to financial market fluctuations, risking erosion. Limiting such risk is of paramount importance for society. This thesis sets out to explore asset management solutions that can help Novo Holdings1manage its wealth effectively.
In the first place, this thesis uses machine learning to label historical periods as either “growth” or “contraction”. Then, it calculates transition probabilities that are used to carry out Monte Carlo simulations. By simulating wealth paths, we develop a framework to evaluate the performance of different asset allocations. Ultimately, this paper compares the risk and return characteristics of three strategies: (1) the actual asset allocation of Novo Holdings, (2) a Flexible Indeterminate Factor-based Asset Allocation (FIFAA) and (3) a Regime-based Asset Allocation (RBAA). We provide evidence that a multi-regime approach captures the tail-risk of returns better than a single-regime one. It follows that long-term investors are better off if they let a multi-regime analysis inform their decisions. When comparing the asset mix of selected foundations, Wellcome Trust stands out. Its choices differ from those of other peers and are forecast to deliver better performance. As for the simulation experiments, we conclude that the FIFAA generates some of the highest wealth multiples in the Base-case scenario. However, in three out of four scenarios, the RBAA guarantees the best protection against capital erosion. De facto, in a scenario characterized by persistently high volatility, the benefits of dynamic asset allocation, like the RBAA, are offset by the rebalancing costs.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2020
Number of pages96
SupervisorsNicholas Skar-Gislinge