This paper was co-authored by Stephan Meschenmoser, Senior Portfolio Strategist, Official Institutions Group, BlackRock; Anthony Chan, Portfolio Strategist, BlackRock Investment Institute; Grace Qiu Tiantian, Senior Vice President, Economics and Investment Strategy, GIC; and Ding Li, Senior Vice President, Economics and Investment Strategy, GIC.
The introduction was written by Kevin Bong, Director, Economics and Investment Strategy, GIC and Alexander Brazier, Deputy Head, BlackRock Investment Institute.
Introduction
The significant impact of Covid on the economic and financial markets landscape has brought into focus the importance of incorporating uncertainty into any investment process. The unusual, stop-start nature of activity has no historical precedent, meaning lessons from the past are unlikely to be very helpful. In addition, the world faces several structural changes such as the challenge of combating climate change, the implications of unprecedented monetary-fiscal coordination and the growing role of China.
The uncertain environment we are in warrants some humility around expected asset returns – the building blocks of strategic asset allocation. It is also important to acknowledge that there is no “optimal” portfolio for the wide range of significantly divergent yet plausible economic outcomes. Yet traditional portfolio techniques, such as mean variance optimization, take the approach of achieving an “optimal” asset allocation by assuming too much certainty in the economic outlook and expected asset returns. This is a significant risk at the current juncture that could impede investors from achieving their objectives.
Both GIC and BlackRock believe strongly in incorporating uncertainty from the outset of any portfolio construction process. In this paper, we set out two alternatives to traditional methods. We study an explicit scenario-based approach and a simulation-based one, and explore ways the two could potentially be combined. The two approaches share a common philosophy – both allow for uncertainty, acknowledge that there is no “optimal” portfolio for all outcomes, and are flexible in a way that an investor’s aversion to uncertainty can be reflected in the portfolio.
This paper provides an overview of how we are looking beyond traditional portfolio construction approaches to prepare for an increasingly uncertain world. It aims to push forward the conversation and stimulate debate around portfolio construction.
Summary
Uncertainty is always an acute investment problem, particularly when several economic scenarios are sufficiently different and plausible. The post-Covid world faces uncertainty on a number of fronts – from the monetary and fiscal policy outlook and its impact on the U.S.-China strategic rivalry to structural forces such as sustainability.
Asset prices and risk profiles are likely to diverge depending on the economic scenarios. Relying on point return estimates to drive strategic asset allocation under such circumstances can be misleading as the “mean” cannot be known with certainty when both the economic scenario and return expectations under that scenario are unknown. Traditional approaches such as “mean variance optimization” (MVO) that rely on mean estimates assume certainty around how the world may evolve as well as certainty around how asset prices may react.
The past 18 months have proved that reality can stray considerably from expectations. In this paper we discuss two alternatives to MVO – a scenario-based and a simulations-based approach. Both approaches stem from the same core philosophy that explicitly allows for uncertainty to help overcome some of the shortcomings of MVO.
A scenario-based approach aims to minimize the opportunity cost – or ‘regret risk’ – of a macro outlook unfolding differently from its base case. This approach involves an investor contemplating several alternate scenarios, assigning a probability to each and using probability-weighted scenario outcomes to construct a portfolio that is tailored to maximize diversification across macro regimes.
A simulation-based approach aims to maximize portfolio returns under potential “worst-case” economic and market outcomes around a single base-case macro scenario. The uncertainty is directly incorporated in the capital market assumptions (CMAs) – or the long-run asset return estimates – by simulating multiple pathways for asset returns and choosing the worst set of return estimates that reflect the desired aversion to uncertainty. To be sure, portfolios resulting from a simulation-based approach are also always being stress-tested on specific historical or market-driven scenarios and different sets of simulations calibrated on alternative scenarios can be produced as well. The stress test around emerging market crises we discuss in this paper is one example of using scenarios within a simulation-based approach.
We adopt a case study of potential emerging market crises to compare and contrast the two approaches using a hypothetical, U.S.-dollar based multi-asset, institutional portfolio investing on a 10-year horizon. We use BlackRock CMAs for the third quarter of 2020 in this case study to reflect a period of high uncertainty in global markets when the path out of the Covid shock was particularly unclear.
One key difference between the approaches is that the simulation-based approach focuses more on portfolio resilience, while the scenario-based approach is designed to minimize regret risk – that captures both potential underperformance and missed upside opportunities.
What’s next? Both approaches have their advantages and disadvantages yet we believe both offer a higher chance at achieving portfolio resilience for the years ahead than an approach that ignores the uncertainty inherent in investing. Ongoing work at BlackRock in evolving the CMAs involves more direct linkages to a macro scenarios and blending in elements of a scenario-based approach in allowing for multiple scenarios, while using simulation techniques.
These are the Introduction and Executive Summary of the report. Please click on “Save As PDF” for the full paper.
