Asset Allocation: From Theory to Practice and Beyond, Second Edition. 2021. William Kinlaw, CFA, Mark Kritzman, CFA, and David Turkington, CFA. Willy.
To build a robust investment process, asset allocators must meet a long list of issues, including:
- which assets to choose,
- how to predict risk and return, and
- how to manage currency risk.
William Kinlaw, CFA, Mark Kritzman, CFA, and David Turkington, CFA offer advice on these and a wide range of other asset allocation topics, backing their recommendations with solid quantitative analysis. Along the way, they dispel a few myths and tackle some of the most difficult aspects of investing.
The authors identify seven essential characteristics of each asset class:
- Their composition must be stable (not static).
- They are directly investable.
- The components are similar to each other.
- The asset class is different from other asset classes.
- Investing in the asset class increases the expected utility of the portfolio.
- Selection skill is not a requirement for investing.
- Investors can access the asset class profitably.
(I would add an eighth: investors must be able to offer credible forecasts of return, risk and correlations with other assets, to implement inclusion in an optimization process. This requirement would exclude, for example, cryptocurrencies.)
What do these criteria mean in practice? Global equities are not internally homogeneous and therefore cannot be considered as a single asset class. Instead, the authors identify three equity asset classes: domestic equities (meaning US equities for the authors), foreign developed market equities, and foreign emerging market equities. Excluded from the asset classes defined by the authors are works of art (not accessible in size), dynamic stocks (unstable composition) and, more unconventionally, high-yield bonds, which are not externally heterogeneous as they are similar to investment grade bonds and therefore fall within the corporate bond asset class.
Ironically, the first myth the book tackles is the importance of asset allocation. A much-cited 1986 paper by Gary P. Brinson, L. Randolph Hood, and Gilbert L. Beebower found that asset allocation determines over 90% of performance. This book argues, however, that the methodology of this study is flawed because it assumes a starting point of an uninvested portfolio. In practice, the authors show that once investors have made the decision to invest, asset allocation and security selection are likely to be equally important (depending, of course, on the investment approach adopted). “In the absence of skill, effort, or special attention,” they write, “investors may simply default to a broadly diversified portfolio such as 60-40 stocks and bonds.”
The outputs of mean-variance optimizers are hypersensitive to small changes in the inputs. Yet the authors dispel the myth that this sensitivity leads to the maximization of error. It is true that small changes in estimation between assets with similar risk and return characteristics can lead to large shifts in the allocations between them. However, since the assets concerned are close substitutes, these reallocations have little impact on the distribution of portfolio returns. On the other hand, a pronounced sensitivity to input changes is do not observed with assets that have dissimilar characteristics. In particular, small changes in the estimates for stocks and bonds do not lead to large variations in the optimal allocation between them.
Asset allocation covers all the key ingredients of its subject, such as forecasting returns, optimization, and currency hedging. The chapter on rebalancing gives a good overview of what practitioners will find: a mix of detailed quantitative analysis and practical guidance, with room to draw one’s own conclusions. Investors must weigh the trade-off between the cost of rebalancing their portfolios to target and the cost of sticking to a sub-optimal mix. A section on a dynamic programming methodology concludes that this approach is computationally impossible. The authors then present an optimal rebalancing methodology, the Markowitz-van Dijk heuristic approach. Its costs (5.4 bps) are compared to the costs of calendar rebalancing (5.5 bps to 8.9 bps), tolerance band rebalancing (5.8 bps to 6.9 bps) and no rebalance (17.0 bps). This detailed analysis supports a simpler conclusion for those of us who deal with individual customers, for whom behavioral bias poses the biggest threat to long-term success: Have a long-term plan, rebalance your portfolio accordingly. of this plan, but don’t swap too often.
The book presents high-level quantitative analysis to explore some of the most challenging aspects of asset allocation. For example, the authors assess the likelihood of forward-looking scenarios using a technique originally developed by Indian statistician PC Mahalanobis to characterize human skulls. They use a hidden Markov model to develop a regime-switching approach. Additionally, they identify the fundamental drivers of stock-bond correlations using statistically filtered historical observations.
Despite its reliance on such sophisticated techniques, this new edition of Asset allocation is accessible to those of us who work with quantitative teams rather than in them. Each chapter offers an independent analysis of one of the 24 aspects of asset allocation. I return to this book regularly for its framing of the issues I face, the authors’ analysis, and their concise presentation of the essentials.
Warning: Please note that the content of this site should not be construed as investment advice, and the opinions expressed do not necessarily reflect the views of CFA Institute.
Editor’s note: The summary bullet points for this article were chosen by the Seeking Alpha editors.