In his article, Ibbotson (2010) investigates the impact of a long-term asset allocation policy on portfolio performance against that of active management with respect to security selection and market timing. The paper draws its inferences from reviewing existing literature on empirical studies examining the same. The issue of contention arises from conclusions made by Brinson, Hood, and Beebower (BHB, 1986) that asset allocation policy was responsible for more than 90% of the return variations of an average fund over time, with market timing and security selection playing minor roles. In contrast, the critics of the BHB model (e.g. Ibbotson and Kaplan, 2000; Xiong, Ibbotson, Idzorek, and Chen, 2010) hold the view that market movement is the chief determinant of the variations in return amounts. The data used in the selected articles comprised of time-series and cross-sectional data on the returns of individual and aggregate funds, which were either balanced portfolios or equity funds. The methodology applied in the scrutinized studies was regression analysis after adjusting for benchmarks and controlling for interaction effects. While the findings of the study reiterated the significance of asset allocation policy in determining portfolio performance, they considered the 90% weight attached to it a gross overstatement. Consequently, Ibbotson (2010) concluded that market movement was the key determinant of portfolio performance variations, whereas the remainder percentage was shared somewhat equally between asset allocation policy and active management.
While the author raises some crucial aspects of the asset allocation debate, he fails to account for certain inherent flaws in both his reasoning and empirical analysis. These shortcomings underplay the role of asset allocation policy and attaches undue weight to market forces. Thus, I disagree with the conclusions made by the author for several reasons. First, the asset allocation debate stems from a gross misunderstanding about the 93.6% figure quoted by BHB. Many investors and financial practitioners erroneously believe that 93.6% of the return level of a portfolio comes from the asset allocation policy of a given fund. This inaccurate assumption stems from the fixation of investors on return amounts over return variability. The truth is that the figure says nothing about return amounts even though the majority of practitioners and investors believe the contrary to be true. The high regression values (R2) obtained in the BHB study only indicate how the return variability in the asset-class factors effectively explain the return variability of the fund under investigation (Idzorek, 2010). For instance, a factor such as interest rate that causes great variations in the prices of bonds is likely to cause deviations of similar magnitude in the overall fund in question. However, the weighted-average policy benchmark may produce a different return level than the actual fund. BHB made similar implications by finding that the average annualized return for the funds under study was 9.01% while that of the corresponding policy portfolios was 10.11%, regardless of the 93.6% return variations (Idzorek, 2010). The ratio between the policy portfolio return and actual fund return divulged that approximately 112% (1.122) of the return levels stem from asset allocation policy. Therefore, the correct statement about return levels that is evident in the BHB study is that 100% of portfolio returns arise from the asset allocation policy – an empirical fact that emphasizes the superiority of the asset policy mix over the market movement.
Secondly, the studies used by Ibbotson (2010) to refute the BHB conclusions used cross-sectional regressions instead of time-series regressions. The empirical approaches and underlying assumptions employed by these methods are divergent. Hence, their results on different data sets cannot be logically compared. The BHB research used time-series regression, whereas the majority of the opposing research used cross-sectional regression. For instance, Ibbotson and Kaplan (2000) used cross-sectional regression for the 40% return variability attributable to asset allocation policy, time-series regression for the 90% variability, and the ratio between policy return and actual fund return for the 100% variability (as cited in Ibbotson, 2010). The different time horizons existing between the two methods render the comparisons and conclusions drawn from the inferences made by the author invalid.
Thirdly, concerning the differences in data type, the BHB study used broadly diversified portfolios of pension funds, with limited market-timing and less active management, while the opposing studies used balanced portfolios of mutual funds with more active management. Generally, such broadly diversified portfolios with limited timing tend to move concurrently with broad financial markets over a period of time, leading to high time-series R2s (Davis, Kinniry, & Sheay, 2007). The reason for the high regressions is the exposure of such portfolios to systematic risk factors such as interest rates and business cycles. Hence, in the BHB study, the high exposure of the pension funds to systematic risk resulted in high regression values between the policy portfolio returns and the actual returns of the funds. According to the study, active management was irrelevant and merely a zero-sum game because the high costs (e.g. high skill hurdle and less predictable returns) it generated far outweighed the possible additional gains that could accrue to the fund (Idzorek, 2010). This view was confirmed by other studies such as Ibbotson and Kaplan (2000), which found that the inclusion of active management created significant dispersion in performance across portfolios, leading to lower R2s obtained in the cross-sectional regression studies.
On average, active management increases volatility but decreases return (Davis, Kinniry, & Sheay, 2007). Unless investors are sure about the ability of active managers to deliver higher returns, they are better off selecting broadly diversified portfolios with limited active management and market-timing. This analysis on data type differences raises two critical issues. One, the decision on whether or not to include active management in the studies used in the article is responsible for the differences between the high R2s in the time series BHB regression and the lower R2s in the opposing cross-sectional studies. Thus, the discrepancies in the underlying assumptions of the data types make the conclusion drawn from their comparison void. Two, the fact that active management decreases returns lends credence to BHB’s assumption that it is a zero sum game and that asset policy allocation is the determinant factor of portfolio performance (100%).
Some new terms highlighted in the article include time-series analysis, cross-sectional analysis, active management, passive management, beta return, and alpha return. Beta return refers to the return that an investor receives depending on the market volatility. The higher the beta coefficient, the higher the market variations. On the other hand, alpha return refers to the additional return an investor receives after outperforming the market through actively taking advantage of market inefficiencies or making accurate predictions regarding market movements. Active management refers to the making of conscious decisions by a manager or an investor regarding a portfolio with the aim of increasing the risk-adjusted return by taking advantage of market inefficiencies or making correct predictions of market changes. Conversely, passive management involves adopting a buy and hold strategy that does not vary with market interpretations. Time-series analysis refers to comparisons between the performance of a policy represented by suitable market indices and the actual portfolio performance over several successive periods. Conversely, cross-sectional analysis measures the dispersion in returns across several funds but within the same period.
In conclusion, the article contributes to the asset allocation debate by highlighting the role played by market forces and active management in determining portfolio performance. However, it fails in its endeavor by downplaying the dominant role of asset allocation policy over the other two factors due to gross misinterpretations regarding the 93.6% figure, and the incompatibility between the data types and regression methodology used in the studies analyzed in the article. Thus, the BHB study remains a vital research in explaining the superiority of asset allocation policy in the decision-making processes of investors.
References
Davis, J. H., Kinniry, F. M., & Sheay, G. (2007). The Asset Allocation Debate: Provocative Questions, Enduring Realities. Retrieved from http://www.vanguard.com/pdf/icradd.pdf
Ibbotson, R. G. (2010). The Importance of Asset Allocation. Financial Analysts Journal, 66(2), 18-20. doi:10.2469/faj.v66.n2.4
Idzorek, T. M. (2010, April/May). Asset Allocation Is King. Morningstar Advisor, 28-31. Retrieved from https://corporate.morningstar.com/ib/documents/MethodologyDocuments/IBBAssociates/AssetAllocationIsKing.pdf
