Asset Allocation Fund-of-Funds Product Specification and Discussion The Smart Approach to Multi-Asset Managed Account Investing

Transcription

1 Asset Allocation Fund-of-Funds Product Specification and Discussion The Smart Approach to Multi-Asset Managed Account Investing The Case for Actively Managing Asset Allocation Exposure The most significant investment decision is asset allocation. A significant swing in relative exposure to stocks can overwhelm the significance of decisions on manager or investment selection in a portfolio. Active asset allocation as an investment strategy to add value (increase return) has inherent appeal because of the continued large return differences over an investment cycle in financial markets stocks vs. bonds, value vs. growth, high yield vs. government, equity sectors, and international vs. domestic. The opportunity to enhance return though active exposure shifts is unlikely to dissipate. Asset Allocation funds can exploit a variety of investment disciplines simultaneously, while addressing quite distinct investment objectives by adjusting risk profiles and investment universe. While the opportunity to enhance returns through active asset allocation is certainly compelling, the most intriguing aspect of the strategy is that it can be combined with traditional active management of stock and bond portfolios to yield a dual alpha or excess return. By combining multiple sources of excess return into one investment strategy, we can compound active return without significantly increasing total risk or introducing leverage. Tactical asset allocation provides a compelling, yet often untapped, source of supplemental value added in portfolios. The product addresses a chronic problem in self-managed accounts of drifting or even unmanaged asset allocation exposure. Recent interest in professionally managed account and internet-based investment advice services highlight this important need, heretofore relatively inaccessible to retail mutual fund, IRA, and defined contribution investors with more than $5 trillion in discretionary investments. A Challenge: Exploiting Skill and Practical Considerations Portfolio managers seeking to add value through asset allocation must address several critical challenges to maximize the probably of success. Of course, demonstrable investment skill and several investment decisions (strategy breadth) are necessary conditions 1. Portfolio risk must be managed in a disciplined way considering a variety of dimensions to control for unintentional exposures that could lead to adverse performance. Portfolio managers who rely heavily on top-down asset allocation using physical securities must overcome a significant transaction cost hurdle to exploit opportunities, capping returns in many strategies historically. The most difficult asset allocation decision of allocating between stocks, bonds, and cash is also the most significant. The significance of this investment opportunity is evident in the chart below with differential returns approaching 60% over a given two- 1 Success in investment management is a function of: skill in forecasting...[and] the breadth of your strategy Richard Grinold, The Fundamental Law of Active Management (JPM, Spring 1989) 7/11/2005 Why Asset Alloc doc Page 1

2 year period. Various academic studies have observed that more than 90% of the wealth differential between portfolios is attributable to the asset allocation decision 2. We believe it is possible to demonstrate skill in tactical asset allocation using fundamental factors to forecast returns governed by a disciplined approach to manage portfolio risk. Relative asset class returns are correlated with rational response to measured changes in security valuation, economic factors, interest rates, behavioral sentiment, and related financial variables (i.e., currency impact) to varying degrees. The multi-factor approach to forecasting has been used on live assets over 14 years and demonstrated an ability to add value through tactical shifts in portfolio exposure. 80% Rolling 24-month Relative Performance: Stock vs. Bonds 60% US Stock-Bond Return % 40% 20% 0% -20% -40% -60% -80% Beyond the difficult stock/bond/cash decision lies other dimensions that have proven as valuable in adding value. Our models identify opportunities to differentiate US and international equity markets, as well as regional exposures (Asia vs. Europe), or even between individual countries and their respective currencies (Japan vs. UK). We look for opportunities in equity style and size, another source of exploitable differential return. Sectors within the equity (Energy vs. Industrial) and bond markets (Corporate or High Yield vs. Government or Agencies) provide certain opportunities, as well. A rigorous risk controlled investment discipline in all these different dimensions can exploit such pricing anomalies, particularly in volatile financial markets that are as slow to correct as we observe. In addition to investment skill and strategy breadth, a practical implementation approach that is cost effect is necessary to maximize return to the strategy. If our strategy can add 1-2% annualized excess return, but it costs 1% to implement, there may not be sufficient benefit. Most asset allocation funds shift exposure in the cash market, meaning that either through the transfer of money between portfolios (requiring simultaneous buying and selling within the underlying funds) 2 Asset allocation explained 93.6% of variation in return between pension plans according to a study by Brinson, Beebower, and Hood, FAJ, Other studies confirmed similar dominance of asset allocation. 7/11/2005 Why Asset Alloc doc Page 2

3 or the buying and selling of actual securities to achieve the desired shift. Such transactions introduce a high hurdle rate to implement asset allocation strategies effectively. We can achieve similar economic exposure using equity index and bond futures importantly targeted toward the specific exposures we wish to adjust at a mere fraction of the cost. Asset allocation turnover can be very disruptive to underlying funds or investment portfolios. Consider the impact of selling some auto and energy stocks to raise cash to purchase corporate bonds. By reducing turnover in the constituent portfolios using derivatives, we also minimize and postpone realizing taxable gains. This example highlights important lessons in unintended consequences for balanced or asset allocation fund investors. We can identify opportunities for excess return, but also must be able to trade cost effectively and minimize turnover in cash securities to maximize return. Portfolio transaction costs are a function of turnover and the type of securities traded or held. While a focus on keeping turnover low can improve performance, actual transaction costs play an important role as well. Portfolio managers utilizing derivatives to reallocate country and currency exposure will undercut performance far less than managers trading only cash securities. Combining skills in both security selection and asset allocation, our investment objective seeks to add value through two independent strategies -- dual alpha. The search for multiple investment sources of excess return derived simultaneously without leverage can be an elusive goal. It is inefficient to try to do so using just cash securities. Our investment process must be able to actually run two strategies, not side-by-side, but overlaid on top of one another and completely independent. The important cash-futures arbitrage in equity index and fixed income futures enables Portfolio Structure Value Added Active Asset Allocation 1.0% Asset Allocation Fixed Equity Cash Income Managers Managers $20 Million Portfolio 2.0% Security Selection 8.0% Benchmark Return 11.0% Portfolio Return 3.0% Active Return There is a tendency to include too many different investments seeking to improve performance and diversification. More decisions won't necessarily improve return 7/11/2005 Why Asset Alloc doc Page 3

4 nor necessarily reduce risk, but can increase confusion and complexity. If allocations are small to certain asset classes, the impact may be diminimus. Decisions like stocks versus bonds and cash have the greatest impact on portfolio returns. To a lesser degree, equity style (value vs. growth and size), as well as some fixed income sector decisions (corporate, high yield, government) are meaningful, as well. Investing in taxable versus nontaxable accounts can and should affect asset class return expectations. Keeping the number of distinct investment decisions to a minimum should be a goal for financial advisors. Small divergences from target policy allocations can be a challenge to rebalance with too many moving pieces, even when they add up to significant risk exposure overall. Monitoring and rebalancing complex portfolio structures requires a more active role of fiduciaries often without providing much additional benefit. Many so-called asset classes or sub-classes are difficult to distinguish long-term. Asset allocation is a dominant force in retirement and wealth management. Choosing appropriate long-term strategic policy benchmarks can be as important for our investors as the tactical opportunities we exploit to add value in the shorter-term. Strategic Asset Allocation Tactical Asset Allocation Product Selection and Performance Long-term Monte Carlo simulation to Strategic Policy Benchmarking HighMark utilizes a stochastic process based on determine the optimal portfolio composition for select points along the efficient frontier. Our empirical resampling methodology is an alternative to calculating discrete mean-variance efficient portfolios from uncertain estimates of expected Annual Return (%) Historical Asset Class Return Scatter ( ) 80/20 Value Blend Equity Small-cap 30/70 S&P 500 Blend Int'l 60/40 Growth Emerging Equity Blend Equity Markets Bonds Short Bond Cash Risk (σ) Source: Thomson Datastream & HighMark Capital Management 7/11/2005 Why Asset Alloc doc Page 4

5 return and covariance. As many have observed, small adjustments to our capital expectations can have a meaningful impact on solutions. Instead, the efficient frontier is determined by a proprietary stochastic optimization methodology designed to rely on the empirical sample distribution of capital market returns, without assuming normal return distribution typical of traditional mean-variance analysis. HighMark s empirical resampling or bootstrapping technique is more adept at identifying robust portfolio efficient points. Selected portfolio allocation solutions derived in this manner tend to be more stable and intuitive. Statistical representation of the behavior of capital markets is more realistic, including return skewness (markets tend to appreciate), kurtosis (higher occurrence of outliers), versus embracing typical normality assumptions imposed by traditional modern portfolio theory (MPT) techniques. Furthermore, cross-sectional correlation between asset classes and time series autocorrelation are preserved in this analysis. It is not necessary with our methodology to impose arbitrary constraints on optimization or otherwise flawed assumptions that decrease portfolio efficiency, but often necessary to generate palatable solutions. Mean-Variance solutions, used to derive a simultaneous asset allocation and investment selection recommendation, yield solutions that are often unintuitive, sensitive to small changes in inputs, and may need to be constrained, providing less efficient solutions. Aggregate strategic policy decisions are polluted by the influence of more volatile fund specific characteristics and greater uncertainty. By separating the asset allocation decision and investment selection, much as an investment consultant would, it is possible to isolate the forces determining these independent decisions. The strategic asset allocation policy frontier is determined through a proprietary stochastic optimization methodology based on the empirical distribution of capital market returns. Our methodology does not assume a normal distribution of returns typical of more traditional mean-variance analysis. This approach accounts for higher occurrence of outliers (kurtosis) and skewness observed in asset returns that often drive instability in mean-variance solutions. Furthermore, any observed nonstationarity in assumed normal return correlations is addressed explicitly by the empirical distribution sampling process. Forecasting future capital market returns over a long time horizon is a difficult and uncertain proposition. Small differences in the rate of return can have a meaningful impact on mean-variance derived asset allocation conclusions. Our investment conclusions are rooted in empirical observations supported by financial, economic and investment theory. Asset classes correspond to the fundamental market risk factors that we determined are most relevant to managing portfolio risk for individual investors. 7/11/2005 Why Asset Alloc doc Page 5

6 Our asset allocation frontier is derived from analyzing the empirical distribution of market returns. The procedure used to derive our asset allocation frontier is a form of stochastic optimization utilizing empirical bootstrapped sampling. By sampling contiguous historical blocks of time, preserving the correlation between asset classes, we are able to model the behavior and interaction between markets. Serial autocorrelation observed in market returns is represented in the methodology utilized to derive our asset allocation policy. Security Universe Our investment universe can tap a wide range of investment vehicles that in combination will achieve our objective. Specifically, we would utilize: HighMark mutual fund shares Exchange traded funds Stock index and Government Bond futures Individual securities The appropriate class and fund level charges are yet to be determined. Fund Descriptions See Appendix A. In this paper we have shown strategic policy benchmarks based on a risk profiles consisting of 20%, 40%, 60%, 80% equity. Alternatively, we might actually want to target 30%, 60%, and 80% equity, plus a global version targeted at 60% equity as shown in Model #4. 7/11/2005 Why Asset Alloc doc Page 6

7 Appendix A: Allocation Portfolio Benchmarks Core Portfolio Allocations (Rounded) Asset Class Benchmark Model#1 Model#2 Model#3 Model #4 US Equity S&P US Value S&P/Barra Value US Growth S&P/Barra Growth US Small-cap Russell Int l Equity MSCI EAFE (US$) US Bonds Lehman Aggregate US Short Bonds Lehman Govt 1-3 yrs US Cash US Treasury Bill (3m) Total Equity Expected Return Expected Risk Opportunistic Investment Allocations US Sectors & REITS Mid-cap Equity Small-cap Value/Growth Regional Equity Emerging Market Real Estate (Traditional) Municipal Bonds Corporate Bonds High Yield Bonds Mortgage Bonds International Bonds Note: Asset allocation policy weights (rounded) derived from proprietary optimized empirical re-sampling methodology. 7/11/2005 Why Asset Alloc doc Page 7

8 Appendix B: Asset Class Correlations CORR LCOR LVALU LGRTH SCOR SVLU SMGRW EAFE EGMK REIT SHTB INTBD LGBD CPBD MORT HIYLD NOUS CSH 1 LRGECAP LGVALUE LGGWTH SMCORE SMVALUE SMGWTH EAFE EMGMKT REIT SHTBND INTBOND LONGBND CORPBND MORTBND HIGHYLD NONUS CASH Real Retn Return Risk Real Retn Risk Large Equity Small-cap EAFE Bonds Cash /11/2005 Why Asset Alloc doc Page 8