The reader should also consider the gap between investor behavior and market logic. For example, Brandes points out that studies by Dalbar Inc. have shown that although the stock market averaged 16.3% annual returns from 1984 through 2000, the average fund investor made an average of only 5.3% annually on equities. To make matters much worse, average fund retention was only 2.6 years, in spite of most investors' long-term investment horizons. My personal experience is that many investors seek great managers but then second-guess them regularly. We've all heard the one about how most of the legendary Peter Lynch's fund investors made no money because they didn't keep their shares. I believe that value investing holds the key to changing this kind of behavior. Value investing, when properly explained and adhered to, takes the investor's eye off prices and refocuses them on staying in for the long haul, like a business owner. The results are worth having, and it is our job to convince our clients to change their thinking.
Portfolio Construction Basics
Six model portfolios were constructed to evaluate risk and return relationships for various value-driven parameters. Portfolios have been modeled using March 31, 2004, data on Morningstar Principia Pro software. Principia Pro portfolios are automatically rebalanced on a monthly basis. The portfolios are all set to the same general allocation, 70/30 stock/bond, using a model allocation somewhat modified from PIE Technologies MoneyGuide software. Mutual funds and stocks were chosen for each model portfolio using various guidelines discussed below. The initial list consisted of 100 funds with at least ten-year records, and good relative performance in the three-year period of the recent bear market. (The funds chosen for this study are not recommended for investors; rather, they are used merely for evaluation purposes.)
The baseline portfolio for comparison purposes uses a 70/30 mixture of index funds, which is labeled here as the Baseline Index Aggregate. Equity index funds for each market capitalization class included in this portfolio are all "blended" with respect to growth vs. value styles. A small proportion of a real estate investment trust (REIT) fund is included in some of the models to partially account for the role played by alternative value investments. Eight passive index funds are included in the Baseline Index Aggregate, six of them from Vanguard Group. The model allocation is: large-cap blend (LCB) using Vanguard 500 Index, 32%; mid-cap blend (MCB) using Vanguard Extended Market, 10%; small-cap blend (SCB) using Vanguard Small-Cap Index, 8%; international stocks (ITL) using Schwab International Index, 10%; real estate equities (REE) using Vanguard REIT Index, 10%; short-term bonds (STB) using Vanguard Short-Term Treasury, 5%; intermediate-term bonds (ITB) using Fidelity Spartan Government Income, 10%; and long-term bonds (LTB) using Vanguard Long-Term Bond Index, 15%. This model approximates the results gained in the last ten years from passive index investing, hence the term baseline.
A High Dividend Stock model was prepared by screening 34 stocks paying dividends greater than 3.50% from a list of high-dividend stocks, combining that with three bond funds in a 70/30 allocation and ignoring the mid-cap, small-cap and international asset classes. Real estate is partially covered however by the inclusion of three REITs in the list of 34 stocks used. Thus the equity portion of this model uses individual large-cap stocks and no mutual funds. Model allocation is as follows: LCV-34 stocks-70%, STB-PIMCO Low Duration A-5%, ITB-PIMCO Total Return A-10%, and LTB-PIMCO Long-Term U.S. Government A-15%. This model approximates a low activity level of value investment that is very tax efficient and involves little tactical management once the initial stocks are chosen.
The Value Funds Only model was constructed using nine funds that passed muster on at least five out of eight value measures available or easily derived from Morningstar Advanced Analytics software. These value measures are: Morningstar's best fit classification, dividend yield greater than 1.50% (S& P 500 avg.), P/E less than 25, P/B less than 3.50, P/CF less than 14.00, earnings yield greater than 4.25% (ten-year Treasury yield), turnover ratio less than 25%, and assets in top ten holdings greater than 40%. All funds also were required to have good relative numbers for the "worst three-year return" calculated by Morningstar. The model allocation is as follows: LCV-American Funds Washington Mutual A-32%, MCV-Ariel Appreciation-10%; SCV-Fidelity Low-Priced Stock-8%, ITL-Tweedy, Browne Global Value and American Funds Capital World-10%, REE-Cohen & Steers Realty Shares-10%, STB-PIMCO Low Duration A-5%, ITB-PIMCO Total Return A-10%, and LTB-PIMCO Long-Term U.S. Government A-15%. This model attempts to show a variation on asset allocation using value funds only, but without strict rules about which of the eight value parameters should be dominant.
A Best Diversification model was constructed using the lowest (R2 less than 0.70) three-year R-squared correlation statistics for funds in each asset class other than large cap. In addition, except for large-cap growth, funds having at least four of the eight value measures mentioned above were used to fill out the rest of the asset allocation. All of the funds were also required to have good relative numbers for the Morningstar "worst three-year return." Nine funds were used in a model allocation as follows: LCV-American Funds Washington Mutual A-16%, LCG-American Funds Growth Fund-16%, MCV-Ariel Appreciation-10%, SCV-Heartland Value-8%, ITL-Tweedy, Browne Global Value-10%, REE-Cohen & Steers Realty Shares-10%; STB-PIMCO Low Duration A-5%, ITB-PIMCO Total Return A-10%; and LTB-PIMCO Long-Term U.S. Government A-15%. This model attempts to balance diversification with value-driven investing, producing a low correlation and thus lower risk allocation.
The Smaller Market Caps model was constructed using only funds with mid-cap or small-cap average market capitalizations (less than $10 Billion). In addition, each fund was required to have at least four of the eight value parameters mentioned above, and the lowest average turnover possible. Again, all of the funds were required to have good numbers for Morningstar's "worst three-year return" relative to the index. Ten funds were used in a model allocation as follows: MCV-Franklin Rising Dividends A, Lord Abbett Mid-Cap Value A and Ariel Appreciation-30%, SCV-Heartland Value, Fidelity Low-Priced Stock, T. Rowe Price Small-Cap Value and Franklin Balance Sheet-40%, STB-PIMCO Low Duration A-5%; ITB-PIMCO Total Return A-10%, and LTB-PIMCO Long-Term U.S. Government A-15%. There is no allocation for real estate in this model. The model attempts to show the results of trying to harvest both the "value premium" and the "small-cap premium" noticed in studies of returns by Ibbotson Associates and others.
The last model presented here, the Focus Mix model, was constructed using as a screen a combination of low average turnover (less than 25%) and high, even unorthodox (greater than 40%) average percentages of "top ten" assets held in funds. All funds also were required to have at least five of the eight value measures, and to have good relative performance for Morningstar's "worst three-year return." The weighted average turnover for equities in this portfolio is only 14.8%. Nine funds were used in this model, as follows: LCV-Sequoia Fund and Legg Mason Value Prime-42%, MCV-Franklin Rising Dividends A-10%, SCV-Franklin Balance Sheet- 8%, ITL-Tweedy, Browne Global Value and American Funds Capital World-10%, STB-PIMCO Low Duration A-5%, ITB -PIMCO Total Return A-10%, and LTB-PIMCO Long-Term U.S. Government A-15%. There is no allocation in this model for real estate. The model shows the impact of combining asset allocation with a somewhat unorthodox value investing style that involves true long-term "buy & hold" discipline.
Results And Interpretation
A summary of the results appears in Table 1. From the table you can see the wide range of results produced for a ten-year period, compared with results for the Baseline Index Aggregate and for an S & P 500 index fund alone. The Baseline Index Aggregate has an average ten-year annualized return of 10.55%, compared with 11.61% for the S & P 500 index fund alone. The bear market dip produced a "worst three-year return" of -5.68% for this portfolio, compared with -14.4% for the S & P 500 index fund alone. The ten-year standard deviation and beta are quite low, alpha is modestly positive, the Sharpe Ratio is moderately positive, and the R-squared correlation is high.
The High Dividend Stock model has an average ten-year annualized return of 12.11%, beating the Baseline Index Aggregate by 1.56% annually, and beating the S & P 500 index fund alone by 0.50% annually. The "worst three-year return" for this portfolio is +4.76%, considerably better than that for the Baseline Index Aggregate. The ten-year standard deviation is low, beta is very low, alpha is moderately positive, the Sharpe Ratio is moderately high, and the R-squared correlation is very low.