Even More MIPY

by Craig L. Israelsen and Patrick McDonough
Reprinted by permission from Financial Planning Magazine, April 2006

n the November, 2004 issue of Financial Planning, we first introduced a novel approach to rebalancing by reallocating assets to index funds on the basis of an index's performance in the previous year. We found that annually reallocating assets from to the middle-performing index in the prior year (MIPY) among a portfolio of three value indexes, generated an excess return of more than 200 basis points for the 18-year period from 1987 to 2004. (Financial Planning, May 2005). We also tried this strategy with portfolios of growth indexes (July 2005). The current study expands on those analyses by looking at a seven-index, multiple-asset portfolio to further illustrate the potential of a MIPY-based reallocation strategy.

We analyzed the 24-year period from 1982 to 2005 using raw performance data from Morningstar Principia. We found that a MIPY strategy outperformed the annually rebalanced, equally-weighted, seven-index portfolio by 233 basis points and a buy-and-hold, seven-index portfolio by 276 basis points. But in both cases, the risk of the MIPY strategy as measured by standard deviation was higher. But we were able to create similar results with a comparable risk level, by applying the MIPY strategy to only 40% of the assets in a portfolio and equally allocating the remaining 60% to the remaining six indexes. This tactic outperformed an annually rebalanced portfolio by 86 basis points with significantly less risk. (The returns discussed in this article are pre-tax and exclude commissions. Annual allocation should of course take advantage of capital gains tax rates, or preferably take place within a tax-sheltered account. As always, past performance is no guarantee of future results.)

Our aim in this study is to show some simple approaches to annual rebalancing using a realistic portfolio containing the following indexes: the Russell 1000 Value Index (R1000V), the Standard & Poor's Midcap 400 Index (S&P400), the Russell 2000 Value Index (R2000V), the Wilshire REIT Index (REIT), the Morgan Stanley Capital International Europe, Australasia, Far East Index (EAFE), the Morgan Stanley Capital International Emerging Markets Index (EM), and the Lehman Brothers Aggregate Bond Index (BOND).

CORRELATING COUPLES

All other things being equal, weakly correlated indexes are better portfolio choices. Among the seven indexes there were a total of 21 possible pair correlations. Over the 24-year period, 17 of those correlations were weak (.33 or lower) or moderate (.38 to .60). All of the U.S. equity indexes are weakly or moderately correlated with both the foreign equity indexes and with the bond index. Not surprisingly, the strongest correlations were among the U.S. equity indexes. The highest correlation (of .88) was between the Russell 1000 Value Index and the S&P Midcap 400 Index.

Of course, index correlation isn't static. Many recent studies by both academics and investment managers have shown that the correlations between U.S. and foreign equities vary over time. Some suggest these changes are part of a trend toward higher correlations. But some analysts suggest that increasing openness in EM markets will lead to more rather than less economic specialization, resulting in lower index correlations over time.

Admittedly, our previous findings bias us in favor of value indexes, which tend be less volatile than growth indexes. Academic research finds evidence of a "value" premium as well as a "small-cap" premium. Thus, of our four U.S. equity indexes in this analysis, two are explicitly value oriented (Russell 1000 Value and Russell 2000 Value).

The Russell 1000 Value Index includes the Russell Midcap Value Index. Thus, we selected the S&P MidCap 400 for a blend of growth and value stocks. Moreover, it offers a different mid-cap value exposure since the Russell and the S&P are constructed differently. The Wilshire REIT Index adds an additional value orientation and has a relatively weak or moderate correlation with U.S. large caps, foreign and bond indexes. Among fixed-income indexes, the intermediate-term Lehman Aggregate Bond index is commonly used as a proxy for the broad, investment-grade U.S. fixed income market. (The Lehman 1-3 Year U.S. Government Bond Index is also a good choice, but it had a lower return than the Aggregate Bond Index.)

FOREIGN INFLUENCE

The MSCI EAFE covers almost all of Siegel's 42.8% of the market, which is non-U.S. developed countries. EAFE is the most popular foreign developed-country equity index. MSCI Emerging Markets (EM) covers most of the 4.9% of the market in developing countries. MSCI EM improves foreign returns and has mostly weak correlation with other indexes, but since it only started in 1988, we used annual returns from the MSCI Pacific ex. Japan Index from 1981 to 1987 and then the EM Index from 1988 to 2005. The Pacific ex Japan Index has a .85 correlation to the EM Index and a .74 correlation to the EAFE Index over the past 10 years.

Using this seven-asset portfolio, we tested seven annual portfolio management protocols from 1982 to 2005 against a baseline buy-and-hold (B&H) approach. The seven strategies were:

1) Annually rebalancing (AR) at the start of each year, giving an equal weight of one-seventh, or .1429, to the return of each index. 2) Reallocating all portfolio assets into the prior year's middle-performing index at the start of each new year, i.e. the MIPY strategy. 3) Reallocating all portfolio assets into the prior year's best-performing index at the start of each new year, a strategy we call Best Index in Prior Year (BIPY). 4) Reallocating all portfolio assets into the prior year's worst-performing index (WIPY). 5) Reallocating 40% of portfolio assets into the prior year's middle-performing index at the start of each year and the remaining 60% of portfolio assets equally allocated into the remaining six indexes (i.e. 40% MIPY). 6) Reallocating 40% of portfolio assets into the prior year's best-performing index at the start of each year and the remaining 60% of portfolio assets equally allocated into the remaining six indexes (i.e. 40% BIPY). 7) Reallocating 40% of portfolio assets into the prior year's worst-performing index at the start of each year and the remaining 60% of portfolio assets equally allocated into the remaining six indexes (i.e. 40% WIPY). COLOR CODING

In "The Seven," ON PAGE TK, the color coding shows how this works. We used the 1981 annual return for each asset class to initiate the MIPY, BIPY, and WIPY protocols in 1982. Among the seven indexes in the portfolio, the bond index return of 6.25% was the middle performing index in 1981, therefore it was chosen as the BIPY asset (as noted by the yellow shading) meaning that all portfolio assets were reallocated into it at the start of 1982. Blue coding refers to the indexes selected using the BIPY protocol, while the pumpkin color shows the WIPY indexes. If the box has no color, it was not used in the MIPY, BIPY, or WIPY protocol in that particular year.

So, to extend the example, based on the MIPY protocol at the beginning of 2006, all portfolio assets would have been reallocated to the S&P MidCap 400 because its 12.55% return in 2005 was the middle performing return among the seven indexes. Under the BIPY protocol, all assets would have been reallocated to the Emerging Markets Index, and using WIPY all assets would have been reinvested in the Aggregate Bond Index. Clearly these particular strategies are extremely aggressive in that they often require complete portfolio reallocation at the beginning of each year. However, in some cases, the assets are left in place for several years. The EAFE Index, for example, was the selected BIPY index from 1986-1988.

As shown in "Beyond Buy-and-Hold" ON PAGE TK, the 24-year annualized return of the MIPY-based portfolio was 15.89%, beating the annually rebalanced portfolio (AR) by 233 basis points and a buy-and-hold (B&H) approach by 276 basis points. However, MIPY generated a higher return at the cost of greater volatility. MIPY's 24-year standard deviation of annual return was 20.2% compared with 12.66% for AR and 17.99% for B&H. Nevertheless, MIPY turned an initial $10,000 investment in 1982 into $344,564 by the end of 2005. By contrast, the ending account value for a B&H protocol was only $193,137 and that of an AR portfolio was $211,448. MIPY also outperformed the BIPY protocol by 311 basis points and WIPY by nearly 1,000 basis points. In both cases, MIPY had a lower standard deviation of return than BIPY and WIPY.

If we measure the differences in portfolio performance using averaged three-year rolling returns over the 24-year period, a MIPY portfolio outperformed a B&H portfolio by 185 basis points, an AR portfolio by 200 basis points, a BIPY portfolio by 236 basis points and a WIPY account by over 800 basis points (see "Rolling Action" DIRECTION TK). RISK ADJUSTMENTS We also tried a series of hybrid approaches by reallocating only 40% of the portfolio according to the MIPY, BIPY or WIPY protocol with the remaining 60% split equally among the other six indexes. The annualized return of the 40% MIPY strategy was only 147 basis points below 100% MIPY (see "Beyond Buy and Hold"), but the reduction in standard deviation of return was 627 basis points. In addition, 40% MIPY outperformed 40% BIPY and 40% WIPY. A 40% allocation to the MIPY index outperformed AR by 86 basis points.

You can see in "Rolling Action," ON PAGE TK that the primary impact of these hybrid approaches is a significantly better risk-adjusted return (as measured by Return/Standard Deviation and the Sharpe Ratio). When you measure the performance using the average of 22 three-year rolling returns, a 40% allocation to the MIPY index and a 10% allocation to each of the six other indexes outperformed a B&H strategy by 57 basis points and an AR strategy by 72 basis points, but with greater volatility (by 61 and 83 basis points respectively when comparing the standard deviations). The 40% MIPY approach had a higher average three-year rolling return compared to both the 40% BIPY and 40% WIPY protocols. The BIPY portfolio underperformed the AR portfolio by 78 basis points on an annualized-return basis and 36 basis points using averaged three-year rolling returns. BIPY tends to perform better with a portfolio of growth-oriented indexes, which are more volatile than value indexes. This is consistent with academic studies of the momentum approaches, which tend to succeed with more volatile stocks. So, BIPY may deliver higher returns than MIPY if using more volatile growth-oriented indexes, but investors will have a rougher ride.

The WIPY strategy was clearly the worst approach-it underperformed AR by 761 basis points using annualized returns and 614 basis points using the average three-year rolling return. WIPY's poor performance is also consistent with academic findings. Investors and advisers sometimes confuse WIPY-type strategies with mean-reversion approaches. The WIPY approach bets that last year's loser will be this year's winner. However, long-term mean-reversion cycles typically last four to eight years, not one or two years. Therefore, WIPY probably has nothing in common with a mean-reversion approach.

Overall, MIPY generated the bigger payoff. The 14.42% annualized return of the 40% MIPY protocol over the 24-year period was 86 basis points better than the 13.56% return of AR. We highlight this comparison because AR is a well-recognized portfolio management protocol. It is important to note that this is 86 risk-adjusted basis points of improvement. You can see in "Rolling Action," that the two measures of risk-adjusted return ("Return/Standard Deviation" and "Modified Sharpe Ratio") are nearly identical using annualized returns for AR and 40% MIPY and comparable if using averaged three-year rolling returns. So, why not make life simple and just do AR? Or, why not simply invest solely in the Wilshire 5000, which averaged a healthy 12.83% between 1982 and 2005?

The case for AR rather than the Wilshire 5000 seems strong: 73 basis points more annual average return with 20% less volatility (12.7% standard deviation for AR compared with 16.0% for the Wilshire 5000). This diversification away from one index to a non-cap-weighted portfolio of seven indexes is clearly valuable. As Robert Arnott points out in his "Editors Corner" article in the March/April 2005 issue of Financial Analysts Journal, "We should not assume that, even in an efficient market in the pricing of individual stocks, we have an efficient market for the cap-weighted indexes."

The case for 40% MIPY rather than AR would be more compelling if we better understood why MIPY works. In our defense, the phenomena of value and small-cap premiums, momentum, and mean reversion were also poorly understood 25 years ago. Furthermore, both academic researchers and investment managers continue to debate whether these phenomena are exploitable effects or merely annoying anomalies. The answer makes a difference: are these phenomena real problems to be turned into investment opportunities, or simply paradoxes to be explained away? MIPY could turn out to be a similar kind of phenomenon.

Is an 86 basis point improvement in return significant? In his "Editor's Corner" article in the July/August 2004 issue of the Financial Analysts Journal, Robert Arnott suggests that an investor can boost return by as much as 250 basis points through systematic rebalancing, asset selection, and tactical asset allocation. The 40% MIPY portfolio management approach includes these three elements and captures almost 35% of this hypothesized 250 basis point premium. We understand that many will reject the notion that 250 basis points can be added to an equity portfolio as Arnott has suggested. However, we have demonstrated a portfolio management protocol (MIPY) that generated a 24-year annualized return that is129 basis points higher than B&H and 86 basis points higher than AR. In an era when many analysts are predicting single-digit asset returns, an extra 86 basis points of annual return may be very attractive to long-term investors.

If nothing else, our research identifies which of these eight portfolio management protocols is closest to the much desired northwest corner of a risk/return map, where Investing Nirvana resides (see "Go Toward the Light" on page TK). It's worth observing that moving from WIPY to BIPY to MIPY increases return significantly while reducing risk slightly. The same general relationship is found between 40% WIPY, 40% BIPY, and 40% MIPY, but with reduced magnitude. B&H is at the center of the portfolio management constellation, while AR moves one closer toward Nirvana. The efficient frontier among these eight strategies (as shown by the line) extends from AR, to 40% MIPY, to 100% MIPY.

For the risk-averse client, AR may be the most appropriate strategy. Clients with higher risk tolerance (or those who don't micro-manage their portfolio) a 40% MIPY protocol shows promise. A 100% MIPY protocol is reserved for those who are willing to make dramatic shifts in their portfolio each year. Over time it seems to pay off, but the path won't be smooth.

The Seven

Beyond Buy-and-Hold

Buy-and-Hold Results:  24-year annualized return 13.13%, standard deviation of 17.99%, Growth of $10,000 = $193,137

Rolling Action

Go Toward the Light

____________________________________________________________________________________
Craig L. Israelsen, Ph.D. is an associate professor at Brigham Young University. He teaches family finance in the Department of Home and Family Living. His research interests include mutual fund analysis. He writes monthly for Financial Planning magazine. Learn more about Craig Israelsen at http://familyliving.familylife.byu.edu/faculty/israelsen.htm

Patrick McDonough is a chief financial officer for high-tech startups and lives in California.