| |
 July 2004 Ready for the Pros
Benchmark comparisons aren't unique to quants. In fact with the rise of mutual funds and Morningstar's rating system, categorizations and comparisons are quite the rage.
It's not always been that way, but became so in the late 1990s when the S&P trounced most active money managers. Investors began to question paying a management fee when an "unmanaged" index offered better results. Professional managers objected that the S&P 500 wasn't the appropriate standard for all portfolios. It is, after all, a large cap core measure while many managers employed a value, growth, or small cap style. Those who didn't become "closet indexers" started comparing their results to other more appropriate benchmarks and the era of relative performance was born. After almost four full years, we've collected a lot of data comparing P3 and P4 with the S&P 500, but how do they compare against professionally managed portfolios with the same benchmark? Is that management really worth paying for or could you do just as well applying our quant portfolios' formulas to a list of stocks? Investment firms will tell you there's no way a quant "black box" can possibly match the skill and experience of their professional managers. Are they right? Let's see. Less than PerfectMutual funds are a good proxy for professionally managed portfolios. Performance and analytical data for them is readily available from Morningstar. The only question is, which Morningstar category is most appropriate for P3 and P4?In actual practice, they tend towards large cap growth (see Apples to Apples), but they weren't supposed to. The data upon which they were based was drawn from the ten years of the 1990s, a period when growth handily outperformed value (see Chart 1). As a result, both models favor value over growth even though that wasn't the intention.
But both were created to track and exceed the S&P 500. To change benchmarks simply because of out of sample performance would be data mining, an anathema of quantitative and statistical analysis. The S&P 500 -- not a pure growth benchmark -- is still the appropriate measure, so Morningstar's "large cap blend" is the proxy for active management. Yet even this isn't a perfect comparison. There are several reasons why: •Expense Ratios -- The Morningstar return data is net of all expenses. The large blend category has an expense ratio of 1.30% which has been subtracted from the fund managers' gross returns. P3 and P4 don't have specific expenses since they're only models. You can calculate what the expenses would be, but that's only an approximation. For example, based on the actual number of trades in each portfolio from inception through April 30, 2004, transaction costs utilizing a discount broker such as Ameritrade at $10.99 per trade with 100-share lots would be 0.73% for P3 and 0.44% for P4. Since discount commissions are per trade rather than per share, these ratios would fall if you used larger lots and rise with smaller. With no management fee per se, these figures would represent the models' total expense ratios. Given this variability, it's not possible to make a simple adjustment to either set of returns to make them completely compatible. To some extent, it may all come out in the wash because of the differences in •Total Return vs. Capital Appreciation -- An investment's total return is the combination of its current yield (dividends) and its change in market value (capital appreciation). Morningstar reports total return while we only measure capital appreciation for our quant portfolios. Because of this, some of the differences in expenses may be minimized. According to Morningstar, funds in the large cap blend category have a 12-month yield of .50%. As of mid-June, Portfolio 3 which is heavily dependent on tech stocks had a yield of .10%, while more broadly diversified Portfolio 4 yielded .90%. As a means of comparison, the S&P 500 yielded about 1.50%.
Chart 2 shows the ranges of the index's yield for the past four years. As you'll notice, dividends can be a major component of return -- especially in the 2001-2002 bear market when stock prices were actually falling. •Portfolio Composition -- All of our equity portfolios are just that -- all equity portfolios. The 500 stocks of the S&P 500 comprise the entire universe of potential holdings for P3 and P4. Regardless of market conditions or trends, both portfolios are always 100% invested in stocks from the benchmark index. That's not the case for the mutual funds in the large blend category (or any category for that matter). Unlike our model portfolios, funds are constantly receiving new subscriptions and redemptions from departing shareholders. The former brings in uninvested cash while the latter requires a certain amount of liquidity. Cash -- or money market investments -- diminish returns when stocks are rising while minimizing losses when stocks fall. Either way, mutual funds' returns don't totally reflect those of the manager's underlying investments. Fund managers' own decisions move their portfolios even further away from their purported benchmarks. As market conditions change, they often let their styles and holdings stray from their stated objectives. We've documented this before, and according to Morningstar, as of April 30 only 89% of "large blend" funds was invested in U.S. domestic equities. The rest was split between foreign stocks, bonds, cash, and other holdings such as derivatives (see Chart 3). And even that 89% may not be in large cap stocks if small or midcaps are outperforming. Many managers purposely try to game the benchmark index by going outside of it for better returns. For example, if small caps are doing better, large cap managers will load up on them rather than see their results suffer with large caps. That's how many funds beat their benchmarks, particularly in a down market. (For a detailed discussion, see Active Management's Dirty Little Secret.)
To be sure, the SEC has set some minimal guidelines for the percentage of large cap stocks a fund must hold if it claims to be a large cap fund in its prospectus. Even so, it's not 100% but that is the requirement for P3 and P4. When they beat the index they beat it at its own game, not by gaming the index. So there's no concise way to adjust the quant portfolios or the large cap blend universe to assure an accurate comparison. Some of the differences favor one over the other and vice-versa, and perhaps to some degree, may tend to cancel out. Despite these discrepancies, let's compare them anyway. Who's the Real Indexer?Over the 46 months from July 2000 through April 2004, stocks experienced a 3-year bear market and then a sharp recovery. Despite the rally of 2003-early 2004, stocks still finished the period down.Before looking at returns though, let's first consider portfolio analytics. They can go a long way towards explaining ultimate results. Chart 4 compares characteristics of P3 and P4 with large blend funds and the benchmark index. There are at least two surprises here. The first comes in regard to correlation. Despite the greater latitude exercised by large blend fund managers, their returns are closer to that of the S&P 500 than are P3's and P4's. Apparently their cash, small or foreign stocks, and bonds don't keep the benchmark index from accounting for almost 93% of their performance. On the other hand, P3 and P4, drawn completely out of the S&P 500, have much lower correlations. Movements in the S&P 500 only explain 61% and 76% of their returns, respectively. This difference in correlations probably stems from the fact that regardless of their other holdings, most large cap portfolio managers tend to hold 40-50 of the largest-cap stocks in the index. The S&P 500 is a capitalization-weighted so the largest stocks carry a much greater weight. Back in 1999, the top 50 stocks contributed all the return of the S&P 500, so it's no wonder most large cap funds perform alike and very close to the index. That's why whether they intend it or not, most end up being closet indexers. But this isn't the case with the quant portfolios. P3 has the freest reign to concentrate in only a few of the 10 S&P sectors. When it focuses on the correct ones, it can do extremely well (as in 2003), but when it relies the wrong ones, it can really suffer (as in 2001). It's no wonder then that it has the lowest correlation with the benchmark.
P4 is more broadly diversified, always having at least some representation in all sectors. Yet it's never held more than 54 stocks and these were never only the largest. If the index is dominated by the biggest constituents, P4 will always track differently giving it a lower correlation than the funds. That's also why the quant portfolios have higher betas than the average large cap blend fund. Despite the difference in composition, P4's beta is only slightly above that of the index while P3's reliance on growthier issues makes it 50% more volatile than the benchmark. Fund managers' reliance on those largest stocks coupled with their use of cash and bonds, make their results slightly less volatile than the index. Again as you'd expect, while roughly the same, the standard deviations of P3 and P4 are considerably higher than that of the index and large cap blend funds. The relative concentration of the quant portfolios leaves them inherently more volatile while the non-equity assets in the funds make them more stable. Turnover is almost non-existent in the S&P 500; just averaging 4% over the past four years. That's understandable since most large companies don't go bankrupt and aren't purchased by others. Once in the index, their stocks tend to stay put. Many mutual funds have actually seen a decline in turnover, especially after the bear market. In the go-go '90s, hotshot fund mangers often traded actively but once stock returns fell off, the tax impact of that approach became painfully evident to shareholders and turnover cooled. Morningstar reports it averaged 79% over the measurement period. There are two things of note regarding the quant portfolio's turnover. First, it's significantly greater than that of the funds and the index. That's to be expected since the portfolios are much smaller. As a result, any changes represent higher percentages. Secondly, it's worth mentioning that P3's turnover has fallen dramatically -- to 23% -- over the past year since making a minor change to its model in July 2003. As time goes by, the overall figure should also reflect this reduction. (For more details, please see Taming Turnover.)
Which brings us to the second surprise in this comparison: mean returns. As mentioned above, the measurement period, July 1, 2000 - April 30, 2004, captured the entire bear market. Despite strong performances in 2003, the quant portfolios, the funds, and even the index itself averaged negative annual returns. But look where the best mean is found: P4. On the other hand, the worst belongs to P3, but still it's not far from that of the benchmark index (-5.7 vs. -5.1%, respectively). Of course these are gross figures versus the net for the large cap funds. Nevertheless, if adjusted for the transaction costs suggested above, P3 would fall further behind by P4 would still have the best average. What a surprise -- and no management fee. You Don't Have to Draw a PictureSo let's look at the periodic returns in a little more detail. The bars in Chart 5 show the annual returns for the portfolios, funds, and index while the lines show the cumulative results. Bear in mind the 2000 results are only for the last half of the year while the 2004 are January - April.As the portfolio analytics suggested, when P3 is bad, it's really bad. That's precisely what happened in 2000, 2001, and 2002. By the end of 2002, a $100 initial investment would have been worth about $42. That same investment in the S&P 500 and P3 would be about $60. Large cap blend managers using bonds and cash were able to retain about 70% of their initial value. The tide turned in 2003 when the additional risk of P3 and P4 abruptly changed from a liability to an asset. They rocketed 60.1% and 35.2%, respectively. Despite the best efforts of large cap blend managers to goose their returns with small cap stocks, they barely outpaced the index, gaining 26.9% vs. 26.4%.
Even so, P3's stellar 2003 still couldn't bring its return up to that of the benchmark. But here's something to think about: Timing is everything. If the portfolio had the same periodic returns just in reverse order, by April 30, 2004 that same $100 initial investment would have been worth $99.77, a mere 0.23% decline. While that certainly doesn't vindicate the actual return, it does underscore the unfortunate timing in the launch of both P3 and P4. In volatile market conditions you'd always prefer to have gains early in the sequence to build a cushion for the future. It's due to the inequality of percentages. While you might assume a percentage gain in one year might be offset by the same percentage loss the following year, it won't. Consider for example, a $100 initial investment with a 20% loss in the first year followed by a 20% gain the second. At the end of the second year you aren't back to where you started with $100, but instead have $96, a 4% loss. Something similar happened to P3 even though its gain in 2003 was markedly greater than any of its annual losses. Just imagine how happy you would be if you had waited until January 1, 2003 to invest your $100 in P3! Yes, timing is indeed everything. P4 fared much better. By April 30, 2004, it had inched ahead of large cap blend funds by 0.1% on a cumulative basis. Again, including expenses would change this, but so would adding in P4's dividends. Despite greater risk (standard deviation) and a rough first 2-1/2 years, it still provided the best return. Investing's a lot like golf in this regard. Your ball may have bounced off a tree or two and rolled along the cart path, but if you're on the green you don't have to draw a picture of how you got there. The number of strokes is all that counts. P4 may have had some ups and downs along the way, but it still ended up on top, and that's what you ultimately take to the bank. P3 and P4 don't look exactly like the benchmark index -- they aren't supposed to. They are designed to be a proxy for large cap stocks in asset allocation models. They are supposed to trade like and outperform the S&P 500 over time. Looking back over almost five years worth of data, they've lived up to expectations -- at least as well as "professionally managed" large cap blend funds. Sure, it would have been nice if P3 could have started with a 60+% gain rather than having three down years first, but even that will probably work itself out over time. P4 has already begun to prove itself with its cumulative return. This suggests it is possible to construct a model to approximate and exceed a benchmark without having to game it or become a closet indexer in the process. The quant models take the first approach. What about the managers of your mutual funds? Search this site! Just enter you key word or words:
Get current quotes or follow your own custom portfolio,
courtesy of E-Line Financials:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
VER SINCE THEIR INCEPTION BACK in the summer of 
