| |
 September 2005 Wrong Place, Wrong Time The Effects of Sector Weighting on Portfolio 3
To be sure, market timing and security selection do impact returns, but so does something that's not so frequently mentioned: Asset allocation. In fact, as we've pointed out before (most recently here), asset allocation is actually the major factor. A few months ago, we considered the effect of turnover -- essentially market timing -- on P3. We've also examined its quantitative selection process as well as some of its specific securities (see "Work in Progress" archives). Oddly enough, we've never really considered its asset allocation. Until now. With five years worth of data on P3, it's high time we looked at how it's been shaped by asset allocation. After all, if it's so important, asset allocation could be the reason P3 has tended to trail the S&P 500.
Bass AckwardsGenerally you think of asset allocation as the division of a portfolio between different asset classes such as stocks, bonds, and cash. Portfolio 3, however, is always 100% invested in large cap domestic equities. In fact, it's limited to the 500 stocks of the S&P 500 Index. In that light, it's not immediately obvious that asset allocation applies at all.
But it does. Although limited in capitalization and country of origin, P3 is not limited by sector. Standard and Poor's divides the stocks in their domestic indexes into the ten distinct sectors listed in Chart 1. At any given time, P3 can hold stocks in any or all sectors. Even so, P3 has tended to favor a few sectors over others. Chart 1 shows a clear preference for Healthcare, Technology, and Industrials. These are usually considered "growth" sectors. On the other hand, Utilities and Materials rarely appeared in the model. These are "value" sectors. That's unfortunate since value outperformed growth throughout P3's five-year lifespan. This is clearly illustrated by Chart 2 which summarizes results from the period July 1, 2000 through June 30, 2005. The red bars give P3's average sector weightings for each sector as measured on the left scale. Notice the minimal weights given to Utilities, Materials, and Energy. The dots on the green line are the cumulative returns for each sector as measured on the right scale. Again look at Utilities, Materials, and Energy -- three of the best performers. On the other hand, consider P3's two heaviest weightings, Technology and Healthcare. They turn out to be two of the three worst performers. Indeed, if you compare P3's sector weightings with cumulative performance, it's pretty obvious that the model overweighted the poorest performers while underweighting the best -- not a formula for success. The End, Not the MeansHow could this have happened? How could P3's sector-weighing process get it so backwards? Actually, it's somewhat misleading to blame P3's sector-weighting process since it doesn't have one.To see why, let's briefly review how P3 works. It's based on a regression analysis of annual return to ten fundamental factors such as P/E, Return on Invested Capital, and Debt/Capital. The top 30 stocks in the S&P 500 are selected for inclusion. They remain in the model until their rating falls below 100. P3 is adjusted six times a year following the close of trading on the 15th day of each even-numbered month. (For a more detailed explanation of P3's construction, please see The Starting Point.)
Sector weighting doesn't play into this, it's just a means of rating and selecting the top stocks across all sectors. In effect, the model's sector weighing is simply the result of the process, not part of it. The model picks stocks, not sectors. The bias toward growth stocks (and derivatively growth sectors) is also easily understandable. The data that went into P3's regression formula came from the decade of the 1990s -- a period in which growth dominated. It's no wonder the model favors growth. So P3's sector weighing is a result of its growth-biased stock picking. Even so, by comparing sector returns to that of the model, it's still possible to get a feel for how much this bias has impacted results. The Wrong Place vs.
If P3 was based on Exchange Traded Funds (ETFs) like Portfolios 5 and 6, it wouldn't be picking individual stocks. Instead it would be holding entire sectors and its return would be based on their weights. By creating portfolios with P3's sector weightings and actual sector returns, we can approximate the effect of the model's growth bias. A comparison of these returns to those of P3 will separate the effects of allocation and individual stocks.
|
|||||||||||||||||||||||||||||||||||||||||||||||
|
||||
Based on this, it would appear that roughly 75% of P3's decline was attributable to its sector allocation. With its emphasis on growth, the sector returns understandably trailed those of the overall index.
P3's individual equity selection didn't help -- in fact, given that it trails the sector returns, it evidently hurt performance. That's not too encouraging.
But it's also not the whole story, either. Without a doubt the five-year period was disappointing for P3, yet improvements may already be underway.
Early on we noted that a number of stocks entered P3, were dropped, and subsequently re-entered. Not only did this lead to excessive turnover, it brought into question the criteria for inclusion and deletion.
At that point, when a stock's ranking fell below 40 it was deleted from the model. That was probably too restrictive and the source of all the turnover. When we reviewed the models in July 2003, the ranking cutoff was raised to 100. In other words, once a stock was included in the portfolio, it would remain as long as it was in the top quintile of the S&P 500. That seemed reasonable.
As we noted in September 2003, turnover immediately decreased. A more recent examination showed a significant reduction in turnover since the 2003 change.
Performance also improved. This can easily be seen on Chart 4. Like Chart 3, it shows the cumulative performance of the S&P 500, P3, and a portfolio consisting of P3's weighted sectors. Instead of covering the entire five-year period, Chart 4 shows only the results from July 1, 2003 (the time of the retention ranking change) through June 30, 2005.
|
||||
What a difference! Throughout the past two years, P3 has led both the S&P 500 as well as the sector-weighted portfolio. Not only that, all three are well into positive territory.
Notice that the sector-weighted portfolio trailed the S&P 500 for most of the period. This is probably the result of value's continued dominance over growth. Regardless, P3 is ahead of the index, suggesting this is the result of timely stock picking. Apparently the model was able to achieve superior returns even when relying upon stocks from underperforming sectors. Maybe talk of the model's failure was a bit premature.
Of course there's another factor that should also be considered: The demise of the bear market. After two down years, 2003 and 2004 marked the return of the bull market. Although stocks remained well below their pre-bear levels, they posted strong gains in both years. Not only does this account for the positive returns of all three series, it might at least partially explain P3's turnabout.
Despite growth's continued lagging performance, P3's algorithm may work better in up markets than in down. This isn't uncommon for quantitative models. Nevertheless, P3's stock selection does appear to have helped it beat the sector returns. This may be the result of the 2003 retention tweak, allowing the model to benefit from the recovery of stocks that stumbled in the short-term -- something missing in the early version when stocks were quickly jettisoned after a hiccup.
Although five years can seem like a long time, it really isn't when testing a model. From an historical standpoint, the equity markets haven't really acted "normally" in the past 10-15 years, so it's hard to justify any conclusions drawn from the past five. This raises more questions than it answers.
First, although P3 has performed substantially better over the past two years, it will be interesting to see how it does should the markets again head downward. Will it be able to maintain its lead over the S&P 500, or will it again fall behind?
In the process, it's also worth watching its performance relative to the sector-weighted portfolio. Can it hold its lead or will it again fall behind -- especially if the market goes south?
Finally, and probably most importantly, there's the question of the sector weighting itself. As it presently stands, P3 makes not effort to sector weight or diversify its holdings. In a narrow market like the late 1990s when the top 50 stocks of the S&P 500 accounted for all of the index's returns, this can be a blessing. But under more typical circumstances, market leaders come from a wider array of options as well as styles. Is P3 destined to only mine the growth sectors and if so, will this ultimately restrict its long-term performance? From the looks of Chart 3, it sure has so far.
The original intent was to test the models and keep meddling to a minimum. Changes are only made at the end of three-year periods, the last being in July 2003. The next opportunity won't come until July 2006 and there's no guarantee anything will be done then.
By the time 2006 rolls around, there'll be another year of data to analyze. Perhaps some of the questions above will be answered, but maybe they won't. The minor change in 2003 seems to have enhanced performance, so maybe an equally minor change regarding sector diversification would give it a boost, too. That extra year's worth of data might help -- it certainly won't hurt.
Search this site! Just enter you key word or words:
Get current quotes or follow your own custom portfolio,
courtesy of E-Line Financials:
|
|||||||||||
| Homepage | Return to Top |
NVESTMENT RETURN COMES FROM a number of different sources. If you heed the financial press, you're well aware of impact of market timing ("Why It's Time to Get Out of the Market") and security selection ("Ten Stocks You Need to Own Now"). Perhaps you're even too aware.
