Beware of Simplified Models
(August 27, 2006)
Dear Subscribers and Readers,
Our 50% long position in our DJIA Timing System that we initiated on the afternoon of July 18th (at a DJIA print of 10,770) was exited on the morning of August 10th at a DJIA print of 11,060 – giving us a gain of 290 points. A real-time email was sent to our subscribers announcing this shift – and the justification for this shift was discussed in our August 10th commentary (“Is Our Short-Term Scenario Busted?”). In retrospect, this call was definitely too early or wrong, but at that time, this author was convinced that the market was making a turn for the worst (see our August 10th commentary for further clarification). At this time, this author is still long-term bullish on the U.S. domestic, “brand name” large caps – names such as Wal-Mart, Home Depot, Microsoft, eBay, Intel (which, as I have discussed over the last few months, will regain a significant chunk of microprocessor market share from AMD), GE, American Express, Sysco (“Sysco – A Beneficiary of Lower Inflation”), etc. The market action in these large caps has also been very favorable thus far.
In the meantime, however, we are still sitting on the sidelines waiting for signs of a more favorable market environment. Since volume is going to continue to dissipate this upcoming week (it always does during the week before Labor Day Weekend), we will most probably not find out until sometime after the Labor Day Weekend. For now, this author is still sitting on the sidelines, but am cautiously bullish. However, should the market continue to correct on weak downside breadth and low volume in the coming days, don't be surprised if this author chooses to shift back to a 50% long position in our DJIA Timing System.
This upcoming week is promising to be a very slow week. In light of the strengthening of Hurricane Ernesto last Friday, this author had wanted to devote a few paragraphs (mostly as a refresher since we had already discussed this in our June 11th commentary: “Oil Shortages this Hurricane Season - Last Thing to Worry About”), but given that Ernesto has already been downgraded and is now projected to veer away from the Gulf of Mexico, this author does not really see the point. But let us now remind our readers our position on the 2006 Hurricane Season: Given the current NOAA forecasts (which have actually been weakening since the initial forecasts back in May) of hurricane activity this year, there is virtually no chance for a Katrina or Rita-style natural gas/crude oil supply interruption this year. The mainstream media has always liked to sell newspapers or attract more traffic to their websites, and playing on the public's fears is one tool they would like to use. Quoting from the EIA: “Although Hurricanes Katrina and Rita caused long-lasting effects on oil and natural gas operations in the Gulf of Mexico, severe weather in the region has historically had only a relatively minor impact.” Moreover, this “lack of interruption” was not accompanied by lack of hurricane activity in the Gulf either, as evident by the following chart (courtesy of the EIA) showing the major hurricanes in the Gulf of Mexico from 1995 to 2005:
Source: National Oceanic and Atmospheric Administration (NOAA) Coastal Services Center
Given that many oil and natural gas drilling rigs have “beefed up” their defenses in light of Hurricanes Katrina and Rita, there is a strong likelihood that the chance of another significant supply interruption in the Gulf has further diminished going forward. In other words – while anything is possible – this author would definitely not bet on a further rise in either the price of crude oil or natural gas due to a supply interruption in the Gulf of Mexico this year.
In our last weekend commentary, we also discussed the continuing evolution of the global economy (with a brief mention on ASEAN's fast-tracking plans to create an EU-style single market by 2015) – with an explicit discussion of the virtual online community of “Second Life.” We also discussed why the structural story of both lower corporate and income tax rates around the world will generally remain intact. Nevertheless, there is one continuing trend we did not mention (although we have been discussing this on-and-off in both our commentaries and in our discussion forum) – and that is the structural story of deflation will continue to persist – at least in the consumer goods world anyway. To illustrate briefly, read no further than the huge increases of capital spending in Japan over the last two years. Combined with an aging domestic economy and a persistence in the lack of consumer spending (not to mention a weak Yen), and there is no doubt that Japan will continue to export deflation for the rest of this decade. Or consider Vietnam – a country with 84 million whose workers (with an average wage approximately half the amount of China's) are rapidly displacing Chinese manufacturers and works at the margin. These two “deflation stories” are what the mainstream media have been missing. Moreover, since the end of this year, China has again been exporting consumer good deflation (after exporting inflation for most of 2004) – something the media still has not caught on yet at this point.
Let us now get to the gist of our commentary. In our last mid-week commentary, we strongly cautioned against looking into too much the popular “forward-looking” market indicators out there, such as the NAHB Index (this author has shown that the movement of the NAHB index has really no predictive value), the percentage of auto sales increases over the last 12 months, etc. In this commentary, we would like to quickly go over the concept of trending in the stock market, including the concept of “reversion to the mean.”
On examining the performance of the stock market since the beginning of the 20th century, any stock market historian should be able to tell you this: Since that time, the U.S. stock market has been steadily trending on an upward path – with some occasional significant overshooting (such as the late 1990s boom) and undershooting (such as the 1973 to 1974 crash). A typical chart showing the history of the Dow Industrials from the beginning of the 20th century to the present (and we use the Dow Industrials for our purpose since it has such a long and rich history vs. the S&P 500) along with its trend may look like the following:
Please note that the above is a monthly chart (showing month-end closing prices) and the y-axis is in logarithmic scale – which is the way that it should be shown. The black line shows the trend – which is essentially the “line of best fit” according to standard regression rules.
So Henry, what are you trying to say or “disprove” here? Isn't the above chart showing that despite the crash of 2000 to 2002, the Dow Industrials is still above the “trend” and should therefore revert to the mean at some point (the uptrend line is sitting at approximately 6,000 right now)?
Yes, that is what the above chart shows. Moreover, the tendency for something to revert to the mean has been observed in the financial markets for centuries and is one of the most reliable ways to make money (as opposed to virtually all other technical indicators or even momentum investing). However, here is the rub: The above results can and will change depending on your starting point. Instead of taking January 1901 as the starting point, how about if, we instead take June 1932 (bottom of the Great Depression) as the starting point? Following is the monthly chart showing the Dow Industrials from June 1932 to July 2006 along with its trend line:
While the current level of the Dow Industrials is still over the trend line, readers can now see that it is not as far above trend compared to the previous chart. While the trend line in the previous chart was sitting at approximately the 6,000 level, the trend line in the above chart is more than 30% above the previous trend line – at approximately 8,000.
Obviously, the Dow Industrials still look very much overvalued in the above chart – and based on this and the concept of “reversion to the mean,” this author would no doubt be very bearish now. However, what if we change the starting point to October 1974 – right at the bottom of the last secular bear market? Again, following is the monthly chart of the Dow Industrials from October 1974 to July 2006 along with its trend line:
Suddenly, the Dow Industrials is below its uptrend line stretching from October 1974. In fact, it has been below its uptrend line since early 2002 and is “on par” with oversold levels that were reached during the October 1990 and December 1994 bottoms. Based on the above chart, one should now be extremely bullish.
So are we thoroughly confused yet? What can we conclude from the above? Based on the above “analysis,” readers should be very wary of simple or “simplified indicators” that attempt to rationalize where the market is heading by painting an overly simple picture. When it comes to communicating to retail investors, the mantra has always been to “keep it simple,” but in the stock market, this is a dangerous practice – given that the stock market is impacted by millions of (both independent and dependent) variables on any given day, not to mention any given minute in every trading day. To believe that one can find a “simple, but foolproof” indicator is simply hubris in the highest form.
Interestingly, the period from the end of 1996 to the end of 2001 is above the trend line in all of the above three charts, but investors who got out of the markets at the end of 1996 based on the above charts most likely missed one of the greatest stock market booms in U.S. history. As an aside, however, the investor who relied on the above charts most probably got out of the market much sooner than the end of 1996, given that the subsequent upward action in 1996 also caused the trend line to be much steeper than it was at the time. To illustrate, this author in the following chart has removed the monthly closes of the Dow Industrials from the above chart from January 1997 and beyond:
I now rest my case. Based on the above chart, the investor who based his market-timing moves on a typical trend analysis would have gotten out by the end of 1995, if not sooner.
Another example of an overly simple study is one that has been presented by Dr. Hussman at the Hussman Funds. Note that I have the highest respects for Dr. Hussman. Moreover, his trading record speaks for itself – in that his fund has overperformed the S&P 500 by a wide margin over the last three and five years. In fact, the Hussman Strategic Growth Fund is ranked in the top 1% of all equity mutual funds by Value Line over the last three years. In his March 20, 2006 commentary, however, he presents a chart showing the growth in earnings of the S&P 500 from 1950 to today – arguing that the current earnings of the S&P 500 is now hitting the upward trending resistance line (which has historically grown at 6%) and in all likelihood, should head down in the near future.
First of all, it is interesting to note that one can always make a bearish case out of everything. In this case, Dr. Hussman is bearish simply because the S&P 500 earnings have accelerated too quickly since the 2002 bottom. His thesis is that earnings are now topping out, and if earnings top out, then the S&P 500 should decline relatively soon as well.
What Dr. Hussman ignores is the fact that the S&P 500 does not necessarily move along with earnings. In fact, as this author mentioned last week, we have been witnessing P/E contractions in the S&P for three years in a row (assuming nothing changes for the rest of this year). Since the beginning of the 20th century, we have only witnessed four consecutive years of P/E contractions twice – one from 1934 to 1937 (during the Great Depression) and the other from 1976 to 1979 (during the near hyperinflation we experienced in the late 1970s). That means the stock market could rise even as earnings decline (and an earnings decline may not necessarily occur either) should we happen to break the trend in declining P/Es later this year or next year. At the very worst, going forward, this author would expect more of the trendless markets we have witnessed over the last couple of years.
For further illustration, readers can go to Dr. Hussman's March 20th commentary and see the chart for yourself. Note that the stock market did well off the bottom in 1990 even as earnings continued to decline. On the other hand, the stock market crashed anyway even as earnings continued to power ahead during 1973 and 1974. Interestingly, John Mauldin has also mentioned the bearish implications of this chart, even though Mr. Mauldin in the past has explicitly claimed that there were no significant correlation of the earnings growth of the S&P 500 and its subsequent performance (note that this analysis does not apply to individual stocks). In fact, Victor Niederhoffer, in his book “Practical Speculation” actually claims that there is historically a negative correlation between earnings growth of the S&P 500 and the subsequent 12-month performance of the S&P 500.
One more fault with Dr. Hussman's chart: It does not break down earnings growth by industry. For example, what if the price of crude oil drops back to the $50 level? Surely, the oil component of the S&P 500 would not be making as much money as it has over the last 12 months. It may even cause the overall earnings of the S&P 500 to decline, thus validating Dr. Hussman's prediction. However, for other investors that are holding stocks other than oil stocks (interestingly, oil shares have taken a beating over the last few months even as crude oil prices stayed above $70 a barrel), such a decline in S&P 500 earnings may actually turn out to be a benefit – especially for retailers and other industries that are very energy dependent (such as the food and chemical industries). In other words, “nothing is obvious” – so going forward, please keep in mind that most analyses usually involves further study (should any of you have interesting research ideas, please send them to this author's way). The market is always full of deception and intrigue. Beware – and always strive to independently come up with your own conclusions when investing!
In last weekend's commentary, this author also discussed the importance of the Homebuilders ETF, the Dow Jones Utility Average, as well as the NASDAQ Daily High-Low Differential Ratio. Since a week ago, the action of these three indicators have been hopeful – with the homebuilders ETF holding above support levels (despite the warning from Toll Brothers) and the Dow Utilities closing at a record high on Tuesday of last week. By definition, a new record high in the Dow Utilities extends the bull market just a little further. As for the NASDAQ Daily High-Low Differential Ratio, it is still vacillating along near the zero line, so it is nowhere near conclusive just yet. I will continue to update the developments of these three indicators for our readers both in our discussion forum and in our mid-week commentary this week.
Let us now discuss the Dow Theory. Following is the most recent action of the Dow Industrials vs. the Dow Transports, as shown by the following chart from July 1, 2003 to the present:
For the week ending August 25th, the Dow Industrials and the Dow Transports both declined – the former declining 97 points while the latter 147 points. Given that this decline was accompanied by weak breadth and relatively low volume, there is no reason to believe that this is any further than a correction of the rally during the week before last. Any upcoming follow-through will most likely not occur until after Labor Day Weekend – as a strong rally cannot be sustained unless we have strong volume. Readers please stay tuned.
I will now end this commentary with a quick discussion of our popular sentiment indicators – those being the bulls-bears percentages of the American Association of Individual Investors (AAII), the Investors Intelligence, and the Market Vane's Bullish Consensus Surveys. The latest four-week moving average of these sentiment indicators bounced from an extremely oversold level of 1.7% in late June to 8.6% in the latest week. Following is a weekly chart showing the four-week moving average of the Market Vane, AAII, and the Investors Intelligence Survey Bulls-Bears% Differentials from January 1997 to the present week:
From a probability standpoint, there is a good chance that this survey has now bottomed and reversed – suggesting higher prices over the near future. However, as I have mentioned before, we still remain cautiously bullish, and will not buy these markets in an aggressive way until we have received some confirmation from the action after Labor Day Weekend.
Conclusion: As this author has said so many times in both our commentaries and in our discussion forum, (poster rffrydr has always emphasized this as well): “Nothing is obvious.” And certainly when a simple indicator or study is presented in a simple way (such as the NAHB Index as a leading indicator of the stock market or the 12-month growth in auto sales as a leading indicator of a recession), readers should sit up and start asking questions, such as: Have the facts changed since the last recession? While the indicator may have always worked (incidentally, the NAHB index has not worked all the time, as this author had already discussed in our mid-week commentary), will it and should it continue to work going forward? In terms of the 12-month growth in auto sales, how did last year's huge incentives from the Big Three (or 2 ½ as other folks would like to call it) play a role in jacking up auto sales last year? In such a scenario, should we not experience a sizable dip in auto sales this year – even if we ultimately do not experience a recession?
As for the short-run, last week's decline in the major market indices was most likely just a slight correction of the huge rally during the week before last – especially since it was accompanied by relatively low volume and weak downside breadth. For now, this author is still choosing to sit on the sidelines and will continue to observe the action going into and immediately after Labor Day Weekend. But should the market corrects on relatively weak downside breadth and low volume in the coming days, don't be surprised if this author chooses to shift back to a 50% long position in our DJIA Timing System. Again, readers please stay tuned.
Henry To, CFA