Progress of the secular bear market: position as of October
The value for R is 1390 as of October 2006. For S&P500 of about 1360 this gives P/R of 0.98
Stock Cycles: Part V. Plotting a course for a secular bear market
In this final article in the series, I use historical patterns in P/R to sketch out a rough path for the S&P500 index during this secular bear market. What I present here is based on chapter 7 of my book Retiring Rich, which was written in early 2003. I leave the presentation as it was given then to give an idea of how the ideas have panned out so far.
It is impossible to predict the market's precise future behavior. What one can do is use the general properties of secular bear markets to identify approximate boundaries within which the market will likely wander. This idea is then combined with the general tendency of the market to move in trends, which suggests that lengthy bear market periods during which the general trend is down will be followed by bull market periods when the trend is up. Approximately three to five such oscillations will make up the complete secular bear market era.
So far, we are in the third year of the first ordinary bear market in the post-2000 secular bear market. At some point this bear market will end and a bull market will begin. This bull market will in time also end and be followed by a second bear market. Several of these bull/bear market cycles will be completed until a final bear market bottom sometime around 2018 will end the secular bear market era.
Figure 1 Bull/bear market oscillations in P/R during past secular bear markets
Figure 1 shows the progress of previous secular bear markets in terms of maximum and minimum P/R values during the bull/bear market cycles embedded within the secular trend. P/R declines during the entire secular bear market era. It begins at the highest level and ends at (close to) the lowest level of the entire secular bear market era. With one exception, the bull/bear cycle lows early in the secular bear market era are higher than are those later in the era. The same is true of the bull/bear cycle highs. With one exception, P/R at the end of the secular bear market is quite low (around 0.3), especially when compared to the level at the beginning of the secular bear market.
One can combine the data in Figure 1 with a projected trend for R to obtain an estimate of the approximate ranges for the tops and bottoms of the bull/bear market cycles to come during the present secular bear market. Figure 2 shows a plot of the S&P500 from 1997 through the end of 2002. Also shown are projected bull market highs and bear market lows based on those seen in three of the four most recent secular bear market eras. The Great Depression secular bear market was not used. These projections were obtained by extrapolating the 4.5% growth rate for R over the past three years (2000-2003) into the future, and then applying the high and low values for P/R from Figure 1 to obtain lower and upper bounds for bull/bear cycle highs and lows.
The first set of limits were those used to time the bull market peak in the late 1990's. As I described earlier, I sold out in December 1998 through September 1999 as P/R reached and exceeded previous all-time highs. I had waited for all-time highs because of the stimulatory effect I expected from the 1997 capital gains reduction. In actuality, stocks went somewhat higher than the projections. Since peaking in 2000, stocks entered a bear market and (so far).
Figure 2. Projected bull/bear market extremes for the near future.
have fallen to levels consistent with the initial bear market of three of the four previous secular bear markets. This was discussed in Part III of this series. At some point this bear market will end and a new bull market begin. Based on the P/R levels reached in the second top from these same three secular bear markets (see Figure 1) a window for the top of the coming bull market can be projected. One could use these levels to begin a selling program similar to that which I executed in 1999.
Figure 3. Projected bull/bear market extremes for the entire secular bear
Figure 3 extends the analysis of Figure 2 for the rest of the secular bear market. The same trend for R was extended to 2021, and the product of projected R and P/R used to obtain P values. For each ordinary bull/bear market cycle the projected "high" and "low" regions are identified. A hypothetical market path is then drawn through these extremes. The timing of the peaks and troughs made use of the four-year cycle1 in the stock market which holds that bear market bottoms then to occur in the fall of non-presidential election years.
Use of projected paths to aid asset allocation
Note: A four year strategy that involves selling at the end of presidential election years and buying in early fall for non-presidential election years was discussed in Retiring Rich and is referenced below.
Based on Figure 2, it was expected that purchasing the S&P500 within the "low" guidelines (below ~940 on the S&P500 for early 2003) would have provided an adequate return going forward. Eventually a new bull market will get underway and the S&P500 will move well above the "low" region in Figure 2. As the market moves into the "high" region in Figure 2, purchases of stocks are no longer a good idea, although selling existing stocks is not necessarily warranted. This has indeed happened.
Since one cannot know when a minor decline is really the start of something bigger, decisions to sell cannot easily be made after stocks have started to decline. For example, one could have sold stocks in October 1998, based on the fact that the S&P500 index had already reached the "high" level and had started a serious decline (see Figure 2). This would have been a mistake. Selling in 1999 (as I did) or in December 2000 in accordance with the four-year strategy would have been a better decision.
One selling strategy would be to reduce stock allocation in December 2004, when the four year cycle suggests one do so. Selling allows one to take advantage of future buying opportunities. The higher the index moves in 2004, the greater the likelihood that the buying opportunity in 2006 will be good. Selling in accordance with the four-year rule would allow one to book paper capital gains at the cost of missing out on further rises. I would have done well had I refrained from re-entering the market until fall 2002 as called for by the four-year rule. Instead, being ignorant of the four year rule until fall 2002, I had bought the declining market on a scale and so held the S&P500 index at an average value considerably higher than the fall 2002 value.
Not only that, but I did not plan to sell in December 2004. I believed the market had fallen so far in 2002 and was coming back so slowly, that by late 2004 it would simply not be very overvalued and so not likely to dip much subsequently. So I figured I would hold my stocks through the 2006 bear cycle--unless the market got "high enough".
After all, there was no certainty that the four-year rule will "work" for each cycle (history suggests it fails 30% of the time). What the historical evidence suggests is if pursued as an overall strategy, it will work often enough to give a better return than buy-and-hold. It was entirely possible that the S&P500 could be higher in October 2006 (when the four-year method would signal a buy) than it was in Dec 2004 (when the method signals a sell).
Another selling strategy would be to sell if a likely superior alternate investment is available. Assuming that Figure 2 gave an approximate picture of what to expect, one can conclude that levels above 1400 on the S&P500 are likely to occur sometime in this decade. Assuming the 1400 level is reached by Dec 2008 (the next four-year cycle sell point after 2004), one can calculate an estimated return based on the current date and level of the S&P500. For example, consider the situation in which the S&P500 were to reach 1200 in late 2004. This level is within the "high" levels in Figure 2 and late 2004 is close to a "sell" point called for by the four-year rule. Assuming a level of 1400 should be achievable within four years and adding 2% dividends implies a likely return of 6% could be obtained from this level. Unless bond yields are well above 6% in 2004, one would likely be better off in stocks and one would not sell. Should stocks rise further, say to 1300 by late 2004, the anticipated return now falls to a level at which bond yields as low 5% can compete. Obviously, if an extreme level like 1400 were to be reached in 2004, shifting out of stocks would be recommended almost regardless of bond yields.
So I set a rule that I would start moving our of stocks in my 401(k) were the S&P500 to approach the 1400 level before the projected fall 2006 bottom. Since this did not happen I did not reduce by stock allocation, and continue to hold the same 80% allocation reached in summer 2002.2
1. Michael Alexander, "Generations and Business Cycles - Part I", Safehaven, November 6, 2002
2. Michael Alexander. "Progress Update on the Current Secular Bear Market", Safehaven, July 23, 2002.