Charting the Course of the Secular Bear Market

By: Michael A. Alexander | Mon, Feb 21, 2005
Print Email

Four years ago, I introduced the Stock Cycle and the price to resources (P/R) valuation measure I use to track the cycle.[1] Figure 1 shows a plot of P/R defining the previous Stock Cycles.

Figure 1. The Stock Cycle in the American market over time

The Stock Cycle can also be identified using other valuation measures such as Tobin's Q and price divided by 10-year average earnings, a measure I call Shiller's P/E because it was described by economist Robert Shiller in his book Irrational Exuberance. The most basic idea behind using the Stock Cycle as an investment tool is to detect when the secular trends that make the bull and bear halves of the cycle are close to their ends. The most recent turning point was the shift from a secular bull market to a secular bear market in 2000. My book Stock Cycles was written in early 2000 to forecast this turning point explicitly using the P/R concept.[2] This forecast was based on P/R reaching an all-time high in January 1999, indicating that the secular bull market had reached levels at which all previous secular bear markets had ended. This same argument could be made (and was) using Shiller's P/E.[3] In the case of P/E, the forecast would be based on P/E reaching an all-time high in January 1997, two years before the P/R signal. As I discussed in a 2002 article, the P/R-based forecast was more useful than the P/E-based forecast:

Stocks have outperformed money markets for the entire period since January 1997 (as of June 2002). Thus, exiting the market in early 1997 in accordance with Shiller's P/E-based warning would not (so far) have given a prospective market timer an edge. In contrast, money markets have strongly out-performed stocks since Jan 1999. Thus, exiting the market in early 1999 in accordance with P/R has already provided a market-timer employing P/R an edge. From these observations, one might conclude that for the present market, P/R has given a better indication of overvaluation than Shiller's P/E.[4]

One of the advantages of P/R is its simple construction, which allows simple analyses to be performed that can give insight in market dynamics. In a 2001 article I described why the signal given by Shiller's P/E did not work:

A given amount of R will produce about 40% less earnings it did a century ago. This implies that for equal peak market valuations (as measured by P/R), peak P/E values of today should be about 50% higher than those of the past. Thus, whereas in the 19th and early 20th century a P/E of 20 would be considered very high, today a P/E of 30 would be considered as equivalently high. This is the reason why P/R gave a closer prediction of the timing of the bull market peak than did 10 year P/E. [5]

What this says is historical P/R is still a relevant valuation guide today, while P/E is not. This is important because current P/E values of about 20 are still high compared to the long-term average of 14, whereas P/R is not high with respect to its historical levels at a corresponding point in the Stock Cycle. Not only that, but P/E today is not high relative to recent market experience. Today's P/E is the lowest in seven years, and stands at the 23rd percentile of all P/E's over the past decade. If things are different today, as the P/R-based analysis suggests, then Shiller's P/E will not provide a useful assessment of valuation. I explicitly tested this idea in a post made to the Longwaves discussion group in August 2002 [6, 7]:

The current market provides a good test of the three valuation methods. Suppose the July 23 (2002) low ends up being THE bottom? This would invalidate Shiller's P/E and probably also Tobin's Q as useful valuation tools. Only P/R would have given a valid reading of valuation. On the other hand, suppose the S&P500 drops to an ultimate bottom below 550 without a deflationary depression? In this case, P/R valuation would be invalidated. Such a low level of P/R this early in a secular bear market has only occurred once, during the early 1930's. It is inconceivable that it happen again without an accompanying depression. Shiller's P/E says such a development is to be expected, no depression is needed.

The test was simple, if P/E valuation still holds, the SPX would likely fall below 550 before the 2000-2002 bear market came to an end. Investments made at the summer 2002 lows would be bad investments and one should sell in August 2002. If P/R valuation holds then the 2000-2002 bear market could have already ended in July 2002 and investments made at the summer 2002 low would be good investments. One could even buy in August 2002.

As it turned out, the SPX fell to a marginal new low in early October 2002 that was 2.5% below the July 23 low; retested the July low in March 2003; and rose strongly after that. So the signal given by P/R in 2002 was more accurate than that given by P/E, just as P/R had been more accurate in forecasting the end of the secular bull market. With these two results, it is reasonable to conclude that P/R does work. This article describes the use of P/R and other shorter cycles [8] to draw a picture for a likely course of the secular bear market.

Figure 2 shows a plot of P/R for the current and four previous secular bear markets. P/R declines at an average rate of 6-7% per year. For the average secular bear market length of 18 years this comes to a 70% decline in P/R over the entire secular bear market. This downwards trend is countered by the rising trend in R, which over the long term averages about 5%. Thus, R grows some 2.4-fold over the average secular bear market. Combining the 70% decline in P/R with the 140% increase in R yields a ~30% decline in P for the "typical" secular bear market. This decline is roughly equal the average decline in an ordinary bear market.

Figure 2. P/R during previous secular bear markets

A secular bear market begins at an ordinary bull market peak and ends at an ordinary bear market bottom. Thus, the "size" of the decline over a secular bear market consists of one ordinary bear market plus the change in price due to the secular bear market trend. Since the projected decline in price for the entire secular bear market is about the same as that for an ordinary bear market, this means that the net trend in a secular bear market is zero.

Table 1 shows the overall decline for the seven previous secular bear markets. Also shown are the declines for the first bear market of each secular bear market. Excluding the 1929-32 disaster, the average decline in the first bear market has been about the same as the decline for the entire secular bear market. Thus, the most likely location of the bottom at the end of the secular bear market in 2018 should be in the low 800's. This is consistent with the theoretical idea that the trend during a secular bear market is zero.

Given these observations, a secular bear market should be thought of as an interruption in the long-term rising trend of the stock market and not as some kind of dramatic decline. What this means is the dominant structure in a secular bear market will be the ordinary bull and bear markets. The secular bear market itself, being trendless, doesn't exert a significant independent effect of its own. In this way secular bear markets are different from secular bull markets. In secular bull markets the dominant structure is the rising trend. The effects of even sizable bear markets like 1987 are quickly overcome by the rising trend. Thus, one can ignore all the other cycles and simply ride the secular trend.

Table 1. Size of previous secular bear markets

Period Initial level1 End level1 Decline First bear1 Decline
1802-1815 3.40 2.45 28% 2.78 18%
1835-1843 3.53 1.81 49% 2.33 34%
1853-1861 3.41 1.58 54% 1.27 63%
1881-1896 6.58 3.81 42% 4.24 36%
1906-1921 10.03 6.88 31% 6.25 38%
1929-1949 31.3 13.55 57% 4.4 86%
1966-2000 94.03 109.4 -16% 77.1 18%
2000- 1485 800's? ? 838 44%
Average -- -- 43% -- 41%2
1Values are monthly averages of the S&P500 index and precursors
2The 1929-32 bear market was left out of the average

Because there is no trend during a secular bear market, there is no special way to exploit it. Instead, an appropriate investment strategy for a secular bear market is to exploit the ordinary bull and bear market cycles. The average size of an ordinary bear market is a 30% decline. On average, they last about 17 months and are spaced about four years apart. This means that during secular bear periods, the market spends an average of 17 months falling into a 30% hole, and 31 months climbing back out, roughly every four years. Ideally, if one were invested only during the recovery periods (the ordinary bull markets) and not during the bear market portions, one could obtain an average capital gains return of 9% from stock investments during a secular bear market. Since typical capital gains returns during a secular bull market are larger than this, about 12%, attempting to play the ordinary bull and bear market cycles during a secular bull market is generally not worthwhile; the long-term trend provides a more than adequate return. During a secular bear market the long-term trend is essentially zero and so capturing even a portion of this 9% annualized capital gain may be worthwhile.

To explore the idea of capturing some of this 9% annual gain we must further develop the topic of shorter-term stock market cycles, specifically the ordinary bull/bear market cycle. Since the 1930's the bull/bear cycle has been nominally about four years long [8]. Furthermore, they have been correlated with election cycles. Fourteen of the 21 bear market troughs since 1933 (see Table 2 in reference 9) fall in non-Presidential election years. The probability of this arising from chance is 0.001%, providing strong evidence for an election-linked four year cycle in the stock market. The correlation between election and business cycles is thought to reflect government economic management designed to help re-elect the party in power.

A possible timing strategy that exploits this cycle would be to buy stocks in the fall of non-presidential election years, and sell at the end of presidential election years. To test this idea I calculated total returns for index purchases made at the average index closing price in September of non-presidential election years and sales made at the average index close in December of presidential election years.

Before 1933 the strategy was distinctly inferior to merely holding the stock index continuously. But this is not surprising since the four-year cycle only emerged after 1933. [8] For the period starting in Dec 1936 and ending in September 2002, a fully invested stance would have returned 10.0% annually, whereas use of the four-year cycle would have given slightly better results at 10.8%. It is interesting to see how these two strategies fared during the secular bull and bear periods (Table 2).

Table 2 Results of different timing strategies during secular bull and bear markets

Period Always invested Invested during
Sec bulls only
4-year cycle 4-year cycle
sec bears only
1937-1949 3.2% 0.27% 4.4% 4.4%
1949-1966 16.7% 16.7% 11.0% 13.8%
1966-1982 4.6% 6.9% 14.5% 13.2%
1982-2000 18.6% 18.6% 13.9% 18.0%
2000-2002 -17.8% 3.4% -3.1% -3.1%
Overall 10.0% 10.9% 10.8% 12.3%

The four-year strategy handily beat the always-invested strategy during the secular bear markets, most noticeably in the current secular bear market. But it lost relative to the always-invested strategy during the secular bull markets, giving up most of the relative gains made during the bear period. This observation suggests, why not apply the four year timing cycle only during the secular bear market periods? This approach gives superior results during the secular bear markets, yet preserves most of the gains during secular bull markets.

The way it would work is that when P/R signaled that the end of the secular bull market was near in January 1999, one would shift to the 4-year rule. The four-year rule would have one invested in stocks in early 1999, with a sell to come in December 2000. So one should stay fully invested in the market even after P/R reached record levels until December 2000. One would move out of stocks then and only re-enter the market in September 2002.

December 2000 was not a better time to exit the stock market than was September 1999, when I made my final exit from stocks in my 401(k). The level of the market was about the same, and I would have given up the gains to be had in the income fund. But re-entering in September 2002 would have been an excellent re-entry point. Unfortunately, I was ignorant of this four-year cycle until fall 2002, and had already increased my 401(k) stock allocation to 80% on a gradual scale by that time [10, 11, 12, 13].

Over the long haul, use of this four-year method provides a return 2.3% greater than that obtained with the fully-invested strategy. Since the entire benefit of this strategy accrues during the secular bear markets, which comprise only half of the time, this 2.3% advantage amounts to a 4.6% advantage relative to buy-and-hold over the entire secular bear market. This 4.6% advantage relative to buy-and-hold comes from capital gains as dividend returns would be earned by buy-and-hold investments. Because average capital gain returns over secular bear markets are ~0%, use of the four-year cycle provides about a 4-5% capital gain return during the secular bear market. Thus, use of the four-year rule can allow one to capture about half of the 9% potential capital gain obtainable from fully-exploiting the bull/bear market cycle.

Figure 3. Development of a theoretical model for secular bear markets

The above example shows that the idea of four-year cycles in the stock market is a real useable thing. We can combine the concept of a four year cycle related to the election cycle with the idea of an 18-year secular trend to produce a simple model for the structure of a secular bear market. We have seen that a secular bear market is a period of falling stock valuations that can be characterized by a downwards trend in P/R. This trend in P/R was shown as the heavy black line in Figure 2. It is reproduced in Figure 3. We have also seen that electoral politics has produced a four year cycle that typically bottoms in the fall of non-presidential election years and peaks around the end of presidential election years. This cycle is shown in Figure 3 as the dotted line. Combining these two trends together gives a path for P/R for a "standard" secular bear market. This path is denoted by the thin black line in Figure 3.

As described in Stock Cycles, the growth rate in R is equal to the long-term rise in the stock index in real terms or about 2%. It is also roughly equal to the long-term growth rate in GDP per capita. If we add the average 3% inflation seen in the United States over the long run, we can say that the "standard" trend in R is a 5% per year rise. Combining the falling trend in P/R in Figure 3 with a 5% rising trend in R, gives a hypothetical value for P, the stock market index value over the course of the secular bear market. This hypothetical value is shown as the bold gray line in Figure 4. Also shown in Figure 4 is the actual path of the S&P500 since 2000.

Figure 4. The theoretical model applied to the current secular bear market

So far it appears that the present bear market is roughly conforming to "typical" secular bear market behavior as described by the model. Figure 5 shows the application of the model to the last secular bear market. The model does a fairly good job of describing the cycles in that secular bear market. There are some discrepancies that are interesting to consider. The first is that the market was unusually low at the end of 1964 and the first bear market in the secular bear market was quite shallow. The 1949-66 secular bull market ended at the lowest value of P/R ever. It was so low the end of the secular bull market would not have been detected prior to 1966; one would have remained fully invested until Dec 1968.

Part of the reason for the weak start of the secular bear market in January 1966 was the fact that it was not accompanied by an economic recession. The expansion continued on for three more years, yet stocks (particularly the Dow) made no sustainable progress in real, inflation-adjusted terms after January 1966 (which is why the secular bull market is considered to have ended in 1966). The first significant bear market was the one that followed the December 1968 stock peak and which was associated with the recession of 1970. Thus, the pattern in the last secular bear market was a weak correlation with the model for the first four-year cycle followed by a strong correlation with the second four-year cycle.

Unlike last time, the current secular bear market began with a powerful bear market that was associated with a recession. The recovery from this recession and bear market was delayed and the S&P500 did not approach anywhere close to its old highs in 2004 as called for by the model. It is likely that the second four year cycle of this secular bear market will not fit the actual market behavior very well just as the first four year cycle did not fit well for the last secular bear market. The reason for this is yet another cycle, the Juglar cycle, which is a cycle in fixed investment. The Juglar is not as regular as the other two cycles employed in the model and so one cannot build a useful model around it.

Figure 5. The 1965-1983 secular bear period versus the theoretical model

But it does matter at times. This cycle is 8-10 years long in the modern economy and is manifest by the long expansions the 1960's, 1980's and 1990's. The reason the 1966 bear market was so mild was that the Jugular cycle was entering its up phase driving the expansion onward. As a result there was no recession in 1966 and the market quickly recovered from its 1966 decline. Similarly, the reason why the 2002-2004 stock market recovery was so weak is that the Juglar cycle was still in its down phase, putting a damper on economic recovery from the recession. Given a weak recovery from the 2002 lows, I would expect a weak decline from 2004-2005 highs. This is further underlined by the gathering strength in business investment to be expected after 2004 when the up phase of the Juglar cycle kicks in.

This assumes no surprises. The Juglar that began in 1970 was cut short to one four-year cycle by the surprise Arab oil embargo in 1973. As a result, a very serious bear market closely followed the sizable 1970 decline. Rising oil prices could threaten the recovery, cutting short the Juglar cycle as happened in the early 1970's. I don't believe this is likely because high prices reflect strong demand rather than supply shock and so are really an indication of robust economic performance. Nevertheless, high oil prices do represent a potential spoiler for the scenario I am developing.

Baring special circumstances, I do not expect the downturn heading into fall 2006 to be particularly severe when compared to the downturn in the next four-year cycle, the one which should bottom around the fall of 2010. This four-year cycle bottom will correspond to a Juglar bottom as well and should be a major bear market like the one we just saw. Such a powerful bear will likely begin from a high point, so I expect stocks to reach levels equal to or higher than the 2000 highs by the end of 2008, the next four year cycle peak. Finally, we should look at the later behavior of the index during the last secular bear market (see Figure 5). Although the four-year cycle was still present, it was greatly attenuated. I expect the cycle in the decade after 2010 to be similarly attenuated.

So far I have constructed an abstract model to describe the twists and turns of the stock market over the course of this secular bear market. Another approach would be to look at the historical record to determine the actual ranges of P/R observed in bull/bear market cycles in past secular bear markets. Figure 6 presents these ranges. These ranges can be used to forecast the range of P that can be expected, based on historical precedent, for future ordinary bull/bear market turning points. Of the four plots shown in Figure 6, the one for 1929-49 is clearly different from the others in its early stages. This difference reflects the huge size of the 1929-32 decline. If we were to use all four plots, the forecasted ranges would be so wide as to be useless. If we ignore the 1929-1949 plot, a useful result can be obtained.

Figure 6. P/R extremes in bull/bear market cycles in past secular bear markets

I chose to ignore the 1929-1932 decline as a model for reasons I explained in spring 2002:

Both the 1980's fall to plateau and the 1940's trough region showed economic characteristics different from those of similar periods in past cycles. This discrepancy suggests that the present fall from plateau may also play out differently than previous ones. This is important because many bearish commentators who hold Kondratiev views similar to mine (i.e. the 2000 stock peak is cycle-analogous to 1929) invoke the dire prospects facing people in 1930 as a template for the future. It is very unlikely that the next few years will be anything like 1929-32, even though they will be "cycle analogous". [14]

We now know that no depression developed after the bursting of the tech bubble and that the post-1929 example is not really relevant to today. With this assumption I can use the range of P/Rs at previous bull/bear market extremes (excluding 1929-1949) to calculate projected ranges for the location of the market in future cycle turning points. This was done in Figure 7 for the current ordinary bull market. Figure 7 was originally produced in March 2003 for my third book Retiring Rich, which has recently come out as a paperback under the title Investing in a Secular Bear Market.

Shown in the figure was the original projection of a top around 1999-2000, which was discussed in Stock Cycles. Also shown is when I moved down from 100% stock allocation and to 0% stock allocation in response to that projection. Shown in the figure is the projection of the first ordinary bear market bottom of the post-2000 secular bear market era, discussed in reference [6]. Also shown is when I moved to 80% stock allocation as described in reference [13]. Finally, the projection of the coming bull market peak is shown and the course of the market since the original figure was created is shown in red.

Figure 7. Projected levels for bull/bear market turning points in this secular bear market

As can be seen in the figure, the S&P500 has yet to move into the projected region for the next ordinary bull market high, despite the fact that we are already past the average location of the top of the four year cycle. This suggests that any decline into the 2006 four-year cycle low from today's levels will likely be small. On the other hand, this decline may not have started yet, in which case the index may rise further in 2005, penetrating into the projected zone before beginning a decline into 2006. For these reasons I have not reduced my 401(k) stock allocation; it remains at 80%. If the four-year cycle peak is already in, I expect the next bull market peak (in ~2008) to be sufficiently high to justify holding positions at current levels. If the four-year cycle top has not yet appeared, there may still be opportunities to reduce stock exposure in the future as P/R rises into the "sell" zone.

References:

1.      Alexander, Michael, A, "Secular Market Trends", Safehaven, March 16, 2001.

2.      Alexander, Michael, A, Stock Cycles, Lincoln NE: Writer's Club Press, 2000.

3.      Campbell, John Y and Robert J Shiller, "Valuation Ratios and the Long-Run Stock Market Outlook", Journal of Portfolio Management, Winter 1998.

4.      Alexander, Michael, A, "Progress of the Current Secular Bear Market", Safehaven, July 1, 2002.

5.      Alexander, Michael, A, "P/E, P/R and Irrational Exuberance", Safehaven, April 30, 2001.

6.      Alexander, Michael, A. "How low must we go?", post made on the Longwaves board, August 25, 2002.

7.      Alexander, Michael, A. "How Low Can We Go? What Several Valuation Methods Have To Say", Safehaven, October 20, 2002.

8.      Alexander, Michael, A. "Generations and Business Cycles", Safehaven, November 06, 2002.

9.      Alexander, Michael, A. "Generations and Business Cycles Part II", Safehaven, November 06, 2002.

10.  Alexander, Michael, A. "Use of Secular Trend Concept for Asset Allocation in 401K", Safehaven, May 26, 2001.

11.  Alexander, Michael, A. "Use of Secular Trend Concept for Asset Allocation in 401K, Update", Safehaven, April 26, 2002.

12.  Alexander, Michael, A. "Use of Secular Trend Concept for Asset Allocation in 401K, Part 3", Safehaven, July 15, 2002.

13.  Alexander, Michael, A. "Progress Update on the Current Secular Bear Market", Safehaven, July 23, 2002.

14.  Alexander, Michael, A. " The Kondratiev Cycle Revisited: Part Two, Economic Implications", Safehaven, May 6, 2002.


 

Michael A. Alexander

Author: Michael A. Alexander

Michael A. Alexander
Stock Cycles

Mike Alexander is the author of four books: (2000) Stock Cycles: Why stocks wont beat money market over the next 20 years; (2002) The Kondratiev Cycle: A generational interpretation; (2003) Retiring Rich: The ultimate IRA and 401(k) investing guide (now available in paperback under the title Investing in a Secular Bear Market) and (2004) Cycles in American Politics: How political, economic and cultural trends have shaped the nation.

Michael is not a registered advisor and does not give investment advice. His comments are an expression of opinion only and should not be construed in any manner whatsoever as recommendations to buy or sell a stock, option, future, bond, commodity or any other financial instrument at any time. While he believes his statements to be true, they always depend on the reliability of his own credible sources. Of course, we recommend that you consult with a qualified investment advisor, one licensed by appropriate regulatory agencies in your legal jurisdiction, before making any investment decisions, and barring that, we encourage you confirm the facts on your own before making important investment commitments.

Copyright © 2000-2015 Michael A. Alexander

All Images, XHTML Renderings, and Source Code Copyright © Safehaven.com