Cycles in American history

This paper is the first of a series exploring the ideas of historical cycles operative in Anglo-American history from the distant past right down to the present time.

Economic cycles: the stock cycle

My initial interest in cycles was piqued in 1995 when I was did a simple analysis of the Dow Jones Industrial Average that implied that the index, then at 5000, would continue to rise until it reached a level between 10000 and 15000 in the 2000-2004 window. As the market continued to rise I refined the analysis and projected the end of the bull market for 1999-2001. I finally sold out in late 1999. The analytical tool I used was a stock valuation parameter I called P/R. P/R is the ratio of the S&P 500 index value to business resources (R). R is the cumulative retained S&P 500 index earnings in constant dollars from 1871 to the present plus the value of R in 1871, which is assumed equal to the index value. For example, R in fall 1999 was $950. Of this value, $880 represented accumulated retained earnings from 1871 to 1999 (in 1999 dollars). The other $70 represents the initial value of the index in 1871, also in 1999 dollars. P/R defines a ca. 30-year cycle in the stock market (see Figure 1). The value of P/R had reached an all-time high in 1999, which is why I sold out of the market.

I wrote about these ideas during the first three months of 2000 and published them that fall in a book called Stock Cycles: Why stocks won't beat money markets for the next twenty years. I made an explicit prediction in Stock Cycles that made use of the stock cycle¹ and which can be used as a test of the validity of the idea that a cycle in the stock market exists.

the current upwards trend in stock index levels will end, most likely this year (2000) but almost certainly by 2004. After it stops going up, the stock index will not go higher (in constant-dollar terms) for a long time, most likely 20 years or more.

On a monthly average basis, the S&P500 index reached its secular bull market peak in August 2000 at about 1485, which would correspond to about 1750 in fall 2007 and 2100 today (Jan 2014). The S&P500 has not reached this level so far and the prediction is still valid. The second test of stock cycles was a prediction made in fall 2002.² Three valuation models were compared: the P/E used by Robert Shiller in his book Irrational Exuberance,³ the Q ratio⁴ invented by James Tobin, and P/R valuation.¹ The market valuation in 2002 according to P/E was:²

Since its 2000 peak P/E has fallen dramatically, but it is still well above the levels it has seen at this stage in previous secular bear markets. The S&P500 would have to fall into the 400's for P/E to be within the historical range. Thus, advocates of this valuation method recommended exiting the stock market in 1997, and would caution against re-entry until after another 50% drop from current (Oct 2002) levels.

The market valuation in 2002 according to Q was:²

Since its 2000 high, Q has also fallen dramatically. Like P/E it too is fairly high relative to its historical values in early stage secular bear markets. The S&P500 would have to fall to the 600's for Q to be in its historical range. Thus, advocates of Tobin's Q would have recommended exiting stocks in late 1997 or early 1998, and would caution against re-entry until after about a one-third drop from current levels.

The market valuation in 2002 according to P/R was:²

Since its 2000 high, P/R has fallen like the other two measures. Unlike the other two, P/R is currently not particularly high relative it its historical values in previous secular bear markets. It is not low either. Thus, although the index certainly could continue to fall, and P/R would remain within its historical norms, it does not have too. Thus, advocates of P/R would have recommended exiting stocks in 1999, and would not caution against re-entry during the recent declines.

As it turned out the market rose from the October 2002 lows. Only P/R, the valuation tool I developed to characterize the stock market cycle gave the correct prediction. The third test was made on 20 November 2008:⁵

Based on the 752 close on 20 November 2008 (see blue cross in figure) we are essentially already at the bottom. Based on the latter assessment, I pulled the trigger and deployed most of my cash into the S&P500 on 20 November 2008.

The S&P500 has risen over 1000 points since 20 November 2008, confirming the prediction that the market was at a low point in the ongoing secular bear market. A fourth test was made in July 2013.⁶

I sold my last position last month when the index was in the low 1600's. The P/R graph above shows that the market has reached roughly the same position relative to past secular bear markets as it had in 2007. There is still room to rise and the current bull market will likely run quite a bit longer, particularly as the length of the current business cycle, at less than 6 years, is still quite short. I had made a similar argument in December 2007, and was wrong. So this time I decided to sell as soon as P/R indicates a relatively overvalued market.

The bet I am making is that there will be another downturn as there was in the past and this downturn will send P/R below its previous low of 0.49. If I assume the next downturn will have happened by 2020, seven years from now, I can extrapolate the R growth curve to estimate a value of about 2550 then, which when multiplied by 0.49 gives a value of 1250 for the next bottom, well below today's levels, at which to move back into stocks.

This experiment is in progress, the results will not be known until the next recession. With three successful predictions so far it appears that the concept of cycles in stock market valuation as measured using P/R is valid. One of the ideas explored in Stock Cycles was the idea that the stock cycle was a binary harmonic of the Kondratiev cycle (i.e. two stock cycles exactly fit into one Kondratiev cycle). Before going further with this idea an introduction to the Kondratiev cycle is useful.

Economic cycles: the Kondratiev cycle

The Kondratiev cycle is a ca. fifty year cycle in prices, interest rates and other economic variables that was first described by N. D. Kondratiev in the 1920's. Kondratiev cycles are most readily apparent in monetary data such as prices and interest rates. Figure 2 shows a plot of the US producer price index over the period 1800 to 2000. Prior to WW II, the Kondratiev cycle could be seen clearly as periodic peaks and troughs in prices spaced about 50 years apart.

Figure 2 shows three Kondratiev peaks in producer prices in 1814, 1864, and 1920. In between the peaks are two Kondratiev troughs, one in 1843 and the other in 1896. The spacing of these peaks and troughs average 53 years, which provides an estimate for the length of the Kondratiev cycle. The period of rising prices between the Kondratiev trough and the Kondratiev peak is called the Kondratiev upwave, or just upwave. Conversely, the decline from the peak to the trough is the downwave. Inspection of each downwave shows that after each Kondratiev peak there is a sharp drop and then a leveling-off in prices, producing what is sometimes called the (price) plateau. The plateau ends with a second precipitous drop, or what is sometimes called the fall from plateau. The first three cycles following 1800 showed plateaus ending in 1819, 1873, and 1929 (see Figure 2).

Kondratiev, working in the 1920's, predicted a deflationary depression using his cycle theory. He was right, but subsequent attempts to apply his theory have been unsuccessful due to the absence of deflation (falling prices) after 1932. The disappearance of deflation over the past 80 years is the result of stimulatory economic policy that began with the New Deal and continues to this day.

In order to examine the post-1932 period I detrended the price series using a concept I call reduced price.⁷ Reduced price is the ratio of the actual price to the value predicting using a simple linear model:

1. Price = A + B * S

2. S = (M3+CumDef)/GDP

S is a parameter called monetary stimulus. It is the sum of cumulative federal government deficits (CumDef) and M3, a standard measure of money supply that once was provided by the Federal Reserve. A and B are the constants obtained when price was regressed against S (see reference 7 for more details). Figure 3 shows a plot of reduced price. The figure shows that the Kondratiev cycle apparent in prices before 1932 is also shown by reduced price, but the reduced price cycle continues on past 1932 with a clear-cut trough in 1946 and a peak in 1981. Another variable that shows Kondratiev-like cycles is interest rates. Table 1 summarizes the Kondratiev cycle in terms of these three variables.

Table 1. Kondratiev cycle turning points in nominal price, reduced price and interest rates

Description	Price	Reduced price	Interest*	Consensus
Peak	1814	1813	1814	1814
Fall from Plateau	1819	1819	--	1819
Trough	1843	1842	1830	1843
Peak	1864	1864	1861	1864
Fall from Plateau	1873	1873	--	1873
Trough	1896	1895	1899	1896
Peak	1920	1918	1920	1920
Fall from Plateau	1929	1929	--	1929
Trough	--	1946	1946	1946
Peak	--	1981	1981	1981
Fall from Plateau	--	2008	--	2008

*Gov. bonds to Civil War⁸, Corporate bonds after⁹

The data in Table 1 form a consensus for the dating of Kondratiev cycle in the US over the past two centuries. Turning points in the Kondratiev cycle are compared to turning points in the stock cycle using the rule that one stock cycle falls into the upwave and one falls into the downwave. This means that Kondratiev peaks and troughs should be closely aligned with stock cycle bottoms. Stock cycle peaks (secular bull market tops) then occur somewhere in the middle of the upwave and downwave. There is no visible Kondratiev turning point in the upwave, but there is one in the downwave, the fall from plateau.

Table 2 compares Kondratiev peak, trough, and fall from plateau turning points with what should be their corresponding Stock Cycle turning points. The data show that Kondratiev turning points precede the correspond Stock Cycle points by 2.7 years on average. There is considerable variation as indicated by the large standard deviation of 5.7 years.

Table 2. Alignment between stock market cycle and Kondratiev cycle

Kondratiev Cycle		Stock Cycle		Difference
Description	Date	Date	Description
Peak	1814	1815	Bottom	-1
Fall from Plateau	1819	1835	Peak	-16
Trough	1843	1843	Bottom	0
Peak	1864	1861	Bottom	3
Fall from Plateau	1873	1881	Peak	-8
Trough	1896	1896	Bottom	0
Peak	1920	1921	Bottom	-1
Fall from Plateau	1929	1929	Peak	0
Trough	1946	1949	Bottom	-3
Peak	1981	1982	Bottom	-1
			Average (std. dev.)	-2.7 (5.7)

Assuming that an alignment between the Stock Cycle and the Kondratiev cycle exists then it should be possible to predict when the next Kondratiev turning point should occur. By early 2002 it appeared that the stock cycle had peaked in 2000, According to Table 2 a fall from plateau event was predicted to occur 2.7 ± 5.7 years before 2000. I employed the 10 differences in Table 1 to construct a Student's t distribution with 9 degrees of freedom in order to estimate the cumulative probability that the fall from plateau will have occurred as a function of years past 2000. Values of 33%, 58%, 74%, 85%, 91% and 95% were obtained for 2002, 2004, 2006, 2008, 2010, and 2012, respectively. Based on this analysis the fall from plateau could happen as late as 2012 and the hypothesis that the two cycles are linked would not be rejected. So I had potentially a long time to wait.

The pre-Civil War data in Table 1 are not very reliable. The stock index contained a relatively small number of firms, most of which were banks. American debt markets had not yet developed fully and the relevance of interest rate time series is not clear. In Britain, which had a well-developed debt market, interest rates showed a trough in 1844 and peaked in 1866, much closer to the price turning points. So I focused on the post-Civil War era. I did the same calculation using only the data since 1860 and obtained values of 49%, 76%, 90% and 95% for 2002, 2004, 2006 and 2008, respectively. With this analysis the hypothesis would be rejected if no fall from plateau had been observed by the end of 2008. Not only that, but the probability was high that it would be soon after 2000.

Because a recession and serious bear market were underway in 2002 I strongly suspected that the fall from plateau had already begun in 2001 (see Figure 1 in reference 10). At this time I believed that the future fall from plateau would not be accompanied by a financial panic (as the previous three had been) because the 1946 trough had no panic, unlike previous troughs. I believed that modern economic management had learned from the 1929-1933 experience and that panics were a thing of the past. So I expected the fall to be a more gradual event, like the decline from the 1981 to the plateau that occurred over five years.

By 2005 it was clear that a fall the from plateau event detectable by my standard reduced price measure had not happened yet:¹¹

At the beginning of 2001, PPI-based reduced price began a sharp decline that I believed (at the time) was the start of the fall from plateau. The fall from plateau is another visual feature of historical price plots of Kondratiev downwaves. In the past, the early downwave has shown two distinct waves. The first wave of falling prices begins at the Kondratiev peak and ends at the plateau. After the plateau there is a second wave down (the fall from plateau). The decline that began in 2001 lasted just one year. As Figure 2 shows, no discernible fall from plateau is evident. On the other hand, the CPI-based trend reduced price has shown a change in trend at the beginning of 2001.

I toyed with the idea that perhaps the Kondratiev features are now revealed in consumer prices and not producer/commodity prices. About a year later the Fed stopped publishing M3 values and I stopped tracking the reduced price. After what seemed like a panic occurred in 2008 I wondered if this might have been the anticipated fall from plateau. After all, the first statistical analysis described above gave a 20% probability that the event would occur in 2008 or even later. I obtained M3 estimates from charts at the Shadow Statistics website and used it to update my standard reduced price plot. As shown in Figure 3, the fall from plateau did indeed occur in 2008 and the hypothesis that the Kondratiev cycle is a real cycle that is aligned with the stock cycle was not rejected.

Based on this result and the evidence of coincident cycles in prices, interest rates and the stock market in the past, I concluded there is good reason to believe that long cycles in the American economy have been and are still relevant.

Economic cycles: The panic cycle

From the late 18^th century to the early 20^th century the American economy featured periodic financial crises called panics. Panics associated with major economic downturns occurred in 1819, 1837, 1857, 1873 and 1893. These panics were associated with major movements in real estate/land prices.¹⁴ The recurring panics define an economic cycle that can be dated by the stock market bottoms associated with the post-panic slump (see Figure 2 in reference 14). This real estate/panic cycle was first identified by the Russian-American economist Simon Kuznets¹⁵ and was studied by Homer Hoyt¹⁶ in the early 1930's. They are sometimes referred to as Kuznets cycles. There were also politically important financial panics in 1772, 1792 and 1907 that were not associated with a real estate cycle and which did not lead to a major economic downturn. There was also a financial panic in 1932-3 that also was not real-estate linked, but was associated with a major downturn. After 1933 there were no panics until 2008. This panic led to a major downturn and was associated with real estate like those of the 19^th century. Whether or not this panic represents a resumption of the panic cycle remains to be seen.

The Saeculum

In their book Generations, William Strauss and Neil Howe introduce a fascinating theory that interprets U.S. history in terms of a repeating series of four basic types of generations.¹⁶ Generations create history and history creates generations. In their follow-up work, The Fourth Turning, they propose that history moves in long cycles, each four generations long, which they call the saeculum, after the ancient Etruscan cycle of similar length.¹⁷ The saeculum contains four periods, called turnings, each of which is associated with the birth dates for a particular type of generation.

What defines a generation is its peer personality. Peer personality is a generational persona recognized and determined by (1) common age location; (2) common beliefs and behavior; and (3) perceived membership in a common generation. Each generation has an archetype which displays the peer personality for that generation.

Table 3. Descriptions of peer personalities during the four phases of life¹⁸

Peer Personality	Youth	Rising Adult	Maturity	Elder	Archetype
Idealist	indulged	narcissistic	moralistic	visionary	prophet
Reactive	criticized	alienated	pragmatic	reclusive	nomad
Civic	protected	heroic	powerful	busy	hero
Adaptive	suffocated	conformist	indecisive	sensitive	artist

Table 3 shows one-word characterization of each archetype/peer personality in each phase of life as given by Strauss and Howe.¹⁸ The prophet archetype is an indulged youth who grows into a narcissistic young adult who has all the answers. As he ages, he sours on the loose standards to which he was held as a youth, becoming moralistic in midlife. Self-assured to the end and full of the wisdom that comes from introspection, he acts as a visionary elder in old age. The nomad archetype is a criticized youth who grows into an alienated young adult. He learns the lessons of life from the school of hard knocks, becoming pragmatic in midlife, and aging into a reclusive elder. The hero The Civic archetypesarchetype, disciplined in youth, becomes equipped for heroic performance in rising adulthood. The experience of purposeful action in rising adulthood prepares his generation to act as powerful institution-builders in mid-life and to continue a lifetime of purposeful activity into old age as busy elders. Finally, the artist archetype is a suffocated youth who grows into a conformist adult. As he enters mid-life he experiences a crisis in which he begins to question his uncritical acceptance of behavioral standards set by others and becomes an open-minded explorer of new ideas and experiences. Indecisive in mid-life leadership roles, he ages into a sensitive elder.

The peer personality of a particular generation is shaped by the generation's historical location relative to a social moment. A social moment is an era, typically lasting about a decade, when people perceive that historical events are radically altering their social environment. Thus, a generation's peer personality (what makes it a particular kind of generation) depends on when they were born relative to particularly eventful periods in history. There are two types of social moments: secular crises, when society focuses on reordering the outer world of institutions and public behavior; and spiritual awakenings, when society focuses on changing the inner world of values and private behavior. Social moments are embedded in slightly longer periods called turnings. Thus, the secular crisis is part of a Crisis turning and the spiritual awakening is part of an Awakening turning. The period in between define turnings also. The turning after a Crisis and before an Awakening is a High. That between the Awakening and the Crisis is an Unraveling. The four turnings together comprise a saeculum.

The driver for the saeculum is the regularly-repeating series of social moments of alternating kinds. If social moments occurred sporadically, a regular series of generations would not be created and there would be no saeculum. Strauss and Howe list six spiritual awakenings and five secular crises (Table 4) spaced an average of 88 years apart. They propose generations reflect the experience of living through a social moment at a particular phase of life. The phases of life are youth (age 0-21), rising adulthood (age 22-43), maturity (age 44-65) and elderhood (age 66-87). They are 22 years in length and four of them comprise an 88-year saeculum, which neatly dovetails with the average spacing of social moments of the same type.

Table 4. Social moments in American history

Cycle	Spiritual Awakening	Secular Crisis
Pre-Colonial	Reformation (1517-1539)	Spanish Armada (1580-1588)
Colonial	Puritan Awakening (1621-1640)	Glorious Revolution (1675-1692)
Revolutionary	Great Awakening (1734-1743)	American Revolution (1773-1789)
Civil War	Transcendental Awakening (1822-1837)	Civil War (1857-1865)
Great Power	Missionary Awakening (1886-1903)	Depression & WWII (1932-1945)
Millennial	Boom Awakening (1967-1980)	--

The last three secular crises in Table 4 are easily recognized as momentous times in American history; it is self-evident that they constitute secular crises. The two before them are not so clear. Why, for example, isn't the English Civil War included with the American Revolution and Civil War as secular crises? Author Kevin Phillips does just this in his monumental Cousin's Wars.¹⁹ Phillips contends that the three great civil conflicts of modern Anglo-American history exerted an enormous influence in shaping nearly every aspect of American and British life. That is, they are secular crises.

The situation for spiritual awakenings is also problematic. The spiritual awakenings in Table 4 roughly correspond to periods of religious fervor identified by historian William McLoughlin in his book Revivals, Awakenings and Reform.²⁰ McLoughlin defines awakenings as periods of cultural revitalization caused by a crisis in beliefs and values that produces a reorientation in those values and beliefs. He identifies awakenings in 1610-40, 1730-60, 1800-30, and 1890-1920.²¹ Comparison of these dates with those for spiritual awakenings in Table 1 shows a rough correspondence. The spiritual awakenings are subsets of the McLoughlin periods that are located midway between secular crises so that a regular pattern of alternating social moments is evident.

Table 5 illustrates this by comparing Strauss and Howe awakening turnings with McLoughlin awakenings. The Strauss and Howe Awakening turnings are located 16-27 years from the nearest secular crisis with an average spacing of 23 years, close to their standard 22 year generation. In contrast, the McLoughlin dates are located 6-35 years from the nearest secular crisis and can hardly be said to be spaced a generation apart from crisis eras. That is, a saeculum defined by McLoughlin Awakenings isn't very regular, suggesting that such a regular cycle may not exist, or at least cannot be revealed by a survey of history. Strauss and Howe did not base their cycle on external events but rather on their interpretation of biographies of members of the various generations. A close correspondence between their periodization and external events is not implied by their model.

Table 5. The spacing of spiritual awakenings relative to crises for two authors

Secular Crises*	Spiritual Awakenings
	Strauss and Howe*	McLoughlin^17
1569-1594	1621-1649	1610-1640
1675-1704	1727-1746	1730-1760
1773-1794	1822-1844	1800-1830
1860-1865	1886-1908	1890-1920
1929-1946	1964-1984	1960-0000

*Values for associated turnings are given

The Strauss and Howe turnings-cycle or saeculum is defined in terms of an extensive vocabulary: generations, peer personalities/archetypes, social moments, phases of life, but has little empirical foundation. Unlike stock market or Kondratiev cycles, which can clearly be seen in a plot of stock market valuation, or reduced price, the saeculum has no objectively-determined indicator. Its presence must be inferred indirectly through its influence on other social, political or economic trends.

Table 6 compares the Kondratiev Cycle turning points with the saeculum turnings. The Kondratiev downwave is split in two: Fall, from the Kondratiev peak to the fall from plateau and Winter, from the fall from plateau to the trough. The upwave is likewise split into Spring, from the Kondratiev trough to the stock cycle peak, and Summer, from the stock cycle peak to the Kondratiev peak. This division of the Kondratiev cycle into seasons has been proposed by other workers.²² Kondratiev turning points for the 18^th century were obtained from Joshua Goldstein's base dating scheme presented in his excellent book on long cycles.²³ Goldstein's scheme represents a consensus of the results of 33 Kondratiev scholars whose work was reviewed in his book. Also shown in Table 6 is the panic cycle in terms of associated market bottoms. Included here are the minor panics in 1792 and 1907 and the 1932-33 and 2008 crises.

Table 6. Comparison of Strauss and Howe Saeculum with economic and financial cycles

Panic Bottoms^14	Saeculum		Kondratiev Cycle*
	Turning Type	Dating	Description	Dating
2008-	Crisis	2008-	Winter	2008-
--	Unraveling	1984-2008	Fall	1981-2008
--	Awakening	1964-1984	Summer	1966-1981
--	High	1946-1964	Spring	1946-1966
--	Crisis	1929-1946	Winter	1929-1946
1907-1932	Unraveling	1908-1929	Fall	1920-1929
1896-1907	Awakening	1886-1908	Summer	1906-1920
1877-1896	High	1865-1886	Spring	1896-1906
1857-1877	Crisis	1860-1865	Winter	1873-1896
1842-1857	Unraveling	1844-1860	Fall	1864-1873
1819-1842	Awakening	1822-1844	Summer	1853-1864
1792-1819	High	1794-1822	Spring	1843-1853
--	Crisis	1773-1794	Downwave	1762-1790
--	Unraveling	1746-1773	Upwave	1747-1762
--	Awakening	1727-1746	Downwave	1720-1747
--	High	1704-1727	Upwave	1689-1720

*Spring/Summer turning points from stock cycle; 18^th century points from Goldstein²³

Examination of Table 6 shows that the turnings align very well with the Kondratiev seasons after 1929 and with the panic cycle between 1792 and 1929. Turnings during the 18^th century and first half of the 19^th century also aligned with Kondratiev waves rather than seasons. This alignment with empirically-defined economic cycles provides some support for the relevance of the Strauss and Howe saeculum as an historical cycle. It also raises some interesting questions. The post-1929 alignment between turnings and the Kondratiev seasons implies that both cycles have been of the same length since 1929. The fall from plateau in 2008 confirms the length of the most recently completed Kondratiev cycle as 79 years. Assuming 2008 is the beginning of a crisis turning, as Howe proposes, then the most recent complete saeculum is also 79 years long. Previous saecula were of similar length: 76, 78, and 81 years for the saecula ending in 1984, 1964 and 1946, respectively. In contrast, the Kondratiev cycles ending in 1981, 1966, and 1946 were 61, 60, and 50 years in length, and the average length of the 11 before them was 52 years. Clearly the length of the Kondratiev cycle has increased in recent times, while that of the saeculum has not.

On the other hand, the apparent association between turnings and Kondratiev waves before the Civil War implies that the saeculum was twice as long as the Kondratiev cycle back then. Since the latter has an average length of over 50 years historically, this implies a 100+ year-long saeculum in the past compared to 75-80 years today. The Strauss and Howe turnings begin in 1435. The first complete saeculum ran 107 years to 1542. The next ran 106 years to 1649. The one after that ran 97 years to 1746, and the next one 98 years to 1844. Thus, saeculum length averaged 102 years (a length consistent with two 51-year Kondratiev cycles) up through the early 19^th century.

Something happened to saeculum length in the 19^th century, which Strauss and Howe acknowledge as the Civil War anomaly. They hypothesize that the hero generation that should have emerged from the Civil War secular crisis did not do so and as a result the saecula containing the Civil War are associated with only three instead of four generations. If this is true the expected pattern would be a sudden shortening of the century-long pre-war saeculum to about 75 years and then restoration to its former length of about a century. But this is not what happened. The average length of the four saecula containing the Civil War averaged 67 years in length, while those beginning in 1865 or later averaged 79 years. Table 6 suggests an alternative explanation. The two-in-one relation of the Kondratiev cycle to the pre-Civil War saecula implies one saeculum mechanism then. This mechanism was replaced by a different mechanism which featured a one-to-one relation between turnings and a new recurrent economic phenomenon.

In the 19^th century, when real estate was the most important asset class, the recurrent phenomenon was boom/bust cycles in land prices, as manifested in the periodic financial panics and subsequent depressions. As the country industrialized, capital became the most important asset class and the recurrent feature was cycles in the capital market as shown by cycles in stock index valuation and interest rates, both manifestations of the Kondratiev/stock cycle.

Political cycles: The Schlesinger cycle

The American historian Arthur Schlesinger Sr. describes a cycle in what he called the political zeitgeist, or spirit of the times.²⁴ It appears as an oscillation between liberal and conservative eras, Table 7 shows dates for the Schlesinger eras.²⁵ Also shown is an economic cycle obtained from a combination of the Kondratiev and panic cycles in Table 6. For 1772 to 1896, the panic cycle dates were used. Dates for Kondratiev waves were used before 1772. Kondratiev season dates were used after 1920. With a single exception the economic cycle closely corresponded to the political cycle, which is not surprising as economic factors are usually very important in elections, the outcome of which impact the mix of policy and political beliefs that make up the political zeitgeist. The two cycles were used to form a consensus political-economic (PE) cycle that will be used for future work.

Table 7. Correlation of political and economic cycles

Schlesinger eras	Economic cycle	Political-economic (PE) cycle	Turnings (type)
--	1720-1747	1720-1747	1727-1746 (A)
--	1747-1772	1747-1774	1746-1773 (U)
1776-1788 (L)	1772-1792	1774-1792	1773-1794 (C)
1788-1801 (C)	1792-1819	1792-1824
1801-1812 (L)			1794-1822 (H)
1812-1829 (C)
1829-1841 (L)	1819-1842	1824-1842	1822-1844 (A)
1841-1861 (C)	1842-1857	1842-1859	1844-1860 (U)
1861-1869 (L)	1857-1877	1859-1873	1860-1865 (C)
1869-1901 (C)	1877-1896	1873-1896	1865-1886 (H)
1901-1919 (L)	1896-1920	1896-1919	1886-1908 (A)
1919-1931 (C)	1920-1929	1919-1930	1908-1929 (U)
1931-1947 (L)	1929-1946	1930-1946	1929-1946 (C)
1947-1962 (C)	1946-1966	1946-1964	1946-1964 (H)
1962-1978 (L)	1966-1981	1964-1981	1964-1984 (A)
--	1981-2008	1981-2008	1984-2008 (U)
--	2008- ?	2008- ?	2008- ? (C)

Two different kinds of zeitgeist have been discussed so far. One is the political zeitgeist indirectly measured by the Schlesinger cycle dating. The other is the social zeitgeist that gives rise to generational peer personalities, which is indirectly measured by the dating of the Strauss and Howe turnings. Since social moments are times of social upheaval, one might be able to track this dynamic though the incidence of popular unrest such as strikes, race conflict, civil strife, and political movements. I assembled a timeline of events of these types (see Appendix A). These data were used to produce a time series of unrest event frequencies. A trend line for this series was calculated using a centered 100 year moving linear regression. The event frequency data was divided by the trend to obtain a normalized unrest event frequency. The results are plotted in Figure 4. Also shown is a 15-year moving average that helps visualize the cyclical characteristics of the data.

The political zeitgeist was characterized by collecting a timeline of political events which appears in Appendix C. These events were classified as either liberal or conservative using rules given in Appendix C. The definition of liberal and conservative was constructed to reflect Schlesinger's notions of what is liberal and conservative and was expected to correlate well with the Schlesinger periodization. If a close correlation is obtained, this dataset should provide a way to measure the Schlesinger zeitgeist. As with unrest the data were normalized by dividing by a trend value obtained from a moving 100-year regression analysis. Average values of the normalized unrest event frequencies were calculated for the PE periods and Schlesinger eras given in Table 8. Average values of the frequencies of conservative and liberal events are also given in Table 8.

Table 8. Normalized frequency analysis for political and unrest events for Schlesinger and P-E periods

PE period	Political Events		Unrest Events	Schlesinger era	Political Events		Unrest Events
	Con	Lib			Con	Lib
1720-1746 (A)	--	--	1.38	1776-1788 (L)	0.16	0.79	0.48
1747-1773 (I)	--	--	0.97	1789-1800 (C)	0.93	0.29	0.91
1774-1792 (A)	0.42	0.54	1.07	1801-1815 (L)	0.49	0.50	0.23
1793-1823 (I)	0.57	0.40	0.45	1816-1828 (C)	0.83	0.34	0.86
1824-1842 (A)	0.43	0.95	1.96	1829-1840 (L)	0.25	1.09	2.17
1843-1859 (I)	0.41	0.05	0.43	1841-1860 (C)	0.45	0.18	0.57
1860-1873 (A)	0.44	0.93	0.50	1861-1868 (L)	0.37	0.81	0.59
1874-1896 (I)	0.53	0.17	0.92	1869-1900 (C)	0.53	0.33	0.85
1897-1919 (A)	0.63	0.58	1.59	1901-1918 (L)	0.67	0.67	1.43
1920-1930 (I)	0.59	0.16	0.54	1919-1930 (C)	0.62	0.18	1.02
1931-1946 (A)	0.29	0.98	1.46	1931-1946 (L)	0.29	0.98	1.46
1947-1964 (I)	0.46	0.31	0.61	1947-1961 (C)	0.52	0.24	0.46
1965-1981 (A)	0.27	0.69	1.51	1962-1978 (L)	0.18	0.75	1.51
1982-2008 (I)	0.70	0.41	0.80	--	--	--	--
Inactive era avg.	0.54	0.25	0.66	Cons. Era avg.	0.64	0.26	0.78
Active avg.	0.41	0.78	1.31	Liberal era avg.	0.35	0.80	1.12
p <=	4.0%	0.05%	0.3%	p <=	0.8%	0.1%	12.7%

As expected, liberal events were more frequent during Schlesinger liberal eras and conservative events were more frequent during conservative eras. This result was highly significant: over 99% confidence for both. This confirms that the political event analysis captures the zeitgeist Schlesinger had in mind. Table 8 shows the same analysis for the PE cycles. The same result is obtained, although the level of statistical significance is lower, mostly because of the incorporation of the 1801-1815 liberal era into a larger conservative PE period.

The PE period also shows a statistically significant correlation with popular unrest with liberal periods showing higher levels than conservative eras. The alignment of politics and unrest in the PE cycle suggests the names active (A) and inactive (I) for the alternating periods of high unrest/liberal politics and low unrest/conservative politics, respectively. When the same unrest analysis is done for the Schlesinger eras, a similar pattern is seen but it is not statistically significant. The reason again lies in the 1789-1828 period, in which the conservative era from 1789-1800 had high unrest relative to the adjacent liberal era. The PE cycle does a better job of representing changes in political and socioeconomic conditions in America than does the Schlesinger cycle.

The PE cycle also aligns well with the Strauss and Howe turnings before the Civil War and after WW I. The active periods in the PE cycle correspond to social moment turnings. Both cycles show a similar degree of shortening during the 19^th century (Figure 5). Turnings were analyzed in terms of unrest and political events. High unrest and liberal zeitgeist were associated with social moment turnings. Low unrest and conservative zeitgeist we associated with the other turnings. The association with unrest was statistically significant while that with political zeitgeist was not. The PE cycle did a better job of representing both than did the saeculum.

Furthermore, the PE cycle is explicitly based on economic cycles and so has an explicit economic component. It appears to be the best consensus of political, social and economic long cycles in America.

References.

1. Alexander, Michael A. Stock cycles: why stocks won't beat money markets over the next twenty years, New York: Writers Club Press, 2000.

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

3. Shiller, Robert J. Irrational Exuberance, Princeton NJ: Princeton University Press, 2000.

4. Smithers, Andrew and Wright, Stephen, Valuing Wall Street: Protecting Wealth in Turbulent Markets, New York: McGraw-Hill, 2000.

5. M. A. Alexander, Progress Update on the Current Secular Bear Market, Safehaven, November 20, 2008.

6. M. A. Alexander, Progress Update on the Current Secular Bear Market, Safehaven, August 23, 2013.

7. M. A. Alexander, The Kondratiev Cycle and Secular Market Trends, Safehaven, May 12, 2001.

8. Mark Phillips, U.S. Long Term Bond Yields, Back to 1800, Wall Street Journal MarketBeat, August 18, 2011.

9. F. R. Macauley, Some Theoretical Problems Suggested by the Movements of Interest Rates, Bond Yields and Stock Prices in the United States Since 1856, New York: National Bureau of Economic Research, 1938, A142-A161.

10. M.A. Alexander, The Kondratiev Cycle Revisited: Part One, Current Position in Cycle, Safehaven, May 5, 2002.

11. M.A. Alexander , Inflation, Monetary Stimulation and Interest Rates - A Cycle Perspective, Safehaven, March 7, 2005.

12. John Williams, Inflation, Money Supply, GDP, Unemployment and the Dollar - Alternate Data Series, Shadow Government Statistics

13. M.A. Alexander , Generations and Business Cycles - Part I, Safehaven, November 6, 2002.

14. Simon Kuznets, Secular Movements in Production and Prices, 1930.

15. Homer Hoyt, The Urban Real Estate Cycle--Performances and Prospects, Urban Land Institute Technical Bulletin No. 38, June 1960, in According to Hoyt: 53 years of Homer Hoyt, no publisher, 1970

16. Strauss, William and Neil Howe, Generations: The History of America’s Future 1584 to 2069, New York: Quill William Morrow 1991.

17. Strauss, William and Neil Howe, The Fourth Turning, New York: Broadway Books, 1997.

18. Strauss and Howe, 1991, p 79.

19. Phillips, Kevin P., The Cousins' Wars: Religion, Politics, and the Triumph of Anglo-America, New York: HarperCollins, 2000.

20. McLoughlin, William Gerald, Revivals, Awakenings and Reform, Chicago: University of Chicago Press, 1978.

21. Carlin, David R., review of The Fourth Great Awakening and the Future of Egalitarianism, First Things, (www.firstthings.com/ftissues/ft0011/reviews/carlin.html)

22. The Kondratieff Wave, KondratieffWinter.com

23. Goldstein, Joshua S. Long Cycles, New Haven: Yale University Press, 1988.

24. Schlesinger, Arthur, Paths to the Present, New York: Macmillan Co., 1949.

25. L. Patrick Hughes, Cycles of American History, July 2000.