Generations and Business Cycles Part II
The generational interpretation of the post-depression era
The generational model holds that the K-cycle has shifted from one-half to a full saeculum in length as a result of industrialization and is now about 72 years long. The cause of this lengthening is the emergence of government economic management, which itself is a direct effect of industrialization as mediated through the generational saeculum cycle. The rise of the industrial economy did more than simply introduce the Kitchen cycle. It also increased the intensity of the generationally-related Kuznets, Stock, and Kondratiev cycles, all of which had already been part of the pre-industrial economy.
Thus, while the Kuznets-related Panic of 1819 was the first panic to make it into the history books, it was a pretty mild bear market. The Panic of 1837 was worse and the one in 1857 worse yet. The Panic of 1873 ushered in the second worst bear market of all time. The depression following the Panic of 1893 was the worst up to that time. This depression was the first to take place with a majority of the population involved in non-agricultural occupations. Although hard times on the farm were a frequent occurrence, depressions did not usually mean hunger. Yet for the large numbers of urban workers thrown onto "the industrial scrap heap" the depression of the 1890's produced a level of suffering unprecedented for a business fluctuation.
Despite the severity of the depression, little government action was taken. President Cleveland resisted pressure to provide relief with expanded public works programs, as did many in Congress. Senator James Berry of Arkansas, voicing the dominant mind-set, declared that "It is not the purpose of this government to give work to individuals throughout the United States by appropriating money which belongs to other people and does not belong to the Senate." Power was still held by the Gilded generation (born 1822-42), the last to come of age in a pre-industrial America. This generation was still steeped in the Jeffersonian concept of America as a nation of sturdy yeoman individualists, who reject the sort of collectivist politics then popular in Europe.
Yet the very next panic in 1907, although considerably milder in its effects, brought reform in the guise of the Federal Reserve System. The Progressive generation (born 1843-59) then in power had come of age in the midst of industrialization, and was more cognizant of the new realities of industrial America. The Progressive reforms occurred during one of the recurrent periods of political liberalism, and were followed by a speculative conservative era. As it turned out, they were not enough to prevent another panic and depression in 1929-33, this one by far the worst of them all. This panic triggered the Crisis era of the Strauss and Howe saeculum, (and another liberal era) the first time a business cycle downturn had done so. Crisis eras produce massive changes in society, and this one was no different. The generation in power, the Missionary (b 1860-82), was the first completely industrial generation. The response they crafted to the problems of the Depression (Keynesian economics) and the dictators it produced (expansion in size of government) completely transformed the role of government in the economy, diminishing the Kuznets real estate cycle and producing a lengthened Kondratiev cycle now fully aligned with the generational saeculum cycle.
The effectiveness of the crisis solutions reached its apogee during the period after crisis, called the High, which ran from 1946 to 1964. The High was a conservative era, yet the liberal policies (e.g. high tax rates) introduced in the Crisis were retained. These served to dampen speculative juices. As a result the post war High was free of anything resembling a panic and the Kuznets cycle was muted. Business cycles continued to be aligned with periodic bear markets as they had been before 1933, except the period between them had increased to four years, in accordance with the SMECT model. According to the generational model, the fiscally conservative policies and financial self restraint of the 1950's is a natural consequence of the generational peer-personalities of the adult generations: cautious, risk-adverse Nomads in elderhood, team-playing, disciplined Heroes in mid-life, and compliant Artists in rising adulthood.
By the late 1960's "self-expression" and the rise of the individualistic ethic (coming with the arrival of a new generation of Prophets) encouraged deficit spending, tax revolts and speculation. Strauss and Howe call such periods Awakenings. They are times when "people stop believing that social progress requires social discipline" and develop "a high tolerance for risk-prone lifestyles". People disdain the prosperity and security of a High, though covertly they are taken for granted. The development of the Awakening mirrored a political shift from conservative to liberal as well as an erosion in financial rectitude. Taxes were cut in the early 1960's and the decision to finance the Vietnam war through deficit-spending rather than increased taxes contributed to an inflationary environment that would make speculation more attractive in the 1970's. War stimulus was probably responsible for extension the 1960's expansion beyond 1966, when it should have ended according to SMECT.
Kondratiev spring ended in 1966, just two years after the beginning of the Awakening, as indicated by the beginning of a secular bear market. Yet economic good times continued until 1973. The credit crunch in 1966 failed to bring on a recession; economic conditions remained favorable for the first half of Kondratiev summer. On the other hand, the 1966 bear market signaled a shift to a highly speculative era seeing first the "go-go years" of the late 1960's and the "nifty-fifty" era of the early 1970's. This same period saw the re-emergence of a full-blown Kuznets cycle. The persistence of good times over the 1966-1973 period reflected the efforts of government policy makers to maintain full employment using Keynesian stimulus in the face of Kondratiev summer. It worked for a while, but the inevitable result was stagflation. The experience of stagflation did not lead to an abandonment of government economic management, but rather to a change in style, from liberal Keynesian fiscal methods to conservative supply-side monetary methods. This shift mirrored a political shift from liberal to conservative. This shift was called the Reagan revolution, but another interpretation of this era is what Strauss and Howe call an Unraveling. The Unraveling "begins as a society-wide embrace of the liberating cultural forces set loose by the Awakening". People "vigorously assert an ethos of pragmatism, self-reliance, laissez faire, and national (or sectional or ethnic) chauvinism".
In the early 1980's, the Federal Reserve, under its new monetarist focus, hiked interest rates to their highest level in the history of the United State, breaking the back of inflation. Taxes were cut substantially again, especially those for capital gains, and as inflation subsided, a great bull market in stocks ensued. This policy had the same effect as that during the early 1920's when the decline of WW I inflation, a capital gains tax cut, and a belief that the Fed had made financial panics a thing of the past unleashed speculative forces. The 1920-29 boom is cycle-equivalent to the 1982-2000 boom. The speculative spirit reborn in the Awakening became sanctioned by official policy during the Unraveling. For example, the 1987 stock market crash was met by copious liquidity, which allowed the market to recover its losses in just two years, resulting in extension of the already-long 1980's expansion to the next Kitchen cycle. In 1997, Congress passed a capital gains tax cut in order to stimulate further stock price rises in an already overvalued market, and the Fed executed a well-timed surprise rate cut in 1998, which ignited an explosive stock rise, even more extreme than the late 1920's blow-off. Both of these policies served to extend the already long 1990's business cycle to three Kitchens, and as a result Kondratiev fall was extended well beyond its length in previous cycles. The net result of these various interventions has been to subtly increase the length of the K-cycle from the 64-year length called for by SMECT to a 72 year length fully aligned with the saeculum.
SMECT does a good job of explaining and correlating the cycles over the 1896-1982 period, but it breaks down for the most recent Kuznets cycle from 1982 to 1998. The 1998 event (cycle-equivalent to 1932 in the SMECT scheme) was very mild--no recession and a bear market milder than any other over this period. In particular, the present bear market, which reflects simply a Kitchen cycle, should be milder than what was seen in 1998--but clearly this is not the case. An argument can be made that policymaker intervention has shifted the effects of Kondratiev winter forward from 1998 (when they should have occurred) to now. No such intervention occurred for the last cycle and so the 1929-1932 equivalent decline is playing out one Kitchen cycle later than it normally would in the absence of policymaker interventions. This argument is a "cycle lengthening" argument, but SMECT already accounts for cycle lengthening though the mechanism of Kitchen cycle expansion. Further delays by policymakers should already be part of any model for the post-1933 economy, when such interventions are now normal. Thus, SMECT's failure to account for the extension of the boom beyond 1998 is a problem.
Another feature of the SMECT model not yet discussed is that the lower order cycles are necessarily in phase with the longer cycles. That is, a BAAC Supercycle turning point (like 1998) is necessarily also a turning point in all the lower-order cycles (Kuznets, Juglar and Kitchen). The empirically-derived cycles shown in Table 6 do not support this alignment. Of course, the SMECT cycles do not refer to the economic phenomenon associated with the cycles of the same name, they are simply names used to classify cycles composed of geometric multiples of Kitchen cycles--which do have a mechanistic interpretation in SMECT (they follow the four-year electoral cycle). On the other hand, the higher order cycles do have meaning. Sixteen Kitchens are supposed to define a K-cycle, which can be independently assessed by long-term trends in monetary variables like interest rates Eight Kitchens are supposed to define a BAAC supercycle and four a secular stock market trend, both of which can be independently assessed using measures like Tobin's Q. The only cycle in SMECT which does not have a specific independent meaning is the Juglar cycle. The Juglar in SMECT does not correspond with the longer category of business cycles since 1933 (see Table 6). The Juglar troughs in 1998, 1966 and 1942 fall into the middle of the 1990's, 1960's and WW II business cycles rather than align with their troughs (see figure).
As an alternative to SMECT, I present the generational scheme shown in Table 8. The same four-year Kitchen cycles found in SMECT are retained. However no fixed connection between these cycles and the longer ones is assumed. Although each longer cycle will necessarily contain a whole number of Kitchen cycles this number is not necessarily constant. The organizing principle behind these cycles is Strauss and Howe's generational cycle, or more specifically the political manifestation of these cycles. What seems to have happened initially after 1933 was expansion of K-cycle length to ~64 years according to the Kitchen cycle expansion concept from the SMECT model. The mechanism for this expansion was a rising impact of politics on economics. That is, economic cycles began to map onto political cycles. The most obvious political cycle is the four year electoral cycle, and so the four year cycle in the stock market appeared almost immediately.
Table 8. Comparison of SMECT with generational scheme for the regulated era
|Cycle||SMECT (w/o generations)||Generational scheme|
|Bull/Bear market cycle||4 years (same as Kitchen)||4 yrs (same as Kitchen)|
|Business cycle||Kitchen or Juglar||Kitchen or Juglar|
|Kitchen||4 years (electorally-based short business cycle)||4 years (same as SMECT)|
|Juglar||8 years (2 Kitchens)||9 yrs avg (long business cycle, 1/2 K-season, usually 2 Kitchens)|
|Kuznets cycle||16 years (4 Kitchens)||18 yrs--not same as K-season|
|Kondratiev season||Same as Kuznets cycle||18 years--1 psychological generation|
|Supercycle/Stock Cycle||32 years (1/2 of K-cycle)||36 years--2 psychological generations|
|Kondratiev Cycle/Saeculum||64 years--no relation to saeculum||72 years--4 psychological generations|
More subtle was the effect of the generationally-influenced liberal-conservative cycle (also known as the Schlesinger cycle) which influenced the nature of government economic management. During the conservative era of the 1950's the emphasis was on the tenets of fiscal conservatism, balanced budgets and financial probity, even at the cost of maintaining the high tax rates from WW II. Although conservative politics tends to favor wealth creation, the style of management during the High was not conducive to speculation, reflecting the generational dynamics of the High. The secular bull market from 1949-1966 ended at the lowest valuation level of any secular bull market, showing the depressed "animal spirits" of the time. As the High gave way to the Awakening, things "loosened up" financially, taxes were reduced and not raised when the nation went to war, and speculative forces began to stir. By the time of the next conservative, pro-wealth creation era in the 1980's, a free-wheeling, pro-speculation style of economic management was in vogue, reflecting the generational dynamics of the Unraveling. The secular bull market peak associated with this era had the highest valuation level in history, showing extremely powerful animal spirits--exactly the reverse of the previous secular bull market peak.
The timing of this process is set by generational dynamics--as mediated through politics on the economy, and not by economic forces themselves. Hence the fundamental large-scale economic cycle, the Kondratiev, and its subharmonics, the Stock Cycle and K-season, adapted to reflect the length of the saeculum, which itself changed to reflect a psychological driver for the cycle. (The classical K-cycle and saeculum had been related to biological generations.). The key timing element of the longer cycles today is the 18 year psychological generation and the 18-year political trends associated with them. This timing determines the length of the secular market trends and the modern business cycle, whereas the short stock market cycles reflect the electoral cycle. Standing aside is the Kuznets cycle, which appears to have retained its traditional 18 year length.
The Kuznets cycle is not aligned with the K-cycle. Although of the same average length, it is not the same as the Kondratiev season as it is in SMECT (see Table 6). The two cycles are not in phase. Thus, although Kondratiev season changes can exert a strong effect on downturn severity, so can the unaligned Kuznets cycle. Hence, the present recession, although "K-cycle equivalent" to 1932, is not as severe as even the mild 1990 recession because it lacks a real-estate (Kuznets) cycle downturn.
One might point out that current strength in real estate reflects the 11 rate cuts made by the Federal Reserve since the beginning of 2001. But these 11 rate cuts have done nothing to prevent the stock market from falling 35% since then. The stock market fell because it had gotten too overvalued, that is, it had reached its peak in the Stock Cycle and thus had to start coming down, regardless of the what the Fed did. On the other hand, real estate valuation has not reached extreme levels as shown by the ratios of price to income in Figure 5. That is, the Kuznets peak has yet to be reached. Thus, rate cuts have supported real estate, just as they did for stocks in 1998. By 2004 we may well have reached a Kuznets peak and then one would expect rate cuts by the Fed to be ineffective in preventing a real estate bust in the next recession. This interpretation suggests that the current downturn is largely stock market driven, being the result of the inevitable downturn following a peak in long-term stock market valuation (P/R). Indeed, current weakness almost entirely comes from weak business investment spending, reflecting the poor business outlook generated by continuing weakness in the stock market. Consumer demand has remained strong.
Thus, if the cycle scheme suggested here is correct, no double-dip recession is to be expected and the stock market may well rally over the next few years, in accordance with P/R valuation.
1. Clement Juglar, Des Crises commerciales et leur retour periodique en France, en Angleterre, et aux Etats-Unis, 1862.
2. Joseph Kitchen, "Cycles and Trends in Economic Policy," Review of Economic Statistics, Jan. 1923.
3. Simon Kuznets, Secular Movements in Production and Prices, 1930.
4. Mason Gaffney, "Privatizing Land Without Giveaway", Delivered at Conference on Social Collection of Rent in the Soviet Union, New York City, August 22-24, 1990.
5. 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
6. David McFadden, "Demographics, the housing market, and the welfare of the elderly", in David A. Wise (ed) Studies in the Economics of Aging, Chicago: University of Chicago Press, 1994.
7. Before 1967 median wages are used as a proxy for household income.
|Dec 1802||Oct 1805||3.2||19||Nov 1919||Dec 1920||3.4||14|
|Dec 1809||Aug 1812||1.9||24||Mar 1923||Oct 1923||0.65||40|
|Dec 1813||Sep 1816||3.1||20||Feb 1926||Mar 19261||0.12||53|
|June 1818||Jul 1819||0.9||38||Sep 1929||Nov 1929||0.66||39|
|Apr 1825||Jul 1829||3.4||16||Apr 1930||Jul 1932||22.8||1|
|Aug 1835||Jun 1837||4.0||12||Feb 1934||Jul 1934||0.63||42|
|Sep 1838||Jan 1843||10.2||3||Mar 1937||Mar 1938||3.7||13|
|Dec 1845||Jan 1848||0.9||37||Nov 1938||Apr 1942||9.3||4|
|Aug 1847||Nov 1848||1.7||28||May 1946||Oct 1946||0.63||43|
|Dec 1852||Jan 1855||3.4||17||Jun 1948||Jun 1949||1.45||30|
|Jul 1855||Oct 1857||7.8||7||Jan 1953||Sep 1953||0.58||46|
|Mar 1858||Jul 1859||1.4||31||Mar 1956||Oct 1957||1.87||25|
|Oct 1860||May 1861||0.9||36||Jan 1960||Oct 1960||0.61||44|
|Apr 1864||Jun 1865||1.8||26||Dec 1961||Jun 1962||0.97||34|
|Aug 1869||Dec 18691||0.17||51||Feb 1966||Oct 1966||0.99||33|
|Feb 1873||Jun 1877||13.9||2||Dec 1968||May 1970||3.4||15|
|Jun 1881||Jan 1885||8.2||6||Jan 1973||Oct 1974||6.2||11|
|May 1887||Jun 1888||0.92||35||Sep 1976||Mar 1978||1.72||27|
|May 1890||Dec 1890||0.61||45||Feb 1980||Mar 19801||0.15||52|
|Aug 1892||Aug 1893||1.7||29||Nov 1980||Aug 1982||2.82||23|
|Sep 1895||Aug 1896||1.12||32||Oct 1983||Jul 1984||0.65||41|
|Apr 1899||Sep 1900||2.9||21||Aug 1987||Oct 1987||0.35||48|
|Jun 1901||Nov 1903||7.7||8||Jul 1990||Oct 1990||0.28||50|
|Jan 1906||Nov 1907||6.3||10||Jan 1994||Dec 19942||0.39||47|
|Nov 1909||Sep 1911||3.3||18||Jul 1998||Oct 1998||0.29||49|
|Sep 1912||Dec 1914||6.6||9||Mar 2000||9.3||5|
|Nov 1916||Dec 1917||2.9||22|
1These bear markets are included only because they are associated with a recession.
2This bear market is included because it represents a major trend change (see Fig A.1)
|Kondratiev Season||Seasonal GDP Cycle1||Business Cycle1||Kondratiev Season||Seasonal GDP Cycle1||Business Cycle1|
|Spring||1787-1802||1787-1792 (E)||Spring||1897-1910||Jun 1897-Jun 1899 (E)|
|(expansionary)||1792-1793 (R)||(expansionary)||Jun 1899-Dec 1900 (R)|
|1793-1797 (E)||Dec 1900-Sep 1902 (E)|
|1797-1798 (R)||Sep 1902-Aug 1904 (R)|
|1798-1802(E)||Aug 1904-May 1907 (E)|
|Summer||1802-1818||1802-1805 (R)||May 1907-Jun 1908 (R)|
|(recessionary)||1805-1806 (E)||Jun 1908-Jan 1910 (E)|
|1806-1809 (R)||Summer||1910-1921||Jan 1910-Jan 1912 (R)|
|1809-1810 (E)||(recessionary)||Jan 1912-Jan 1913 (E)|
|1810-1814 (R)||Jan 1913-Dec 1914 (R)|
|1814-1816 (E)||Dec 1914-Aug 1918 (E)|
|1816-1818 (R)||Aug 1918-Mar 1919 (R)|
|Falll||1818-1836||1818-1819 (E)||Mar 1919-Jan 1920 (E)|
|(expansionary)||1819-1821 (R)||Jan 1920-Jul 1921 (R)|
|1821-1826 (E)||Fall||1921-1929||Jul 1921-May 1923 (E)|
|1826-1829 (R)||(expansionary)||May 1923-Jul 1924 (R)|
|1829-1836 (E)||Jul 1924-Oct 1926 (E)|
|Winter||1836-1843||1836-1838 (R)||Oct 1926-Nov 1927 (R)|
|(recessionary)||1838-1839 (E)||Nov 1927-Aug 1929 (E)|
|1839-1843 (R)||Winter||1929-1949||Aug 1929 -Apr 1933 (R)|
|Spring||1843-1853||1843-1848 (E)||(recessionary)||Apr 1933-May 1937 (E)|
|(expansionary)||1848-1849 (R)||May 1937-Jun 1938 (R)|
|1849-1853 (E)||Jun 1938-Feb 1945 (E)|
|Summer||1853-1867||1853-Dec 1854 (R)||Feb 1945-Oct 1945 (R)|
|(recessionary)||Dec 1854-Jun 1857 (E)||Oct 1945-Nov 1948 (E)|
|Jun 1857-Dec 1858 (R)||Nov 1948-Oct 1949 (R)|
|Dec 1858-Oct 1860 (E)||Spring||1949-1968||Oct 1949-Jul 1953 (E)|
|Oct 1860-Jun 1861 (R)||(expansionary)||Jul 1953-May 1954 (R)|
|Jun 1861-Apr 1865 (E)||May 1954-Aug 1957 (E)|
|Apr 1865- Dec 1867 (R)||Aug 1957-Apr 1958 (R)|
|Fall||1867-1882||Dec 1867-Jun 1869 (E)||Apr 1958-Apr 1960 (E )|
|(expansionary)||Jun 1869-Dec 1870 (R)||Apr 1960-Feb 1961 (R)|
|Dec 1870-Oct 1873 (E)||Feb 1961-Dec 1969 (E)|
|Oct 1873-Mar 1879 (R)||Summer||1969-1982||Dec 1969-Nov 1970 (R)|
|Mar 1879-Mar 1882 (E)||(recessionary)||Nov 1970-Nov 1973 (E)|
|Winter||1882-1897||Mar 1882-May 1885 (R)||Nov 1973-Mar 1975 (R)|
|(recessionary)||May 1885-May 1887 (E)||Mar 1975-Jan 1980 (E)|
|Mar 1887-Apr 1888 (R)||Jan 1980-Jul 1980 (R)|
|Apr 1888-Jul 1890 (E)||Jul 1980-Jul 1981 (E)|
|Jul 1890-Apr 1891 (R)||Jul 1981-Nov 1982 (R)|
|Apr 1891-Dec 1893 (E)||Fall||1982-2001||Nov 1982-Jul 1990 (E)|
|Dec 1893-Jun 1894 (R)||expansionary)||Jul 1990-Mar 1991 (R)|
|Jun 1894-Dec 1895 (E)||Mar 1991-Mar 2001 (E)|
|Dec 1895-Jun 1897 (R)||Winter||2001-||Mar 2001- (R)|
1 Years in red denote Kondratiev peaks, those in blue, Kondratiev troughs
Bear markets are usually considered "large" declines in the stock index over a fairly lengthy period of time. A common rule of thumb is any decline of 20% or greater. The market analyst Robert Bronson has developed novel measure of bear market severity which I will make use of here. This method accounts for both the magnitude and the length of the bear market decline and is defined as follows:
A.1 Severity = log(index top/index bottom) x duration in months
For example, the current bear market, which began at the S&P500 intra-day peak of 1552.87 on March 24, 2000 has reached (so far) an intra-day low of 768.63 on October 10, 2002. Through the October low, the bear market has lasted 30.5 months. From this data I can calculate a severity as follows:
A.2 Severity = log(1552.87/768.63) x 30.5 = 9.3.
Bronson has listed thirty bear markets since 1895 as assessed by this criterion. Already the present bear market falls into the #2 position in his list, and it may not yet be over. The nature of this measure can make fairly small declines significant bear markets--if they last long enough. For example, the short, but severe 1987 bear market showed a decline 150% greater than the much longer 1960 bear market yet was 40% less severe by the Bronson criterion. This makes identification of major bear markets more difficult than the use of the simple 20% rule. Some of the bear markets in Bronson's list seem pretty small. For example the 1926 bear market has a severity of only 0.12. One can find declines not on his list with greater severity. For example, the 12% Dow decline from July 15, 1943 to November 30, 1943 (severity = 0.25); the 10.8% S&P500 decline from Sept 25, 1967 to March 5, 1968 (severity = 0.26); or the 13.9% S&P500 decline from April 29, 1971 to November 24, 1971 (severity = 0.45).
None of these declines fits the 20% standard, but they all are more severe (by the Bronson definition) than the 1926 bear market, yet are not included in the list of top bear markets. If we focus more closely on the 1926 bear market, we see that it was associated with an NBER recession dated from October 1926 to November 1927. It is reasonable to assign special significance to a market decline that is associated with an economic decline. None of the declines mentioned above were associated with recessions. If we look only at non-recession bear markets that did not show 20% declines, we see that only one of Bronson's bear markets (1994) is of similar magnitude to those mentioned above.
Figure A.1, which shows a plot of index value relative to ten-year trend, has 1994 displayed as a prominent low, reflecting the sea change in stock returns that occurred at this point. None of the declines mentioned above feature prominently in Figure A.1. If we exclude 1994 on the basis of this significance, the rest of the bear markets have severity greater than 0.5 or show a 20% decline. From these observations the following "rule" for bear markets was developed:
"A bear market is any decline of 20% or greater, or a severity of 0.5 or greater. Declines smaller than 20% and with severity between 0.1 and 0.5 are also considered bear markets if they occurred at the same time or just before an NBER recession, or form a major low on a plot of stock index versus 10-year trend."
All of the bear markets listed by Bronson fall into these criteria. None of those I listed earlier do, and thus should not be included in the list. There is one more issue with which I need to deal before a complete methodology for defining bear markets has been defined. One of Bronson's declines, the 1929 bear market, is part of a larger decline from 1929 to 1932--yet it is considered a separate bear market. Apparently, it seems the rally from Dow 195.4 on 11/13/29 to 297.3 on 4/16/30 was "big enough" to be considered a separate bull market. I can calculate a "severity" for this bull market as follows:
A.3 Bull market size = log (297.3/195.4) x 5.0 months = 0.91
A rally with severity of 0.9 or better, it would seem, constitutes a bull market. In comparison, I note the recent rally from S&P500 944.75 on September 21, 2001 to 1176.97 on January 7, 2002 scores an intensity of only 0.34:
A.4 Bull market size = log (1176.97/944.75) x 3.6 months = 0.34
This is much smaller than the 1929-30 rally and is consistent with the general sense that the 2001-2002 rally was "just" another bear market rally and not a bull market.
The market nearly reached the January top on March 11. If I use this later date for the end of the rally I obtain a severity of 0.53, still well below the size of the 1929-30 rally. Nothing is changed. However, suppose the market had reached a peak of 1373 or higher on March 11, instead of forming a double top? In this case, the rally would have been as large as the 1929-1930 rise. Would such a large rally have been considered a bull market? It is impossible to know, but it seems reasonable to me that the vast majority of market participants would have hailed a rise of this magnitude (before it peaked) as a new bull market.
After this discussion one can add the definition of intervening bull market as a rise in a larger decline that is 0.9 or larger. With this, our rules for locating bear markets are complete.
 The ten year trend is obtained by regression of log(index) versus time over the ten year period centered on the point of interest. Thus the trend for a point in 1990 is regression of data over 1985 to 1995. For points after 1997 or before 1807, the regression equation for the 1992-2002 and 1802-1812 period is used.
The National Bureau of Economic Research (NBER) has identified US business cycles back to 1854. Prior to that one can use the GDP relative to its trend as a crude estimate for cycles. I already used this approach with a ten year trend to help identify stock bear markets in Figure A.1. This time I use a 25 year trend which is shown in Figure B.1 for data between 1790 and 1890. Also shown in the figure as the thick light-gray line is the Kondratiev seasonal cycle. This cycle is a binary subharmonic of the Kondratiev cycle (two seasonal cycles per K-cycle) and is correlated with the stock cycle. It is the same cycle that Bronson calls the BAAC Supercycle. I used a plot analogous to Figure B.1, but using a 100 year trend, to show this cycle in my book The Kondratiev Cycle (p 67).
Figure B.1 GDP relative to 25-yr trend showing K-seasonal and ordinary business
Within the larger season cycles, there are smaller perturbations, which presumably reflect ordinary business cycles. In The Kondratiev Cycle I smoothed these to de-emphasize them and bring out the longer cycle. Now I wish to examine these shorter cycles. Recession troughs are marked by visual inspection. Four troughs (marked in red italics) are the recessions associated with the four big "panics" that occurred during the period shown in the chart: the panics of 1819, 1837, 1857 and 1873. The others are either "major" troughs or isolated ones. Figure B.2 shows more cycles for the period 1880-1950.
Figure B.2. K-seasonal and ordinary business cycles 1880-1950
To test the validity of these troughs I compare them with those obtained by NBER. Table B.1 shows troughs from Figures B.1 and B.2 compared to the NBER recession bottoms. The correspondence is quite good. All but three NBER recessions have a corresponding trough and only one trough doesn't have an associated NBER recession. This suggests that the business cycles obtained from Figure B.1 can be used to provide estimates for business cycle dates before 1854.
Table B.1 GDP troughs from Figures B.1 and B.2 compared to NBER recession
|Dec 1854||--||Jun 1894||1894||Mar 1919||1917|
|Dec 1858||1857||Jun 1897||1897||Jul 1921||1921|
|Jun 1861||--||--||1899||Jul 1924||1924|
|Dec 1867||1865||Dec 1900||1901||Nov 1927||1928|
|Dec 1870||1871||Aug 1904||1904||Apr 1933||1933|
|Mar 1879||1876||Jun 1908||1908||Jun 1938||1938|
|May 1885||1886||Jan 1912||1911||Oct 1945||--|
|Apr 1888||1889||Dec 1914||1915||Oct 1949||1949|