Kondratiev-Zyklen

verfasst von Ricardo, 23.03.2002, 13:09

Check out my new book: Stock Cycles: Why stocks won't beat money markets over the next twenty years

The Kondratieff cycle (or long wave) theory holds that capitalistic economies grow in a cyclic fashion with a average periodicity of about 53 years. This cycle is shown most clearly by the behavior of prices (or inflation) and interest rates which rise and fall over time. The maximum in prices or inflation occurs at the Kondratieff peak (K-peak) and the minimum, at the Kondratieff trough (K-trough). The period of rising prices (inflation) leading up to the K-peak is called the upwave. The converse is the downwave.
In this webpage, I will attempt to put specific dates on previous cycles both in the US and in Great Britain. The strategy taken will be to examine the history of prices and, more recently, inflation rates to determine long term maxima and minima, which are then potential K-peaks and K-troughs. I will then use a number of arguments to determine which of these truly are Kondratieff points. A great many of the K-peaks show up during wartime. Since government fiscal policy during most wars is highly stimulatory, high inflation is to be expected at these times. The Kondratieff cycle purports that oscillations in inflation arise naturally. Thus, inflationary behavior should arise when stimulatory policies are absent. Hence, the trends in prices during times of no stimulation (e.g. during times of basically balanced budgets or surpluses) should give evidence of the underlying trends in place at various times. For example a price peak that occurs during a war between two periods of non-stimulated rising prices is probably due to stimulation and is not a K-peak. On the other hand, a price peak that occurs during a war between a period of non-stimulated rising prices and falling prices probably is a K-peak. That is, the peak in prices would have occurred at about the same time even if the war hadn't happened.

The Kondratieff cycle for the US

Figure 1 shows the US price/inflation data. For the period 1720 to 1934 the line in the figure presents the level of producer prices. Producer prices were selected because data exists for them further back than it does for consumer prices. Also in an agricultural/industrial economy, producer prices are more relevant. After 1934, the US economy has experienced secular inflation with very little price deflation. A plot of prices would show a monotonic rise. Thus, for the years after 1934 the annual rate of inflation in consumer prices is plotted. The plotted inflation rate was smoothed by summing twice the rate for the current year with those of the year before and the year after and then dividing by four. Consumer price inflation was selected because after the 1930's the economy became increasingly dominated by services for which consumer prices are more relevant. The price and inflation plots were spliced together and the scales of each adjusted to give a similar degree of fluctuation throughout the entire period. Interpretation of the figure is simple. Peaks represent maximum values in prices (before 1934) or inflation rate (after 1934). Similarly, troughs represent minimum values in prices or inflation. Prior to 1934 the rising portions of the graph represent inflation and the falling portions deflation. After 1934, the rising portions represent accelerating prices (rising inflation) whereas the falling sections represent decelerating prices (disinflation).
Six price peaks are labeled in the figure, five of them marking Kondratieff peaks. For each of the K-peaks one can see rising prices before the peak and falling ones afterward that happened during periods of peace. This is what makes them K-peaks. Note that the 1947 inflation peak associated with World War II inflation is not considered a Kondratieff peak. Prices in the 1920's and early 1930's were falling. Inflation in the 1950's and 1960's was rising. If 1947 is to be a K-peak there would have to be two K-troughs in the period between 1932 and 1954. This would give an entire K-cycle only 22 years long at most.
Don Roper advances the hypothesis that what makes the long wave isn't the inflationary episode per se, but rather the response of the monetary system following the inflation. During wartime, monetary restraint is relaxed in order to meet the national emergency. Afterward high real interest rates drive up the value of money to restore pre-war levels of monetary rectitude. Is is these high interest rates that cause the downwave to begin, a process that typically takes decades to be completed. Figure 2 shows real interest rates for US government bonds for the period around five price peaks in Figure 1. In all cases except 1947, real interest rates rose substantially after the peak. The average real interest rates for the eight years after the peak were 10.3% for 1814, 6.6% for 1864, 4.3% for 1919 and 5.8% for 1980. These rates are high relative to the average value of 3.1% over the entire period since 1800. In contrast, the average real rate after 1947 was only 0.4%, much lower than the long-term average. In the first four cases, there was a significant amount of monetary tightening. In the last case there was not. Clearly, 1947 is not a Kondratieff peak according to Roper's definition.
Next we consider the seven troughs in Figure 1. The 1745, 1787, 1843 and 1897 troughs are generally considered to be K-troughs. The next trough is more problematic. It would seem that 1932 is the obvious candidate. It was a bona fide price trough and inflation was nearly -10%. There are also troughs in 1939, 1949 and 1954. Schumpeter favored the first of these, denoting the K-trough as 1940. Brian Berry favors 1954, which appears natural enough in the smoothed trends shown in Figure 1. Figure 3 shows running 12-month inflation rates for the period 1947-1959. This figure shows that inflation was lower in 1949 than in 1954 supporting the choice of 1949.
The price structure of the 1932-1954 period is complicated by the massive governmental stimulus of the economy produced by the New Deal and World War II, which produced inflation during what would ordinarily be the downwave. If we look at those (few) years when the government ran small deficits or even a surplus, we can get an idea of what the nonstimulated economy was doing. The US government ran the smallest deficit of the New Deal era in 1938 at 1.3% of revenue. Prices fell 2% in that year, suggesting that the downwave was still in operation. This observation rules out 1932 as a K-trough and makes 1939 the earliest candidate for K-trough. This observation provides another line of argument for 1947 not to be a K-peak. Upwaves typically take more than seven years to complete, and the massive fiscal stimulus during World War II strongly suggests that the 1947 peak was just a war-induced peak during what would otherwise have been a downwave.
The government ran a surplus from 1947 to 1949, during which time inflation collapsed and deflation appeared (see Figure 3 ). As discussed above, this collapse in inflation occurred without the need for high real interest rates to reverse inflationary tendencies. Simply the removal of stimulus did the job. This strongly suggests that a deflationary tendency (i.e. a downwave) was still in operation in these years. The government ran a basically balanced budget from 1954-58 (0.8% surplus) indicating little fiscal stimulus during this period. Figure 3 shows that during this time of non-stimulation, inflation and interest rates rose, implying we had entered an upwave. The evidence of both inflation and interest rates is consistent with a rising trend in prices starting around 1949.
So far we have established a deflationary trend in the late 1930's and a disinflationary trend in the late 1940's and an inflationary trend in the 1950's. The simplest interpretation of this data is that the downwave lasted until 1949 and then a new upwave began, making 1949 the K-trough. The rate of inflation reached a lower low in 1949 than it did in 1954. Also, the stock market began a 17 year secular bull market in that year as well. The stock market began 10 year secular bull markets in 1842 and 1896, very close to the 1843 and 1897 K-troughs. All told, the preponderance of the evidence is in favor of a 1949 date for the most recent K-trough.
The five US troughs have an average spacing of 51 years. The five peaks have an average spacing of 50 years. All together, the average cycle length for the US is 51 years.

The Kondratieff Cycle for Great Britain after 1700

Figure 4 shows a price index and nominal government interest rates for the British economy over the period 1700-1900. This figure shows that price peaks were commonly seen during the 18th century. Most of these occurred during a war. Prices and interest rates peaked in 1711, during the War of the Spanish Succession (1702-13). Government deficits reached 192% of income in 1711. Prices peaked in 1741 and interest rates peaked in 1748, both during the War of the Austrian Succession (1739-48). Government deficits reached 67% of revenues in 1749. The Seven Years War (1756-63) gave a price peak in 1757 and an interest rate peak in 1762. Deficit spending peaked at 120% of revenues in 1761. The American Revolution (1776-1783) led to an interest rate peak in 1782. Deficit spending peaked at 112% of revenues in that same year. Interestingly, there was no price peak during this war, prices peaked in 1774, before the war. Finally, there was a price peak in 1801 and an interest rate peak in 1798, closely associated with 170% deficit spending in 1797, resulting from expenditure for the first Napoleanic War (1793-97). These deficits may seem excessively high when compared to all the concern about deficits in the Reagan-Bush years, which peaked at 35% of revenues in 1983. During wartime deficits are much larger: US deficits reached 209% of revenues in 1943 and a whopping 350% of revenues in 1918.
Massive borrowing can lead to increases in the money supply and inflation so the presence of price peaks during wars is not unexpected. As discussed above, not all war-induced peaks are Kondratieff peaks (e.g. WW II). Also price peaks not associated with wars can occur as well (e.g. 1980). It is likely that not all the peaks shown in Figure 4 are Kondratieff peaks. The question is, which ones are?
If the Kondratieff cycle is to reflect some underlying economic reality, the price peaks should show up in the absence of governmental stimulation during wars. We note that US prices were rising well before the peak-related wars (see Figure 1 ). Prices rose substantially from the 1787 low to 1812, before the wartime stimulus. Similarly prices rose from 1843 to 1860 and from 1897 to 1915. To establish which peaks truly are Kondratieff peaks we should examine price trends during those periods that were not affected by wartime spending.
Since the history of Britain during the 18th century is so replete with wars there are not lengthy periods of peace like in the US. Fortunately, we have the record of deficit spending. If we examine the price trend during those (relatively few) years during which the British government ran surpluses we may gain some insight into price movement in the absence of distortions. The period from 1717 to 1734 was the longest period of surpluses. The price trend during this time was downward, culminating in a minimum in 1744. Leaving out the spike associated with the collapse of the South Sea bubble in 1720, interest rates also trended down about 1.25% during this period. It would seem likely that the 1711 peak was a Kondratieff peak and the 1711-1744 period was the downwave. This suggests that the 1741 price peak was a war-induced peak during a downwave (not a K-peak), much the same as WW II.
The next period of surpluses was 1750-54. During this period prices rose about 5%, so we conclude that an upwave had started from the 1744 low, making it a K-trough. This trough is consistent with a K-trough in 1745 for the US (see Figure 1 ). The next period of surpluses was 1767 to 1775. Prices rose 9% from 1767 to a peak in 1774, suggesting that the upwave that started in 1744 continued at least until 1774. This means that the price peak in 1757 was a war-induced price/inflation peak during an upwave (not a K-peak) much like the Korean War inflation peak in 1953 (see Figure 3) or the Vietnam War price peak in 1969 (see Figure 1). Prices fell from 1774 to 1789 (Figure 4) despite the strong fiscal stimulus of the 1776-83 war, during which deficits averaged 80% of revenues (interest rates rose nearly 2% during the war). This implies a strong deflationary trend was in place after 1774. The rising trend up to 1774 in the absence of stimulation and the downward trend after 1774, in spite of stimulation, makes 1774 a K-peak. A 1774 peak for Great Britain is also consistent with the 1779 US K-peak. Finally we have the 1801 and 1812 peaks. Since the period from 1794 until 1815 was one of almost continuous warfare and fiscal stimulation, we choose the absolute price peak of 1812 for the next K-peak by default. By default, the 1789 trough becomes the K-trough. We note that a 1789 trough is consistent with the 1787 trough in the US and the 1812 peak is consistent with an 1814 peak in the US.
The 19th century is the inverse of the eighteenth. Instead of a confusion of peaks there are only two: 1812 and 1873. The price behavior in the decades after the Napoleanic wars clearly shows a downwave terminating in 1849. Similarly rising prices from 1857 to 1873, in the absence of war, show an upwave, making 1873 the next Kondratieff peak. Similar arguments also show 1897 as a K-trough, exactly in line with the US data.

The Kondratieff Cycle for Great Britain before 1700

Figure 5 shows the British price index for the five centuries previous to 1750. Major price peaks and price troughs are labeled in this figure. There was no war in 1557. Nor was there in 1650 (Cromwell's Anglo-Dutch trade wars started later in 1652-54). Elizabeth I (1558-1603) managed to avoid war during the early years of her reign, but by the 1580's England was engaged in hostilities against Spain. During the last years of her reign England was fighting the Irish. So the 1597 price peak did occur during a time of war. It is not clear that government stimulus was a factor this early in history. Borrowing moneys in the form of specie (gold and silver) is not inflationary (relative to specie) since no increase in the money supply (relative to the supply of specie) is produced. Although a great deal of inflation did occur during the sixteenth century, this likely reflects a fall in the value of specie relative to other commodities, caused by increased supply from the massive influx of silver from Spain's colonies in the new world.
The sort of monetary inflation seen in the 18th century probably wasn't a factor in the 16th century. For example, the military excursions of Henry VIII in 1513, 1522-3, 1528 and 1544 did not give rise to significant price spikes like the conflicts of the 18th century did. Hence is likely that the price peaks shown in Figure 5 reflect"natural" or Kondratieff price peaks. Since 1557, there have been 8 Kondratieff peaks in the British price series separated by 38 to 63 years with an average separation of 52 years. Since that time there have also been 7 troughs separated by 45-63 years, with an average of 54 years. Taking both together gives an average spacing of 53.3 years. Here we see some of the evidence used by others for a quasi-periodic longwave going back for centuries.
Going before 1500, I have labeled peaks in 1316, 1370, 1439 and 1483. The price peak in 1316 almost certainly reflects the great famine of 1315-16, caused by torrential rains in 1315 that destroyed much of the crop. The 1370 price peak reflects an inflation caused by a rise in wages, reflecting the shortage of workers following the Bubonic Plague (1348-50). The 1439 and 1483 peaks are not associated with any events with which I am familiar.

Characteristics of the Kondratieff Cycle Length

Thus far, we have looking to historical fluctuations in prices and interest rates that appear to repeat themselves in a cyclic fashion every 50 years or so. I have endeavored to put specific dates on the peaks and troughs of each cycle as far back as feasible. I have identified four cycles for the US and the UK economies since 1740 with what I believe to be a reasonable degree of accuracy. The four cycles line up fairly close to each other. Prior to 1740 I have been able to identify four and one half additional cycles that I could be reasonably sure weren't caused by an external stimulus. Prior to 1557 there are four more putative K-peaks, but no clear-cut K-troughs. I have identified an external cause for two of the peaks and have not been able to rule out external causes for the others so far. Hence, I have restricted my identification of Kondratieff cycles to just those since 1557. For these 11.5 cycles (7.5 British and 4 American) the average length is 52.4 years with a standard deviation of 8.2 years. Figure 6 shows the distribution for the 23 separate assessments (peak to peak, and trough to tough) of Kondratieff cycle length obtainable from these 11.5 cycles.
Although the cycle length is definitely centered on the"standard" cycle length of 53 years there is considerable variation in cycle length. Some workers prefer to use 1974 instead of 1980 as the most recent US Kondratieff peak since the distance from the 1919 peak of 55 years is much closer to the standard length of 53 years than is 61 years. Figure 6 shows that cycle lengths over 60 years occur one quarter of the time, and so are hardly uncommon. Another issue is whether the Kondratieff cycle changes in length over time. Figure 7 shows a plot of the British and American cycle lengths as a function of mid-point date. The cycle lengths are quite variable before 1600, but the dates used to calculate these cycle lengths are questionable as described above. The cycles since 1600, for which I have greater confidence as to the dates, show more consistency Both the US and UK cycles were unusually short right around the time of the industrial revolution. These cycle dates were determined carefully, as described earlier, so I have some confidence that the cycle length was really shorter then.
Since the industrial revolution, the average cycle length for both countries has been 54 years. If one constructs a composite cycle for both countries by averaging their cycle dates together, one obtains a fairly consistent cycle length of 53.5 years with a standard deviation of only 3.3 years for the period 1800 to 1925. This was approximately the period Kondratieff examined when he discovered the cycles that bear his name. The particularly narrow range of cycle lengths observed over this period has led some to advance the idea that the cycle length is fixed, or nearly so at about 53 or 54 years. A broader examination of the data, making use of the cycle behavior since Kondratieff's time and data from the 18th century, which was not available to him, shows that the cycle lengths are considerably more variable than the 1800-1925 period would suggest.

Economic character of the Kondratieff Cycle

An alternate way to characterize the K-cycle is to divide the cycle into to growth (or Kuznets) cycles, each consisting of an expansion and a recessionary portion. The expansion during the upwave can be called the inflationary growth period. The recessionary period might be called the stagflation period. The expansion during the downwave can be called the deflationary growth period and the recession can be termed the depression. These periods have also be given seasonal names, e.g. spring, summer, fall and winter.
Once the K-peaks and troughs had been determined, I simply picked the GDP minima closed to the K-trough to determine the beginning of the inflationary growth period and the peak in GDP prior to the K-peak to market the end of the inflationary growth period. I chose the GDP minima following the K-peak to denote the beginning of the deflationary growth period and a maxima in either GDP (1836, 1929) or in stocks (1881) to denote its end. There was a clear-cut GDP maxima in 1873, just before the Panic of 1873. But GDP was 36% higher in 1881 than in 1873, and grew very slowly after 1881. Since stocks peaked in 1881, just as the economy was slowing, I used the market as my indicator in this case. The stagflation and depression periods are the periods in between the growth periods. Average growth rates in GDP per capita, wages, prices and stock investments are given for each of these periods in Table 1.
The data in Table 1 are summarized in terms of a composite cycle in Table 2. This cycle is 56 years long and contains about 16 years of inflationary growth, 9 years of stagflation, 16 years of deflationary growth and 15 years of depression.

Fine structure of the Kondratieff cycle

Figure 8 shows the fine structure of the US price/inflation history since 1800. After the K-peak there is a rapid drop-off in prices (or inflation). After a few years, the fall slows, or even reverses and a small peak may appear. This period is called the plateau. The plateau is follow by a second drop in prices (inflation) culminating in a low called the vortex by Berry. After the vortex low there is a second rise in prices (inflation), this time stronger, that is sometimes termed the"false spring". This period peaks at what I call the deflationary (disinflationary) growth peak or DG-peak. Following the DG peak prices (inflation) fall again, reaching a low at the K-trough.
Table 3 shows specific dates for these features from six complete past cycles (3 US and 3 UK). The current (partial) cycle is shown as well. The length of these cycles range from 45 to 63 years, with and average of 53.5 years. The upwave duration has run from 20 to 31 years with an average of 24.4 years. The K-peak to plateau period has lasted from 4 to 12 years, with a median value of 6 years. The fall from the plateau to the vortex has taken from 2 to 11 years, with a median of 4 years. The rise from the vortex to the DG peak has taken from 3 to 15 years, with a median of 5 years. The fall from DG peak to the K-trough has lasted from 2 to 16 years with a median of 11 years.
The labeling in Figure 8, suggests that in the year 2000 we are two years past the vortex and are now rising towards the DG peak. If we apply the median value of 5 years for this period, we would expect a DG peak around 2003. If we then add the median value of 11 years for the fall to K-trough, we get an estimate of 2014 for the K-trough, which would make the current cycle about 65 years long.
There is an alternate interpretation of the current cycle. If we adopt 1974 as the K-peak (1974 was the peak in producer price inflation), we would then assign 1980 as the plateau rather than the K-peak. In this case 1986 would then become the vortex, 1990 the DG-peak and 1998 the K-trough. This last date is tentative since it is not year certain that we have entered a sustainable rise in inflation. This count gives an upwave 25 years in length, exactly in line with the historic average. The six year period from K-peak to plateau would also be exactly in line with the historical median value. The 6 year fall from plateau would is closer to the median value of 4 years than is the 8 year value from the other count. Finally, the 8 year fall to K-trough is shorter than the median value for this period, but well within the historical range. The length of the current cycle would then be 49 years, a little shorter than average, but far closer to the standard value of 53 years than is the alternate cycle length of 65 years.
The principal argument for this count is that it doesn't stretch the cycle length way beyond its standard length. In fact, most proponents of this count maintain that the K-trough is still a couple years away, in order to obtain an even better fit to the standard cycle length. But unless inflation falls below its 1998 level between now and then, it will be difficult to argue for say a 2002 K-trough instead of 1998. Since the government is currently running surpluses, a stimulation argument could hardly be applied. As shown above, the concept of a fixed cycle length developed because of an artifact, fairly consistent cycle lengths happened to occur over the 1800 to 1925 period, but not before nor afterward. The principal argument against the alternate account is that both consumer price inflation and interest rates rose from 1974 to 1980. Also, the high real interest rates that mark the beginning of a downwave, according to Roper, did not occur until after 1980.


Table 1. Historical US economic cycles



Period Type GDP growth Wagegrowth InflationRealstockreturns

1787-1806 inflationary growth +6.1% +0.5% +2.3% --

1806-1814 stagflation -1.3% -3.4% +5.1% 2.5%

1814-1836 deflationary growth +6.4% +3.7% -3.5% 8.2%

1836-1843 depression 2.1% +4.0% -2.6% 2.6%

1843-1853 inflationary growth 6.5% +0.5% +0.5% 9.2%

1853-1864 stagflation -0.7% -1.0% +4.7% 3.6%

1864-1881 deflationary growth +7.1% +1.5% -1.5% 10.1%

1881-1896 depression +3.0% +1.9% -1.3% 3.6%

1896-1912 inflationary growth +4.6% +0.6% +1.6% 8.0%

1912-1921 stagflation -0.3% +2.4% +6.8% -4.5%

1921-1929 deflationary growth 5.7% +1.7% -0.4% 25%

1929-1954 depression +2.0% +3.0% +1.7% 1.0%

1954-1973 inflationary growth +4.0% +2.3% +2.7% 9.9%

1973-1982 stagflation +2.3% -1.0% +8.8% -2.9%

1982-present deflationary growth +3.5% -0% +3.3% 15.0%





Table 2. Average economic performance during a composite cycle


Period Length (yrs) GDP growth Wage growth Inflation Real stock return

Inflat. growth 16 5.2% 1.3% 1.9% 9.2%

Stagflation 9 0.2% -0.7% 7.0% -0.2%

Deflat. growth 16 5.8% 1.8% -0.8% 12.6%

Depression 15 2.4% 2.8% 0% 2.2%

Overall 56 3.8% 1.4% 1.5% 6.8%








Table 3. Dates of specific features in historical Kondratieff Cycles


Feature UK UK UK US US US US

K-trough 1628 1691 1744 1787 1843 1897 1949

K-peak 1650 1711 1774 1814 1864 1919 1980

Plateau 1662 1720 1778 1818 1872 1925 1990

Vortex 1672 1723 1780 1829 1878 1932 1998

DG peak 1675 1729 1783 1836 1882 1947?

K-trough 1691 1744 1789 1843 1897 1949?








Figure 1. Producer price index (PPI) and consumer price inflation (after 1934).


Figure 3. Inflation rates over rolling 12 month periods from 1947 to 1959.







Figure 5. Peaks in British producer prices (1264-1750)




Figure 6. Distribution of cycle lengths for the US (1745-1980) and the UK (1557-1977)



<center>

<HR>

</center>

Thread gesperrt

 

gesamter Thread:

• Kondratiev-Zyklen - Ricardo, 23.03.2002, 13:09
  • Re: Kondratiev-Zyklen Das im Jahre 2000 erschienene Buch von Michael A. - Galiani, 23.03.2002, 15:23
    • Re: Kondratiev-Zyklen Das im Jahre 2000 erschienene Buch von Michael A. - Diogenes, 23.03.2002, 16:53
    • Zustimmung. - El Sheik, 23.03.2002, 18:48
    • Re: Kondratiev-Zyklen Das im Jahre 2000 erschienene Buch von Michael A. - Ricardo, 23.03.2002, 18:49
    • Re: Kondratiev-Zyklen Das im Jahre 2000 erschienene Buch von Michael A. - kingsolomon, 23.03.2002, 22:51