The Kondratiev Cycle in Prices
This is the first of a series of articles describing the progress of the Kondratiev cycle over a long period of time. It is an extension of a previous article that described the cycle over the most recent two centuries. It shows how the cycle can be detected using the method of reduced prices that I describe in my recent book on the Kondratiev cycle. In a future article I will present the cycle as shown by real economic data such as production indices. Finally, I will present my best estimate for our current position in the cycle using as close to real-time data as I can.
Figure 1 shows a plot of US producer prices over the 1785-1925 period. A striking pattern of rising and falling price trends is evident that are spaced roughly 50 years apart. Similar patterns can be seen in price indices from other countries and also in interest rates. Nikolai Kondratiev, a Russian economist, performed the first detailed investigation of these patterns in a series of articles in the 1920's. Kondratiev presented a variety of analyses, the most convincing of which were price charts like this one which clearly showed fifty year oscillations over the late 18th through early 20th century. These wavelike patterns in economic variables have come to be called the Kondratiev cycle or long wave.
This striking pattern, occurring across a number of national economies, was suggestive of some underlying cyclical process. A number of economists after Kondratiev have sought to expand upon and further refine his original observations in an attempt to elucidate what sort of process may underlie the long wave. They have not been very successful. A big part of the problem has been the inability of most economists to clearly show the long wave pattern in economic data other than that originally presented by Kondratiev. For example, suppose one looks at prices over a longer period than Kondratiev. Figure 2 shows prices from the late 12th century through the early 20th century. The price waves noted by Kondratiev are visible in the upper right corner. But the Kondratiev pattern is not evident elsewhere in the figure.
Figure 2. The long-term price trend 1162-1925
What is evident are much larger price waves that are called price revolutions. The American historian David Hackett Fischer recently wrote a book called The Great Wave that discusses these price revolutions. He notes that unlike the more familiar Kondratiev wave, the Great Wave is evident without any special treatment of the data. Of course if we look at Kondratiev's original price observations, the waves he noted were just as obvious as those Fischer describes. They are diminished only when a much longer price series is examined. For all we know, the Great Wave may also shrink to insignificance were we to examine several millennia worth of price data. Since such data is not available, we shall never know if this is the case.
Some longwave researchers, including Kondratiev himself, have hypothesized that the long wave is an emergent property of the postindustrial economy and, as such, is not visible before the industrial revolution in the late 18th century. In this case, examination of data for evidence of long wave patterns should be restricted to just that data after 1790 or so. Since it has been some 70 years since Kondratiev's original work, even if we restrict ourselves to the post-1790 period, there is still some 50% additional data in the subsequent years that we can examine for further evidence of the long wave pattern. Figure 3 shows the US PPI from 1785 to the present in bold black. The dates of the peaks and troughs seen by Kondratiev are labeled. It is clear from the figure that after 1925 the "Kondratiev pattern" is no longer evident.
Figure 3. US producer prices and reduced price showing the long
The last seven decades have shown what appears to be another price revolution. The regular pattern of peaks and valleys in producer prices seen by Kondratiev has been replaced by one enormous upward trend. The long wave pattern seen by Kondratiev can no longer be detected, suggesting that it may simply have been a random fluctuation that happened to occur during the 140 years after 1790. It is the disappearance of the clear-cut long wave pattern after the early 1930's that has prompted many observers to conclude that the Kondratiev wave, if it ever did exist, no longer does. Other researchers have sought to transform the price data in various ways in an attempt to visualize the long wave despite the presence of the price revolution. One such method is the reduced price, which was described in my previous article. Reduced price, shown in Figure 3 in red, captures the same wave pattern seen by Kondratiev over the 1790-1925 period. In addition, it shows a further peak and trough after 1925 that is not visible in the raw price index. That is, the pattern of peaks and valleys seen by Kondratiev in raw prices over 1790-1925 can be seen in reduced prices over the 1790-2000 period. Whereas Kondratiev only counted three cycles, today we can see four through the use of reduced price.
It would appear that the method of reduced price allows one to see Kondratiev waves that are otherwise obscured by price revolutions. One might wonder what the use of such a method might reveal about the existence of long waves before 1790, when long waves were not evident in the raw price index. Figure 4 shows a plot of the reduced price from 1510 to 1825 as the thin black line. Also shown is a nine year moving average in bold black to bring out the trend in reduced price.
Figure 4. Reduced price trends before 1825 showing apparent long
The reduced price can be thought of price relative to monetary stimulation. That is, a rising reduced price means that price is rising faster than monetary stimulation is driving it. A falling price means prices are rising slower than stimulation is driving them. Thus, it is possible for reduced price to fall even when ordinary prices are rising. This is why the method of reduced price can show valleys and troughs during periods of price revolution. This property of "subtracting out" the secular trend produced by a price revolution is demonstrated by the lack of secular trend shown in the above figure.
Examination of the black lines does not show the clean cycles that are evident in the post-1790 plot of reduced price. Additional information concerning the general trend in price inflation relative to stimulation is useful. This information is given by the bold red line, which shows the average inflation rate over a trailing 25 year period minus the amount of inflation that would be expected to happen due to the monetary stimulation over that same 25 year period. A peak in the red line denotes the end of a 25 year period when prices have been rising faster than stimulation has been driving them. A trough denotes the end of a 25 year period when prices have been rising slower than stimulation has been driving them. A local peak in reduced price that occurs near a peak in the red line is likely to reflect a true peak in the reduced price cycle. Similarly, a trough that occurs close to a trough in the red line should reflect a true valley in the reduced price cycle.
Using the red line as a guide, major spikes in reduced price in 1557, 1598, 1650, and 1711, all of which are close to a peak in the red line, can be considered as (Kondratiev) peaks in the reduced price cycle. A major price spike in 1802, is not so considered because it does not occur at the peak in the red line. A later peak in the smoothed reduce price (bold black line) in 1813 is considered the Kondratiev peak instead because it occurs right at the peak in the red line. The 1813 K-peak in Figure 3, which was derived from British data, agrees well with the 1814 K-peak in reduced price from Figure 2, which was derived from U.S. data.
Similarly, downspikes in reduced price in 1621 and 1688, which occur very close to a trough in the red line are labeled as (Kondratiev) troughs in reduced price. A trough in the bold black line in 1736 that occurs at the second of a double bottom in the red line is also selected as a trough. Finally we note that the red line indicates that somewhere around 1575 there should be another trough. In my book I select 1583 for this K-trough based on the beginning of an uptrend in the raw prices in that year. The following table summarizes the Kondratiev peaks and troughs obtained using both raw prices and reduced prices:
Table 1. Kondratiev peaks and troughs identified from raw prices and from reduced prices
|Reduced Price||Raw Prices||Reduced Price||Raw Prices|
In summary, the method of reduced prices allows the Kondratiev wave-like pattern to be visualized in the presence of an otherwise obscuring price revolution as the demonstration of a clear peak and trough after the 1930's showed. Use of reduced prices also allowed a less clear, but still discernible, pattern of peaks and troughs to be seen in the pre-1790 period.