The Question of Periodicity

While many models of fluctuation patterns suggest possible mechanisms for endog­enously driven repetition, if only through “echo effect” bunching of investments in similarly depreciating capital stocks, probably a majority of economists are sceptical regarding regular periodicities occurring with respect to most macroeconomic fluc­tuations.

Although he would be followed soon by Jevons (1878) with his exogenously driven but regular sunspot cycles working through agriculture, the first to identify a possible regular periodic macroeconomic cycle was Juglar (1862). He argued that these were driven by cycles of fixed capital investment along echo boom lines, replacement waves of invest­ment coinciding with a periodicity of about 7-11 years, depending on time period and place. Kitchin (1923) would label this the “major cycle” and posited a shorter 40-month cycle that he called the “minor cycle”. He found this looking at commodity prices, inter­est rates, and bank clearings, and explained it as due to mass psychology, and argued that the major cycle was simply an aggregation of two or three minor cycles.

Cobweb models themselves (Ezekiel 1938) usually focused on agricultural cycles such as 5-6 year corn-hog cycles and 14-17 year cattle cycles, with Rosen et al. (1994) finding the cattle cycle to hold empirically for 1875-1990 in the US. Ezekiel also recognized the possibility that such cycles might be erratic, and Artstein (1983) showed they could be chaotic. Hommes (1991) has studied chaotic dynamics in macroeconomic cobweb models based on the Hicks (1950) nonlinear-accelerator model that assumes floors and ceilings for investment.

Next up in periodicity is the “long swing”, first argued for by Simon Kuznets (1930), with a period of 15-17 years, possibly twice the length of a Juglar major cycle. Kuznets focused on demographics and residential construction as the main sources of this hypothetical cycle, although infrastructure investments have been suggested as well.

Probably the most controversial of hypothesized cycles is the “long wave”, usually associated with Nikolai D. Kondratieff (1926), although he was preceded by van Gelderen (1913), with others making vague arguments along such lines even earlier (van Duijn 1983). Kondratieff would be executed by the Soviet government, at least partly for his work on this idea. The cycle was argued to be about 50 years in length, and Kondratieff focused on prices and interest rates like some others in looking for cycles, with these variables more observable than gross domestic product (GDP), which was not being estimated officially prior to the early twentieth century. From the beginning there was debate about what might be a driving mechanism, with many arguing that even if it was there in the data it was an artefact with no clear basis for a consistent regular­ity (Kuznets 1940). Some argued that it was a monetary phenomenon, with the major discoveries of gold in California and Alaska a half-century apart responsible for it, or even a sort of long period accelerator tied to investment waves in the capital goods sector responding slowly to those in the consumer goods sector (Forrester 1977).

Unsurprisingly, the empirical debate on this has been lengthy itself without a clear con­clusion. Typical is the exchange between Solomou (1986) and Bieshaar and Kleinknecht (1986) with many issues involved in this. One of the most recent efforts to consider these matters has been by Korotayev and Tsirel (2010) who test for all four of the hypothetical cycles mentioned above at the level of global GDP. They claim to find both the Juglar and the Kondratieff, but neither the Kitchin nor the Kuznets.

While they emphasize elements that go beyond economics, and economists have largely ignored their work, others have posited even longer cyclical patterns. Modelski (1987) sees economic developments tied to larger political ones, particular the rise and fall of hegemonic powers at the global level, with the economies of the hegemons rising and falling along with their power. He argues that such a cycle may be around 120 years.

Even longer is the “longue duree” or “geographical cycle” of Fernand Braudel (1949). Associated with the Annales school of history, he sees deep cycles driven by demographic forces in relation to the geographical environment that work themselves out on a scale of 300-400 years, with collapses in Europe in the sixth, the tenth, the fourteenth, and the seventeenth centuries. Note that the dynamics described by Braudel are somewhat similar to those modelled by Day and Walter (1989). Note also that if one wishes to be pessimistic about future economic growth in the coming decade or so, some have argued that we could be experiencing a simultaneous downturn of the fifth Kondratieff wave since the Industrial Revolution (Korotayev and Tsirel 2010) with a downturn of a Braudel longue duree that began its early upturn in the late seventeenth century, with the pessimists pointing to a possibly substantial slowdown of serious new innovations as the key to this secular stagnation (Cowen 2011).