Cyclical industrial dynamics
From Wikipedia, the free encyclopedia
Industrial dynamics is the study of the means and processes through which industries change over time, through their own processes of evolution – as first analyzed by Joseph Schumpeter. It is the complementary study to that of an industry’s comparative statics, which still dominates economic analysis. Industrial dynamics, as studied by scholars such as Carlsson and Eliasson,[1][2] reveal the basic underlying forces driving industry evolution. Some industries, particularly those with rapid product turnover or high levels of capital expenditure, reveal special dynamics moving through intrinsic upturns and downturns that are not necessarily related to the wider economic fluctuations. These are known as cyclical industrial dynamics. They have recently come under investigation in the specialized literature.[3][4]
Commodities
Almost all industries exhibit cyclicality to some extent. Below are some examples where industry cycles have been particularly examined.
The research by John Sterman finds that the so-called commodity cycles arise in many commodity markets.[5] For example, price and production cycles in markets of hog, cattle and copper span 4 years, 10–12 years and 8–10 years each in average respectively. Margret Slade estimates that cycles in prices of metals including aluminum, copper, iron, lead, silver, tin, and zinc are about 10–14 years in duration, which is twice as long as the investment periods.[6] The study by Thomas Stanback indicates a persistent cycle of approximately two years duration in the textile industry in the 1950s and 1960s.[7] Alajoutsijärvi and colleagues report that the cycle in the German paper industry has shortened in duration since 1990, and is now about two or three years in length.[8]
Durable goods
Based on his observations of the explicit cyclical movement in the shipbuilding industry, Nobel Prize-winning economist Jan Tinbergen believes the so-called ‘durable goods cycle’ is a result of the lag of the upstream industries such as shipbuilding in response to the cycles in end users markets such as that in freight rates.[9] He further suggests that the cycle length of the upstream industry is four times the lag, and is about 17 years in the shipbuilding industry.[10] Besides the traditional industries, many new industries such as semiconductors, flat panel displays, computers and telecommunications also exhibit strong cyclicality.[11]
Services
In the service sector, for example, Choi and colleagues date the industry cycles of the US hotel industry and restaurant industry.[12] They establish the mean duration of the business cycle, measured with aggregate activities in absolute level, as about 7.3 years for the U.S. hotel industry; and for the U.S. restaurant industry, as about 8 years.
Industry cycles versus business cycles
Cyclical dynamics at the level of individual industries may present rather different patterns from those of the general business cycles. For example, while the fluctuations of many industries correlate with those in the aggregate economy, there were also many industries that are not sensitive to business cycles — such as the pharmaceutical, educational service, insurance carriers and public service industries; some other industries such as the health service industry even enjoy higher growth during recessions.[13] In fact, it was estimated that "in any one recession [during the 1980s and the 1990s] only 60% of all industrial sectors were actually in a downturn."[14]
The timing, duration and amplitude of industry cycles can vary widely. Durable goods industries in the US are approximately three times more cyclical than nondurable-goods industries.[15]
Identification and measurement of industry cycles
Industry cycles have been identified using either the ‘classical cycle’ approach or the ‘growth cycle’ approach. The former approach uses time series in levels of economic activities to define cycles; and only absolute decline in the activities qualifies as a ‘downturn’.[16]
While this approach is consistent with the approach used by the National Bureau of Economic Research (NBER) to identify business cycles in the US, cycles at the industry level are usually concealed by strong industry trends which dominate industrial series. Therefore industry cycles are more commonly identified using the ‘growth cycle’ approach, by separating the cyclical component of a time series from the underlying trend.
Combining the growth cycle approach and other econometric techniques such as the Hodrick-Prescott filter, the industry cycles in the global semiconductor, PCs and flat panel display industries in the past decades are identified.[17]
Industry cycles can be further measured using techniques such as the Fourier analysis. For example, three most powerful cycles of the global semiconductor shipment data in the frequency domain are identified, with an average period of 4 years, 2.29 years, and 1.03 years accordingly (reverting to the original time domain).[18]