Power Laws and Economic Cycles
There is a very nice "conversation starter" essay on McKinsey's site today regarding the fact that natural disasters and economic cycles seem to occur with the same kind of frequency and intensity. The actual pattern isn't a normal distribution (the kind of bell curve that we all studied in statistics) but a "power law distribution" - which means, essentially, that there is a "long tail" of outliers. In a power law distribution an event that is twice as severe, for instance, might occur just 1/4 as often (power = 2), or 1/8 as often (power = 3), or 2.83 times as often (power = 1.5). The bigger the power number, the steeper the fall-off in frequency as the size or import of an event increases.
Power Law distributions characterize complex systems in which the actions of individual "agents" making up the system impact other agents, which can create a cascading effect. In nature, one burning tree might ignite another tree, sometimes causing a forest fire. Individual rocks and underground fault lines might affect other rocks and fault lines, sometimes causing an earthquake. In economics and marketing, complex systems with their power law distributions are all around us. A Web site's popularity today among some visitors impacts its popularity among others. One company's advertising budget affects its competitors' budgets. One investor's sentiments influence other investors' sentiments.
The key is how the agents in a system interact, and in our economic system new technologies mean interactions are getting more frequent, more cost-efficient, and faster. Networks of economic agents respond to one another now at Blackberry speed, rather than at the speed of the daily news cycle. This means not only that our economic feedback loop is accelerating and intensifying, but also that our lives are governed more by randomness and chance than most of us thought.
But will an increased volume of interaction make future economic cycles even more pronounced? Or might the increasing speeds of interaction have the effect of "dampening" down these cycles, or perhaps making the swings between them shorter in duration? This is clearly a topic worthy of more research and study.
