Why Should Economics Give Chaos Theory Another Chance?

Abstract

Economic data provide little evidence -if any- of linear, simple dynamics, and of lasting convergence to stationary states or regular cyclical behavior. In spite of this the linear approach absolutely dominates mainstream economics. The problem is that mainstream economics is now in deep crisis. The recent financial crisis clearly showed that orthodox economics was quite unprepared to deal with it. Most mainstream economists not only did not foresee the depth of the current crisis, they not even consider it possible. It is well known since the famous contribution of Mandelbrot (1963) that many economic and financial time series have fat tails, i.e. that the probability of extreme events is higher than if the data-generating process were normal. However, the usual practice among orthodox economists has been to assume-implicitly or explicitly- a normal distribution. Orthodox economists represent the economy as a stable equilibrium system resembling the planetary one. The concept of equilibrium plays a key role in traditional economics. This approach is useful in normal, stable times. However, it is incapable of dealing with unstable, turbulent, chaotic times. The crisis has clearly showed this. Heterodox contributions shed much more light on what happens during these crucial periods in which a good part of the economy is reshaped; they provide powerful insights towards what policies to follow in those extraordinary circumstances. However, they remain as theories mainly suitable for those periods of instability and crisis. The challenge is to arrive at a unified theory valid both for normal and abnormal times. In this respect, the complexity approach with its use of non-linear models offers the advantage that the same model allows to describe stable as well as unstable and even chaotic behaviors. Although the results of chaos tests do not prove so far the existence of chaos in all economic variables they are consistent with its existence. The detection of chaos in economic time series faces three types of difficulties: (1) the limited number of observations such series contain; (2) the high noise level in economic time series; and (3) the high dimension of economic systems. However, topological methods for chaos detection seem to be a highly promising tool. On the other hand, in economics, there are no such things as crucial experiments. Economists seldom practice the falsificationism they preach. Confidence in the implications of economics derives from confidence in its axioms rather than from testing their implications. Therefore, non-linear dynamics and chaos theory should not be subject to more stringent rules than what is usual for the rest of economic theory.

Helpful comments by an anonymous referee are deeply thanked. The usual caveats apply.