Reconsidering ergodicity and uncertainty17 February, 2016 at 18:03 | Posted in Economics | Leave a comment
The concept of fundamental uncertainty is a centerpiece of much of Post Keynesian economics. The foundation of this concept in Keynes’s own work and in broader intellectual foundations and considerations has generated much debate and discussion in recent decades, most recently between Rod O’Donnell (2014-15) and Paul Davidson (2015). We have reviewed their arguments, finding some grounds for O’Donnell’s criticism of Davidson’s claim that rejection of the ergodic axiom is the strongest ontological foundation for a solid theory of Keynesian uncertainty. Probably the strongest argument made by O’Donnell involves the problem of time horizons. That ergodicity is only known or determined as one approaches an infinite time horizon not only makes it a deep epistemological problem even to determine if a system is ergodic or not, but also this approach appears to contradict some of Keynes’s own strongly held views in favor of looking at shorter time horizons in studying economies.
On the other hand, Davidson can make some claims that whether or not Keynes knew of the discussions regarding “the ergodic hypothesis” (which he never mentioned in any of his writings, even while being in milieus where it is quite likely it was discussed), the understanding of ergodicity in the time period he lived and wrote in was such that it was closely linked with concerns about homogeneous data and stationarity of time-series, which we know he was concerned with, even if we now know these links to be both weaker and more complicated than was thought in the 1930s. Of course, it is true that Keynes’s own definition of uncertainty as involving processes not based on probability distributions would certainly not be ergodic. Davidson can also be defended for using the nonergodicity argument at a time period when advocates of the rational expectations hypothesis were arguing against pretty much any form of Keynesian economics by asserting the axiom of rational expectations, and asserting the nonergodic axiom was a useful way to directly combat this argument, despite its ultimate flaws.