Do wealth distributions follow power laws? Evidence from "rich lists" · HackerPapers discussions

Do wealth distributions follow power laws? Evidence from "rich lists" — mentioned by defertoreptar

Mon, 21 Jan 2019 15:05:31 GMT

What I'd like to see more in these articles is an acknowledgement of how power laws work (or perhaps log-normal distributions or stretched exponential distributions https://arxiv.org/abs/1304.0212).

Paper and discussion history. A mention is not an endorsement.

Do wealth distributions follow power laws? Evidence from "rich lists" — mentioned by nopinsight

Thu, 15 Mar 2018 17:00:08 GMT

A key purpose of the original paper is to inform policy to improve public funding of research [1]. Thus, to a first-order approximation, it should reflect realities. However, ... The number of lucky and unlucky events in the paper (figure 5, page 8) are each around 5-7 events per agent on average, where 'lucky' is defined as doubling the capital (or success) accumulated so far and 'unlucky' as halving it. In the real world, how many regular people/scientists have that many career-defining events? Most are quite risk-averse, especially later in their careers. Some entrepreneurs could be…

Paper and discussion history. A mention is not an endorsement.

Do wealth distributions follow power laws? Evidence from "rich lists" — mentioned by nopinsight

Thu, 08 Mar 2018 11:01:52 GMT

Correction for above: Each factor needs not be normally distributed, just independent. Evidence that suggests Log-Normal is at least as plausible as Power Law as a model for wealth distribution: "Using recently developed empirical methodology for detecting power-law behaviour introduced by Clauset et al. we have found that top wealth values follow the power-law behaviour only in 35% of analysed cases. Moreover, even if the data do not rule out the power-law model usually the evidence in its favour is not conclusive – some rivals, most notably the log-normal and stretched exponential…

Paper and discussion history. A mention is not an endorsement.