This article, and most reviews of Piketty, have to be misrepresenting or misunderstanding something. The idea that r > g is simply mathematically impossible, and I don't think it's actually what Piketty is pushing.

That's because r is part of g. If the whole economy looks like E(t) = exp(rt)+rest, and exp(rt) grows faster than the rest, then eventually E(t) ~ exp(rt).

Since g = log(E(t))/t, g and r must eventually be equal.

The only way I can think of to resolve this is to assume a permanent redirection of wealth from investment to consumption. This could potentially result in investment returns being r, but the growth of invested capital being only g.

It's also worth noting that Piketty's numbers significantly underestimate income for the bottom 90%.

http://www.forbes.com/sites/scottwinship/2014/04/17/whither-...

See also Doug's comment on Robin Hanson's blog post: http://www.overcomingbias.com/2014/03/hidden-taxes-must-be-h...

Perhaps Piketty merely means that E[r] > E[g] (where r and g are both random variables). But that doesn't actually imply long term concentration of wealth at all, which seems to contradict the spirit of Piketty.