Another topical colloquium here at the Kavli Institute for Theoretical Physics…
“Money, It’s a Gas”
New Developments in Statistical Mechanics of Money, Income, and Wealth (podcast, video, slides)
by Victor Yakovenko, University of Maryland.
Money, of course, is not conserved.
It can be created, with some work,
and, famously, it can be destroyed.
Makes for interesting statistical mechanics, eh?
It turns out that putting a lower bound on net money per person was unphysical…
there is some interesting literature out there on the instability of systems with unbounded lowest state.
Hm. Actual data – in 1997 the income distribution in the US was a two piece fit, a Boltzman distribution for < $120k p.a. and Pareto (power-law [sic]) for the 3% with annual income above $120k. No evidence for a distinct middle class - in income - might be in asets. Boundary of course moves - but the low end is stable when rescaled. Power-law segment is not stable - it changes secularly. So, low end income is a diffusive process, Swedish data shows the power-law tail is capital unearned income; interstingly social transfer looks almost like a Fermi distribution, very sharp cutoff. I suspect this data comes with a narrow definition of "social transfer" - kinda pre-AIG definition. Heh. Physicists rediscover Marx. I think I need to read "Classical Econophysics", sounds amusing. Fokker-Planck eqn for diffusive income kinetics. Look for stationary solution. ` essential difference is that for small Δ$ the $Delta; is independent of $ and for high income, it is not - clearly an approximation. In F-P terms, the high end of the income distribution is non-local, changes in income are not perturbative, they scale with existing income, per unit time. I wonder what happens when the proportionality parameter goes negative - be fun to model... I have a very handy dandy Fokker-Planck code with a non-local non-perturbative term added. “The next great depression will be from 2008 to 2023″
p. 16 in “The Great Boom Ahead” by H.S Dent (1993)
based on demographic data.