Statistical Mechanics of Money, Income, and Wealth Adrian Drăgulescu and Victor Yakovenko Department of Physics, University of Maryland, College Park, MD 2742-4111, USA http://www2.physics.umd.edu/~yakovenk/ Outline Boltzmann-Gibbs distribution in physics distribution of money (theory) The roles of debt and time-reversal symmetry distributions of income and wealth (data) Conclusions Publications European Physical Journal B 17, 723 (2), cond-mat/1432 European Physical Journal B 2, 585 (21), cond-mat/835 Physica A 299, 213 (21), cond-mat/13544
Distribution of Individual Income U.S. Census data, 1996 (histogram and points A) Panel Study of Income Dynamics (U. Michigan), 1992 (points B) 18% 16% 14% 12% 1% 8% 6% 1% 1% B 1% A.1%.1% 2 4 6 8 1 12 Individual annual income, k$ Cumulative probability 4% 2% % 1 2 3 4 5 6 7 8 9 1 11 12 Individual annual income, k$ The solid lines are fits to the exponential distribution P 1 (r) = exp( r/r)/r
Distribution of Individual Income Internal Revenue Service data, 1997 (main panel and inset A) and 1993 (inset B) Cumulative percent of tax returns 1% 9 8 7 6 5 4 3 2 1% A 1% B 1% Cumulative probability.1% 2 4 6 8 1 Adjusted gross income, k$ 1 1 2 3 4 5 6 7 8 9 1 Adjusted gross income, k$ The solid lines are fits to the exponential cumulative distribution N 1 (r) = r P 1 (r ) dr = exp( r/r)
Lorenz Curve for individual income: Horizontal axis x(r) the cumulative fraction of population with income below r: x(r) = r P (r ) dr. Vertical axis y(r) the fraction of income this population accounts for: y(r) = r r P (r ) dr / r. Gini Coefficient is the measure of inequality: Area between the diagonal and the Lorenz curve G = Area of the triangle beneath the diagonal 1% Cumulative percent of income 9 8 7 6 5 4 3 2 1 1.75.5.25 Gini coefficient 1 2 198 1985 199 1995 Year 1 2 3 4 5 6 7 8 9 1% Cumulative percent of tax returns For the exponential distribution: Lorenz curve: y = x + (1 x) ln(1 x), the solid line. Gini coefficient: G 1 = 1/2.
Income distribution for two-earners families r = r + r P 2 (r) = dr dr P 1 (r ) P 1 (r ) δ(r + r r) 8% 7% 6% 5% 4% 3% = r P 1 (r ) P 1 (r r ) dr = (r/r 2 ) exp( r/r) 8% 6% 4% 2% r r % 2 4 6 8 1 12 Family annual income, k$ r r 2% 1% % 1 2 3 4 5 6 7 8 9 1 11 12 Family annual income, k$ distribution of income for families with two adults, U.S. Census data, 1996.
Lorenz curve and Gini coefficient for families U.S. Census data, 1947 1994 1% Cumulative percent of family income 9 8 7 6 5 4 3 2 1 1.8.6.375.2 Gini coefficient 3 8 195 196 197 198 199 Year 1 2 3 4 5 6 7 8 9 1% Cumulative percent of families The solid curve is for P 2 (r) = (r/r 2 ) exp( r/r). The calculated Gini coefficient: G 2 = 3/8 =.375
Scattering plot of two incomes for families 1 PSID data for families, 1999 Labor income of another earner, k$ 8 6 4 2 2 4 6 8 1 Labor income of one earner, k$ Every family is represented by two points (r 1, r 2 ) and (r 2, r 1 ). The absence of significant clustering of points (along the diagonal) indicates that the two incomes are approximately uncorrelated.
World distribution of Gini index World Bank data, 1988 and 1993 World distribution of Gini index, 1988 and 1993.6 Gini index for households.5.4.3.2 1988 1993.375.1 W.Europe Asia S.America Africa E.Europe N.America FSU In Western Europe and North America, the Gini index is close to 3/8=.375, in agreement with our theory. Other regions have higher Gini coefficients, i.e. higher inequlity. A sharp increase in the Gini coefficient is observed in Eastern Europe and former Soviet Union (FSU) after the collapse of communism.
Conclusions Conservation law of money leads to the exponential, Boltzmann-Gibbs probability distribution of money: P (m) exp( m/t ). Deviations from the Gibbs law exist in the models where the time-reversal symmetry is broken. Tax and census data for USA and UK show that distributions of individual income and wealth of more than 95% of population are exponential. The high-end tails have power-law distributions and account for the top 1% of population and 16% of income and wealth ( Bose condensate ). Income distributions in UK and different states of USA collapse to a single curve, when rescaled using their individual temperatures. Temperature varies by ±25% within USA. The US temperature is twice higher than the UK one. A thermal machine can extract profit from the temperature difference. Income of families with two earners has the gamma distribution. The Lorenz curve and the Gini coefficient 3/8=37.5% agree well with the US census (1947 1994) and World Bank data.