Wealth distribution: To be or not to be a Gamma?
We review some aspects, especially those we can tackle analytically, of a minimal model of closed economy analogous to the kinetic theory model of ideal gases where the agents exchange wealth amongst themselves such that the total wealth is conserved, and each individual agent saves a fraction () of wealth before transaction. We are interested in the special case where the fraction is constant for all the agents (global saving propensity) in the closed system. We show by moment calculations that the resulting wealth distribution cannot be the Gamma distribution that was conjectured in Phys. Rev. E 70, 016104 (2004). We also derive a form for the distribution at low wealth, which is a new result.
pacs:89.65.Gh 87.23.Ge 02.50.-r
The distribution of wealth or income in society has been of great interest for many years.
As first noticed by Pareto in the 1890’s Pareto (), the wealth distribution
seems to follow a “natural law” where the tail of the distribution is described
by a power-law . Away from the tail, the distribution is better described by a Gamma or Log-normal distribution known as Gibrat’s law Gibrat (). Considerable investigation with real data during the last ten years revealed that the power-law tail exhibits a remarkable spatial and temporal stability and the Pareto index
is found to have a value between and Dragulescu2001 (); Aoyama2003 (). Even after 110 years the origin of the power-law tail remained unexplained but recent interest of physicists and mathematicians
in econophysics has led to a new insight into this problem (see Refs. Yakovenko2009 (); Arnab2007 (); Chakrabarti2010 ()).
Our general aim is to study a many-agent statistical model of closed economy (analogous to the kinetic theory model of ideal gases) Bennati (); Angle (); Ispolatov1998 (); Dragulescu2000a (); Chakraborti2000a (); Chatterjee2003a (), where agents exchange a quantity , that may be defined as wealth. The states of agents are characterized by the wealth , and the total wealth is conserved. The evolution of the system is then carried out according to a prescription, which defines the trading rule between agents. These many-agent statistical models have basic units , interacting with each other through a pair-wise interaction characterized by a saving parameter , with . We define the equilibrium distribution of wealth as follows : is the probability that in the steady state of the system, a randomly chosen agent will be found to have wealth between and . In these models, if is equal for all the units, is fitted quite well by a Gamma-distribution Anirban2008 (); Patriarca2004a (); Patriarca2004b ()
Ii Many-agent model of a closed economy
We study many-agent statistical models of closed economy (analogous to the kinetic theory model of ideal gases), where agents exchange wealth . The states of agents are characterized by the wealth , and the total wealth is conserved. The evolution of the system is then carried out according to a prescription, which defines the trading rule between agents. At every time step two agents and are extracted randomly and an amount of wealth is exchanged between them,
It can be noticed that in this way, the quantity is conserved during the single transactions: (see Fig. 1), where and are the agent wealths after the transaction has taken place. Several simple models dealing with different transaction rules have been studied (see the reviews Yakovenko2009 (); Arnab2007 (); Patriarca2010 () and references therein). Here we will present a few examples.
ii.1 Basic model without saving: Boltzmann distribution
In the first version of the model, the wealth difference is assumed to have a constant value Bennati (),
This rule, together with the constraint that transactions can take place only if and , provides a Boltzmann distribution, see the curve for in Fig. 2. Alternatively, can be a random fraction of the wealth of one of the two agents,
where is a random number uniformly distributed between 0 and 1. A trading rule based on the random redistribution of the sum of the wealths of the two agents had been introduced by Dragulescu and Yakovenko Dragulescu2000a (),
All the versions of the basic model lead to an equilibrium Boltzmann distribution, given by
where the effective temperature of the system is just the average wealth Bennati (); Dragulescu2000a (). The result (8) is found to be robust; it is largely independent of various factors. Namely, it is obtained for the various forms of mentioned above, for a pair-wise as well as multi-agent interactions, for arbitrary initial conditions Chakraborti2000a (), and finally, for random or consecutive extraction of the interacting agents. For the trading rule (6) one can show the convergence towards the Boltzmann distribution through different methods: Boltzmann equation, entropy maximization, distributional equation, etc.
ii.2 Model with global saving propensity
A step toward generalizing the basic model and making it more realistic, is the introduction of a saving criterion regulating the trading dynamics. This can be practically achieved by defining a saving propensity , which represents the fraction of wealth which is saved – and not reshuffled – during a transaction. The dynamics of the model is as follows Chakraborti2000a (); Chakraborti2002a ():
corresponding to a in Eq. (3) given by
This model leads to a qualitatively different equilibrium distribution. In particular, it has a mode and a zero limit for small , see Fig. 2. Later we will derive a form for an upper bound on at low range. The functional form of such a distribution was conjectured to be a -distribution, as given by Eq. (1) on the basis of an analogy with the kinetic theory of gases. Indeed, it is easy to show, starting from the Maxwell-Boltzmann distribution for the particle velocity in a dimensional gas, that the equilibrium kinetic energy distribution coincides with the Gamma-distribution (1) with . This conjecture is remarkably consistent with the fitting provided to numerical data Anirban2008 (); Patriarca2004a (); Patriarca2004b (). In the following section we will show by two different approaches that the conjecture (1) cannot be the actual equilibrium distribution.
Iii Analytical results for model with saving propensity
iii.1 Fixed-point distribution
Let be a random variable which stands for the wealth of one agent, at equilibrium and in the limit of an infinite number of agents, we can say from Eq. (9) that the law of , is a fixed-point distribution of the equation
where means identity in distribution and one assumes that the random variables and have the same probability law, while the variables and are stochastically independent. It seems difficult to find the distribution of , however, one can compute the moments of . Indeed with (11), one can write immediately
and by developing (12) one can find the recursive relation
Using (13) with initial conditions (normalization) and (without loss of generality), we obtain
The fourths moments (eqs.(16) and (20)) are different so the conjecture that the Gamma distribution is an equilibrium solution of this model is wrong. Nevertheless the first three moments coincide exactly which shows that the Gamma-distribution is strangely a very good approximation. Moreover the deviation in the fourth moment is very small (see Fig. 3, which shows that the two curves can hardly be distinguished by the naked eye).
Finding a function that would coincide to higher moments is still an open challenge. These results are consistent with the ones found by Repetowicz et al. Repetowicz2005 () which will be presented in the following section.
iii.2 Laplace transform analysis
In this section we will confirm the previous result with a different approach based on the Boltzmann equation and along the lines of Bassetti et al. Matthes2008 (). Given a fixed number of agents in a system, which are allowed to trade, the interaction rules describe a stochastic process of the vector variable in discrete time . Processes of this type are thoroughly studied e.g. in the context of kinetic theory of ideal gases. Indeed, if the variables are interpreted as energies corresponding to the -th particle, one can map the process to the mean-field limit of the Maxwell model of particles undergoing random elastic collisions. The full information about the process in time is contained in the -particle joint probability distribution . However, one can write a kinetic equation for one-marginal distribution function
involving only one- and two-particle distribution functions
which may be continued to give eventually an infinite hierarchy of equations of BBGKY (Born, Bogoliubov, Green, Kirkwood, Yvon) type Plischke (). The standard approximation, which neglects the correlations between the wealth of the agents induced by the trade gives the factorization
which implies a closure of the hierarchy at the lowest level. In fact, this approximation becomes exact in the thermodynamic limit (). Therefore, the one-particle distribution function bears all information. Rescaling the time as in the thermodynamic limit , one obtains for the one-particle distribution function the Boltzmann-type kinetic equation
This equation can be written (see Matthes et al. Matthes2008 ()) as
where is a collision operator. A collision operator is bilinear and satisfies, for all smooth functions
where and are the post-trade wealth. With this property the equation can be written in the weak form, for all smooth functions
It is very useful because the choice gives (after some calculations) the Boltzmann equation for the Laplace transform of
For the steady state, and if is drawn randomly from a uniform distribution, the previous equation reduces to
which coincides with results of Repetowicz2005 (). The Taylor expansion of can be derived by substituting the expansion in (26). Since is the moment-generating function we have . With this method Repetowicz et al. Repetowicz2005 () obtained the recursive formula
Now with this formula one can obtain the first four moments and they match the ones found in the previous section eqs.(14-16), which confirms that the Gamma-distribution is not the stationary distribution.
Iv Upper bound form at low wealth range
From equation (11)
we have for all
where means the probability of the event inside the brackets. If the number of agents in the market is large, the distributions of different agents are independent. Then
where is the Heaviside step function. Taking the derivative with respect to in both sides, we have
This equation is an integral equation for . As mentioned earlier, we are not able to solve it in closed form. However, one can simplify the equation, by doing the integral over . Then the -function will contribute only if we have the following constraints
The range defined by these constraints is shown in figure 4. In this range, the derivative of the argument of the delta function with respect to is just . And, hence we get
This immediately gives
We now use the observation that for the numerically determined is a continuous function with a single maximum, say at (see Fig. 2). Then for all , the integrand (38) takes its maximum value at the right extreme point, i.e. when . This then gives us
Iterating this equation, we get
We can set in the above equation, giving
Then taking and rescaling the variables, we get
as , where and are two constants dependent on . The Gamma-distribution decays slower than the rhs in (42) when . The expression (42) gives an upper bound form at low wealth range and confirms again that the distribution of the global saving propensity model is not a Gamma-distribution.
V Discussion and outlook
We have used different approaches to show that the correct form of the wealth distribution cannot be the Gamma distribution. We have also derived an analytical form of an upper bound at low wealth range see Eq. (42). This is an analytically calculated upper bound but the closed form of the solution to Eq. (9) is still an open question.
As a further generalization, the agents could be assigned different saving propensities Chatterjee2003a (); Chatterjee2004a (); Chatterjee2005a (); Repetowicz2005 (); Patriarca2005a (); Chakraborti2009 (). In particular, uniformly distributed in the interval have been studied numerically in Refs. Chatterjee2003a (); Chatterjee2004a (). This model is described by the trading rule
or, equivalently, by a (as defined in Eq. (3)) given by
One of the main features of this model, which is supported by theoretical considerations Mohanty (); Chatterjee2005a (); Repetowicz2005 (), is that the wealth distribution exhibits a robust power-law at large values of ,
with a Pareto exponent largely independent of the details of the -distribution.
Acknowledgements.The authors are grateful to D. Dhar for critical reading of the manuscript and inputs in finding the functional form of the wealth distribution at low range. The authors also acknowledge F. Abergel, N. Millot, M. Patriarca, M. Politi and A. Chatterjee for critical discussions or comments, and B.K. Chakrabarti for pointing out reference Chakrabarti2010 (). AC is grateful to Department of Theoretical Physics, TIFR for the kind hospitality where part of the work was initiated.
- (1) V. Pareto, Cours d’économie politique, F. Rouge, Lausanne (1897).
- (2) R. Gibrat, Les inégalités économiques, Sirey, Paris (1931).
- (3) A. Dragulescu, V.M. Yakovenko, Exponential and power-law probability distributions of wealth and income in the United Kingdom and the United States, Physica A 299 (2001) 213.
- (4) H. Aoyama, W. Souma, Y. Fujiwara, Growth and fluctuations of personal and company’s income, Physica A 324 (2003) 352.
- (5) V. Yakovenko, J. Rosser, Colloquium: Statistical mechanics of money, wealth, and income, Rev. Mod. Phys. 81 (2009) 1703.
- (6) A. Chatterjee, B.K. Chakrabarti, Kinetic Exchange Models for Income and Wealth Distributions, Eur. Phys. J. B 60 (2007) 135.
- (7) A.S. Chakrabarti and B.K. Chakrabarti, Statistical theories of income and wealth distribution, Economics E-Journal 4 (2010) 2010-4.
- (8) E. Bennati, La simulazione statistica nell’analisi della distribuzione del reddito: modelli realistici e metodo di Monte Carlo, ETS Editrice, Pisa (1988); E. Bennati, Un metodo di simulazione statistica nell’analisi della distribuzione del reddito, Rivista Internazionale di Scienze Economiche e Commerciali 35 (1988) 735; E. Bennati, Il metodo Monte Carlo nell’analisi economica, Rassegna di lavori dell’ISCO X (1993) 31.
- (9) J. Angle, The Surplus Theory of Social Stratification and the Size Distribution of Personal Wealth, Social Forces 65 (1986) 293; J. Angle, The inequality process as a wealth maximizing process, Physica A 367 (2006) 388.
- (10) S. Ispolatov, P.L. Krapivsky, S. Redner, Wealth distributions in asset exchange models, Eur. Phys. J. B 2 (1998) 267.
- (11) A. Dragulescu, V.M. Yakovenko, Statistical mechanics of money, Eur. Phys. J. B 17 (2000) 723.
- (12) A. Chakraborti, B.K. Chakrabarti, Statistical mechanics of money: How saving propensity affects its distribution, Eur. Phys. J. B 17 (2000) 167.
- (13) A. Chatterjee, B.K. Chakrabarti, S. S. Manna, Money in gas-like markets: Gibbs and Pareto laws, Physica Scripta T 106 (2003) 367.
- (14) A. Chakraborti, M. Patriarca, Gamma-distribution and Wealth inequality, Pramana 71 (2008) 233.
- (15) M. Patriarca, A. Chakraborti, K. Kaski, Statistical model with a standard gamma distribution, Phys. Rev. E 70 (2004) 016104.
- (16) M. Patriarca, A. Chakraborti, K. Kaski, Gibbs versus non-Gibbs distributions in money dynamics, Physica A 340 (2004) 334.
- (17) M. Patriarca, E. Heinsalu, A. Chakraborti, Basic kinetic wealth-exchange models: common features and open problems, Eur. Phys. J. B 73 (2010) 145.
- (18) A. Chakraborti, Distributions of money in model markets of economy, Int. J. Mod. Phys. C 13 (2002) 1315.
- (19) P. Repetowicz, S. Hutzler, P. Richmond, Dynamics of money and income distributions, Physica A 356 (2005) 641.
- (20) D. Matthes, G. Toscani, On steady distributions of kinetic models of conservative economies, J. Stat. Phys. 130 (2008) 1087; F. Bassetti, G. Toscani, Explicit equilibria in a kinetic model of gambling, http://arxiv.org/abs/1002.3689.
- (21) M. Plischke, B. Bergersen, Equilibirum statistical mechanics, World scientific (1989) 137.
- (22) A. Chatterjee, B.K. Chakrabarti, S. S. Manna, Pareto law in a kinetic model of market with random saving propensity, Physica A 335 (2004) 155.
- (23) A. Chatterjee, B.K. Chakrabarti, R. Stinchcombe, Master equation for a kinetic model of trading market and its analytic solution, Phys. Rev. E 72 (2005), 026126.
- (24) M. Patriarca, A. Chakraborti, K. Kaski, G. Germano, Kinetic theory models for the distribution of wealth: Power law from overlap of exponentials, in: A. Chatterjee, S.Yarlagadda, B. K. Chakrabarti (Eds.), Econophysics of Wealth Distributions, Springer, (2005) 93.
- (25) A. Chakraborti, M. Patriarca, Variational principle for the Pareto power law, Phys. Rev. Lett. 103 (2009) 228701.
- (26) P.K. Mohanty, Generic features of the wealth distribution in an ideal-gas-like market, Phys. Rev. E 74(1) (2006) 011117; U. Basu, P.K. Mohanty, Modeling wealth in growing markets, Eur. Phys. J. B 65 (2008) 585.