cond-mat0203399/12.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%%    12.tex 
3: %%%    written by Atushi Ishikawa, Tadao Suzuki and Masashi Tomoyose
4: %%%    March 20, 2002
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6: \documentstyle[11pt,epsf]{article}
7: %\input {colortextl.mac}
8: %\input 2up
9: %\renewcommand{\baselinestretch}{2.0}
10: 
11: %--------------------------- Latout  -------------------------
12: \setlength{\topmargin}{-45pt}
13: \setlength{\oddsidemargin}{0cm}
14: \setlength{\evensidemargin}{0cm}
15: \setlength{\textheight}{23.7cm}
16: \setlength{\textwidth}{16cm}
17: 
18: %----------------------------- TeX ----------------------------
19: \def\INT{\hbox{\twentyfivepoint $ \displaystyle\int$}}
20: \def\SUM{\hbox{\twentyfivepoint $ \dyesisplaystyle\sum$}}
21: \def\PROD{\hbox{\twentyfivepoint $ \displaystyle\prod$}}
22: \def\Int{\hbox{\seventeenpoint $ \displaystyle\int$}}
23: \def\Sum{\hbox{\seventeenpoint $ \displaystyle\sum$}}
24: \def\Prod{\hbox{\seventeenpoint $ \displaystyle\prod$}}
25: 
26: 
27: %-------------------------------- LaTeX -------------------------
28: \newcommand{\qed}{\hbox{\rule[-2pt]{3pt}{6pt}}}
29: \newcommand{\qqquad}{$\qquad \quad$}
30: \newfont{\subsub}{cmr6}
31: 
32: %%%%%%%%%%%
33: \newcounter{szk}
34: \renewcommand{\theszk}{\Alph{szk}}
35: \newcommand{\szkmark}[1]{$^{\setcounter{szk}{#1} \rm \theszk}$}
36: \newcommand{\tdnmark}{$^{\rm o}$}
37: 
38: 
39: 
40: 
41: %---------------------------
42: \begin{document}
43: 
44: \title{A New Approach to Personal Income Distribution}
45: \author{
46: \footnote{e-mail address: ishikawa@kanazawa-gu.ac.jp} Atushi Ishikawa$^a$, 
47: \footnote{e-mail address: tadao@nanao-c.ac.jp} Tadao Suzuki$^b$ and
48: \footnote{e-mail address: tomoyose@hep.s.kanazawa-u.ac.jp} Masashi Tomoyose$^c$\\ \\ 
49: $^a$ The Organization of Core Curriculum Studies, Kanazawa Gakuin University\\
50: Kanazawa 920-1392, Japan\\
51: $^b$ Department of Management Information, Nanao Junior College\\ 
52: Nanao 926-8570, Japan\\
53: $^c$ Institute for Theoretical Physics, Kanazawa University\\ 
54: Kanazawa 920-1192, Japan}
55: \date{\today}
56: \maketitle
57: 
58: 
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60: %%%       ABSTRACT
61: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
62: \begin{abstract}
63: The results of $R^2$ dynamical random surface model
64: (2-dimensional quantum gravity with a $R^2$ term) 
65: are applied to explain the personal income distribution.  
66: A scale invariance exists if there is not the $R^2$ term in the action.  
67: The $R^2$ term provides a typical scale and breaks the scale invariance explicitly 
68: in the low and middle income range.  
69: A new distribution, Weibull distribution, is deduced from the action analytically
70: in the low income range, and  
71: a consistent fitting is obtained in the whole income range.  
72: Also, we show that the lognormal distribution 
73: in the middle income range can be understood in this framework.  
74: \end{abstract}
75: 
76: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
77: %%%       SECTION
78: %%%       INTRODUCTION
79: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
80: \section{Introduction}
81: The analysis of the personal income distribution is one of 
82: the important subjects in the econophysics \cite{MS}.
83: The first step concerning this research is almost established.
84: There are some empirical knowledge of it.  
85: It is believed that the distribution follows a power law in the high income range 
86: and follows a lognormal distribution in the low-middle income range.
87: The purpose of this paper is to explain this behavior
88: from a viewpoint of $R^2$ dynamical random surface (DRS) model.
89:  
90: The distribution in the high income range is first discussed 
91: by V. Pareto in 1897 \cite{Pareto}.  
92: The probability density function of the personal income $x$ is given 
93: by the following;
94: \begin{eqnarray}
95:  p(x)=Cx^{-(1+\alpha)},  
96:  \label{eq-pareto}
97: \end{eqnarray}
98: where $C$ is a normalization factor. 
99: Also $\alpha$ is the parameter which characterizes
100: the profile of the income distribution, so-called Pareto index,
101: and this power law behavior is known as Pareto law.  
102: 
103: On the other hand, 
104: it is indicated by R. Gibrat in 1931
105: that the probability density function in the low-middle range follows a lognormal
106: distribution as follows \cite{Gibrat};  
107: \begin{eqnarray}
108:  p(x)=\frac{1}{x\sqrt{2\pi\sigma^2}}\exp\biggl[ 
109:  -\frac{\log^2\bigl(\frac{x}{x_0}\bigr)}{2\sigma^2}\biggr],
110:  \label{eq-gibrat}
111: \end{eqnarray}
112: where $x_0$ is a geometric mean value of $x$ and $\sigma^2$ is 
113: a geometric variance.  
114: 
115: In the early '80s, the above two knowledge are rediscovered \cite{Badger}.  
116: The income distributions in United States 1935-1936 are analyzed 
117: in these papers.  
118: Most of the data follows the lognormal distribution, while the higher 
119: 1\% is governed by the power law.  
120: Recently, this fact is reconfirmed \cite{ASNOTT} by using the detailed 
121: Japanese data \cite{URL}.  
122: 
123: The personal income distribution has been an interesting object of study.  
124: What seems to be lacking is a model
125: from which we can deduce the features in the whole income range.  
126: For example there are models employed a stochastic evolution equation. 
127: It is easy to derive a power law, however,
128: it is hard to lead a lognormal distribution by using these models \cite{LS}.  
129: Because the basic idea of these models is 
130: that they contain no typical scale in the fundamental equations.  
131: We should comment an exception.  
132: In recent paper \cite{SFA}, the authors employed a concept of 
133: complex networks \cite{Watts} and a stochastic evolution 
134: equation with some interactions.
135: As a consequence, they succeed to construct a power law and lognormal 
136: distribution in the whole range.  
137: There is, however, not an analytic consideration for creation of the lognormal 
138: distribution but only a phenomenological one by using computer simulations.   
139: 
140: Our viewpoint is simple as follows.  
141: There is no typical scale concerning to money in the high income range.  
142: The distribution of it, therefore, should obey a power law.  
143: On the contrary, there is a typical scale in the low-middle income range, 
144: and it causes the breaking of the power law distribution.  
145: As a toy model which can realize above two cases in the same framework, 
146: we employ $R^2$ DRS model which is one of the methods for studying 
147: 2-dimensional quantum gravity with a typical scale.
148: 2-dimensional gravity without a $R^2$ term has no typical scale, 
149: unlike 4-dimensional gravity. 
150: Here $R$ means the scalar curvature of a 2-dimensional surface.  
151: The dimension of $R$ is (length)$^{-2}$, namely the $R^2$ term 
152: is dimensionfull.  
153: The $R^2$ term, therefore, provides a typical scale to 2-dimensional gravity. 
154: It is well known that this model has two ranges.  
155: A fractal and self-similar structure, $i.e.$ a power law, is seen in the range
156: where the typical scale does not affect the theory.
157: On the other hand, the fractal breaks down
158: in the other range where the scale has the sense. 
159: This peculiarity is similar to the profile of the personal income distribution. 
160: 
161: In order to confirm our view in the framework of $R^2$ DRS model, 
162: it is straight to adopt the money itself as the fundamental variable of theory
163: not the money possessed by individuals.   
164: We have once forgotten the concept of an individual 
165: and pay attention only to the connection of the money in a certain unit.
166: This connection forms a 2-dimensional surface in DRS model,
167: and later the money of an individual is formed geometrically.  
168:  
169: The outline of this paper is as follows. 
170: Section \ref{sec-minbu} is a brief review of the formalism upon which 
171: our work is based.  
172: $R^2$ DRS is explained and MINBU analysis that is one of the methods 
173: for studying the property of random surface is described.  
174: The aspects of continuum quantum $R^2$ gravity in 2-dimensions 
175: and the two asymptotic forms of the partition function are presented 
176: in section \ref{sec-R2Gin2D}.  
177: These two forms correspond to the distribution functions in the high 
178: and the low income ranges, respectively.   
179: The interpretation and relation between quantities in $R^2$
180: DRS model and ones in the personal income are discussed 
181: in section \ref{sec-map}.  
182: In section \ref{sec-fitting}, the practical fitting is shown.   
183: The personal income distributions of United States in the year 1935-1936 and 
184: Japan in the year 1997-2000 are employed for its purpose.  
185: Summary and discussion will be given in section \ref{sec-summary}.  
186: 
187: 
188: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
189: %%%       SECTION
190: %%%       R^2  RANDOM  SURFACE  MODEL  AND  MINBU  ANALYSIS
191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
192: %\section{$R^2$ dynamical random surface model and MINBU Analysis}
193: \section{$R^2$ DRS Model and MINBU Analysis}
194: \label{sec-minbu}
195: In this section, we review DRS and explain MINBU analysis which 
196: is one of the methods 
197: for studying DRS \cite{JM}.  
198: DRS model is known as dynamical triangulation (DT) \cite{KAD} in particle 
199: physics, and DRS with a $R^2$ term 
200: has a fractal and a non-fractal ranges \cite{ITY}.    
201: In DRS, we consider a 2-dimensional surface.  
202: This surface is constituted 
203: by attaching the sides of simplexes (for instance equilateral triangles), 
204: and all possible constitutions are statistically averaged. 
205: 
206: The typical surface is shown in 
207: Fig.~\ref{fig:2-dim Random Surface}.  
208: This surface has a self-similar structure.  
209: A simply connected region of area with one boundary
210: is called a baby universe.
211: In the dynamically triangulated surface,
212: a minimum neck between baby universes is composed of three sides of triangles
213: and can be identified uniquely.
214: The distribution of minimum neck baby universe (MINBU) 
215: is, therefore, one of the important observable quantities in DRS (DT).
216: In order to analyze it,
217: we consider the case that a whole universe (area $A$) is divided 
218: into two baby universes
219: (area $A-B$, $B$),
220: paying attention to one of minimum necks (Fig.~\ref{fig:Divided MINBs}).
221: We denote that $Z(A+1)$ is the fixed area $A+1$ partition function,
222: and that $Z(A, l)$ is the partition function for surfaces (area $A$) 
223: with one boundary (length $l$).
224: MINBU (for minimum boundary loop length $l=3$) can be recognized as a surface 
225: from which one triangle is removed from a sphere.
226: We can correlate these two partition functions,
227: by counting ways which triangle is chosen and removed,
228: as follows;
229: \begin{eqnarray}
230:  Z(A, 3)\sim (A+1)Z(A+1).
231: \end{eqnarray}
232: The partition functions of two MINBUs in Fig.~\ref{fig:Divided MINBs} are
233: given by
234: \begin{eqnarray}
235:  Z(B, 3)&\sim& (B+1)Z(B+1),
236:  \label{B}
237: \\
238:  Z(A-B, 3)&\sim& (A-B+1)Z(A-B+1).
239:  \label{A-B}
240: \end{eqnarray}
241: We define, here, that $n_A (B)$ is the statistical average number of MINBUs
242: of fixed area $B$ on a closed 2-dimensional surface 
243: of area $A$.
244: If the 2-dimensional surfaces once divided into two MINBUs 
245: are made to be one universe again, the following expression
246: can be obtained; 
247: \begin{eqnarray}
248:  n_A(B)\sim \frac{Z(B, 3) Z(A-B, 3)}{Z(A)}.
249: \end{eqnarray}
250: This formula works well as long as we treat MINBUs.
251: 
252: 
253: 
254: It is well known that DT is equivalent to the continuum theory \cite{Takusan}.
255: If we take the continuum limit 
256: which brings every sides of triangles to the infinitesimal and
257: the number of triangles to the infinity simultaneously
258: keeping the area of a 2-dimensional surface constant, 
259: it is thought that this limit becomes equivalent to 2-dimensional gravity
260: defined by the continuous variables.  
261: The explicit forms of the partition function $Z$ and $n_A(B)$ is obtained 
262: by using the knowledge 
263: of 2-dimensional continuum quantum gravity.  
264: In next section, we consider it.  
265: 
266: 
267: 
268: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
269: %%%       SECTION
270: %%%       CONTINUUM QUANTUM  R^2  GRAVITY  IN  2-DIMENSIONS
271: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
272: \section{Continuum Quantum $R^2$ Gravity in 2-dimensions}
273: \label{sec-R2Gin2D}
274: 2-dimensional quantum gravity with a $R^2$ term is investigated 
275: in the continuum 
276: framework in ref.\cite{KN}.
277: The partition function for fixed area $A$ is given by
278: \begin{equation}
279: Z=\int \frac{{\cal D} g %{\cal D} \chi 
280:     {\cal D} X}{\rm vol(Diff)}
281:     {\rm e}^{-S(g)   %_{\chi}(\chi; g)
282:      -S_M(X^i; g)} 
283:     \delta \left(\int {\rm d^2}x \sqrt{g}-A \right),
284: \label{partition}
285: \end{equation}
286: where $g_{\mu \nu}(\mu, \nu=0, 1)$ is a metric, $X^i(i=1,2,\cdots ,d)$
287: is a scalar matter field
288: and $d$ is the number of scalar fields.
289: Also $R$ is the scalar curvature of the 2-dimensional surface.  
290: The $\delta$-function fixes the area to $A$.  
291: The actions are defined as
292: \begin{eqnarray}
293: S(g)&=&\frac{1}{8 \pi} \int {\rm d^2}x \sqrt{g} \left(
294:     \frac{1}{4 m^2}R^2 + 4 \mu_0 \right),
295: \label{action chi}
296: \\
297: S_M(X^i;g)&=&\frac{1}{8 \pi} \int {\rm d^2}x \sqrt{g} g^{\mu \nu} 
298: \partial_{\mu} X^i \partial_{\nu} X_i,
299: \label{action X}
300: \end{eqnarray}
301: where $\mu_0$ and $m$ are a bare cosmological constant 
302: and a coupling constant of length dimension $-1$, respectively.  
303: The actions (\ref{action chi}), (\ref{action X}) and the integration measure 
304: ${\cal D}g, {\cal D}X$
305: in the partition function (\ref{partition}) are invariant under 2-dimensional 
306: diffeomorphisms.
307: The measure, therefore, should be divided by the volume of the diffeomorphisms,
308: which is denoted by $\rm vol(Diff)$.
309: The partition function (\ref{partition}) carried out by an appropriate 
310: gauge fixing has the exact expression.   
311: The partition function, however, contains 
312: infinite excited modes and is difficult to integrate the full functional 
313: completely.
314: In Ref.~\cite{KN}, instead of the full integration, the two asymptotic 
315: forms are shown:
316: \begin{eqnarray}
317: Z(A;m,\mu)&\sim& {\rm const.}~A^{\gamma_{\infty }(d)-3} 
318: {\rm e}^{-\frac{\mu A}{2 \pi}}
319:                             ~~~~~~~~~~~~~~~~~~
320:                             {\rm for}~ m^2 A\rightarrow \infty,
321: \label{infty}                          
322: \\
323: &\sim& {\rm const.}~A^{\gamma_0(d, m^2 A)-3} {\rm e}^{-\frac{\mu A}{2 \pi}
324:                     -\frac{2 \pi}{m^2 A}}
325:                             ~~~~~~{\rm for}~ m^2 A\rightarrow 0,
326: \label{0}                               
327: \end{eqnarray}
328: where 
329: $\mu$ is a renormalized cosmological constant, and
330: \begin{eqnarray}
331: %\gamma_{\infty }(h, d)&=&\frac{d-25-\sqrt{(25-d)(1-d)}}{12}(1-h)+2,
332: \gamma_{\infty }(d)&=&\frac{d-25-\sqrt{(25-d)(1-d)}}{12}+2,
333: \label{gamma_infty}
334: \\
335: \gamma_0(d, m^2 A)&=&2+\frac{(d-12)}{6},
336: \label{gamma_0}
337: \end{eqnarray}
338: are string susceptibilities of 2-dimensional gravity.
339: 
340: As for $m^2 A\rightarrow \infty$,
341: the asymptotic form of the partition function (\ref{infty})
342: represents that the surface is fractal.
343: The similar phenomenon can be seen in 2-dimensional gravity 
344: without the $R^2$ term \cite{BIPZ},
345: where no typical length scale exists.
346: In this $A \gg 1/{m^2}$ range, even if the model contains the $R^2$ term,
347: at an area scale much larger than $1/{m^2}$ surfaces are smoothly fractal.  
348: On the other hand, as for $m^2 A\rightarrow 0$,
349: the partition function (\ref{0}) is highly suppressed by an exponential factor
350: ${\rm exp}[-\frac{2 \pi}{m^2 A}]$, hence, the fractal structure
351: of 2-dimensional surface is broken.
352: This phenomenon can be understood as follows.
353: In the $A \ll 1/{m^2}$ range,  
354: at an area scale much smaller than $1/{m^2}$ surfaces
355: are affected by the typical length scale, and they are not fractal.
356: 
357: This is the model for which we were asking, however,
358: the partition function (\ref{infty})
359: cannot be made to correspond to Pareto law (\ref{eq-pareto}) directly. 
360: Because area $A$ in Eq.(\ref{infty}) or (\ref{0}) 
361: should be correspond to the whole money,
362: on the other hand, $x$ in Eq.(\ref{eq-pareto}) or (\ref{eq-gibrat}) 
363: is the individual income. 
364: 
365: In order to connect this model to the personal income, we should consider
366: MINBU analysis in the previous section.
367: By substituting the asymptotic forms (\ref{infty}), (\ref{0}) to 
368: the partition function (\ref{B}), (\ref{A-B}),
369: We can find the following two asymptotic formulae;
370: \begin{eqnarray}
371:  n_A(B)&\sim& {\rm const.}~
372:  %A~
373:  \biggl[\Bigl(1-\frac{B}{A}\Bigr)B\biggr]^{\gamma _\infty -2}
374: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
375: \mbox{ for $m^2(1-\frac{B}{A})B\rightarrow \infty $},
376:  \label{eq-n_A-hight}\\
377: &\sim& {\rm const.}~
378: \biggl[\Bigl(1-\frac{B}{A}\Bigr)B\biggr]^{\gamma_0-2}
379:       %\!\!\!
380:       \exp\biggl[ -\frac{2\pi}{m^2} \frac{1}{\Bigl(1-\frac{B}{A}\Bigr)B}\biggr]
381: ~~   \mbox{ for $m^2(1-\frac{B}{A})B\rightarrow 0 $}.
382:  \label{eq-n_A-low}
383: \end{eqnarray}
384: From these expressions, we can recognize that
385: the distribution of MINBUs obeys a power law in the limit 
386: $m^2(1-B/A) B\to\infty$,
387: on the other hand, it does not follow any power law in the limit
388: $m^2(1-B/A) B\to 0$.  
389: 
390: 
391: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
392: %%%       SECTION
393: %%%       THE  MAP  BETWEEN  R^2  DRS
394: %%%       AND  THE  PERSONAL  INCOME
395: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
396: \section{The Map between $R^2$ DRS and the Personal Income}
397: \label{sec-map}
398: In the previous section, we analyze how the 2-dimensional surface 
399: (the whole universe)
400: is distributed to MINBUs.
401: Let us consider a relation between the quantities of $R^2$ DRS
402: and those of the personal income.
403: By comparison between Pareto law~(\ref{eq-pareto}) and 
404: Eq.~(\ref{eq-n_A-hight}),  
405: we propose a correspondence relation;
406: \begin{eqnarray}
407:  \Bigl(1-\frac{B}{A}\Bigr)B&\Leftrightarrow & x,
408:  \label{eq-map01}\\
409:  \frac{n_A(B)}{N}&\Leftrightarrow & p(x),
410:  \label{eq-map02}
411: \end{eqnarray}
412: where $N$ is a normalization factor.
413: We should note that MINBU is, in general, defined as the smaller universe
414: when the whole universe is divided into two parts, therefore, $B \le A/2$
415: ($x \le A/4$). From this correspondence, 
416: we can correlate the string susceptibility to Pareto index
417: as follows;
418: \begin{eqnarray}
419:  \gamma_{\infty}=1-\alpha.
420:  \label{eq-map_gamma-pareto}
421: \end{eqnarray}
422: Since Pareto index $\alpha$ is reported as $1.5$ in \cite{MS, Pareto, Badger}, 
423: we should take the value of $\gamma_{\infty}$ as $-\frac{1}{2}$.  
424: This means that we should chose the corresponding 2-dimensional surface 
425: as a sphere without any matters ($d=0$),
426: and we can conclude that $\gamma_{0}=0$ by using Eq.~(\ref{gamma_0}).    
427: Two asymptotic distributions are, in the end, acquired;  
428: \begin{eqnarray}   
429: p(x)&\sim& {\rm const.}~x^{-\frac{5}{2}} 
430: \qquad\qquad\qquad\qquad\qquad\qquad   
431: \mbox{ for $m^2x\rightarrow \infty $},
432:      \label{eq-distribution-high}\\
433: &\sim& 
434:     {\rm const.}~x^{-2}
435:   \exp\biggl[-\frac{2\pi}{m^2}\frac{1}{x}\biggr]
436: \qquad\qquad\qquad~   \mbox{ for $m^2x\rightarrow 0 $}.
437:      \label{eq-distribution-lowmiddle}
438: \end{eqnarray}
439: Here $m$ is a typical area scale.  
440: We should notice that Eq.~(\ref{eq-distribution-lowmiddle}) is known as 
441: Weibull distribution.  
442: 
443: The asymptotic distribution (\ref{eq-distribution-lowmiddle}) is correspond to
444: the low income one, and it is not a
445: lognormal distribution which is offered by many researchers.  
446: This fact must be paid attention.  
447: The form of Eq.~(\ref{eq-distribution-lowmiddle}) obtained by our argument
448: agrees with the practical data in the low income range.  
449: Particularly its fitting is better than the lognormal one 
450: in the low income range.  
451: 
452: 
453: 
454: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
455: %%%       SECTION
456: %%%        THE  DATA FITTING  OF  THE  INCOME  DISTRIBUTIONS
457: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
458: \section{The Data Fitting of the Income Distributions}
459: \label{sec-fitting}
460: In this section, we attempt to fit the personal income distributions of Japan 
461: in several years 1997-2000 \cite{ASNOTT, URL}, 
462: and the United States in the year 1935-1936 \cite{Badger}.   
463: For this purpose, we define a cumulative probability 
464: %to deal with cumulative number 
465: as follows:
466: \begin{eqnarray}
467:  P(x\leq)=\int_x^{\infty}dy\ p(y).
468: \end{eqnarray}
469: Substituting Eqs.~(\ref{eq-distribution-high}) and 
470: (\ref{eq-distribution-lowmiddle}) 
471: for this definition, 
472: we obtain below formulae:
473: \begin{eqnarray}
474:  P(x\leq)&\sim& C_{\infty}~x^{-\frac{3}{2}} 
475:   \qquad\qquad\qquad\qquad\qquad~~~~   \mbox{ for $m^2x\rightarrow \infty $},
476:   \label{eq-distribution-high-cum}
477: \\
478: &\sim&  
479: 1- C_0~ \exp\biggl[-\frac{2\pi}{m^2}\frac{1}{x}\biggr]
480:   \qquad\qquad\qquad  \mbox{ for $m^2x\rightarrow 0 $},
481:   \label{eq-distribution-lowmiddle-cum}
482: \end{eqnarray} 
483: where $C_{\infty}$ and  $C_{0}$ are normalization factors.
484: 
485: Firstly, let us consider the distribution in the high income range.  
486: In our model, Pareto index $\alpha$ takes the value of $1.5$. 
487: We find that this index is almost consistent with 
488: the result of \cite{MS, Pareto, Badger} 
489: (see U.S. data in Fig.~\ref{fig:1997-99-japan_1935-36-U.S.}).
490: Also we should notice that this value is little larger than 
491: that in Japan for the year 1997-2000 \cite{ASNOTT}
492: (see the Japanese data in 
493: Figs.~\ref{fig:2000-japan} and~\ref{fig:1997-99-japan_1935-36-U.S.}).  
494: 
495: Secondly, the profile in "the low" income range is considered.  
496: This paper gives the weight to analyze this distribution.  
497: It must be noted that "the low" does not mean the low and middle.  
498: This point of view is one of the main differences from previous works.  
499: The plots of the data and the best fitting are also presented in 
500: Figs.~\ref{fig:2000-japan} and~\ref{fig:1997-99-japan_1935-36-U.S.}.  
501: The parameter $2\pi/m^2$ in the figures can be obtained as the slope 
502: in the graph 
503: where the horizontal axis is taken 
504: as $1/x$ and the vertical axis is taken as 
505: $\log\bigl(1-P(x\leq)\bigr) $
506: in the limit $m^2x\to 0$.  
507: 
508: In our model, 
509: the parameter $m$ specifies the aspect of the breaking of a power law, 
510: and we are able to make a best fitting by changing the value of it.  
511: The best fit values of $2\pi/m^2$ in Japan 1997-2000  
512: and in U.S. 1935-36 are given as 
513: $9.48$, $10.8$, $8.43$, $7.93$,  
514: and $2050$, respectively (Figs.~\ref{fig:2000-japan} and 
515: \ref{fig:1997-99-japan_1935-36-U.S.}).
516: We have employed the coefficient of determination $r^2$ for the parameter that 
517: indicates 
518: the appropriateness of practical fitting:
519: \begin{eqnarray}
520:  r^2 = \frac{\Sigma(f_i -\langle F\rangle)^2}
521: {\Sigma(F_i - \langle F\rangle)^2}.
522: \end{eqnarray}
523: Here $F_i$ is the $i$th value of data, $f_i$ is the value of 
524: the fitting function 
525: and $\langle F\rangle$ is the mean value of $F_i$.  
526: 
527: It remains an unsettled question 
528: what distribution function should be employed in the middle income range.  
529: We have to inquire, to some extent, into this subject.  
530: It is believed that the distribution function is a lognormal one 
531: in this range.  
532: It is difficult to deduce a lognormal distribution 
533: from the partition function (\ref{partition}) analytically.  
534: We employed, therefore, a computer simulation of DT with a $R^2$ term 
535: $\beta_L \frac{4\pi^2}{3} \sum_{i} \frac{(6-q_i)^2}{q_i}$ \cite{ITY}
536: to investigate a profile in the middle income range, 
537: $i.e. x$ is not so high neither low.
538: Here $\beta_L$ is a coupling constant and $q_i$ is the number 
539: of links at the vertex in DRS.  
540: The result is shown in Fig.~\ref{fig:Log_simu} as the data of the 
541: distribution of MINBUs
542: which is not cumulated. 
543: The horizontal and the vertical axes are
544: the logarithm of $(1-A/B)B$ and $n_A(B)$, respectively.  
545: As the figure indicates, 
546: the profile in the middle range can be well approximated with 
547: a secondary curve,
548: and this behavior deduces a lognormal distribution.  
549: It seems that we can obtain the whole personal income distribution 
550: from the partition function (\ref{partition}).  
551: 
552: %Fig:: fig:Log_simu
553: 
554: %\newpage
555: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
556: %%%       SECTION
557: %%%       SUMMARY  AND  DISCUSSION
558: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
559: \section{Summary and Discussion}
560: \label{sec-summary}
561: 
562: In this paper we have proposed the $R^2$ dynamical random surface (DRS) 
563: model as a toy model to explain the profile of the personal income distribution.
564: The money in a certain unit have been identified with a certain unit area 
565: (simplex) of a 2-dimensional surface.
566: The connection of the unit money have been corresponded to the attaching 
567: of simplexes' sides which forms a 2-dimensional surface.  
568: On the basis of this correspondence, we have analyzed the distribution of the 
569: 2-dimensional surface statistically as the personal income distribution 
570: by using MINBU analysis.  
571: 
572: The point is that $R^2$ DRS model has a typical scale $1/m^2$, 
573: which is the coefficient of the $R^2$ term, 
574: and that the distribution of MINBUs exhibits two distinct phases.
575: One is fractal in the large MINBU range where 
576: the typical scale does not affect the DRS, and
577: the other is not fractal in the middle-small MINBU range
578: where small MINBUs are suppressed by the effect of the $R^2$ term.
579: On the other hand, this fact could be confirmed by the continuum theory
580: which analytically presented the two asymptotic forms in these two region.  
581: 
582: Moreover, we have proposed that
583: the personal income $x$ should be corresponded to $(1-B/A)B$,
584: where $A$ is the whole area of a 2-dimensional surface and $B$ 
585: is the area of a MINBU. 
586: From this assumption, the power law distribution has re-obtained in the high 
587: income range similarly to Refs.~\cite{Pareto, Badger, ASNOTT, LS, SFA}.  
588: In addition, it has deduced directly from the partition function 
589: in our method.  
590: At the same time, we have identified that Pareto index 
591: $\alpha = 1-\gamma_{\infty}$,
592: where $\gamma_{\infty}$ is the parameter which characterizes the profile of
593: the 2-dimensional surface in the limit $x \to \infty$, so called the 
594: string susceptibility.   
595: We have also concluded that we should chose the corresponding surface
596: as a sphere without any matter fields ($d=0$). 
597:   
598: We have argued, in particular, the distribution functions in the low and
599: the middle income range in detail.  
600: The new distribution function so-called Weibull function has been proposed 
601: in the low income limit.  
602: This function has well fitted to the data rather than the lognormal 
603: distribution.   
604: A method of least squares has been employed to evaluate a validity 
605: of the data fitting.
606: Also we have employed the computer simulation to discussed 
607: the distribution function  in the middle income range.  
608: Our whole results were displayed in Fig.~\ref{fig:2000-japan} and 
609: Fig.~\ref{fig:1997-99-japan_1935-36-U.S.}.  
610: 
611: In our argument, we assume %there was a logical jump 
612: that the personal income $x$ was identified with the area $(1-B/A)B$. 
613: We try to explain this identification below.
614: In MINBU analysis, a boundary of a MINBU is a minimum neck,
615: where two universes are correlated through only three sides of 
616: simplexes each other (See Fig.~\ref{fig:Divided MINBs}).
617: This means that the connection between universes is very weak 
618: at that point, therefore,
619: it is natural to consider that the minimum neck separates 
620: the money which an individual possesses. 
621: 
622: We should notice that the area of a small MINBU is double counted 
623: as the area of bigger 
624: one by definition.  
625: This may be because why the income which an individual obtains
626: should be added to the income of those who have obtained more income.
627: The effect $-B/A$ can be neglect when $B$ is much smaller 
628: than $A$.  
629: We must, however, take into account of it when $B$ is close to $A/2$.
630: In our model, there is the huge mother universe
631: from which all MINBUs are removed (See Fig.~\ref{fig:2-dim Random Surface}).
632: It is hard to find the object corresponding to this mother universe 
633: in the real economics, however,
634: the theoretical distribution obtained has well expressed the real data.
635: There may be virtual heat bath of the money in the real world. 
636: These arguments are left behind as a future subject.
637: 
638: 
639: In this paper, we have shown that the distribution of the personal 
640: income could be understandable as
641: the geometry of a 2-dimensional surface of a sphere without no matter field.
642: This suggests that the technique developed for the analysis of 
643: spacetime structure 
644: can be applied also to the analysis of our social structure.
645: In 2-dimensions, above geometry is simplest one and
646: the geometry of a 2-dimensional surface with genus and matter 
647: fields is also well understood.
648: There are many fractal objects in the econophysics, for instance 
649: the distribution of the company profit, 
650: and the extended geometry may suit these objects.
651: We are pleased if this analysis becomes a first step for the research 
652: which considers our society geometrically.  
653: 
654: 
655: 
656: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
657: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
658: 
659: \section*{Acknowledgements}
660: 
661: The authors would like to express our gratitude to Dr. W. Souma,
662: Dr. Y. Fujiwara and Professor H. Terao for valuable advices and discussions.  
663: We are also grateful to Professor H. Kawai for useful comments
664: especially about his work \cite{KN}. 
665: Thanks are also due to members of YITP,
666: where one of the authors (A. I.) stayed several times during the completion 
667: of this work. 
668: One of the author (T. S.) is also grateful to Professor Y. Yamagishi 
669: for useful discussions.
670: 
671:  
672: 
673: 
674: %------------------------------ Bibliography -------------------------
675: \begin{thebibliography}{99}
676: 
677: \bibitem{MS}
678:     R.N.~Mategna and H.E.~Stanley, An Introduction to Econophysics, 
679: (Cambrige Univ. Press, U. K., 2000).
680: \bibitem{Pareto}
681:     V.~Pareto, Cours d'Economique Politique, (Macmillan, London, 1897).  
682: %\bibitem{Mandelbort}
683: %    B.P.~Mandelbort, The Fractal Geometry of Nature, 
684: %(Freeman, San Francisco, 1982).    
685: \bibitem{Gibrat}
686:     R.~Gibrat, Les inegalits economiques (Paris, Sirey, 1931).    
687: \bibitem{Badger}
688:     W.W.~Badger, Mathematical models as a tool for the social science, 
689: ed. B. J. West, 
690:     (New York, Gordon and Breach, 1980), pp. 87;\\
691:     E.W.~Montrll and M.F.~Shlesinger, J. Stat. Phys. 32, 209 (1983).  
692: \bibitem{ASNOTT}
693:     H.~Aoyama, W.~Souma, Y.~Nagahara, H.P.~Okazaki, H.~Takayasu and 
694: M.~Takayasu, 
695: Fractals 8, 293 (2000);\\
696:     W.~Souma, to be appear in Fractals 9, (2001), cond-mat/0011373.
697: \bibitem{URL}
698:     The Japanese Tax Administration, http://www.nta.go.jp.
699: \bibitem{LS}
700:     M.~Levy and S.~Solomon, Int. J. Mod. Phys. C7, 595 (1997);\\
701:     H.~Takayasu, A.H.~Sato and M.~Takayasu, Phys. Rev. Lett. 79, 966 (1997);\\
702:     J.~P.~Bouchaud and M.~Mezard, Physica A282, 536 (2000);\\
703:     J.P.~Bouchaud, arXiv: cond-mat/0008103;\\
704:     D.~Sornette, arXiv: cond-mat/0011088.
705: \bibitem{SFA}
706:     W.~Souma, Y.~Fujiwara and H.~Aoyama, cond-mat/0108482, KUCP0189.
707: \bibitem{Watts}
708:     D.J.~Watts and S.H.~Strogatz, Nature 393, 440 (1998);\\
709:     D.J.~Watts, Samll Worlds: The Dynamics of Networks between 
710: Order and Randomness. (Princeton
711:     University Press, Princeton, New Jersey 1999);\\
712:     M.E.~Newman, cond-mat/0001118;\\
713:     S.H.~Strogatz, Nature 410, 268 (2001);\\
714:     R.Albert and A.-L.~Barabasi, cond-mat/0106096;\\
715:     S.N.~Dorogovtsev and J.F.F.~Mendes, cond-mat/0106144. 
716: \bibitem{JM}
717:     S.~Jain and S.D.~Mathur, Phys. Lett. B286, 239 (1992);\\
718:     J.~Ambijorn, S.~Jain and G.~Thorleifsson, Phys. Lett. B307, 34 (1993).
719: \bibitem{KAD}
720:     V.A.~Kazakov, I.K.~Kostov and A.A.~Migdal, 
721:     Phys. Rev. Lett. 66, 2051 (1991);\\
722:     J.~Ambjorn, B.~Durhuus and J.~Frohlich, Nucl. Phys. B257, 433 (1985);\\
723:     F.~David, Nucl. Phys. B 257, 543 (1985).
724: \bibitem{ITY}
725:     S.~Ichinose, N.~Tsuda and T.~Yukawa, Int. Jour. M. Phys. A12, 757 (1997).
726: \bibitem{KN}
727:     H.~Kawai and R.~Nakayama, Phys. Lett. B306, 224 (1993).
728: \bibitem{BIPZ}
729:     E.~Brezin, C.~Itzykson, G.~Parisi and J.B.~Zuber, Commun. 
730: Math. Phys. 59, 35 (1978);\\
731:     F.~David, Mod. Phys. Lett. A3, 1651 (1988);\\
732:     J.~Distler and H.~Kawai, Nucl. Phys. B321, 509 (1989);\\
733:     V.G.~Knizhinik, A.M.~Polyakov and A.B.~Zamolodchikov, 
734: Mod. Phys. Lett. A3, 819 (1998).   
735: \bibitem{Takusan}
736:     E.~Brezin and V.~Kazakov, Phys. Lett. B 236, 144 (1990);\\
737:     M.~Douglas and S.~Shenker, Nucl. Phys. B 335, 635 (1990);\\
738:     D.J.~Gross and A.~Migdal, Phys. Rev. Lett. 64, 127 (1990);\\
739:     M.~Douglas, Phys. Lett. B. 238, 176 (1990);\\
740:     M.~Fukuma, H.~Kawai and R.~Nakayama, Int. J. Mod. Phys. A6, 1385 (1991); 
741:     Commun. Math. Phys. 143, 371 (1992); 148, 101 (1992);\\
742:     R.~Dijkgraaf, E.~Verlinde and H.~Verlinde, Nucl. Phys. B 348, 435 (1991);\\
743:     A.~Jevicki and T.~Yoneya, Mod. Phys. Lett. A5, 1615 (1990);\\
744:     P.~Ginsparg, M.~Goulian, M.~Plesser and J.~Zinn-Justin, 
745: Nucl. Phys. B342, 539 (1990);\\
746:     T.~Yoneya, Commun. Math. Phys. 144, 623 (1992);
747:     Int. J. Mod. Phys. A7, 4015 (1992);\\
748:     J.~Goeree, Nucl. Phys. B358, 737 (1991).
749: 
750: %\ma
751: \end{thebibliography}
752: 
753: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
754: %%%%%%%%%%%%% FIGURE ( PLOTS )%%%%%%%%%%%%%
755: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
756: 
757: \begin{figure}[htb]
758:  \begin{minipage}{0.49\textwidth}
759:   \begin{center}
760:    \epsfysize=50mm
761:    \leavevmode
762:    \epsfbox{e1.eps}
763:   \end{center}
764:   \vspace{-5mm}
765:   \caption{Fractal 2-dim Random Surface}
766:   \label{fig:2-dim Random Surface}
767:  \end{minipage}
768:  \begin{minipage}{0.49\textwidth}
769:   \begin{center} 
770:    \epsfysize=50mm
771:    \leavevmode
772:    \epsfbox{e2.eps}
773:    \caption{Divided MINBs}
774:    \label{fig:Divided MINBs}
775:   \end{center}
776:  \end{minipage}
777: \end{figure}
778: 
779: 
780: \begin{figure}[htb]
781:  \centerline{\epsfxsize=0.65\textwidth\epsfbox{h12.eps}}
782:  \caption{Cumulative Probability for 2000 Japan}
783:  \label{fig:2000-japan}
784: \end{figure}
785: 
786: %\newpage
787: 
788: \begin{figure}[htb]
789: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
790: %%%%%%%% 1st Line (two pictures)%%%%%%%
791: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
792:  \begin{minipage}[htb]{0.49\textwidth}
793:   \epsfxsize = 1.0\textwidth
794:   \epsfbox{h9.eps}
795:  \end{minipage}
796:  \hfill
797:  \begin{minipage}[htb]{0.49\textwidth}
798:   \epsfxsize = 1.0\textwidth
799:   \epsfbox{h10.eps}
800:  \end{minipage}
801: % \caption{1996 and 1997 Japan Income}
802: % \label{1996-97-japan}
803: %\end{figure}
804: %%%%%%%%%%%% End 1st Line %%%%%%%%%%%%%
805: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
806: %%%%%%%% 2nd Line (two pictures)%%%%%%%
807: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
808: %\begin{figure}[htb]
809:  \begin{minipage}[htb]{0.49\textwidth}
810:   \epsfxsize = 1.0\textwidth
811:   \epsfbox{h11.eps}
812:  \end{minipage}
813:  \hfill
814:  \begin{minipage}[htb]{0.49\textwidth}
815:   \epsfxsize = 1.0\textwidth
816:   \epsfbox{ba.eps}
817:  \end{minipage}
818:  \caption{Cumulative Probability for 1997-1999 Japan and 1935-36 U.S.}
819:  \label{fig:1997-99-japan_1935-36-U.S.}
820: \end{figure}
821: %%%%%%%%%%%% End 2nd Line %%%%%%%%%%%%%
822: 
823: 
824: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
825: %             FIGURE of Simulation
826: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
827: \begin{figure}[htb]
828:  \begin{center}
829:   \epsfxsize=.8\textwidth
830:   \leavevmode
831:   \epsfbox{min.eps}
832:  \end{center}
833:  %\vspace{-5mm}
834:  \caption{Lognormal fitting into the simulation of $R^2$ DRS model}
835:  \label{fig:Log_simu}
836: \end{figure}
837: 
838: %--------------------------------------------------------------
839: 
840: \end{document}
841: 
842: