1:
2: % Use the option doublespacing or reviewcopy to obtain double line spacing
3: % \documentclass[doublespacing]{elsart}
4:
5: \documentclass{elsart}
6:
7: \usepackage{amssymb}
8: \usepackage{latexsym}
9: \usepackage{amsmath}
10: \usepackage{amsfonts}
11:
12: % the epsfig is the package to insert the figures, as an alternative
13: % you can use \usepackage{graphicx}
14:
15: \usepackage{epsfig}
16:
17:
18: \begin{document}
19: \journal{European Physical Journal B}
20: \begin{frontmatter}
21:
22:
23: \title{Wealth distribution and Pareto's law in the Hungarian
24: medieval
25: society}
26:
27:
28: \author[1,2]{G\'eza Hegyi},
29: \author[1,3]{Zolt\'an
30: N\'eda\corauthref{cor1}}\corauth[cor1]{}
31: \ead{zneda@phys.ubbcluj.ro},
32: \and \author[2]{Maria Augusta Santos}
33:
34: \address[1]{Department of Physics, Babe\c{s}-Bolyai
35: University,
36: str. Kog\u{a}lniceanu 1, RO-400084 Cluj-Napoca,
37: Romania}
38:
39: \address[2]{Department of History and Philosophy , Babe\c{s}-
40: Bolyai University,
41: str. Kog\u{a}lniceanu 1, RO-400084 Cluj-Napoca,
42: Romania}
43:
44: \address[3]{Departamento de F\'{\i}sica and Centro de
45: F\'{\i}sica do Porto,
46: Faculdade de Ci\^encias, Universidade do Porto, Rua de Campo
47: Alegre 687,
48: 4169-007 Porto, Portugal}
49:
50:
51: \date{September 2005}
52:
53:
54: \begin{abstract}
55: The distribution of wealth in the Hungarian medieval
56: aristocratic
57: society is reported and studied. The number of serf families
58: belonging to a noble is taken as a measure of the
59: corresponding wealth. Our results reveal the power-law
60: nature of this distribution function,
61: confirming the validity of the Pareto law for such a
62: society. The obtained Pareto index $\alpha=0.92$ is however
63: smaller
64: than the values currently reported in the literature. We
65: argue that the value close to $1$, of the Pareto index is a
66: consequence of the absence of a relevant economic life in
67: the targeted society, in agreement with the prediction of
68: existing wealth distribution models for the idealized case
69: of independently acting agents. Models developed to explain
70: city populations may also be adapted to justify our results.
71:
72: \end{abstract}
73:
74: \begin{keyword}
75: % keywords here, in the form: keyword \sep keyword
76: Wealth distribution \sep Econophysics \sep Pareto's Law
77:
78: % PACS codes here, in the form: \PACS code \sep code
79: \PACS 89.75-k \sep 89.65.Gh \sep 89.75.Hc \sep 87.23.Ge
80:
81: \end{keyword}
82:
83: \end{frontmatter}
84:
85:
86: \section{Introduction}
87:
88:
89: At the end of the XIX century the economist Vilfredo Pareto
90: \cite{pareto} discovered a universal law
91: regarding the wealth distribution in societies. His
92: measurement results on several
93: European countries, kingdoms and cities for the XV-XIX
94: centuries revealed
95: that the cumulative distribution of wealth (the probability
96: that the wealth
97: of an individual is greater than a given value) exhibits a
98: universal functional form.
99: Pareto found that in the region containing the richest part
100: of the population, generally
101: less than the $5\%$ of the individuals, this distribution is
102: well described by a power-law
103: (see for example \cite{ebooks} for a review).
104: The exponent of this power-law is
105: denoted by $\alpha$ and named Pareto index. In the limit of
106: low and medium wealth, the
107: shape of the cumulative distribution is fitted by either
108: an exponential or a log-normal function.
109:
110: The power-law revealed by Pareto has been confirmed by many
111: recent studies on the economy of several corners of the
112: world. The presently available data is
113: coming from so apart as Australia \cite{matteo03}, Japan
114: \cite{aoyama03,fujiwara03}, the US \cite{silva04},
115: continental
116: Europe \cite{fujiwara04,clementi04}, India \cite{india} or
117: the UK \cite{dragulescu01}. The data
118: is also spanning so long in time as ancient Egypt
119: \cite{abul02},
120: Renaissance Europe \cite{souma02} or the $20th$ century
121: Japan\cite{souma01}. Since it is difficult to measure {\it
122: wealth}, most of the available data comes from tax
123: declarations of individual {\it income}
124: (which is assumed to be proportional to wealth). There are
125: however some
126: other databases obtained from different sources, like the
127: area
128: of the houses in ancient Egypt \cite{abul02} or the
129: inheritance taxation
130: or the capital transfer taxes \cite{UK}. The results mostly
131: back
132: Pareto's conjecture on the shape of the wealth distribution.
133:
134: In the present paper, we present and discuss some recent
135: results of wealth distribution in a medieval society -- the
136: Hungarian aristocratic
137: society around the year 1550. The wealth of the nobles is
138: estimated and expressed in the number of owned serf
139: (villein) families, a measure generally used by historians.
140: The case under study offers a somehow idealized example of a
141: society without a relevant wealth-exchange mechanism and our
142: results give further evidence for the universal nature of
143: Pareto's law.
144:
145:
146: \section{Statistical Physics approach to Pareto's law}
147:
148:
149: Typically, the presence of power-law distributions is a hint
150: for the complexity
151: underlying a system, and a challenge for statistical
152: physicists to model
153: and study the problem. This is why Pareto law is one of the
154: main problems studied
155: in Econophysics. Since the value found by Pareto for the
156: scaling exponent was around $1.5$, Pareto law is sometimes
157: related
158: to a generalized form of Zipf's law \cite{zipf} and referred
159: to as Pareto-Zipf law. According to Zipf's law, many natural
160: and social phenomena (distribution of words frequency in a
161: text, population of cities, debit of rivers, users of web
162: sites, strength of earthquakes, income of companies, etc)
163: are characterized by a cumulative distribution function
164: with a power-law tail with a scaling exponent close to $1$.
165:
166: It is however important to notice that, in contrast to what
167: happens with most exponents in Statistical Physics, the
168: Pareto index
169: $\alpha$, may change from one society to another, and for
170: the same society can also
171: change in time depending on the economical
172: circumstances \cite{fujiwara03,souma01}. The measured values
173: of $\alpha$ for the {\it individuals} wealth/income
174: distribution span a quite broad interval, typically in the
175: $1.5-2.5$ range; a recent study of
176: the Indian society \cite{india} reports however, much lower
177: values ($0.81- 0.92$).
178: This large variation of $\alpha$ indicates the absence of
179: universal scaling in this problem -- a feature which models
180: designed to
181: describe the wealth distribution in societies should be able
182: to reproduce.
183:
184:
185: Models for wealth distribution are essentially defined
186: by a group of agents, placed on a lattice, that interchange
187: money following
188: pre-established rules. The system will eventually reach a
189: stationary state where some quantities, as for
190: instance the cumulative distribution of wealth $P_>(w)$,
191: may be measured. Following
192: these ideas, Bouchaud and M\'ezard \cite{bouchaud00} and
193: Solomon and
194: Richmond \cite{solomon01,biham98} separately proposed a
195: very
196: general model for wealth distribution. This model is based
197: on a
198: mean field type scenario with interactions of strength $J$
199: among all the agents
200: and on the existence of multiplicative fluctuations acting
201: on each
202: agent's wealth. The obtained wealth distributions
203: adjust well to the phenomenological $P_>(w)$ curve. Their
204: mean-field
205: results predict that Pareto index $\alpha$ should increase
206: linearly with
207: the strength of interaction between the agents and that
208: $\alpha=1$ for the case of independent agents ($J=0$).
209:
210: Similar conclusions were obtained by Scaffeta et al
211: \cite{scafetta04},
212: who considered a nonlinear version of the model and from
213: other regular
214: lattice based models as those in Refs. \cite{chatterjee03}
215: and \cite{pianegonda03}.
216:
217: Given the known complex nature of social networks (see
218: \cite{dorogovtsev03} for a recent review), models were also
219: considered in which the economic transactions take place on
220: a complex small-world or scale-free network
221: \cite{souma01,iglesias03,garlaschelli04}. More
222: recently, a model was proposed \cite{coelho05} where the
223: network structure is not predefined but rather dynamically
224: coupled to the wealth-exchange rule. This model successfully
225: reproduces both the characteristics of the modern family
226: relation networks and the measured shape of the wealth
227: distribution curves, accounting for variations in $\alpha$
228: in the $1.7-2.0$ interval.
229:
230:
231: From the studies and modeling efforts made by the
232: statistical physics community, one can conclude that the
233: emergence of Pareto law can be understood from the
234: multiplicative nature of the fluctuations of the agents'
235: wealth and from the dynamics of wealth on the underlying
236: complex social network.
237: Both the experimental results and the models suggest that
238: societies in which there is strong wealth-exchange among the
239: agents are characterized by higher values of the Pareto
240: index, whereas societies of isolated agents, where each
241: agent increases or decreases his wealth in a multiplicative
242: and uncorrelated manner, are usually characterized by
243: $\alpha$ close to $1$. Our measurement results on the
244: Hungarian
245: medieval society confirm this rule.
246:
247:
248: \section{Wealth distribution measurements in the Hungarian
249: medieval society}
250:
251:
252: To our knowledge, there is no available data concerning the
253: wealth distribution and the Pareto law for the
254: Central-Eastern European aristocratic medieval societies. A
255: flourishing economic life, barter and wealth exchange
256: developed very slowly in this part of Europe. A centralized
257: and documented taxation system was also introduced
258: relatively late. Of course, this undeveloped economic life
259: makes it hard to collect uniform,
260: relevant and large-scale data on this society. However, once
261: such data is obtained, it could be of decisive importance,
262: since such a society represents an idealized case, where the
263: agents are acting roughly independently. The underlying
264: social network and the wealth exchange on it, are expected
265: to have no major influence, since the aristocratic families
266: were more or less self-supporting and no relevant barter
267: existed.
268: This simplified economic system provides thus an excellent
269: framework to test, in a trivial case, the prediction of
270: wealth distribution models.
271:
272: The first centralized data for 47 districts of the medieval
273: Hungary (10 of them under Turkish occupation)
274: is dated from 1550. The data, in a rough format, is
275: available in a recent book \cite{maksay90}
276: dealing with property relations of the XVI century Hungary.
277: For each district, the nobles are arranged alphabetically
278: and their wealth is grouped in six categories: number of
279: owned serf families and their lands,
280: unused lands, poor people living on their land, new lands,
281: servants and others. Using a method accepted by historians
282: \cite{engel},
283: the number of owned serf families is taken as a common
284: measure of the total wealth of a noble family.
285:
286: After a careful analysis of all districts and summing up the
287: wealth for those families
288: that owned land in several districts, we obtained a dataset
289: that is usable for wealth distribution studies. We also
290: imposed a lower cut-off value, chosen to be 10 serf
291: families, and disregard thus the
292: low and medium wealth aristocrats. This cut-off is necessary
293: since
294: historians suggest that the database is not reliable in this
295: ranges. With the above constraints, our final database
296: \cite{dataset} had data for 1283 noble families and 116
297: religious or city institutions. Considering an average of
298: five persons per family (a generally accepted value by the
299: historians and sociologists specialized in the
300: targeted medieval period) we obtain that our sample contains
301: around 6400 people. This represents the top $8\%$ of the
302: estimated $80000$ aristocrats living in Hungary at that
303: period, and $0.2-0.3\%$ of the estimated total population
304: (2.7 millions) of Hungary in 1550.
305:
306: Ranking the considered families after their decreasing
307: wealth and plotting the rank as a function of wealth for
308: each family,
309: gives the tail of the cumulative distribution function for
310: wealth up to a proportionality
311: constant (equal to $1/N_t$, where $N_t$
312: is the total number of families in the
313: considered society). According to Pareto law, this curve
314: should have a power-law nature, thus plotting it on a log-
315: log scale should give a straight line with a negative slope,
316: which yields the Pareto index. The results obtained from our
317: dataset are
318: plotted in Figure 1.
319:
320:
321:
322: \begin{figure}
323: \begin{center}
324: \epsfysize=100mm
325: \epsffile{fig1.eps}
326: \caption{The rank of the top $8\%$ aristocrat families and
327: institutions
328: as a function of their estimated total wealth
329: on a log-log scale. Measurement results for the Hungarian
330: noble society in the year 1550.
331: The total wealth of a family is estimated as the number of
332: owned serf families.
333: The power-law fit suggests a Pareto index $\alpha=0.92$.}
334: \label{fig1}
335: \end{center}
336: \end{figure}
337:
338:
339: The power-law scaling is nicely visible on two decades in
340: the figure, thus confirming Pareto law with an exponent
341: $\alpha=0.92$. This value is comparable with the ones
342: obtained for the Indian society \cite {india} and much
343: smaller than all the other values reported in the
344: literature.
345:
346: As explained above, a value of $\alpha$ close to $1$ is in
347: agreement with our expectations for the studied society.
348: The validity of the Pareto
349: law and the value of the Pareto index will not change
350: considerably (we get $\alpha=0.95$) if we study only
351: the wealth distribution of the $1283$ aristocratic families
352: and neglect from the database the $116$ institutions.
353: Studying however only the wealth distributions of the
354: institutions will not give a Pareto-like tail at all,
355: and on the log-log scale we will get a constantly decreasing
356: slope (empty circles in Figure 1.)
357: It is also observable from Figure 1, that
358: the Pareto law breaks down in the limit of the very rich
359: families, where the wealth is bigger than
360: 1000 serf families. This is presumably a finite-size effect
361: and such results are observable in other databases
362: too (\cite{fujiwara03,india}).
363:
364: In order to have some information on the time evolution of
365: the Pareto index, the
366: wealth distribution of the Hungarian aristocratic families
367: in the 1767-1773 period was also studied.
368: For this period we had rough data \cite{felho} available
369: only for 11 districts. The sample was much smaller than for
370: the year 1550: we had only 531 families and 65 institutions
371: with total wealth greater than one serf family. However, as
372: a compensation
373: for the smaller database, in this case the wealth of each
374: family is given with three markers: the owned number of serf
375: families, the exact number of owned serfs and the total size
376: of the owned land. To our great surprise, we obtained for
377: each marker
378: that the cumulative distribution function of wealth does
379: not give the expected power-law behavior. Instead of a
380: straight line with a negative slope, we found a constantly
381: decreasing slope in the log-log plot of the rank as a
382: function of wealth (Figure 2). The fact that we used only
383: data for 11 districts cannot be the reason for the breakdown
384: of the Pareto scaling -- we checked that, on the same 11
385: districts, the 1550 data still gives the Pareto power-law
386: tail. We believe the reason why the
387: Pareto law is not valid for this database, is that a large
388: wealthy part of the society is missing. Indeed, in the mid
389: XVIII century in Hungary there were already many wealthy
390: non-noble families of merchants, bankers, rich peasants,
391: whose wealth exceeded the wealth of small or middle class
392: aristocrats. This large category of relatively wealthy
393: families were not landowners and had no serfs, so they are
394: simply not present in the considered database. After our
395: estimates, our wealth
396: distribution data gives a reliable picture of the society
397: only for the wealthier aristocrats,
398: with wealth greater than 100 serf families. It is believed
399: that the wealth of non-noble families
400: that owned no land and serfs could not exceed this
401: threshold. As observable however in the data from 1550,
402: for wealth values larger than 1000 serf families finite-size
403: effects are dominant, and the scaling breaks down.
404: There is thus a very short wealth interval (one decade)
405: where the data is trustful. Fitting a power-law
406: on this interval leads to a Pareto-index $\alpha=0.99$,
407: which is a reasonable value. This value
408: is also bigger than the one measured in year 1550,
409: suggesting thus a more active economic life. From this
410: Pareto index value it is possible to estimate the wealthy
411: part of the society
412: missing from our data (see Appendix A). The results obtained
413: in such a manner are reasonable ones.
414:
415: \begin{figure}
416: \begin{center}
417: \epsfysize=100mm
418: \epsffile{fig2.eps}
419: \caption{The rank of noble families and institutions
420: as a function of their estimated total wealth
421: on a log-log scale. Data for the Hungarian noble society
422: between the years 1767-1773.
423: The total wealth of a family is estimated in the number of
424: owned serf families.
425: The power-law fit suggests a Pareto index $\alpha=0.99$.}
426: \label{fig2}
427: \end{center}
428: \end{figure}
429:
430:
431: \section{Discussion and conclusions}
432:
433: Our study shows that the cumulative wealth distribution of
434: the Hungarian
435: aristocratic families in the year 1550 exhibits a power-law
436: shape with a Pareto index $\alpha=0.92$.
437: This result is a surprise in some sense, since it is
438: generally believed that
439: the Pareto index should be bigger than $1$. The fact that
440: the measured value of $\alpha$ is close
441: to $1$ is however predicted by existing wealth-exchange
442: models in the {\it no trade} limit, which we believe is
443: applicable to the type of society under study.
444:
445:
446: In the model introduced by Bouchaud and M\'ezard
447: \cite{bouchaud00},
448: if we consider the $J=0$ case, the wealth $w_i$ of each
449: family varies in an independent, stochastic multiplicative
450: manner, according to
451:
452: \begin{equation}
453: \frac{dw_i}{dt}=\eta_i(t) w_i + \xi_i(t)
454: \end{equation}
455:
456: where $\eta_i(t)$ is presumably a Gaussian distributed
457: random variable with zero mean ($<\eta_i>_t=0$), and
458: $\xi_i$ is an uncorrelated random noise with zero mean
459: ($<\xi_i(t)>_t=0$ and $<\xi_i(t) \xi_i(t')>=C \delta(t-
460: t')$).
461: This additive noise is necessary to prevent wealth
462: extinction and stabilize the power-law type solution in a
463: finite system. Following
464: the solution given in \cite{bouchaud00} we immediately get
465: $p(w) \propto w^{-2}$ for the tail of the wealth
466: probability density function $p(w)$, which yields a Pareto
467: index $\alpha=1$ for the cumulative distribution function.
468: The fact that the measured exponent is smaller than $1$
469: cannot be explained within this model.
470:
471: The wealth distribution obtained by us has a shape and
472: scaling exponent very similar to the population size
473: distribution of large cities \cite{zipf,zanette}. Zanette
474: and Manrubia \cite{zanette} successfully justified this
475: behavior using a simple reaction-like model (which is a
476: generalization of the Zeldovich model for intermittency
477: \cite{zeldovich}).
478:
479: Inspired by Zanette et al's model, one
480: can consider thus that the wealth of noble $i$ fluctuates in
481: time due to several processes, either exogenous (wars,
482: meteorological conditions affecting harvest sizes,...) or
483: endogenous (good or bad administration, gambling,...). So
484: one
485: can assume that $w_i (t+1) = \lambda_j w_i (t)$ due to
486: process $j$, which
487: occurs with probability $p_j$. If there are $n$ such
488: process ( $\sum_{j=1}^n p_j=1$) it is sufficient to require
489: that $<\lambda>=\sum_{j=1}^n \lambda_j p_j=1$
490: and to add some noise term to prevent collapse to $0$ to
491: obtain (see \cite{zanette,comment}) that
492: $p(w) \propto w^{-2}$ is a stationary solution and thus
493: $\alpha=1$.
494: This simple reaction-like stochastic
495: model is also appropriate thus to understand a Pareto index
496: close to $1$ for a system composed by independently evolving
497: agents.
498:
499: We have thus argued that the obtained Pareto index, close to
500: $1$, can be justified in the context of several simple
501: models, compatible with the socio-economical characteristics
502: of the real system under study. Recalling that a similar
503: value of $\alpha$ was recently presented for contemporary
504: Indian society, one may speculate that this is not a mere
505: coincidence, but rather evidence for the minor role of
506: exchanges in the dynamics of wealthier Indian people.
507:
508: The value smaller than $1$
509: obtained for the Pareto index in the Hungarian medieval
510: society is however still a puzzle,
511: and probably further modeling effort is still necessary to
512: account for it.
513:
514:
515: \section{Appendix A}
516:
517: Let us focus on wealth values greater than one serf family.
518: If we assume that for wealthy families the cumulative
519: distribution is scaling as $P_{\ge} (w) \propto w^{-1}$,
520: the $p(w)$ wealth distribution density function has the
521: form:
522:
523: \begin{equation}
524: p(w)=\frac{C}{w^2}
525: \end{equation}
526:
527: The $C$ constant can be determined by taking into account
528: that we have $83$ families and institutions with wealth
529: between
530: $100$ and $1000$ serfs
531:
532: \begin{equation}
533: \int_{100}^{1000} \frac{C}{w^2} dw \approx \frac{C}{100}=83,
534: \end{equation}
535: leading to $C=8300$. From here it is immediate to estimate
536: the $N_1$ total number of
537: families or institutions in the targeted 11 districts of
538: Hungary that have wealth greater than $1$ serf family:
539:
540: \begin{equation}
541: N_1=\int_1^{W_0} \frac{C}{w^2} dw \approx C = 8300
542: \end{equation}
543:
544: ($W_0$ stands for the biggest reported wealth value,
545: $W_0=6600$). This estimate can be confronted
546: with the results of the census made between 1784-1787. For
547: the studied
548: 11 districts the total population was estimated to be around
549: 1.6 million, with 45 thousand aristocrats
550: and 33 thousand rich burgers. Taking again $5$ members
551: per family, our database shows that
552: only around $531 \times 5=2655$ aristocrats had wealth
553: bigger than the considered $1$ serf family limit. This
554: value is only $6\%$ of the total estimated aristocrats in
555: the society. Most of the aristocrats at that time
556: had thus no considerable fortune except their noble title.
557: The missing $8300-600=7700$ families
558: corresponding to roughly $7700 \times 5=38500$ persons,
559: should be mostly rich burghers or some richer
560: peasants. This estimate is in reasonable agreement with the
561: census from 1784-1787.
562:
563:
564: {\underline{Acknowledgment}}
565:
566: Z. Neda acknowledges a Nato Fellowship and the excellent
567: working atmosphere at the
568: Centro de F\'{\i}sica do Porto. Research funded by EU
569: through FEDER/POCTI.
570:
571:
572: \begin{thebibliography}{10}
573:
574: \bibitem{pareto} V. Pareto, Cours d'\`economie politique,
575: Vol. 2, Macmillan,
576: Paris, 1897.
577:
578: \bibitem{ebooks} R.N. Mantegna and H.E. Stanley, {\em An
579: Introduction to Econophysics:
580: Correlations and Complexity in Finance} (Cambridge
581: University Press, Cambridge, 2001)
582:
583: % Real data
584:
585: \bibitem{matteo03} T. Di Matteo, T. Aste and S.T. Hyde,
586: preprint cond-mat/0310544.
587:
588: \bibitem{aoyama03} H. Aoyama, W. Souma and Y. Fujiwara,
589: Physica A {\bf 324}, 352
590: (2003).
591:
592: \bibitem{fujiwara03} Y. Fujiwara, W. Souma, H. Aoyama, T.
593: Kaizoji
594: and M. Aoki, Physica A {\bf 321}, 598 (2003).
595:
596:
597: \bibitem{silva04} A.C. Silva and V.M. Yakovenko, Europhys.
598: Lett. {\bf 69}, 304 (2005)
599:
600: \bibitem{fujiwara04} Y. Fujiwara, C. di Guilmi, H. Aoyama,
601: M. Gallegati and W.
602: Souma, Physica A {\bf 335}, 197 (2004).
603:
604: \bibitem{clementi04} F. Clementi and M. Gallegati, Physica
605: A {\bf 350}, 427 (2005)
606:
607: \bibitem{india} S. Sinha, preprint, cond-mat/0502166
608:
609: \bibitem{dragulescu01} A. Dr\u{a}gulescu and V. Yakovenko,
610: Physica A {\bf 299},
611: 213 (2001) and {\em Modeling of Complex Systems}, Seventh
612: Granada Lectures, AIP Conference Proceedings 661, NY 2003,
613: pg 180
614:
615: \bibitem{abul02} A.Y. Abul-Magd, Phys. Rev. E {\bf 66},
616: 057104 (2002).
617:
618: \bibitem{souma02} W. Souma in {\em Empirical Science of
619: Finantial Fluctuations: The Advent of Econophysics},
620: Springer-Verlag, Tokyo 2002
621:
622: \bibitem{souma01} W. Souma, Fractals {\bf 9}, 463 (2001).
623:
624: \bibitem{UK} http://www.inlandrevenue.gov.uk/
625:
626: % reviews
627:
628: \bibitem{zipf} M.E.J. Newman, preprint, cond-mat/0412004
629:
630: % generic economical models with Pareto's law outputs
631:
632: \bibitem{bouchaud00} J.-P. Bouchaud and M. M\'ezard, Physica
633: A {\bf 282}, 536
634: (2000).
635:
636: \bibitem{biham98} O. Biham, O. Malcai, M. Levy and S.
637: Solomon, Phys. Rev. E
638: {\bf 58}, 1352 (1998).
639:
640: \bibitem{solomon01} S. Solomon and P. Richmond, Physica A
641: {\bf 299}, 188
642: (2001).
643:
644: \bibitem{scafetta04} N. Scafetta, S. Picozzi and B.J. West,
645: Quantitative Finance
646: {\bf 4}, 353 (2004).
647:
648: \bibitem{chatterjee03} A. Chatterjee, B. Chakrabarti and
649: S.S. Manna,
650: Physica Scripta {\bf T106}, 36 (2003).
651:
652: \bibitem{pianegonda03} S. Pianegonda, J.R. Iglesias,G.
653: Abramson and J.L. Vega,
654: Physica A {\bf 322}, 667 (2003).
655:
656: % Complex networks
657:
658: \bibitem{dorogovtsev03} S.N. Dorogovtsev and J.F.F. Mendes,
659: {\em Evolution of Networks: From Biological
660: Nets to the Internet and WWW}, Oxford University Press
661: (2003).
662:
663:
664: \bibitem{iglesias03} J.R. Iglesias, S. Gon\c calves, S.
665: Pianegonda, J.L. Vega
666: and G. Abramson, Physica A {\bf 327}, 12 (2003).
667:
668: \bibitem{garlaschelli04} D. Garlaschelli and M.I. Loffredo,
669: Physica A {\bf 338},
670: 113 (2004).
671:
672: % Correlation and social networks
673:
674: \bibitem{coelho05} R. Coelho, Z. N\'eda, J.J. Ramasco and
675: M.A. Santos, Physica A
676: {\bf 353}, 515 (2005)
677:
678: \bibitem{maksay90} F. Maksay, {\em Magyarorsz\'ag
679: birtokviszonyai a XVI. sz\'azad k\"ozep\'en},
680: preprint (Budapest 1990)
681:
682: \bibitem{engel} P. Engel, {\em Szabolcs megye
683: birtokviszonyai a 14.-16. sz\'azadban}, in
684: {\em Honor, v\'ar, isp\'ans\'ag, V\'alogatott tanulm\'anyok}
685: pp. 601 (Osiris Kiad\'o, Budapest, 2003)
686:
687: \bibitem{dataset}
688: http://www.phys.ubbcluj.ro/$\sim$zneda/wealth.html
689:
690: \bibitem{felho} I. Felh\H{o} (editor), {\em \'Urb\'eri
691: bortokviszonyok Magyarorsz\'agon. Dun\'ant\'ul.}
692: (Akad\'emiai Kiad\'o, Budapest 1970)
693:
694: \bibitem{zanette} D. H. Zanette and S.C. Manrubia, Phys.
695: Rev. Lett. {\bf 79} 523 (1997)
696: and Phys. Rev. Lett. {\bf 80} 4831(1998)
697:
698: \bibitem{zeldovich} Ya. B. Zeldovich et al., Usp. Fiz.
699: Nauk., {\bf 152}, 3 (1987) [Sov. Phys. Usp. {\bf 30},
700: 353 (1987)]
701:
702: \bibitem{comment} M. Marsili, S. Maslov and Y.-C. Zang,
703: Phys. Rev. Lett. {\bf 80} 4830 (1998)
704:
705:
706:
707:
708: \end{thebibliography}
709:
710: \end{document}
711: