physics0505008/proj.tex
1: 
2: \documentclass{article}
3: \usepackage{times} 
4: \include{psfig}
5: \usepackage{epsfig}
6: 
7: 
8: %\bibliographystyle{apsrev}
9: 
10: \begin{document}
11: 
12: \title{Socioeconomic Development and Stability: \\
13: A Complex Network Blueprint}
14: 
15: \author{Luciano da Fontoura Costa, \\
16: Instituto de F\'{\i}sica de S\~ao Carlos. \\ 
17: Universidade de S\~ao Paulo, S\~{a}o Carlos, \\
18: SP, Caixa Postal 369, 13560-970, \\
19: Fone +55 16 3373 9858, FAX +55 1633 71 3616, \\
20: Brazil, luciano@if.sc.usp.br}
21: 
22: 
23: \date{21st March 2005}
24: 
25: 
26: \maketitle
27: 
28: 
29: {\bf ABSTRACT:} Complex networks provide flexible and powerful
30: resources for characterizing, modeling and simulating a wide range of
31: real world complex systems.  The current work discusses how such a
32: versatile new area can be applied in order to aid economical
33: development and stability at several scales and contexts.  The
34: following activities are involved: (a) compilation of several types of
35: data related to socioeconomic development, including several types of
36: transportation systems, availability of human and natural resources,
37: communication and energy networks, climate and geographical features,
38: as well as endemic diseases, to name but a few; (b) representation of
39: such data in terms of multilayer interacting complex networks
40: (cond-mat/0406369) registered geographically; (c) application of
41: traditional and new methods for complex networks characterization and
42: analysis.  Such an approach allows the identification of bottlenecks
43: and deficits/surpluses, simulation of system development under varying
44: constraints and perturbations, as well as the application of
45: optimization methods in order to help identify the most effective
46: strategies leading to social and economic wealth and stability.  In
47: addition to its practical implications, such an approach also
48: emphasizes several issues of substantial theoretical interest,
49: including the integration of networks of different natures, the
50: interplay between the dynamics of topological and node state
51: evolution, the effects of geographical constraints, community finding,
52: as well as the interesting problem of how to optimize such systems
53: with respect to network topology and dynamics in order to achieve
54: specific objectives expressed by merit figures.  The discussed
55: methodology is particularly interesting for applications in developing
56: countries because of the greater potential for economic evolution in
57: such nations.  This manuscript also includes a brief review of complex
58: networks approaches to socioeconomics modeling.
59: 
60: 
61: \pagebreak
62: 
63: \begin{quotation}
64: '... it is hard to turn ideas into wealth in the absence of social 
65: connectedness, which in the age of the Internet still requires 
66: something more than bandwidth and high-speed connectivity.' \\
67: \hspace{2cm} \emph{(F. Fukuyama, The Great Disruption)}
68: \end{quotation}
69: 
70: 
71: 
72: \section{Introduction}
73: 
74: The main activity in science is the art of model building.  Good
75: models require sound representation of the phenomenon of interest in
76: terms of a reduced set of most relevant variables and parameters.
77: Because all models are necessarily incomplete, it is important to
78: obtain particularly effective representations, which should be able to
79: account for the relevant variables under constraints imposed by the
80: chosen parameters.  By being naturally oriented to representing
81: connections and relationships, graph theory stands out as particularly
82: general and suitable for model representations.  Indeed, almost all
83: discrete data structures can be understood as a particular instance of
84: graphs.  As the relationships between variables and parameters
85: typically change with time, it is also necessary that the adopted
86: representations be capable of expressing dynamical changes in both
87: network topology as well as the dynamics undergone by the states of
88: the nodes.  With this respect, the seminal works by Flory~\cite{Flory},
89: Rapoport~\cite{Rapoport}, Erd\"os and R\'enyi~\cite{Erdos_Renyi} on
90: random networks have been substantially expanded in the recent years
91: mainly through the consideration of principles from statistical
92: physics and dynamic systems, especially regarding the characterization
93: and modeling of scale free network models~\cite{Albert_Barab:2002}.
94: 
95: A direct indication of the impressive success of complex network
96: research is the large and ever increasing related scientific
97: production, accounting for about 1 or 2 new \emph{arxiv} manuscripts
98: per day (\emph{http://arxiv.org/archive/}).  Indeed, the catalisis
99: of scientific investigation, interaction and dissemination allowed by
100: the WWW and internet resources as \emph{arxiv} provides further
101: evidence of the impact of immediate wide dissemination of data,
102: results and knowledge.  One particularly interesting portion of the
103: developments in complex network has been aimed at applying concepts,
104: measurements and models in order to mimic and predict the behavior of
105: real complex systems including the internet (e.g.~\cite{Faloutsos,
106: peering, Barth_internet, Barth_Gondran}), WWW (e.g.~\cite{Bara_WWW,
107: Kahng_WWW}), protein and metabolic interactions
108: (e.g.~\cite{Jeong_protein, Reka_protein}), transportation systems
109: (e.g.~\cite{Bagler, Parongama, Guimera_airport}), opinion formation
110: (e.g.~\cite{Elgazzar, forsta, Stauffer_Hohnisch, L_sznaj:2005}),
111: epydemiology (e.g.~\cite{Masuda_diseases, Szendroi_diseases}), among
112: many other relevant issues.  Because of its flexibility and power for
113: developing good integrative models of complex systems, it becomes
114: particularly interesting to consider the systematic application of
115: this new area as means to represent, model, predict and optimize
116: the behavior of socioeconomic systems.
117: 
118: Socioeconomics environments rank among the most complex systems known
119: to humanity, involving a whole range of entities and relationships.
120: By selecting a set of reference entities (e.g. institutions or
121: cities), it is possible to obtain a series of integrated complex
122: networks~\cite{Solomon} registered by geography, each representing
123: specific types of relationships.  Several concepts, measurements and
124: modeling approaches derived from the area of complex network research
125: can then be applied in order to characterize several topological and
126: dynamical features of such systems, allowing the prediction of the
127: response of the system to modifications, the identification of
128: promising development strategies, the enhancement of the stability of
129: the whole system, optimization in order to achieve specific
130: objectives, as well as its resilience to disruptions (e.g. epidemics
131: and catastrophes), to name but a few possibilities.
132: 
133: This text describes a possible blueprint for such investigations,
134: which are relevant not only for their immediate implications and
135: potential for optimizing commonwealth, but also for poising several
136: interesting theoretical issues arising from the integration of
137: networks and optimization of topology-dynamics under pre-specified
138: optimality indices.  It should be emphasized at the outset that it is
139: the broad integrative perspective of considering several types of
140: evolving geographically registered geographical networks, with
141: emphasis on applications to third-world socioeconomics, which
142: provides the main motivation for the present work.  It should also be
143: noted that any result arising from such a type of simulations need to
144: be treated with the greatest caution because of the several non-linear
145: and unpredictable effects which are known to operate in
146: socioeconomics.  However, it is expected that the systematic and
147: integrative use of concepts and methods from complex network research,
148: itself a very dynamic research area, can provide valuable means for
149: optimizing socioeconomic development and stability.
150: 
151: This article starts by briefly reviewing some of the main complex
152: network approaches to socioeconomics and follows by discussing the
153: representation of the modeled systems in terms of multi-layer,
154: geographically registered networks, as well as simulation and
155: optimization possibilities.
156: 
157: 
158: \section{A Brief Review of Complex Network Approaches to \\
159: Socioeconomics and Related Issues}
160: 
161: Previous developments involving complex network approaches to
162: socioeconomics reported in the literature are briefly reviewed in the
163: following.  One of the first socioeconomics related applications of
164: complex networks was reported by Guardiola et
165: al.~\cite{Guimera_trusts}, where the web of trust scheme between users
166: of PGP was considered as a model for trust networks, leading to good
167: resilience to intentional attack.  The use of small-world models for
168: socioeconomic systems was investigated by Elgazzar~\cite{Elgazzar}
169: through simulation of the Sznajd dynamics~\cite{Sousa, forsta} on
170: small-world networks.  The issue o geographical embedding of
171: networks~\cite{Rozenfeld, Alon, Andersson} as a constraint to
172: U.S. Internet infrastructure and its implications for economy and
173: politics was considered in~\cite{Gorman}.  Spatial constraints have
174: also been found to affect networking and internet
175: architectures~\cite{Gray}, in the sense that computing resources would
176: tend to be placed as close as possible to the source of data in order
177: to avoid expensive network traffic.  The network topology of the
178: Austrian Interbank market was investigated by Boss and
179: collaborators~\cite{Thurner}, indicating that the contract size
180: distribution follow an extensive power law.  The issue of grid
181: computing has been addressed in terms of resource allocation and
182: regulation~\cite{Buyya, Lai} as well as by adopting complex network
183: interconnections~\cite{Costa_grid}. Bonanno et al.~\cite{Bonanno}
184: investigated the possibility to extract meaningful economic
185: information from portfolio of stocks and its implications for
186: comparison of the topological properties of networks.  A model of
187: wealth dynamics and transactions among economic agents by considering
188: different network connectivities was investigated by Garlaschelli and
189: Loffredo~\cite{Garlaschelli}.  The effect of information cascades over
190: economic recessions in the U.S., assuming a random network model was
191: described by Cook~\cite{Cook}, while studies of cascades of failures
192: as a consequence of attacks have been covered in~\cite{Hayashi}.  The
193: possibility of representing economic variables of different nature in
194: terms of multiple interconnected networks, called \emph{Solomon
195: networks}, have been considered for Ising simulations of the dynamics
196: of economic systems~\cite{Solomon}.  Bipartite network models have
197: been used to simulate relationship between countries and currencies in
198: world exchange arrangements web~\cite{Li}, considering assortativity
199: aspects.  The stability in supply/production networks has been
200: investigated by Helbing~\cite{Helbing} who analysed, by considering
201: different network topologies, how networks of damped oscillators tend
202: to be subject to increasing oscillations.  Topological investigations
203: of networks defined by traders exchanging goods have been considered
204: by Reichardt and Bornhold~\cite{Reichardt}, who analysed the 2004
205: pre-Christmas season and identified high modularity.  A review of
206: econophysics has been presented by Di~\cite{Di}, and a review of
207: quantitative modelling of financial markets has been reported by
208: Farmer and Lo~\cite{Farmer}.
209: 
210: 
211: \section{Representation}
212: 
213: One of the first important decisions while modeling a socioeconomic
214: system is to define and represent in a careful way its most
215: representative components.  Such components can be divided into two
216: categories: \emph{states} and \emph{relationships}. In socioeconomics
217: networks, local states correspond to properties of the main considered
218: sites or places (e.g. cities, institutions, etc.).  Examples of states
219: include but are not limited to:
220: 
221: \begin{trivlist}
222:   \item (a) \emph{Human resources}: The involved individuals, possibly
223:   subdivided into workers, consumers, experts, etc.  
224: 
225:   \item (b) \emph{Natural resources}: The existing (or prospected)
226:   energy sources, organic and inorganic assets, rivers and lakes,
227:   climatic features, etc.
228: 
229:   \item (c) \emph{Storage capabilities}: The local potential for
230:   storing raw and processed materials.
231: 
232:   \item (d) \emph{Industrial resources}: The facilities available
233:   locally which can be used to process raw materials as well as high
234:   technological means for obtaining more sophisticate goods.
235: 
236:   \item (e) \emph{Financial resources}: May include the bank and
237:   finance systems which can be found in the locality.
238: 
239:   \item (f) \emph{Endemic diseases}: The epidemics and pathologies
240:   which continuous or periodically affect humans and animals in the
241:   region.
242: 
243:   \item (g) \emph{Cultural and social features}: The cultural
244:   traditions and social features and values.
245: 
246:   \item (g) \emph{Scientific and technological assets}: The level of
247:   scientific and technological development at each locality.
248: 
249: \end{trivlist}
250: 
251: Note that several of such states are not straightforward to be
252: quantified, implying some degree of arbitrariness.
253: 
254: Global states are typically maps of local states into overall
255: properties of the modeled system, such as overall production, debt or
256: surplus, total birth/death rates, etc.  The relationships between the
257: localities follow naturally, defining a complex network for each
258: considered type of interaction~\cite{Solomon}.  A particularly
259: interesting possibility is to integrate such networks geographically,
260: i.e. each site is represented as a node with geographical position and
261: several types of edges are defined between such common nodes, yielding
262: an integration of several geographical networks in a way that reminds
263: the topographical connections between cortical
264: layers~\cite{Luc_topo}. Figure~\ref{fig:exnet} illustrates a simple
265: hypothetical socioeconomic model involving six cities and $p$
266: relationship networks.
267: 
268: \begin{figure}
269:   \begin{center}
270:     \includegraphics[scale=.7]{fig_exnet.eps}  
271: 
272:     \caption{Socioeconomics complex network models may involve
273:     several geographical networks characterized by the fact that the
274:     nodes have definite spatial positions.  Each network $\gamma_i$
275:     represents one of the considered types of relationship.  Note that
276:     the registration between these networks is accomplished through
277:     the spatial congruence of the nodes position. The interactions
278:     between the several layers are not represented in this
279:     diagram.}~\label{fig:exnet}
280: 
281:   \end{center}
282: \end{figure}
283: 
284: Examples of relationships relevant for socioeconomic systems are
285: listed in the following:
286: 
287: \begin{trivlist}
288:   \item (A) \emph{Transportation}: Essential for economic integration,
289:   allowing human, raw materials, and processed goods displacement.
290:   Each type of transportation (e.g. railway, motorways, airways) can
291:   be represented as a specific complex network, facilitating the
292:   analysis of the complementarity between such resources.
293: 
294:   \item (B) \emph{Energy distribution}: The existing network allowing
295:   access to several types of energy (e.g. electrical and gas).
296: 
297:   \item (C) \emph{Communications}: The interconnections allowing
298:   information exchange between the considered localities, possibly
299:   subdivided into networks for cell and fixed telephony, satellite,
300:   optic fibers, etc.
301: 
302:   \item (D) \emph{Financial and political trusts and alliances}:
303:   Corresponding to the network of government and private financial
304:   agencies.
305: 
306:   \item (E) \emph{Borrow/Loan relationships}: The directed network
307:   defined by borrowing and debts between the involved localities.
308: 
309:   \item (F) \emph{Cultural and social links}: Networks
310:   established by common share of common beliefs and traditions.
311: 
312:   \item (G) \emph{Distributed computing}: Including wide range
313:   distributed computing facilities, such as grid computing.
314: 
315: \end{trivlist}
316: 
317: Observe that most such networks are typically dynamical (in the sense
318: that their topology will vary with time) and weighted.  Flow or
319: resources conservation may eventually be observed.  Except for
320: communications networks, all the above networks are typically
321: represented as digraphs (i.e. involving oriented edges).  
322: 
323: 
324: \section{Characterization and Simulation}
325: 
326: Once a part of a socioeconomic system has been represented in terms
327: of geographically integrated networks, a series of measurements can be
328: used to characterize and analyse their topology.  The choice of
329: measurements should be performed with basis on the specific issues of
330: interest. For instance, in the case of transportation systems,
331: statistics of shortest paths are of special relevance.  The reader is
332: reported to a recent survey~\cite{Luc_rev} for a comprehensive review
333: of complex network measurements.  Of special interest are the specific
334: demands implied for the topological characterization of the
335: geographically registered coexisting networks, motivating new
336: measurements capable of expressing the topological interactions
337: between the several layers (e.g.~\cite{Luc_topo}).  Another
338: particularly relevant issue is the identification of communities in
339: the integrated networks.  A particularly interesting possibility is to
340: consider the identification of well-defined communities in one of the
341: layers as a subsidy for identification of communities in the remainder
342: layers, as well as the analysis of overlaps and divergences between
343: such clusters.  Perspectives of special interest also include the
344: development of models describing the topological evolution of the
345: networks at each layer and as a whole.
346: 
347: The simulation of the dynamics of the network states can be performed
348: by assuming several methods including spin dynamics
349: (e.g.~\cite{Solomon, Bose}), cellular automata, and systems of coupled
350: differential equations.  Global feedback can also be considered in
351: such formulations~\cite{Solomon}.  Important related aspects involve
352: the synchronization of events in the networks as well as the
353: appearance of instabilities and oscillations~\cite{Helbing}.  A
354: particularly interesting possibility is to link the dynamics of the
355: network individual states with the dynamics of topological changes in
356: the network structure (e.g.~\cite{L_sznaj:2005}).
357: 
358: While the simulation of the dynamical evolution of the network states
359: can lead to valuable insights about socioeconomic development and
360: stability, it is also interesting to consider the \emph{optimization}
361: of the network architecture and constraints in order to achieve
362: specific goals.  Provided merit figures are clearly established in
363: terms of properties of the modeled system, a series of optimization
364: approaches ranging from linear programming to genetic algorithms can
365: be applied in order to identify improvements to the topology of the
366: network or its dynamical evolution.  For instance, once a government
367: have decided to explore a recently discovered source of raw material
368: or to find the best way to protect indigenous fauna and flora,
369: optimization of the topology of the network under constrains imposed
370: by its states can be performed in order to evaluate possible
371: development strategies.  Of course, such approaches are inherently
372: limited by the suboptimality (i.e. convergence to local minima)
373: characterizing most non-linear optimization methods.
374: 
375: 
376: \section{Concluding Remarks}
377: 
378: This article has discussed how complex network concepts and methods
379: can be extensively used to model socioeconomic systems.  After
380: reviewing briefly the main related literature, a blueprint has been
381: proposed and discussed.  Such a model involves multilayer
382: networks~\cite{Solomon} whose nodes are registered in terms of the
383: geographical positions of the entities which they
384: represent~\cite{Luc_topo}.  Several interesting issues are motivated
385: regarding the characterization and simulation of such networks.  As
386: far as the analysis of the topological features of the multiple
387: networks are concerned, adaptations of traditional measurements and
388: new features capable of taking into account and quantifying the
389: interconnections between different layers are of special interest.
390: Two mains possibilities are defined regarding the dynamics of state
391: evolution: (i) the simulation of the dynamics on static diverse
392: topologies; and (ii) the simulation of state dynamics on networks
393: whose topology undergoes dynamic evolution.
394: 
395: Several are the difficulties constraining such investigations.  To
396: begin with, it is important to gather reliable, uptodated and
397: representative data related to each socioeconomic features considered
398: in the model.  A particularly challenging issue concerns the
399: definition of the merit figures used for the optimization of the
400: topology and weights of the network, which often implies political,
401: ethical and/or arbitrary nature (e.g. should transgenic products or
402: abortion be allowed?).  Also, the high complexity of the involved
403: systems severely constraints the time window for predictions, implying
404: that the greatest caution should be taken when analysing
405: characterization and simulation results.  Despite such difficulties,
406: it should still be possible to adopt a progressive approach starting
407: with only a few layers and gradually increasing complexity.  An
408: example of a particularly feasible and interesting starting point is
409: to investigate the efficiency of coexisting transportation networks
410: (e.g. railways, motorways, airport systems) while trying to identify
411: how such systems can be improved (e.g. minimize the average shortest
412: path) by small topological modifications (e.g. the inclusion of a new
413: railway link).  Such models can be easily upgraded by including the
414: availability of natural and human resources, consequently defining
415: interesting problems of optimizing flow and production-consuming
416: interactions.
417: 
418: The author is currently conducting related efforts considering the
419: Brazilian economy and would highly appreciate to receive comments and
420: suggestions and to consider collaborations in any theoretical or
421: practical related aspects.
422:  
423: \bibliographystyle{unsrt}
424: \bibliography{proj}
425: 
426: \end{document}
427: 
428: 
429: 
430: 
431: 	
432: 
433: 
434: 
435: 
436: 
437: 
438: 
439: 
440: 
441: 
442: 
443: 
444: 
445: 
446: 
447: 
448: 
449: 
450: 
451: 
452: 
453: 
454: 
455: 
456: 
457: 
458: 
459: 
460: 
461: 
462: 
463: 
464: 
465: 
466: 
467: 
468: 
469: 
470: 
471: 
472: 
473: 
474: 
475: 
476: 
477: 
478: 
479: 
480: 
481: 
482: 
483: 
484: 
485: 
486: 
487: 
488: 
489: 
490: 
491: 
492: 
493: 
494: