The Oxford Handbook of Agent-based Computational Management Science

(In Progress)
Friederike Wall (ed.) et al.

https://doi.org/10.1093/oxfordhb/9780197668122.001.0001
Published: 2024 Online ISBN: 9780197668153 Print ISBN: 9780197668122

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Search in this book

CHAPTER

Agent-Based Modelling and Social Network Analysis: A

Review for Economic and Management Studies 
Pedro Campos

https://doi.org/10.1093/oxfordhb/9780197668122.013.16
Published: 21 August 2024

Abstract
Social network analysis o ers a powerful method for comprehending intricate systems created
through agent-based computational models. Scholars contend that intricate agent networks have the
capacity to grasp both the dynamics at an individual level and the overarching characteristics at a
global level within a complex system. Consequently, they can contribute to a deeper comprehension of
the underlying principles governing such systems. Within this endeavour, we undertake a
comprehensive examination of the existing body of literature that establishes a connection between
agent-based models and social network analysis. The focal aim of this exploration is to cater to the
domain of management studies. We de ne a baseline for a network topology for a taxonomy of ‘macro’
and ‘micro’ characteristics of social interaction networks. We apply and extend this taxonomy for
agent-based models and classify existing models in macro structures, macro patterns of interaction,
micro interactions and multilayers, and micro level (clustering and local interaction). We also
emphasize the learning methods for agent-based models of social networks.

Keywords: agent-based models, social network, micro–macro interactions, multilayer networks, social
dynamics, network structure, multiagent learning
Subject: Knowledge Management, Research Methods, Business and Management
Series: Oxford Handbooks
Collection: Oxford Handbooks Online
1. Introduction

Agent-based models (ABMs) allow researchers to specify the behaviour of a number of agents and observe
their interactions over time, and they are especially useful to investigate the aggregate behaviour that
emerges from the interactions of individual agents. Since network science has established itself as an
important approach to modelling social, economic, industrial, and biological processes, its connection with

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
ABMs has become very fruitful. Indeed, almost all solutions to real-world problems involve the network
element (Namatame and Chen 2016). Agents can be seen as nodes of a network (e.g., people, companies),
where links function as interactions (e.g., communications, social relations). Modelling social interactions
among agents may bene t from the fact that they are embedded in networks. Each agent at each node is
characterized by its location and state. The ‘checkerboard’ models, such as the segregation model of
Schelling (1971), where agents are cellular automata, can be considered the earliest integration of agent-
based modelling with networks (Chen 2012; Namatame and Chen 2016).

Social network analysis (SNA) provides a complementary approach for understanding complex systems.
Researchers argue that complex agent networks can capture both individual-level dynamics as well as
global-level properties of a complex system and, as such, may help to obtain a better understanding of the
fundamentals of such systems. Together with ABM, network science can provide a powerful tool for
understanding how social systems operate and evolve—namely, with the study of social networks. Social
networks are systems of individuals or organizations that are connected through social ties, such as
friendship, family, or professional relationships. These ties can have a signi cant impact on the behaviour
of the agents within the network, and understanding the patterns of these connections is crucial for
understanding the dynamics of the network as a whole.

One example of how ABMs and SNA can be used together is in the study of the spread of information or ideas
within a social network. ABMs can be used to simulate the behaviour of agents as they receive and
disseminate information, while SNA can be used to analyse the patterns of connections and interactions
between agents. Together, these approaches can provide insight into how information spreads within a
network and which agents are most likely to be in uential in the spread of information. Another area where
ABMs and SNA can be used together is in the study of social in uence. Social in uence refers to the process
by which an agent’s behaviour or attitudes are a ected by the behaviour or attitudes of other agents within
a network. ABMs can be used to simulate the behaviour of agents as they are in uenced by others, while SNA
can be used to analyse the patterns of connections and interactions between agents. Together, these
approaches can provide insight into how social in uence operates within a network and which agents are
most likely to be in uential.

Eden et al. (2021) use the term agent-based network modelling to name the principle where each person is
simulated at the individual level as an agent of the simulation and uses network generation algorithms to
generate the network of contacts between individuals. In economics, the introduction of SNA in agent-
based approaches is an answer to the limitations of dynamic stochastic general equilibrium (DSGE) models.
These traditional macroeconomic models are restricted in explaining interaction e ects, especially with the
nancial sector, and crises (LeBaron and Tesfatsion 2008). In DSGE models, there is no causal link between
a boom and a bust by construction, except for small self-correcting deviations from the deterministic
steady state, so the stable period can be understood as separate from the unstable period. DSGE models have
been deeply challenged by the Global Financial Crisis of 2008.

In this work, we review the literature having a bridge between ABMs and SNA for the purpose of
management studies. For this purpose, we follow four seminal works that may be seen as in uential
references: the work of Wilhite (2006), who de nes a base framework for agents in networks; Namatame
and Chen (2016), who distinguish the rst from the second generation models of networks (lattices vs.
graphs) and provide a framework for ABMs of social networks; Jackson et al. (2017a), who classify the most
prominent papers using a speci c taxonomy of network characteristics of social interaction networks,
classi ed in ‘macro’ and ‘micro’; and the recent work of Steinbacher et al. (2021), who discuss advances in
the agent-based modelling of economic and social systems, as well as argue how economic networks and
social systems can be modelled using ABM.

We identify some key issues in economics and business studies, where ABMs and SNA play important roles:

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
• Social dynamics and networks (i.e., the general aspects of ABMs that simulate the behaviour of
individuals within a social network and the dynamics that emerges from their interactions, including
opinion dynamics). This includes social in uence, persuasion, and di usion processes, where ABMs
help to study the spread of information or innovation within a social network, including social
in uence in online platforms and social media.

• For the study of network structure, we follow the work of Jackson et al. (2017b), who survey the
literature on the economic consequences of the structure of social networks and develop a taxonomy of
‘macro’ and ‘micro’ characteristics of social interaction networks. Their classi cation of network
characteristics o ers a framework through which to understand how structural properties of a network
impact the behaviours of the agents who comprise the network.

Although this classi cation is applied to scienti c works on networks where, several times, the approach
does not include ABMs, it seems important to follow it, as it is perhaps the most complete classi cation of
network characteristics in the literature.

The chapter is structured as follows: we start by de ning a general framework for networks in section 2,
based on the classical and well-accepted graph theory. Then, in section 3, we address the links between
social dynamics and networks, and in section 4, we classify the relevant literature according to the Jackson
et al. (2017b) taxonomy. In section 5, we address the aspects of learning in agent-based network modelling,
and in section 6, we provide a review of ABMs of social networks based on Namatame and Chen (2016). We
conclude the chapter with a section with nal considerations and provide directions for future work.

2. A General Network Framework

In SNA, networked structures are characterized in terms of nodes, which may represent people or rms
within the network, and the edges, or links, which may represent relationships or social interactions.
Although nodes are agents in the perspective of ABMs, there are di erent expressions for nodes and links
regarding their eld of application (see Table 1).

Table 1. Di erent expressions for each field of nodes and links

Engineering and computer science Mathematics and graph theory Physics Social sciences

ʻPointʼ Node Vertex Site Actor

ʻLineʼ Link Edge Bond Tie

ʻNetworkʼ Network Graph Network Network

Source: Marcelo (2017).

Economists use networks as re ecting a set of choices, or results between economic agents. We now focus
on the di erent topologies of networks as these emerge naturally from the behaviour of agents in economic
and business sciences. Network topology is the study of the arrangement of the elements (links, nodes, etc.)
of a network. In economics, the number of topologies observed in cooperation networks is relatively high,
from simple to more complex ones. A seminal application of the topologies of networks to industry is
provided by Hakansson (1987), who argues that all concerned with technical innovation should bear in mind

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
the importance of ‘networks’ (de ned as the web of contacts that exist between suppliers, customers, and
producers in industry).

For formalizing the network concepts based on graph theory, we follow the work of Wilhite (2006). Wilhite
(2006) looks at the nodes of the network as agents of an ABM and identi es seven di erent types of network
topologies for agents—complete, star, ring, grid (lattice), tree, small world, and power—that we address in
the next paragraphs. We added the random graph topology for comparison purposes.

One can formally de ne a graph as G = (N , E ) , consisting of the set N of agents and the set E of edges,
which are pairs of elements of N . In G, to de ne that there is a link or edge between pairs of agents {i, j},
we write, i~j ∈ G .

2.1. The Random Network

In random networks (see Figure 1), each node pair is connected with a probability p , and the resulting
networks have low heterogeneity (most nodes have the same number of connections). Random networks
are graphs built after the E-R (Erdos–Renyi) model. For modelling p in the E-R model, one may use the
uniform distribution. Following this model, all graphs on a xed vertex set with a xed number of edges are
equally likely. The resulting networks have short average paths (meaning average short distances between
agents) and low clustering (as the nodes tend to be sparsely connected in groups). The road system, for
example, ts with this type of network model. But in other cases of real life, social networks do not have a
homogeneous distribution of degree that regular or random networks have. This model can be formalized as
follows: given two agents, i and j, we can de ne that the probability that a link is formed in {i, j} is given by

P (i~j ∈ G) = p

where the distribution of links can be given by a binomial model Bi (n, p), with n being the number of edges
in the network.
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 1. Representation of a random network.

2.2. The Complete Network

In the complete network, also known as a fully connected network (see Figure 2), every node is connected to
every other node:

i~j ∈ G, ∀i,j, i ≠ j

Few examples of complete networks exist, such as the telephone network, if we consider that almost
everyone can make a phone call to almost everyone else.
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 2. Representation of a complete network.

2.3. The Star Network

A star network is a topology where one agent (s) is connected to others, but no other agents are connected
(see Figure 3):

s~j ∈ G, ∀j,s, s ≠ j

One example of this type of network is the airline industry (Hendricks et al. 1995). Most of an airline’s
customers need to y between two di erent airports of the group of airports served by the airline company,
and the optimal network in this case is the star, since customers rst y to the airline’s hub airport, in the
centre, and then y from there to the nal destination (Wilhite 2006).
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 3. Representation of a star network.

2.4. The Ring Network

Within this topology, each of the nodes of the network is connected to two other nodes (see Figure 4). All the
ow that is transmitted between nodes in the network travels from one node to the next node in a linear
manner. Formally,

j = {(i − m) + n} mod n or
i~j ∈ G, if {
j = {(i + m)} mod n

where m ∈ {1, 2, …,
k

2
} and k is an even number .

Rings are related to the local neighbourhood constraints, because agents interact only with a fraction of the
population. In rings/linear topologies, the ow of information that is transmitted between agents in the
network travels from one node to the next node in a linear manner. In Campos et al. (2013), these topologies
emerge in a network formation strategy for rms’ collaboration, called peer-to-peer complementariness. In
this strategy, for the selection of a partner, a meeting probability between two rms (i andj) is de ned as
mt (i, j) , in which rm i chooses the rm j that minimizes the distance between them.
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 4. Representation of two possible configurations of a ring (or linear) network.

2.5. The Grid or Lattice Network

A simple grid network (see Figure 5) is a matrix or chessboard with agents placed in the corners of each
square. Examples of grid networks are related to the distribution of agents in space, such as the layout of
city blocks.

j = {(i ± i) + n} mod n or
i~j ∈ G, if {
1/2
j = {(i ± n ) + n} mod n

Figure 5. Representation of a grid network.

2.6. The Tree Network

In the tree network (see Figure 6), each agent branches o to k other agents, each of which branches again.
Tree networks are often used to describe hierarchies in social systems, such as organizations and
institutions (e.g., government), or the supply chain of an industry.
 i = 1 and j = 2 + (0, 1, … , k)
i~j ∈ G, if {
i > 1, j = (ki + v), v ∈ {0, 1, … , k − 1}

where k is the number of edges connecting each node.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 6. Representation of a tree network.

2.7. Small-World Network

Small-world networks lie between a ring network and a random network. In small-world networks (see
Figure 7) most of an individual’s neighbours are also neighbours of each other. In this sense, surprisingly,
seemingly distant nodes tend to be just a few steps away (Wilhite 2006).

Watts and Strogatz (1998) investigate these types of networks and de ne then as follows: a particular
connection i~j is rewired by severing that link at j and then randomly selecting a new node (say h) and
creating a new link: i~h. There is a probability p of rewiring: if p equals zero, the ring is undisturbed. If p = 1,
the wiring will be random. But if p ∈ (0, 1) , then intermediate groups emerge. Electric power grids and some
neurologic networks, as well as the coauthoring of scienti c publications, are examples of such small-world
networks.

The structure of small-world networks stands between the regular and random types. Solé and Valverde
(2004) show how di erent types of networks relate to each other in terms of their randomness,
heterogeneity, and modularity. On one extreme of the types of networks, the authors consider grids (regular
lattices) and trees. These are usually humanmade networks, such as streets, energy networks, and water
channels, with the lowest heterogeneity and the lowest randomness. Regular graphs tend to have long
average path lengths and high clustering. On the other extreme, there are the random E-R graphs
introduced above. Small-world networks are very close structurally to many social networks in that they
have a higher clustering and almost the same average path than the random networks with the same
number of nodes and edges.
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 7. Representation of a small-world network.

Finally, there is a large class of so-called scale-free networks characterized by a highly heterogeneous
degree distribution, which follows a ‘power-law’ (Barabási and Albert 1999) that we address in the
following paragraphs.

2.8. Power Networks

Also known as scale-free networks, power networks (see Figure 8) have a degree distribution that follows a
power law. In other words, few agents have many connections with other agents, who are commonly called
hubs, and many agents have few connections with other agents.

The internet is an example of such a power network. In scale-free or power networks, there is a preferential
attachment process (Albert and Barabási 2002), de ned by

jd
P (i~j ∈ G|i) =
nd

where jd is the number of edges reaching node j and nd is the total number of agents in the entire network.
This preferential attachment may be seen as a meeting process between agents (e.g., two rms), in which
rm i chooses the rm j that maximizes a particular value, such as the geographical or technological
distance between the agents. In Campos et al. (2013), the authors use di erent strategies of collaborative
networks that emerge from the creation and di usion of knowledge between rms and analyse the evolving
dynamics of the resulting networks. Power networks emerge from the preferential attachment process,
where agents ( rms) meet easily other agents in their neighbourhood. Preferential attachment is an
attractive idea since it implies an intuitive mechanism whereby individuals follow other people who are
more popular than they are (Watson et al. 2022).
 Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Figure 8. Representation of a power network.

On a parallel and somewhat complementary approach, Namatame and Chen (2016) de ned two generation
models of networks: lattices and graphs. The rst models (lattices) are simple models of cellular automata,
consisting of a regular grid of cells, each in one of a nite number of states, where agents interact spatially
with their neighbours. Conway’s game of life (Gardner 1970) and the Schelling segregation model (Schelling
1971) are examples of ABMs that use lattice structures.

In the second type of models (graphs), the idea of network has been extended beyond the concept of lattices.
Real examples of graphs models include some of the structures identi ed above: the complete network
(implicitly assumed in many economic contexts, such as competitive markets), the random graph (used in
social sciences as a model to capture the randomly local encountering), and the ring (occasionally used to
characterize the geographical restriction) (Namatame and Chen 2016).

3. Social Dynamics and Networks

Social systems are intricate networks of relationships. One of the most important goals of using SNA in
social simulation is that it helps understanding and analysing the complex interactions and relationships
that shape social dynamics within various systems. In addition, SNA provides new tools for the
formalization of interactions between agents.

Steinbacher et al. (2021) discuss advances in the agent-based modelling of economic and social systems and
introduce network theory. They argue that much of today’s research on SNA is actually based on previous
works in sociology, physics, and computer science and provide a review that gives an idea of the bridges
between ABM and SNA.

Acemoglu et al. (2012) realized the presence of interconnections between di erent rms and sectors,
functioning as a potential propagation mechanism of idiosyncratic shocks throughout the economy. And the
possible role of such interconnections in propagation of shocks was highlighted during the debate leading to
the crisis that started with the automobile industry bailout.

Hoertel et al. (2020) built a stochastic ABM model of the Covid-19 epidemic in France, including a social
contact network among the individuals in the population, each with a geolocalized activity sequence over
the day, taking into account co-location probability and duration.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
In the examples above, SNA provides a data-driven foundation for policy and decision-making when
combined with ABMs. Governments and organizations can leverage SNA insights to design more e ective
interventions, allocate resources e ciently, and address challenges that arise from the dynamics of
interconnected systems. We can also see how SNA can be used to assess the e ectiveness of interventions or
initiatives aimed at changing behaviours or improving social outcomes.

Examples are fruitful in the elds of economics and business studies. For example, Schulz and Mayerho er
(2021) propose a network-based evolutionary learning for the industrial dynamics of superstar rms. By
introducing a network structure of localized competition and innovation, they conclude that most intensive
local competition leads to the highest concentration, whilst the lowest concentration appears for a mild
degree of local oligopoly.

Krichene et al. (2019) explain how rms behave and select their suppliers and customers in the Japanese
production network. Their results show that all of the e ects (social and economic e ects) are statistically
signi cant in explaining the ties formation between rms. Nogueira et al. (2023) investigated how
companies react to market change by understanding the e ect of a perturbation on the rest of the
companies that are interconnected through peer-to-peer relationships. An ABM has been developed that
simulates a multilayer network involving di erent types of companies. The authors demonstrate that the
number of links, the volume of sales, and the total pro t of a node in the network have an impact on its
survival throughout time.

Dias et al. (2015) analysed a network model of banking relationships in the interbanking market and with
clients and to study the relationships between di erent agents, accounting, and nancial concepts. Their
goal is to understand how propagation of failures in the banking network occurs in a very short-run
analysis. They conclude that the presence of systemic risk for certain setups and their analysis provide some
insights for policymakers on questions about solvability minimum requirements along with market
regulation.

3.1. Social Influence, Persuasion, and Di usion Processes

SNA can be used to identify key in uencers or potential ‘super-spreaders’ in the network or to study the
spread of information or disease within a social network. Applications to health include investigating the
key factors that in uence the di usion process. Alexander et al. (2022) used two di erent network
strategies in health care. The goal of their work was to measure the impact of seeding social networks on the
adoption of a public health intervention in urban India. Strategies included choosing di erent types of
people as seeds, such as (1) targeting 20 per cent of women at random, (2) targeting friends of such
randomly chosen women, (3) targeting pairs of people composed of randomly chosen women and a friend,
or (4) no targeting. The authors concluded that both targeting algorithms (friendship nomination and pair
targeting) enhanced adoption of a public health intervention. In particular, the targeting of pairs of friends
enhanced adoption of novel practices through both social in uence and social reinforcement.

Block et al. (2020) simulated stochastic infection curves incorporating core elements from infection models,
ideal-type social network models, and statistical relational event models. The authors demonstrate that a
strategic social network–based reduction of contact strongly enhances the e ectiveness of social distancing
measures while keeping risks lower.

There are also several applications to marketing and advertising in the research on how rms can use social
networks to reach new customers and how customers interact with brands on social networks. SNA is used
to study the spread of information, or misinformation, opinions, innovations, or diseases within a social
network, and allows identifying key players and in uencers. These in uencers can shape opinions,

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
decisions, and behaviours of others. Understanding who these key players are is essential for targeted
interventions, marketing strategies, or policy implementations. Tseng and Nguyen (2020) identify and
evaluate the factors that impact the spread of rumours and propose a strategy to counteract these
phenomena using agent-based modelling and social impact theory.

Peralta et al. (2021) investigate opinion dynamics and information spreading on networks under the
in uence of content- ltering technologies. They consider the case where the social platform shows a
predilection for one opinion over its opposite, unbalancing the dynamics in favour of that opinion. In their
ABM, they consider that a fraction of the neighbours of an individual holding disagreeing opinions is
ltered, and thus interactions with those neighbours are not possible.

4. Structures of Social Networks

Jackson et al. (2017b) surveyed the literature on the economic consequences of the structure of social
networks to measure the e ects of di erent network structures on the behaviour of individuals within the
network (patterns’ identi cation, resilience issues and trends in network structures) and to study of the
formation and evolution of social networks. Although it has not been created speci cally for ABM, the
authors o ered a framework through which to understand how structural properties of a network impact
the behaviours of the agents who comprise the network. The framework is based on the fundamental
characteristics of networks: degree distributions, homophily patterns, clustering, and the centrality of
nodes.

The taxonomy includes two levels of analysis—‘macro’ and ‘micro’ characteristics of social interaction
networks—and emphasize that the relationship between social structure and economic behaviour is not
unidirectional, as the relationships that constitute a given network are endogenous and determined partly
by economic behaviours. There is no exact distinction between macro and micro characteristics. However,
we can say that macro questions address issues that are society-wide, such as identifying the conditions
under which a process of contagion is likely to lead to a persistent level of infection or the extent to which
polarized views are likely to coexist in society. On the other hand, the micro questions address issues that
tend to focus on a given individual or a small subset of society, such as how in uential a given agent is in
shaping the opinions of others or whether or not two friends have su cient incentives to exchange favours.

A reduced version of Jackson’s taxonomy has been used in our work where the level ‘macro’ is subdivided
into macro structures (that include community and homophily) and macro patterns of interaction (covering
topics such as propagation of shocks, di usion, contagion). The level ‘micro’ (local) is subdivided into
micro interactions and multilayers and clustering.

In macro structures, the taxonomy includes characteristics such as the density of links or segregation
patterns. Macro questions focus on broad societal issues, such as determining when a contagious process
will maintain a consistent rate of infection, or how polarized opinions might persist within a society
(Jackson et al. 2017b).
Micro/local/individual characteristics include those such as whether some given person’s friends are
friends with each other. The main issues of the micro level tend to focus on a given individual or a small
subset of society, as stated above (Jackson et al. 2017b).

We have identi ed the most recent and signi cant studies of ABMs that can be classi ed according to the
Jackson et al. (2017a) taxonomy (see Table 2).

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
4.1. Macro Structures
Starting with macro structures, Sikder et al. (2020) studied opinion polarization and cascade dynamics.
They used US county-level data on the impact of internet access on the formation of beliefs about global
warming. The authors conclude by discussing policy implications of their model, highlighting the
downsides of debunking and suggesting alternative strategies to contrast misinformation.

Campos et al. (2013) used a combination of SNA and ABM where rms (the agents) can collaborate and
create networks for research and development purposes. Three collaboration strategies for partner selection
have been compared: strategy 1 (peer-to-peer complementariness), strategy 2 (the concentration process),
and strategy 3 (virtual cooperation networks). In strategy 1, a linear topology emerges frequently. Networks
are smaller, having a lower diameter and lower pro t (one of the main characteristics of the study). Star
networks, containing more than eight nodes, on average, are often observed in strategy 2 (see Figure 9).
Clusters of rms have been found inside some networks, identifying groups that are linked together around
a particular rm. These situations are typical in multipolar networks (somewhat similar to power
networks). Strategy 3 shows di erent topologies, although the star topology is the most frequently observed
in these strategies.

Figure 9. Power networks emerging from strategy 2 in a firmʼs collaboration framework (Campos et al. 2013) (numbers inside
the circles are the firmʼs identifiers).

Wilhite (2006) provides some good examples involving ABMs and network topologies. His goal is to see to
what extent the topology of a network whose nodes are occupied by agents a ects the resulting economic
decisions. Coordination and cooperation are essential aspects in a society where conventions guiding
interactions often become too costly to follow. He explored trade on a network with only two goods and used
1000 agents each possessing his Cobb–Douglas utility function. Then he observed how a network’s topology
(complete, star, ring, grid, tree, small world, and power network) would a ect trade. The author concluded
that small world and power network allocate goods to reach equilibrium price. The agents in hubs in the
power networks are more pro table, but these topologies (small world and power network) do not
concentrate earning power into the hands of a few.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
4.2. Macro Patterns of Interaction
Still in the macro level, a di erent approach considers homophily and patterns of interaction, propagation
of shocks, di usion, and contagion. If a network has strong homophily, then there are relatively fewer
bridges across groups, and as a consequence, when a contagion is spread through contact that requires
more than just one interaction, the di usion can remain con ned within certain areas of the network.

Schelling (1971) o ered a simple example of how individual actions can lead to an unanticipated aggregate
behaviour. Schelling’s original model of segregation is de ned in a grid network containing N ⨉ N agents
that are split into two groups (black and whites, for example). Agents check their neighbourhood to see if
the fraction of neighbours that match their groups is greater than a fraction de ned previously. If it has
already happened, then agents are satis ed. Otherwise, agents will choose to relocate to an empty cell in the
grid. The model is especially useful for the study of residential segregation of ethnic groups where agents
represent householders who relocate in the city. Hatna and Benenson (2012) explore whether the Schelling
model can reproduce such patterns in Jewish–Arab ethnic residential distribution in Israeli cities. They
reveal a new type of model pattern in which a portion of one group segregates while another portion
remains integrated with the second group.

4.3. Micro Interactions and Multilayers

Tseng and Nguyen (2020) used a network-based approach to simulate rumours that propagate through oral
communication. One type of model that is often used to model network propagation of a virus is SIR
(susceptible, infected, and recovered), corresponding to three possible states of an agent. These alternative
states correspond to compartments in which any agent may alternate. The mechanisms for controlling
rumours are introduced and an approach for dispelling the rumour e ect in the workplace using an ABM
and SNA with micro interactions is explained. Based on their results, Tseng and Nguyen (2020) have found
that organizations can signi cantly reduce the spread of the rumours by improving the workplace
environment and instituting counselling for those in management positions.

Kandiah et al. (2019) used an agent-based modelling approach to simulate opinion dynamics based on the
social ampli cation of risk and captured changes in perceptions about the risks and bene ts of water reuse.
Application in the town of Cary, North Carolina, has been made using data about new water reclamation
accounts and plans for infrastructure expansion. The authors used a reticulation network (tree-based
network) to distribute reclaimed water to residential and commercial customers.

4.4. Micro Level (Clustering and Local Interaction)

In this level of analysis, information travels quickly and reliably among tightly clustered groups of people
(Jackson 2017b). For instance, if any individual misbehaves and their friends are all friends with each other,
then the news ows through the links in the network, and this can lead many of their friends to quickly hear
that news. Thus, the local patterns of friendships are important in not only how information about
behaviour spreads but also how people are able to coordinate in responding to a behaviour.
Eismann (2021) uses agent-based simulations to illustrate that searchability can hinder actors seeking to
evaluate the trustworthiness of a rumour’s source and hence impede self-correction. The ndings indicate
that exchanges between related users can increase the likelihood that trustworthy agents transmit rumour
messages, which can promote the propagation of useful information and corrective posts. Social media
networks have been used. SNA has been introduced in the research, and the ndings suggest that
su ciently high levels of homophily are crucial to ensure that agents who seek the truth behind a rumour

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
primarily receive trustworthy information from their direct and indirect contacts.

Ushchev and Zenou (2020) analyse the comparative statics, welfare properties, and policy implications of
the local average model. They characterize the Nash equilibrium in the local-average model and show that
individual e orts, social norms, and aggregate e ort are the weighted sums of productivity. Then, they
provide comparative statics of individual and aggregate e orts with respect to the key parameters of the
model, focusing especially on the taste for conformity. SNA is used to study the impact of adding a link on
the equilibrium e orts of all agents in the network. Ushchev and Zenou (2020) observe that all agents in the
network increase their e ort if and only if a link between two agents with su ciently high productivities is
added in the network. This result seems to produce a snowball e ect.

Schulz and Mayerho er (2021) study the e ect of heterogeneous initial productivities arising from locally
segregated markets on aggregate outcomes regarding revenue distributions. They introduce a network
structure of localized competition and innovation that makes it possible to nd two modes of competition:
global competition for sales and localized competition for market power. Their model contests the
conventional wisdom of DSGE models that maximum competition leads to minimum concentration of
revenue.
Table 2. Classification of ABM literature according to the SNA levels of analysis

SNA level Network Economic and management goals Examples of ABM studies
of analysis characteristics

Macro Macro structures Firm selection, social norms emerging, opinion

• Krichene et al. (2019)
(aggregate) polarization, and cascade dynamics

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
(suppliers and customers
selection)
• Nogueira et al. (2023)
(industrial partner selection)
• Sikder et al. (2020) (opinion
polarization and cascade
dynamics)
• Campos et al. (2013) (partner
selection)
• Wilhite (2006) (coordination
and cooperation)

Macro patterns of Poverty traps, propagation of shocks, di usion,

• Schelling (1971) (social
interaction contagion, and social segregation
segregation)
(homophily and
segregation) • Acemoglu et al. (2012)
(propagation of shocks in
financial crisis)
• Dias et al. (2015) (propagation
of shocks in industrial
networks)
• Hatna and Benenson (2012)
(social segregation in Jewish–
Arab cities)

Micro Micro interactions and Spread of information, innovation adoption,

• Cabrera et al. (2021) (disease
(local) multilayers opinion dynamics, and rumour spreading
flows within a network)
• Kandiah et al. (2019) (opinion
dynamics)
• Peralta et al. (2021) (opinion
formation and dynamics)
• Tseng and Nguyen (2020)
(rumour spread)

Clustering and local Reinforcement of behaviours (e.g., disease

• Block et al. (2020) (disease
interaction behaviour, social norms)
infection models)
• Hoertel et al. (2020) (social
contact network for Covid-19)
• Schulz and Mayerho er (2021)
(industrial competition)
• Ushchev and Zenou (2020)
(social norms)
• Eismann (2021) (di usion and
persistence of false rumours in
social media)
 Micro– Micro interactions and Dynamics of consumer credit, timing of
• Boero et al. (2004) (adaptation
macro macro structure retirement, micro behavioural attitudes, and
in industrial districts)
interaction macro technological adaptation
• Axtell and Epstein (2006)
(timing of retirement)
• Francisco et al. (2011) (studies
of science scientific
conferences and symposia)

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
• DʼOrazio and Giulioni (2017)
(economic model with
consumer credit)
• Narasimhan et al. (2017)
(duality between structure
[macro] and agency [micro])

Interactions between micro and macro structures are not explicitly addressed in the work of Jackson et al.
(2017b). So, we extended their taxonomy by considering a new level of micro–macro interaction that we
introduce in the following paragraphs.

4.5. Micro–Macro Interaction

While the examination of interactions between micro and macro levels of analysis is not a novel concept and
has been explored in various contexts, it remains an unresolved matter within the eld of ABM. Boero et al.
(2004) studied micro behavioural attitudes and macro technological adaptation in industrial districts.

Motivated by a policy question related to the timing of retirement, Axtell and Epstein (2006) studied the
connection between individual rationality and aggregate e ciency. In other words, their work focuses on
the links that exist between optimization by individuals and optimization in the aggregate. Their study is
supported by social interactions and social networks. The social networks are set up by individual agents,
with a random size, and the set of all social networks (with the network relations as directed edges)
constitutes a single random graph. The authors argue that with social network interactions, very little
individual rationality is needed for society as a whole to exhibit an optimal behaviour. The nonequilibrium
dynamics will depend on both the size and the structure of networks.

For a bridge between micro and macro studies of science scienti c conferences and symposia, Francisco et
al. (2011) suggest using robotics as a case study of global, cross-cultural, and interdisciplinary scienti c
practices. D’Orazio and Giulioni (2017) developed an ABM populated by heterogeneous consumers, a
productive sector and a banking sector to explore the micro behaviours and the corresponding macro
dynamics of an agent-based economic model with consumer credit.

Narasimhan et al. (2017) developed a model based on the principle of structuration theory: the duality
between structure (macro) and agency (micro). They conceptualized the duality between practitioners and
practices based on theoretical insights from social practices literature.
5. Learning in Agent-Based Network Modelling

In agent-based studies, agents’ preferences are typically xed, appear to be in uenced exogenously, and
remain stable across many time horizons (Namatame and Chen 2016). Although researchers often analyse
the combined actions of agents driven by their inherent individual preferences, it is valuable to attempt
elucidating the origin and formation of these preferences. One feasible method to tackle this matter is to

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
embrace a learning-oriented standpoint (such as reinforcement learning) when examining the evolution of
preferences.

Sometimes the term adaptation is used as a principle of learning (Lucas 1988), when agents review and
adapt decision rules in a trial-and-error process through which the modes of behaviour are determined.
New decisions are tried and tested against experience, and rules that produce desirable outcomes supplant
those that do not. Belew and Mitchell (1996) de ned adaptation as ‘the capacity for change and the
additional requirement that this change signi es an improvement of t’. Learning is adapting to the
constantly changing environment. Individuals adapt their behaviour in response to others’ behaviour and to
the environment. We can view the problem of the adaptation of a group of agents in two di erent ways:
either as an individual characteristic of the agents—learning—or as a collective process bringing
reproductive mechanisms into play—evolution (Ferber 1999).

Multiagent learning is today arguably one of the most fertile interaction grounds between computer science
and game theory. According to Brenner (2006), there are essentially three ways of learning in economic
literature: nonconscious learning, routine-based learning, and belief learning, with some of these ways
being connected with the current machine learning algorithms. Following Dahlke et al. (2020), machine
learning is mainly used in ABM for two main purposes: (1) the modelling of adaptive agents equipped with
experience learning and (2) the analysis of outcomes produced by a given ABM. For an overview of agent-
based modelling and learning in economics, see Campos et al. (2023). For a general structured literature
review to investigate how the ABM process uses machine learning techniques, see Dehkordi et al. (2023).

The literature related to agent learning obviously intersects with network science. The agents in networks
learn two things: with whom to interact and how to behave (Namatame 2007). A crucial ingredient in social
interaction models is the network structure in which individuals interact. Several studies address the topic
of learning in networks (e.g., Namatame 2007; Özman 2007; Seltzer and Smirnov 2015; Xiong et al. 2018). A
new concept of collective evolution in a society of networked agents is proposed by Namatame (2007), who
develops two approaches to model the learning behaviour of agents: the microscopic model (based on
individual learning) and the macroscopic model (based on social learning). In individual learning, agents
are modelled to have some behavioural rules, and they update those rules using the existing rules. In
‘classical’ social learning, agents take decisions based on prescribed behavioural rules. As space imposes
important constrains on interaction, individuals in most social processes interact with the highest
probability with others who are nearby. The probability of interactions drops down as a square of the
distance between the places where each individual lives.

It is possible to consider the adaptive choices of agents by incorporating two factors: (1) individual
reinforcement, in which agents reinforce their preferences based on prior choices, and (2) social
reinforcement, in which agents reinforce choice probability based on the choices of other agents in similar
decision environments (Namatame and Chen 2016). Reinforcement is a stimulus that increases the
probability of a behaviour to be repeated. With a combined model of learning and speci c network structure,
interacting agents are generally driven toward heterogeneous individual preferences. In their work,
Namatame and Chen (2016) de ne di erent levels of social learning and analyse the outcomes in the
perspective of network topologies: with (strong) social reinforcement, most agents reinforce their
preferences when agents are located on the complete network or the scale-free network. With (mild) social
reinforcement, most agents are split into two groups, and most diversities are observed when agents are
located on the scale-free network and the core–periphery network.

In a network setting, agents should evaluate their actions not in isolation but with the expectation that the
world will react to their actions. This means that cause–e ect relationships can become quite subtle.

There is also the random matching model, where each agent is assumed to interact with a randomly chosen

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
agent from the population. Namatame (2007) shows that networked agents collectively evolve e cient
rules that realize a Pareto-e cient outcome of the underlying games. The essential point is that desirable
behavioural rules emerge spontaneously at the collective level from the pairwise interactions of networked
agents.

Seltzer and Smirnov (2015) propose an ABM of a social network in which agents make contributions to
others conditional upon the social distance. There are two di erent models under analysis: a simple
imitation model, where the emerging strategy pro le is characterized by high levels of cooperation with
those who are directly connected to the agent, and a model where agents learn and imitate socially distant
neighbours. The latter increases cooperation for all types of interaction. The authors concluded that
combining greater interaction distance with greater learning distance leads to a positive change in the total
social welfare produced by the agents’ contributions.

In Chen and Gostoli (2017), the agents take their decisions on the basis of their neighbours’ past decisions.
The authors investigate the e ect of two network structures: the circular neighbourhood and the von
Neumann neighbourhood in learning in the El Farol bar problem. The authors retain the best-reply
strategies of the original El Farol bar problem and modify the standard settings by adopting the
informational structure introduced by the works on the minority games with local interaction. They
conclude that by using local information, the agents could overcome the problem of herd behaviour,
generating the use of global information, and consequently, being able to improve their coordination has
proved to be correct.

Chen et al. (2023) studied the emergence of arti cial collective intelligence through massive-agent
reinforcement learning. They propose an open-sourced environment Lux, where hundreds of agents in two
teams compete for resources and build cities.

6. Agent-Based Models of Social Networks

In this section, we focus on the most relevant agent-based models of social networks, based on the work of
Namatame and Chen (2016), in which we emphasize aspects of network structure and learning. Namatame
and Chen (2016) provide a review of the study of social network formation, from both agent-based
modelling and the game-theoretical perspective. They address the question on how the ABMs of social
networks can be related to the standard model of network formation games, in either the cooperative
version or the noncooperative version, since it is assumed that agents know the cost and the payo of each
connection and are neither adventurous nor driven by chance discovery. Besides the game-theoretical
perspective, the authors identify di erent ABMs where di erent typologies emerge as the result of the local
interaction of the agents (see Table 3). Agent-based modelling is applied here as a generation process, as in
the scope of the generative social science of Epstein (2006). Although these models are rather di erent from
the game-theoretical perspective, they also rely on the use of various theoretical economic games, such as
the prisoner’s dilemma game, ultimatum game, trust game, and the public good game. It worthy to note
that the games are not xed in time–space but actually evolve with the networks.
Table 3. Relevant agent-based models of social networks

Agent-based Goal/findings Network topologies Learning

models of methods
social
networks

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Skyrms– Social network structure emerges as a Star network and other Reinforcement
Pemantle consequence of the dynamics of the agentsʼ typologies learning
model (Skyrms learning behaviour (multiarmed
and Pemantle bandit)
2000)

Zimmermann– Explore the coupled dynamics of the internal Small-world network __

Eguiluz model states of a set of interacting elements and the
(Zimmermann network of interactions among them
et al. 2004)

Zschache The network members are engaged in collective Complete network ___
model action. The success of the collective action
(Zschache depends on the initial network structure.
2012)

Bravo– Repeated trust game. The model examines the Fully connected (or complete) ___
Squazzoni– relationship between the personality and the network; small-world network,
Boero model resultant personal networks, as well as the the scale-free network; core–
(Bravo et al. resultant earnings. peripheral network
2012)

Network-based Impact of the technology potential on the Networks with fully connected ____
trust games network formation components
(Chen et al.
2015)

Model of Combining greater interaction distance with Small-world network Learning by

Seltzer and greater learning distance leads to a positive imitation
Smirnov (2015) change in the total social welfare produced by
the agentsʼ contributions.

Model of Chen Investigate the e ect of network structures and Circular neighbourhood (ring) Social learning
and Gostoli learning on the El Farol bar problem and the von Neumann (imitation) and
(2017) neighbourhood individual
learning
(mutation)

Source: Based on Namatame and Chen (2016).

Note: The models of Setlzer and Smirnov (2015) and Chen and Gostoli (2017), introduced earlier, have been added to the original
review of Namatame and Chen (2016).

6.1. The Skyrms–Pemantle Model

Skyrms and Pemantle (2000) consider a dynamic social network model in which agents play repeated games
in pairings determined by a stochastically evolving social network. Individual agents begin to interact at
random, with the interactions modelled as games. The authors conclude that modelling network structure
as dynamic increases realism without rendering the problem of analysis intractable.
6.2. The Zimmermann–Eguiluz Model
In the Zimmermann–Eguiluz model (Zimmermann et al. 2004), the authors analyse a simple setting of an
adaptive and evolving network, in which there is coevolution of the state of the elements in the nodes of the
network and the interaction links de ning the network. This is the earliest prominent example showing
how a social network can be developed through a prisoner’s dilemma game.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
6.3. The Zschache Model
As agents participate in public goods production, processes of social comparison may a ect the decisions
about the individual contributions as well as the endogenously changing structure of an underlying social
network. Zschache (2012) focused on social norms: agent i will be happier if their action is closer to the
centroid of the actions taken by their neighbours and unhappier if their action is further away from the
centroid. The goal of the Zschache model is to trace the formation of public goods while varying network
characteristics, such as density, segregation, or the strength of relationships.

6.4. The Bravo–Squazzoni–Boero Model

In this model (Bravo et al. 2012), the importance of the endogenous selection of partners for trust and
cooperation in market exchange situations is investigated, where there is information asymmetry between
investors and trustees. The authors tested various social structure con gurations and showed that dynamic
networks lead to more cooperation when agents can create more links and reduce exploitation opportunities
by free riders.

6.5. Network-Based Trust Games

In the model of trust by Chen et al. (2015), the authors demonstrate the positive e ects of trust on growth.
The trust considered here is myopic trust, which can a ect agents’ decisions regarding networking,
portfolios, and kickbacks, which in turn a ect network formation, wealth creation, and distribution.

7. Further Issues and Directions for Further Work

There are several issues that should take place in this review and have not been covered in the previous
sections, such as empirical validation and resilience, among others. In this section, we make a short
overview of these topics by considering some relevant studies and provide directions for future work.
7.1. Empirical Validation of ABM
Fischbach et al. (2021) provide an overview of the use of ABM in social sciences and enhance the empirical
validation of the old models. Early simulations, they state, are sometimes based on highly idealized or even
empirically false assumptions, raising the question of the utility of such models. Windrum et al. (2007) raise
a set of issues that are common to all modellers engaged in empirical validation and identify three
alternative methodological approaches for empirical validation of ABM: indirect calibration, the Werker–

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Brenner approach, and the history-friendly approach. The indirect calibration approach rst performs
validation and then indirectly calibrates the model by focusing on the parameters that are consistent with
output validation. The main di erence for the Werker–Brenner approach is that here one tries to pick
empirical parameters directly to calibrate the model. The history-friendly approach constrains parameters,
interactions, and decision rules in the model in line with the speci c, empirically observable history of a
particular industry. It can be interpreted as a calibration exercise with respect to a unique historical trace.

Moss (2008) addresses ‘companion modelling’, a key di erence from conventional social science, including
agent-based computational economics. In ‘companion modelling’, the idea is to represent social processes
as perceived by participating stakeholders. It does not employ representations of behaviour based on utility
functions or social environments cast.

7.2. Resilience in SNA

The resilience of networks is a very important aspect, which has recently been going through social,
political, and nancial crises. In complex systems, resilience is addressed from a structural point of view,
regarding the preservation of the connectivity against the suppression of individual nodes or individual
links. In a study of innovation networks, Lozano and Arenas (2007) associate the resilience concept with the
capability of a social organization to keep acceptable levels of functionality against external socioeconomic
disrupting factors that do not imply necessarily destruction of existing links.

Nogueira et al. (2023) investigate how companies react to market change by understanding the e ect of a
perturbation (such as a business cessation) on the rest of the companies that are interconnected through
peer-to-peer relationships. The ABM produces the network dynamics due to the changes in its
con guration throughout time. The authors demonstrate that the number of links, the volume of sales, and
the total pro t of a node in the network have an impact on its survival throughout time.

7.3. Future Work

One of the novel contributions of this work stands in the links found between ABM and SNA that we
classi ed following the topology of Jackson et al. (2017a). Although this topology has been originally created
for the economic networks’ literature, we applied and extended it for ABMs. ABMs modelled with SNA have
been classi ed in macro structures, macro patterns of interaction, micro interactions and multilayers, and
micro level (clustering and local interaction). The studies and models that we identi ed in this chapter
examine network dynamics, agent interactions, and their implications in various domains, contributing to a
deeper understanding of complex systems and behaviours.

In this chapter, we have room for further considerations. We have touched lightly on some extensions
related to learning, where networks play a fundamental role in modelling social interactions—namely, in
the way that agents interact more easily with agents who are spatially closer in the network. Learning by
agents is a very important component for further studies.
Furthermore, interconnections exist between micro and macro structures, a subject that is not directly
tackled in the research by Jackson et al. (2017b) and not entirely novel within the realm of ABM literature.
Although the investigation of interactions between micro and macro analysis levels is not without precedent
and has been scrutinized in various contexts, there is a need for further exploration within the eld of ABM.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
References
Acemoglu, Daron, Vasco Carvalho, Asuman Ozdaglar, and Alireza Tahbaz-Salehi. 2012. ʻThe Network Origins of Aggregate
Fluctuationsʼ. Econometrica 80 (5): 1977–2016.
Google Scholar WorldCat

Albert, Réka, and Albert-László Barabási. 2002. ʻStatistical Mechanics of Complex Networksʼ. Reviews of Modern Physics 74 (47):

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
47–97.
Google Scholar WorldCat

Alexander, M., L. Forastiere, S. Gupta, and N. Christakis. 2022. ʻAlgorithms for Seeding Social Networks Can Enhance the Adoption
of a Public Health Intervention in Urban Indiaʼ. Proceedings of the National Academy of Sciences 119 (30): 2120742119.
Google Scholar WorldCat

Axtell, Robert, and Joshua Epstein. 2006. ʻCoordination in Transient Social Networks: An Agent-Based Computational Model of
the Timing of Retirementʼ. In Generative Social Science: Studies in Agent-Based Computational Modeling, edited by
Joshua M. Epstein. Princeton, NJ: Princeton University Press.
Google Scholar Google Preview WorldCat COPAC

Barabási, Albert-László, and Réka Albert. 1999. ʻEmergence of Scaling in Random Networksʼ. Science 286: 509–512.
Google Scholar WorldCat

Belew, Richard, and Mitchell, Melanie. 1996. ʻIntroductionʼ. In Adaptive Individuals in Evolving Populations: Models and
Algorithms, edited by R. Belew and M. Mitchell. Sante Fe, NM: Santa Fe Institute in the Sciences of Complexity, Addison Wesley.
Google Scholar Google Preview WorldCat COPAC

Block, Per, Marion Ho man, Isabel J. Raabe, Jennifer Beam Dowd, Charles Rahal, Ridhi Kashyap, and Melinda C. Mills. 2020.
ʻSocial Network-Based Distancing Strategies to Flatten the COVID-19 Curve in a Post-Lockdown Worldʼ. Nature Human Behaviour
4: 588–596.
Google Scholar WorldCat

Boero, Riccardo, Marco Castellani, and Flaminio Squazzoni. 2004. ʻMicro Behavioral Attitudes and Macro Technological
Adaptation in Industrial Districts: An Agent-Based Prototypeʼ. Journal of Artificial Societies and Social Simulation 7.
https://www.jasss.org/7/2/1.html.
Google Scholar WorldCat

Bravo, Giangiacomo, Flaminio Squazzoni, and Riccardo Boero. 2012. ʻTrust and Partner Selection in Social Networks: An
Experimentally Grounded Modelʼ. Social Networks 34 (4): 481–492.

Brenner, Thomas. 2006. ʻAgent Learning Representation: Advice on Modelling Economic Learningʼ. In Handbook of
Computational Economics, edited by Leigh Tesfatsion and Kenneth L. Judd, vol. 2, 895–947. Amsterdam: Elsevier.
Google Scholar Google Preview WorldCat COPAC

Cabrera, Benjamin, Bjorn Ross, Daniel Röchert, Felix Brünker, and Stefan Stieglitz. 2021. ʻThe Influence of Community Structure
on Opinion Expression: An Agent-Based Modelʼ. Journal of Business Economics 91: 1331–1355.
Google Scholar WorldCat

Campos, Pedro, Pavel Brazdil, and Isabel Mota. 2013. ʻComparing Strategies of Collaborative Networks for R&D: An Agent-Based
Studyʼ. Computational Economics 42: 1–22.
Google Scholar WorldCat

Campos, Pedro, Annand Rao, and Pavel Brazdil. 2023. ʻAgent-Based Modeling and Learning in Economics—An Overviewʼ. In A
Machine Learning Perspective of Agent-Based Models—Applications to Economic Crises and Pandemics, with Practical Examples in
R and Python. Springer.
Google Scholar Google Preview WorldCat COPAC

Chen, Hamo, Stone Tao, Jiaxin Chen, Weihan Shen, Xihui Li, Chenghui Yu, Sikai Cheng, Xiaolong Zhu, and Xiu Li. 2023. ʻEmergent
Collective Intelligence from Massive-Agent Cooperation and Competitionʼ. arXiv. https://arxiv.org/abs/2301.01609.
WorldCat

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Chen, Shu-Heng. 2012. ʻVarieties of Agents in Agent-Based Computational Economics: A Historical and an Interdisciplinary
Perspectiveʼ. Journal of Economic Dynamics and Control 36 (1): 1–25.
Google Scholar WorldCat

Chen, Shu-Heng, Bin-Tzong Chie, and Tong Zhang. 2015. ʻNetwork-Based Trust Game: An Agent-Based Modelʼ. Journal of Artificial
Societies and Social Simulation 18 (3). https://www.jasss.org/18/3/5.html.
Google Scholar WorldCat

Chen, Shu-Heng, and Umberto Gostoli. 2017. ʻCoordination in the El Farol Bar Problem: The Role of Social Preferences and Social
Networksʼ. Journal of Economic Interaction and Coordination 12: 59–93.
Google Scholar WorldCat

Dahlke, Johannes, Kristina Bogner, Matthias Mueller, Thomas Berger, Andreas Pyka, and Bernd Ebersberger. 2020. ʻIs the Juice
Worth the Squeeze? Machine Learning (ML) in and for Agent-Based Modelling (ABM)ʼ. Arxiv.
https://doi.org/10.13140/RG.2.2.16497.02403.
WorldCat

Dehkordi, Molood Ale Ebrahim, Jonas Lechner, Amineh Ghorbani, Igor Nikolic, Émile Chappin, and Paulien Herder. 2023. ʻUsing
Machine Learning for Agent Specifications in Agent-Based Models and Simulations: A Critical Review and Guidelinesʼ. Journal of
Artificial Societies and Social Simulation 26 (1). https://doi.org/10.18564/jasss.5016.
Google Scholar WorldCat

Dias, André, Pedro Campos, and Paulo Garrido. 2015. ʻAn Agent-Based Propagation Model of Bank Failuresʼ. In Advances in
Artificial Economics, edited by François Amblard, Fariba Hashemi, Anne-Laure Michel, and Bruno Gaudou, vol. 676: 119–130.
Cham: Springer.
Google Scholar Google Preview WorldCat COPAC

DʼOrazio, Paola, and Gianfranco Giulioni. 2017. ʻFrom Micro Behaviors to Macro Dynamics: An Agent-Based Economic Model with
Consumer Creditʼ. Journal of Artificial Societies and Social Simulation 20 (1). https://doi.org/10.18564/jasss.3260.
Google Scholar WorldCat

Eden, Matthew, Rebecca Castonguay, Buyannemekh Munkhbat, Hari Balasubramanian, and Chaitra Gopalappa. 2021. ʻAgent-
Based Evolving Network Modeling: A New Simulation Method for Modeling Low-Prevalence Infectious Diseasesʼ. Health Care
Management Science 24: 623–639.
Google Scholar WorldCat

Eismann, Kathrin. 2021. ʻDi usion and Persistence of False Rumors in Social Media Networks: Implications of Searchability on
Rumor Self-Correction on Twitterʼ. Journal of Business Economics 91, 1299–1329. https://doi.org/10.1007/s11573-020-01022-9.
Google Scholar WorldCat

Epstein, Joshua M. 2006. Generative Social Science: Studies in Agent-Based Computational Modeling. Princeton, NJ: Princeton
University Press.
Google Scholar Google Preview WorldCat COPAC

Ferber, Jacques. 1999. Multi-Agent Systems: An Introduction to Distributed Artificial Intelligence. Harlow: Addison Wesley
Longman.
Google Scholar Google Preview WorldCat COPAC
Fischbach, Kai, Johannes Marx, and Tim Weitzel. 2021. ʻAgent-Based Modeling in Social Sciencesʼ. Journal of Business Economics
91: 1263–1270.
Google Scholar WorldCat

Francisco, Matthew, Stasa Milojevic, and Selma Šabanovic. 2011. ʻConference Models to Bridge Micro and Macro Studies of
Scienceʼ. Journal of Artificial Societies and Social Simulation 14 (4). https://doi.org/10.18564/jasss.1838.
Google Scholar WorldCat

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Gardner, Martin. 1970. ʻMathematical Games: The Fantastic Combinations of John Conwayʼs New Solitaire Game ʻLifeʼʼ. Scientific
American 223 (4): 120–123.
Google Scholar WorldCat

Hakansson, Hakan, ed. 1987. Industrial Technological Development: A Network Approach. London: Croom Helm.
Google Scholar Google Preview WorldCat COPAC

Hatna, Erez, and Itzhak Benenson. 2012. ʻThe Schelling Model of Ethnic Residential Dynamics: Beyond the Integrated—
Segregated Dichotomy of Patternsʼ. Journal of Artificial Societies and Social Simulation 15 (1).
https://doi.org/10.18564/jasss.1873.
Google Scholar WorldCat

Hendricks, Ken, Michele Piccione, and Guofu Tan. 1995. ʻThe Economics of Hubs: The Case of Monopolyʼ. Review of Economic
Studies 62: 83–99.
Google Scholar WorldCat

Hoertel, Nicolas, Martin Blachie, Carlos Blanco, Mark Olfson, Marc Massetti, Marina Sánchez Rico, Frédéric Limosin, and
Henri Leleu. 2020. ʻA Stochastic Agent-Based Model of the SARS-CoV-2 Epidemic in Franceʼ. Nature Medicine 26: 1417–1421.
Google Scholar WorldCat

Jackson, Matthew, Joshua Conrad, David Rand, Kevin Lewis, Michael I. Norton, and Kurt Gray. 2017a. ʻAgent-Based Modeling: A
Guide for Social Psychologistsʼ. Social Psychological & Personality Science 8 (4): 387–395.
Google Scholar WorldCat

Jackson, Matthew, Brian W. Rogers, and Yves Zenou. 2017b. ʻThe Economic Consequences of Social-Network Structureʼ. Journal
of Economic Literature 55 (1): 49–95.
Google Scholar WorldCat

Kandiah, Venu, Emily Z. Berglund, and Andrew R. Binder. 2019. ʻAn Agent-Based Modeling Approach to Project Adoption of Water
Reuse and Evaluate Expansion Plans within a Sociotechnical Water Infrastructure Systemʼ. Sustainable Cities and Society 46:
101412.
Google Scholar WorldCat

Krichene, Hazem, Yoshi Fujiwara, Abhijit Chakraborty, Yoshiyuki Arata, Hiroyasu Inoue, and Masaaki Terai. 2019. ʻThe Emergence
of Properties of the Japanese Production Network: How Do Listed Firms Choose Their Partners?ʼ Social Networks 59: 1–9.
Google Scholar WorldCat

LeBaron, Blake, and Leigh Tesfatsion. 2008. ʻModeling Macroeconomies as Open-Ended Dynamic Systems of Interacting Agentsʼ.
American Economic Review 98 (2): 246–250.
Google Scholar WorldCat

Lozano, Sergi, and Alexandre Arenas. 2007. ʻA Model to Test How Diversity A ects Resilience in Regional Innovation Networksʼ.
Journal of Artificial Societies and Social Simulation 10 (4). https://www.jasss.org/10/4/8.html.
Google Scholar WorldCat

Lucas, Robert E. 1988. ʻOn the Mechanics of Economic Developmentʼ. Journal of Monetary Economics 22 (1): 3–42.
Google Scholar WorldCat
Marcelo, João. 2017. ʻMining Multivariate Networksʼ. MSc thesis, Faculty of Economics, University of Porto.

Moss, Scott. 2008. ʻAlternative Approaches to the Empirical Validation of Agent-Based Modelsʼ. Journal of Artificial Societies and
Social Simulation 11 (1). https://www.jasss.org/11/1/5.html.
Google Scholar WorldCat

Namatame, Akira. 2007. ʻCollective Intelligence of Networked Agentsʼ. In Emergent Intelligence of Networked Agents, edited by

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Akira Namatame, Shu-Heng Chen, and Hiroshi Nakashima, Studies in Computational Intelligence, vol. 56: 159–176. Berlin:
Springer.
Google Scholar Google Preview WorldCat COPAC

Namatame, Akira, and Shu-Heng Chen. 2016. Agent-Based Modeling and Network Dynamics. Oxford: Oxford Academic.
Google Scholar Google Preview WorldCat COPAC

Narasimhan, Kavin, Thomas Roberts, Maria Xenitidou, and Nigel Gilbert. 2017. ʻUsing ABM to Clarify and Refine Social Practice
Theoryʼ. In Advances in Social Simulation 2015, edited by Wander Jager, Rudi Verbrugge, Ad Flache, Geert de Roo, and
C. K. Hemelrijk, Advances in Intelligent Systems and Computing, vol. 528: 307–319. Cham: Springer.
Google Scholar Google Preview WorldCat COPAC

Nogueira, Ana, Conceição Rocha, and Pedro Campos. 2023. ʻResilient Agent-Based Networks in Automotive Industryʼ. In A
Machine Learning Perspective of Agent-Based Models. Springer (forthcoming).
Google Scholar Google Preview WorldCat COPAC

Özman, Muge. 2007. ʻNetwork Formation and Strategic Firm Behaviour to Explore and Exploitʼ. Journal of Artificial Societies and
Social Simulation 11 (1). https://www.jasss.org/11/1/7.html.
Google Scholar WorldCat

Peralta, António, János Kertész, and Gerardo Iñiguez. 2021. ʻOpinion Formation on Social Networks with Algorithmic Bias:
Dynamics and Bias Imbalanceʼ. Journal of Physics: Complexity 2 (4). https://iopscience.iop.org/article/10.1088/2632-072X/ac340f.
Google Scholar WorldCat

Schelling, Thomas C. 1971. ʻDynamic Models of Segregationʼ. Journal of Mathematical Sociology 1: 143–186.
Google Scholar WorldCat

Schulz, J., and D. M. Mayerho er. 2021. ʻEqual Chances, Unequal Outcomes? Network-Based Evolutionary Learning and the
Industrial Dynamics of Superstar Firmsʼ. Journal of Business Economics 91: 1357–1385.
Google Scholar WorldCat

Seltzer, Nicholas, and Oleg Smirnov. 2015. ʻDegrees of Separation, Social Learning, and the Evolution of Cooperation in a Small-
World Networkʼ. Journal of Artificial Societies and Social Simulation 18 (4). https://www.jasss.org/18/4/12.html.
Google Scholar WorldCat

Sikder, Orowa, Robert E. Smith, Pierpaolo Vivo, and Giacomo Livan. 2020. ʻA Minimalistic Model of Bias, Polarization, and
Misinformation in Social Networksʼ. Scientific Reports 10: 5493.
Google Scholar WorldCat

Skyrms, Brian, and Robin Pemantle. 2000. ʻA Dynamic Model of Social Network Formationʼ. Proceedings of the National Academy
of Sciences 97 (16): 9340–9346.
Google Scholar WorldCat

Solé, Ricard V., and Sergi Valverde. 2004. ʻInformation Theory of Complex Networks: On Evolution and Architectural Constraintsʼ.
In Complex Networks, edited by Eytan Ben-Naim, Hans Frauenfelder, and Zoltán Toroczkai, Lecture Notes in Physics, vol. 650:
189–207. Berlin: Springer.
Google Scholar Google Preview WorldCat COPAC
Steinbacher, Mitja, Matthias Raddant, Fariba Karimi, Eva Camacho Cuena, Simone Alfarano, Giulia Iori, and Thomas Lux. 2021.
ʻAdvances in the Agent-Based Modeling of Economic and Social Behaviorʼ. SN Business & Economics 1, 99.
https://link.springer.com/article/10.1007/s43546-021-00103-3.
Google Scholar WorldCat

Tseng, Shih-Hsien, and Tien Son Nguyen. 2020. ʻAgent-Based Modeling of Rumor Propagation Using Expected Integrated Mean
Squared Error Optimal Designʼ. Applied System Innovation 3 (4): 48.

Downloaded from https://academic.oup.com/edited-volume/55833/chapter/478781720 by St Loye's Schoool of Health Studies user on 04 September 2024
Google Scholar WorldCat

Ushchev, Philip, and Yves Zenou. 2020. ʻSocial Norms in Networksʼ. Journal of Economic Theory 185: 104969.
Google Scholar WorldCat

Watson, Levens, Alex Szorkovszky, and David J. T. Sumpter. 2022. ʻFriend of a Friend Models of Network Growthʼ. Royal Society
Open Science 9: 221200. https://doi.org/10.1098/rsos.221200.
Google Scholar WorldCat

Watts, Duncan J., and Steven H. Strogatz. 1998. ʻCollective Dynamics of ʻSmall-Worldʼ Networksʼ. Nature 393: 440–442.
Google Scholar WorldCat

Wilhite, Allen. 2006. ʻEconomic Activity on Fixed Networksʼ. In Handbook of Computational Economics, edited by Leigh Tesfatsion
and Kenneth L. Judd, vol. 2, 1013–1045. Amsterdam: Elsevier.
Google Scholar Google Preview WorldCat COPAC

Windrum, Paul, Giorgio Fagiolo, and Alessio Moneta. 2007. ʻEmpirical Validation of Agent-Based Models: Alternatives and
Prospectsʼ. Journal of Artificial Societies and Social Simulation 10 (2). https://www.jasss.org/10/2/8.html.
Google Scholar WorldCat

Xiong, Hang, Diane Payne, and Stephen Kinsella. 2018. ʻIdentifying Mechanisms Underlying Peer E ects on Multiplex Networksʼ.
Journal of Artificial Societies and Social Simulation 21 (4): 6. https://www.jasss.org/21/4/6.html.
Google Scholar WorldCat

Zimmermann, Martin, Victor M. Eguiluz, and Maxi San Miguel. 2004. ʻCoevolution of Dynamical States and Interactions in
Dynamic Networksʼ. Physical Review E 69 (6): 065102.
Google Scholar WorldCat

Zschache, Johannes. 2012. ʻProducing Public Goods in Networks: Some E ects of Social Comparison and Endogenous Network
Changeʼ. Social Networks 34 (4): 539–548.
Google Scholar WorldCat