Urban modelling approaches have over the years, co-evolved with theory development as well as advances in computing and geographical information systems (Wise et al., 2016). Consequently, since the early 1970s, there has been a gradual shift from aggregate, deterministic urban models grounded in classical econometric and entropy-based spatial interaction modelling traditions towards disaggregate stochastic modelling approaches broadly classified in the literature as micro-simulation models. Several theoretical propositions including Random-utility theory (McFadden, 1973), Time-geography theory (Hagerstrand, 1970; Chapin, 1974) as well as Systems and Complexity theory (Batty, 2007; Allen, 2012; Moroni, 2015; Forrester, 1993) have driven the transitions towards disaggregate modelling.
The literature on disaggregate modelling is extensive. In Chapter two, the theoretical foundations of this modelling paradigm, the accompanying techniques as well as models that have emerged from it were discussed. Here, a summary of the relative strengths and limitations of aggregate and disaggregate urban modelling is presented.
As summarized in table 6.1, the disaggregate modelling paradigm has several advantages over aggregate approaches. This includes the ability to represent a wide range of urban actors or
decision-makers (i.e. heterogeneous actors) and their complex behaviours which derive from their personal-level attributes as well as the attributes of their surrounding environments (Pinjari and Bhat 2011; Rasouli and Timmermans 2014). Moreover, disaggregate urban modelling approaches helps to overcome and/or relax weak assumptions including spatial homogeneity, monocentricity, spatial equilibrium as well as the presence of unboundedly rational decision-makers with unlimited access to information (Batty, 2017; Acheampong and Silva, 2015).
Table 6.1: Comparison between aggregate and disaggregate urban modelling approaches
Aggregate modelling approach Disaggregate modelling approach
Homogenous urban actors/decision-makers Representation of heterogeneous urban actors/decision makers
Representation of urban landscape as homogenous
Representation of urban landscape as heterogeneous Simplifying and restrictive assumptions
including rationality, unlimited information flow and access
Ability to overcome or relax simplifying assumptions and improve model realism by incorporating
bounded rationality, heuristics and choice situations under conditions of uncertainty
Assumption of urban systems in equilibrium Urban systems are assumed to be in perpetual disequilibrium
Often top-down deterministic specification of urban systems and behavioural rules
Often bottom-up specification of complex and adaptive rules leading to self-organization and emergent behaviour.
Simple, tractable and requires less computational power
Complex, high computational demand, stochastic variation and output uncertainty
6.3.1 Agent-based modelling as a disaggregate urban modelling paradigm
ABM as a disaggregate urban modelling paradigm, has it origins in the micro-simulation approaches to urban modelling. Fundamentally, the concept of micro-simulation is one in which the aggregate behaviour of a system is explicitly simulated over time as the sum of the actions and interactions of disaggregate behavioural units within the system (Iacono, et al., 2008; Miller and Savini, 1998). Micro-simulation models in general, derive their strength from their dynamic nature, which makes it possible to trace model components (e.g., individuals, households, jobs, and dwellings) over time and to observe the modelled processes of change at a level of detail that is not possible in other types of models (Pagliara and Wilson, 2010; Huang et al., 2014).
ABM is a bottom-up computational method that allows for the creation, analysis and experimentation with models composed of autonomous agents that interact with each other and their environment locally (Gilbert 2008, Railsback and Grimm 2011, Railsback et al., 2006,
Silva, 2011). Rooted in general systems theory and complexity theory, ABM as a microscopic modelling paradigm allows for a natural description of a complex system in a flexible and robust manner to capture emergent phenomenon (Batty 2001, Bonabeau 2002, Castle and Crooks 2006, Wu and Silva 2010). The approach allows one to represent heterogeneous agents (e.g., household members or individuals within the simulated population) who can learn, modify, and improve their interactions with their environment (Batty 2007, Pinjari and Bhat 2010, Jin and White 2012,). As an emerging methodology that continues to find new applications in different disciplines, the field of ABM has become established as one of the innovative approaches to represent and simulate multi-scale urban dynamics (Batty, 2005; Arsanjani et al., 2013; Martinez and Morales, 2012; Zhang et al., 2010; Parker et al., 2003; Bithell and Brasington, 2009).
The conceptual model presented in this chapter therefore adopts the ABM approach because of the inherent capabilities and advantages of the technique, and in line with the overall direction of current research in the field of urban land use and travel behaviour modelling. Moreover, the choice of ABM derives naturally from the objective of this research to develop a disaggregate model to simulate residential-job location choice behaviour and the associated mobility patterns of heterogeneous urban households. Such a model, ought to be able to represent the complex micro-level behaviour of individuals’ choice decisions and property market dynamics over time in a spatially explicit framework. ABM simulation paradigm makes this possible by allowing for a bottom-up representation of preferences of heterogeneous agents (i.e. households) for spatially diverse goods (i.e. dwelling units, land parcels and employment locations), the idiosyncratic differences in decision-making processes as well as important feedback relationships (Magliocca et al., 2011; Murray-Rust et al, 2013; Huang et al., 2014). In addition, ABM has proved useful in representing bilateral transactions and competition among different actors in housing and property markets as the basis for demand and supply dynamics and price determination (see e.g. Ettema, 2011; Waddell 2001; Parker and Filatova, 2008; Filatova et al., 2011).
Furthermore, the approach provides a platform to hybridize the strengths of standard urban econometrics with the principles of complexity theory and bounded rationality/decision making under uncertainty. Consequently, the unrealistic assumptions of aggregate agent behaviour, spatial homogeneity and systems equilibrium under conditions of rationality and
perfect information can either be relaxed or completely overcome in ABM models of complex systems (Batty, 2008; Manson, 2006; Rasouli and Timmermans 2014a; Filatova et al., 2011).
Finally, ABM allows for the integration of a wide range of data sources including expert knowledge, empirical research, insights from existing models, survey data and geospatial data to model complex urban processes. Models could also be adapted to different contexts through the modification of model parameters and calibration based on the prevailing realities of the context (Murray-Rust et al., 2013; Janssen and Ostrom, 2006; Robinson et al., 2007).