University of Oxford, Department of Sociology, and Nuffield College
Agent-based (AB) models are structural dynamical models characterized by three features: (i) there are a multitude of objects that interact with each other and with the environment, (ii) these objects are autonomous, i.e. there are no central, or “top-down” coordination devices (e.g. the Walrasian auctioneer), and (iii) aggregation is performed numerically. A crucial role in AB models is played by heterogeneity: this includes partial knowledge of the environment, and limited and differentiated computational ability. Rather than following axioms of logical consistency and computing fixed points, they follow simple heuristics based on psychological plausibility and ecological effectiveness: ecological rationality appears when the structure of the boundedly rational decision mechanisms matches the structure of information in the environment. These heuristics are evolved through learning and selection, with a crucial distinction being whether learning takes place at an individual or a social level.
This methodological stance removes a lot of technical constraints in model building, allowing for more flexibility in model specification. In particular, the use of simple learning mechanisms coupled with evolutionary mechanisms that allow “learning about learning” drastically reduces the difficulty of the choice problem of the agents, and consequently the computational complexity of the model. This permits to introduce more institutional details and more complex interactions.
However, the increased flexibility coming from an evolutionary perspective which departs from the consistency requirements of rational expectations (RE) equilibrium analysis has downsides: (i) assumptions are sometimes deemed arbitrary and disconnected from the literature, suggesting a return to the sort of anarchy that was lamented before the RE revolution; (ii) models often exhibit too many degree of freedoms, and are non-falsifiable; (iii) models are oftentimes poorly documented and hardly replicable; (iv) writing an AB model requires quite a lot of programming skills; code is often not re-usable and projects are not incremental, (v) models often lack a sound empirical grounding; when present, this is often limited to some ad hoc calibration.
Purpose of this short course is to introduce M.Phil. and Ph.D. students to the AB toolbox. In particular, the course will answer the following questions:
1) What are AB models? How did they develop?
2) Why and when should we want to use an AB model?
3) How can we build an AB model?
4) How can we analyse an AB model?
The course took place on March 21-22, 2016 at the Sociology Department, University of Oxford.
Lecturers: Dr. Matteo Richiardi (Institute for New Economic Thinking and Nuffield College), Prof. Francesco Billari (Department of Sociology and Nuffield College), Prof. Jakub Bijak (University of Southampton), Dr. Omar Guerrero (Institute for New Economic Thinking and Said Business School), Dr. Ross Richardson (Institute for New Economic Thinking).
Lecture 1: Introduction: features of the approach and historical development. Lecturer: Dr. Matteo Richiardi.
Laboratory 1: Introduction to Netlogo. Lecturer: Dr. Matteo Richiardi.
Lecture 2: AB modelling for demographic and sociological research. Lecturer: Prof. Francesco Billari.
Laboratory 2: Development of a simple demo model with Netlogo: the Schelling Segregation model. Lecturer: Dr. Matteo Richiardi.
Lecture 3: Agents’ behaviour: Expectations, Bounded rationality, Learning. Lecturer: Dr. Matteo Richiardi.
Laboratory 3: Experimentation and policy analysis with the Schelling Segregation model. Lecturer: Dr. Matteo Richiardi.
Lecture 4: Analysis of model behaviour: AB models as Markov chains, statistical equilibria, sensitivity analysis, estimation. Lecturer: Dr. Matteo Richiardi.
Laboratory 4: Network models of the labour market. Lecturer: Dr. Omar Guerrero.
Lecture 5: Simulation platforms: JAS-mine. Lecturer: Dr. Ross Richardson.
Lecture 6: Design of experiments for AB modelling. Lecturer: Prof. Jakub Bijak.