Agent based Modeling for Macro economyPolicy Design

Silvano Cincotti

Description

EURACE Q B F W o rk sh o p , N ic e , D e ce m b e r 8 -1 0 2 0 1 0 1 Agent-based modeling for macro-economy policy design Silvano CINCOTTI University of Genoa - Italy

• scale agent


• agent


• summary ??


• based computational


• pioneered heterogeneous economic


• ?? policy design


• complex systems ??

Sujets

• Peter Cincotti
• Schelling

Informations

• Publié par: pefav
• Nombre de lectures: 18
• Langue: English
• Poids de l'ouvrage: 2 Mo

Extrait

EURACE
Agent-based modeling for
macro-economy
policy design
Silvano CINCOTTI
University of Genoa - Italy
1
www.eurace.org
QBF Workshop, Nice, December 8-10 2010 Summary
  Economy as a complex system
  The EURACE project
  Technological solutions
  Economic solutions
  Policy design example
  Final remarks
2
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 Economies as
Complex Systems
  Large efforts and interests have been
devoted to “Social” Complex Systems
  Economies are Complex Social Adaptive
Systems where agents make forward-
looking decisions
  There is a world economic crisis which
needs innovative ways of analyzing and
understanding the issues
  Agent-based Computational Economics
(ACE) investigates how aggregate
outcomes arise from the micro-processes
of interactions among many agents
3
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 Why the Agent-based Computational
Economics approach?
  Well suited to study economies as complex adaptive
systems
  Feedback from other disciplines
  mathematics, physics, engineering, biology, computing science
and software engineering
  Widespread dissatisfaction with the mainstream
approaches rooted on the representative agent
hypothesis and equilibrium theory
  Tremendous development in the last 25 years
  Joseph E. Stiglitz, Nobel Laureate 2001 in Economic Sciences,
pioneered heterogeneous economic agents by analysis of
markets with asymmetric information (Quarterly Journal of
Economics 1976)
  Thomas Schelling, Nobel Laureate 2005 in Economic Sciences,
pioneered ACE presenting an agent-based model on racial
segregation (Micromotives and Macrobehavior 1978)
4
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 EURACE Project Grand-Challenge
  Constructing a software platform to
perform large-scale agent-based
simulations on high performance
computer
  Developing a model of the European
economy, implemented on the simulation
platform.
  Reproducing statistical regularities at the
macro-level.
  Investigating/experimenting with
macroeconomic policy scenarios.
5
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 EURACE Context 1/3
  Macroeconomic policy design plays a
fundamental role in social welfare and requires
a coordinated application of economic policy
measures
  EURACE worked in the agent-based
computational economics framework
  Economics is represented through
heterogeneous interacting agents
  Actors
  Households, Firms, Banks, Etc
  Markets
  Goods, Labour, Credit, Monetary, Assets, Etc
  Policy Institutions
  Central Bank, Government, Regulatory Bodies, Etc
6
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 EURACE Context 2/3
regulations Regulatory Central
bodies Firms bank
financing labour reserves
revenues
taxes
interests interests
Government
Labour Assets
Goods market market Banks
market
taxes
savings
goods
wages dividends,
interests interests Households public
spending
7
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 EURACE Context 3/3
  X-Agents are the essential software elements in
EURACE
  To define an economic agent as a X-Agent, we
need to define:
  The markets on which the agent can be
active.
  The activities of the agent on each market.
  The decisions the agent has to make on
each market.
  The messages an agent can exchange.
  The X-agent framework is adopted for modeling
economic systems and extended to run on
parallel clusters of computers
8
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 Why is EURACE different ?
  Its main distinctive and innovative features are:
  its closure
  the presence of real and financial markets
  the wide use of empirically documented behavioural
rules
  the different levels of time and space granularity
  asynchronous decision-making across different agents
  the explicit spatial structure
  the very large number of agents
  the use and development of innovative software
frameworks
  User-transparent scalability on different computational
arcitectures
9
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010 EURACE Objectives
  From a scientific point of view
  The study and the development of multi-agent models that
reproduce, at the aggregate economic level, the emergence
of global features as a self-organized process from the
complex pattern of interactions among heterogeneous
individuals
  From a technological point of view
  The development, with advanced software engineering
techniques, of a software platform in order to realize a
powerful environment for large-scale agent-based economic
simulations
  From a societal point of view
  Outstanding impact on the economic policy design
capabilities, allowing “what-if” analysis in order to optimize
the impact of regulatory decisions that will be quantitatively
based on European economy scenarios
10
www.eurace.org
Quantitative Behavioral Finance, CIF–NICE, December 8-10 2010