3. Classification
• By technique (GA, NN, Monte Carlo,
Stochastic)
• By application area (forecasting, learning,
macro policy)
• Other possibilities?
4. Classification by scientific hierarchy
METHODOLOGY
THEORY
MODELS
DATA
?
Different disciplines use this
differently
What is taken as given
5. New typology
• Casts new light on kinds of simulation
• Allows simulators to orient with respect to
each other: descriptive not prescriptive
• Casts light on the attitudes of different
disciplines to methodology, theory,
models and data and, indirectly, on
expected attitudes to simulation
6. Economic application
• Objective, quantitative data
• Models based on equations linking data
• Theory based on optimisation within a
system of constraints
• Methodology: positivist, individualistic,
humanistic
• Theory/methodology distinction only made
clear by study of history of economics
8. Data
• Calculation(s)
• Estimation (Monte Carlo)
• POLIMOD
• “Aggregation” (SPSS)
• Even given “objective” data, the goals
and success of these tasks are well
defined/uncontroversial
9. Models
• No longer uncontroversial: Can’t take
model as given (“more of the same”)
• Generalisation to n dimensions (but
not “difficult” cases)
• Addition of types
• Addition of noise
• “Safe” applications are an
overwhelming majority … why?
10. Theory and methodology
• Problem separating these at moments in
time
• Uncertainty versus risk
• Altruism
• Bounded rationality
11. New possibilities
• Embodied preference
• Different models, relative success, evolution
• Autonomy of environment, other agent process
(not exogeneity!)
• Alternative knowledge and decision
representations (hybrid agents?)
• Integration of partial disciplinary theories
• Simulation removes technical defences of
traditional economic method. Empirical defences
were always contentious. What is left?