Econometrics and Policy Analysis:
A Critique and an Alternative


CPM Report No.: CPM-07-177
By: Scott Moss
Date: 22nd May 2007


Results from agent based social simulation studies have shown that reasonable and well validated representations of individual behaviour and social interaction generate results that are inconsistent with both economic (or any social) equilibrium and the conditions of application of classical statistics or econometrics. At the same time, both econometricians and other scientists studying climate change recognise that important behavioural and social phenomena can only suitably be described qualitatively and not by numerical representations. Declarative rulebased techniques for representing individuals with software agents supports formal qualitative representations and integrate readily with quantitative biophysical models. This renders such agent based social simulation techniques suitable for such difficult, long run policy analyses as those addressing climate change impacts and social policies for mitigation. Since econometric and other economic modelling techniques cannot appropriately be applied to such problems, the only available means to render the formal components of climate change policy analysis more complete is to use agent based social simulation models. A particular advantage of agent based social simulation is the ability to capture the unpredictability of social process outcomes which will be especially important in the development of climate change scenarios.

econometrics, agent-based social simulation, equilibrium, evidence, policy modelling

Accessible as: