Enhancing ABM into an Inevitable Tool for Policy Analysis
Neo-classical models take into account many aspects of a real economy including commodities , production , growth , and money ( King , Plosser , and Rebelo 1988 ). However , they mostly address centralized market economy and therefore are not suitable for other types of decentralized markets .
The underlying assumptions in NE models such as full rationality of parties , complete information , and certainty also create concerns about their reliability on the insights they provide . Although this line of research is gradually moving toward higher uncertainty approaches , rationality of individuals and complete information are the necessary pillars in equilibrium modeling .
Traditional Game Theory ( TGT )
Game theory is the most frequently applied tool for understanding actor behavior and decision making in policy problems ( Gibbons 1992 ). The fundamental concepts in game theory are players , strategies , and payoffs . A player may be an individual or a composite actor that is capable of making choices . Strategies are lists of consecutive actions for a player , or functions assigning actions to each decision point of a player on the basis of previous actions by the opponents .
The limited number of actors and outcomes , the joint product of separate choices , and the actors being aware of their interdependence , make game theory useful for policy analysis ( Scharpf 1997 ).
However , there are a number of strong assumptions in TGT that make it less suitable for many policy problems ( Scharpf 1997 ): perfectly rational actors , complete information , self-interest , and unlimited computational and cognitive ability . Another limitation of game theory is that it does not provide a macroperspective explanation of policy choices , which is commonly required for policy analysis ( Scharpf 1997 ). One other limitation of TGT is that the number of interactions between actors is very limited ( interactions between three agents ( Moss 2001 )) while for policy problems , hundreds or thousands of actors may be involved .
System Dynamics
System dynamics ( SD ) is a computational simulation approach which has its roots in differential equations . With this approach , a system is described using a system of equations with which future states of the system are derived from its current state . In system dynamics ( SD ), real world problems are represented in terms of stocks , flows , and information . SD ignores single events and entities and takes an aggregate perspective ( Borshchev and Filippov 2004 ).
The ease of use and availability of packages and tools makes system dynamics one of the most popular computer-based analysis tools among policy analysts . However , as the simulations grow bigger , the number of assumptions increases , introducing additional questions of validation to support the reliability of the simulated model .
The high number of assumptions thereby also makes the model less flexible . Essentially , SD is a one-layer approach which means that the focal system is simulated as an indivisible whole . It does not take into account the fact that the actual system consists of individual people and it is their behavior and reaction that actually results in global outcomes .
67