From Agent-Based Models to Network Analysis ( and Return ): The Policy-Making Perspective
quantity of money with the aim of obtaining full employment and stable prices . Implicitly , control requires the possibility of forecasting both the trend and the turning points of economic systems . Sadly enough , the history of control and prediction of economic phenomena is beset with failures . The record of failures is as long as the discussion concerning their causes .
Switches in policies are the consequences of this debate : theories used by economists have been held responsible for the ineffectiveness of their applications . A classic example is the discussion generated by the Lucas critique ( 1976 ) on large-scale macroeconometric models . He raised a crucial issue : the parameters of those models vary with the undertaken policies ( they are structural ) and therefore their predictions are likely to be misleading . Lucas ’ suggestion was to model the micro parameters of the models , that is to say preferences , technology constraints , and so forth , in order to understand what the agent would do as a consequence of a policy . The aggregation of individual responses would have generated the macroeconomic impact of the change in policy . Kidland and Prescott ( 1977 ) developed Lucas ’ thesis by operationalizing the search for micro foundation of macroeconomic models .
Dynamic stochastic general equilibrium ( DSGE ) models constitute the most recent development of this line of research . With respect to previous efforts they try to include historical time and random events . However , in order to assure solvability and simplicity DSGE usually neglects parts of the economic systems such as the financial markets and the banks whose importance has been remarkably highlighted by the last economic crisis .
With the persistence of the current financial crisis and resultant recession that models — especially DSGE — have failed to capture , the discussion concerning the need for new economic theories has gained new vigor . Complexity economics enters the stage by formulating the hypothesis that the cause of the policies failures is not to be found in theories ; rather it resides in their underlying ontology . It is the assimilation of the economy to a machine ruled by equilibrium that deceives economists . If we remove this cognitive habit , the importance of complexity-based policy is evident . It allows for procedural rationality , for explicit institutional settings , for the inclusion of historical time and it permits thorough comparisons among systems . All these features are definitely of immeasurable value to policymakers and their demand for nonconventional tool is now increasing . 8
The joint application of ABM and NA can meet this need by providing a series of information that were hardly available beforehand . By strengthening the connection between micro behaviors and emerging networks , agent-based networks can improve knowledge on how efficient and stable 9 networks come about . It is well known that the sets of efficient and stable networks do not always intersect ( Carayol , Roux , & Yildizo Glu , 2008 ; Jackson & Wolinski , 1999 ). The trade-off between the two is of crucial interest to the policymaker when it is a matter of creating a new or modifying an existent network .
In the absence of a way to model the real process of network emergence , scholars have often focused on notions of stability that do not depend on any particular formation process ( e . g ., pairwise stability ), thereby separating the stability of the network from the stability of internal dynamics . We strongly believe that policy could profit from a deeper knowledge of how stability relates to the rules that generate the network .
Notice that agent-based network models can also explore the tension between stability and dynamics . As explained in
88