An Agent-Based Simulation of the Swiss Labour Market : An Alternative for Policy
argumentation and the ability to calculate the effects of different situations and hypotheses together .
According to Tesfatsion and Judd ( 2006 ), the main potential contributions of an ACE approach to labor research concern the possibility to model employers and workers as autonomous interacting agents , to define a preferential matching process ( simulating endogenous hires , quits , and firings ), to calibrate learning to data , to simulate the evolution of behaviors or interaction networks and , finally , to incorporate realistically detailed structural features as market protocols , policy rules , program eligibility requirements . If ABMs are built with sufficient richness and organization to be applicable to real policy questions , simulation can act as a sort of magnifier that may be used to understand reality in a better way .
However , these new approaches are challenging , because new modeling techniques such as ABMs are not established in the way that traditional ones are ( Lejour , Veenendaal , Verweij , & van Leeuwen , 2006 ). It is not possible , indeed , to assess the macroeconomic consequences of labor market policies using a pure micro-approach ; in the same way , a macro-approach does not permit to look at an individual level , making impossible to evaluate the impact of ALMPs on a precise target group . Agent-based labor market models allow instead the extraction of information based on aggregate outcomes , which are fully explained by the characteristics of the agents and the systemic structure of their actions .
Since these types of models simulate a complete set of individual observations they are also useful in addition to the microeconometric evaluation approach ex post . A final advantage is that the simulation at the individual level can help to formulate hypotheses on the micro-economic agent ’ s behavior using simple and intuitive rules , closer to the reality than the abstraction of rational aggregate models . This greater attention to micro-economic behavior characterizes the current ACE research .
3 - The Model
Following Dahlem ABM documentation guidelines ( Wolf et al ., 2010 ), the features of the model are structured as follows .
a . Overview
The proposed approach has been implemented with an integrated ( micro / macro ) simulation model . The model has been structured in modules ( until now , two modules were implemented : subsidized training and job displacement effects ), which let the user enable or disable some features to explore different policy options . Following the exposed perspective , we have developed a case study to test and validate the application of the proposed methodology and framework .
The implemented ACE model considers a virtual regional labor market with firms and worker agents . Firms are assigned different sectors and have sectorspecific skill requirements .
Let us consider a virtual word populated by a number nWork of worker agents . Each worker has assigned a nationality ( native or foreigner ), an initial random skill S i and an initial productivity
P i
.
There are nSect sectors in the virtual word and nFirm number of firms . The number nSect has to be ≤ nFirm . As in Neugart ( 2009 ) each firm has different skill requirements . Each firm is also assigned a random number of available vacancies .
18