Contribution for Innovations in Financial and Economic Networks

Determining the effects of labor institutions on macroeconomic performance is a central concern of economic policymakers. Differences in labor institutions have been conjectured to be a key explanation for observed crosscountry differences in the level and persistence of unemployment, in the distribution of income and wealth, and in growth rates for labor productivity and GDP. European OECD countries over the past twenty years have tended to rely on administered wages and legislated job protection and have experienced sluggish job growth and persistently high unemployment. In contrast, the United States has had a relatively 2 1 Evolution of Worker-Employer Networks and Behaviors Under Alternative Non-Employm more flexible, less regulated labor market and has achieved much greater job growth and relatively lower unemployment rates. This has led many Euro-pean policymakers to argue the need for reforms in their labor institutions. it is difficult to obtain conclusive empirical results regarding how labor institutions affect economic performance. Regression methods relating changes in labor institutions to economic outcomes quickly tax degrees of freedom. This problem is compounded if one institution's impact depends on the presence or absence of other institutions. Also, labor institutions are inherently endogenous. For example, governments continually revise labor institutions in response to economic and political pressures. This endogeneity makes it difficult to interpret the validity of empirical investigations. In recognition of these difficulties, Freeman (1998, pp. 19-20) suggests that agent-based computational modeling might offer a promising additional way to study the impact of labor institutions, particularly from a market design perspective. Tesfatsion (1998,2001,2002a) reports some preliminary work along these lines. An agent-based computational economics (ACE) framework is used to study path dependence, market power, and market efficiency outcomes for a labor market under systematically varied concentration and capacity conditions. ACE is the computational study of economies modeled as evolving systems of autonomous interacting agents (Tesfatsion,2002b,2003). 1 In this study we conduct an ACE labor market experiment to test the sensitivity of labor market outcomes to changes in the level of a non-employment payment. Our ultimate objective is to understand how the basic features of real-world unemployment benefit programs affect labor market performance. 2 However, as will be clarified below, a human subject experiment was run in parallel to this computational experiment as a check on the reliability of our findings and the adequacy of the learning representations for our computational workers and employers. To facilitate this initial benchmark check, a deliberately …