Agent-Based Models (ABMs) are versatile tools that only recently have begun to find applications in forecasting the unintended consequences of public policy decisions. For example, what is the impact of a 1% increase in the tax rate? What happens if the retirement age is raised to 70? What is the effect of increasing the pension contribution by 1%? ABMs can estimate the outcomes of such changes on labor force participation, individual savings levels, old-age income adequacy, job mobility, consumer expenditures, health care costs, and the government budget. In the context of the judicial system, ABMs can study the effects of harsher sentencing policies on the size of the prison population, the number of recidivists per year, the influx of first offenders, and so forth. And in law enforcement, ABMs can forecast changes in the crime rate that result from shifting police resources away from the traditional mission and into combatting terrorism. More generally, ABMs are ideal for quantifying response distributions in a world of uncertainty and heterogeneity.
This workshop builds upon and extends applications of ABMs that many of the participants have been working upon for decades. More specifically, this program aims to identify the technical, practical, and political challenges of using ABMs as a routine tool for implementing evidence-based public policy and for performing "what if" experiments to determine likely first- and second-order outcomes that might result from potential regulatory changes. The intent is to identify some public sector areas, presumably ones that are less politically sensitive, that are natural places for the academic community to partner with policy-makers to use ABMs to make quantitatively better decisions and to bring both policy-makers and academics together to work intensively on these topics.