Growing Collaborations
Ebenhoeh , 2004 ; Janssen & Anderies , 2007 ), source document analysis ( Zia & Koliba , 2015 ) and available databases ( Zia & Koliba , 2015 ). A growing arm of policy research is now applying ABMs to examine both policymaking and policy implementation ( Axelrod , 1997 ; Lempert , 2002 ; Janssen & Ostrom , 2006 ; Zia & Koliba , 2013 ; Choi & Robertson , 2014 ; Maroulis , & Wilensky , 2014 ). The outputs of these governance ABMs have focused on local level irrigation management decisions ( Janssen , 2007 ; Janssen & Anderies , 2007 ) and the prioritization of transportation projects at the regional scale ( Zia & Koliba , 2015 ). To date , no ABM of governance configurations has led to the generation of a predictive capacity to anticipate changes in governance arrangements over time and as the result of intentional interventions . While , ABM population models of civilizations ( Epstein , 2006 ) and urban centers ( Campbell , Kim , & Eckerd , 2014 ; Kim , Campbell , & Eckerd , 2014 ) have been used to regenerate historical settlement patterns in existing populations or forecast future patterns in simulated populations , these population models have not focused on predicting the evolution of present time , empirically observable social structures .
The ability to simulate the emergence of new arrangements for networked governance has been elusive . Policy systems operate as complex networks ( Weible & Sabatier , 2005 ; Lubell & Fulton , 2007 ; Henry et al ., 2010 ; McGinnis , 2011 ; Ingold , 2011 ; Provan & Lemaire , 2012 ), meaning they often feature extensive sets of human agents who apply complicated and varying decision rules , are subject to path dependency , and are continuously changing as agents review and revise decisions ( Axelrod & Cohen , 2000 ; Ostrom , 2005 ; Koliba et al ., 2010 ). Networks between individuals and organizations provide structure to policy systems . This structure then influences the policy choices and social impacts that emerge from the policy systems ( Provan & Milward , 1995 ; O ’ Toole , 1997 ; Koliba et al ., 2010 ). This relationship between structure and outcome gives governance networks two venues for change : the underlying network structure and the policy decisions and social impacts that are the consequence of that network ’ s functions . Our algorithm speaks to the changes in underlying network structures . As this structure evolves , so too will how the network behaves , such as in how financial resources and information flow through the network . Altering these flows can alter the outcomes for service delivery . However , no accurate measurement of how an intervention alters the network and its behavior can be made without accounting for the network ’ s organic growth and development . By capturing the processes of this organic growth , as measured through homophily , heterophily , transitivity , and preferential attachment , our algorithm provides the baseline necessary to being measuring the impacts of policy interventions to the network ’ s structure and behavior .
Study Limitations and Future Research
Our work is built from an analysis of one governance network that operates in a food systems context . In a strict sense , this places potentially severe limits on the external applicability of our results ; since our analysis is built on one type of network in one context , the results only apply to that type of network in that context . However , none of our research is based on the idiosyncrasies of food systems or governance networks . We parameterize our model using the ERGM from one , specific context . ERGM is applicable for any network context , making our approach flexible enough to apply to any
19