The changing configuration of policy fields: Policy images and institutional effects

The policy subsystem, policy domain, or policy field is a traditional and well-established focus of research on the policy process. Yet the policy subsystem with well-established boundaries does not capture appropriately the manifold contests about the boundaries of fields we observe. Indeed, recent research has begun to take reflect the theoretical challenge of overlapping policy subsystems and of boundary-spanning policy regimes. This is an important contribution to understanding policy dynamics. In my paper, I will consider boundary processes as crucial for policy change in a more fundamental sense: The evolution and institutionalization of policy fields is essentially a process in which the meaning of the boundaries is defined. In my paper, I will analyse the consequences of institutional and discursive effects on the configuration of fields: The establishment of institutions will have an impact on the configuration of policy fields as well as changing policy images - how a policy is understood and discussed. A cross-cutting, boundary-spanning field require issues are persistently linked as a matter of political attention. My analysis will focus on policy fields that have been conceptualized as being “cross-cutting” at some stage in German politics: Environmental policy and immigrant integration policy were declared to be cross-cutting at certain times. The fields have seen different institutional trajectories: A ministry of the environment is well-established within the machinery of government, but a ministry of integration has not been founded. The configuration of the fields is a matter of the allocation of attention: This is not to be expected easily: My guiding hypothesis is that a policy image demanding a cross-cutting configuration will only have the respective effect on political attention if there are institutions that are sufficiently strong. The main empirical challenge will be to measure linking of issues as a matter of political attention. Taking immigrant integration and environmental policy within the German political system as cases, I will use topic modelling techniques to operationalize the links of issue attention. Topic models have been used by computational linguists for more than a decade, but the possibilities to measure topic correlations make it a very useful instrument for the problem at hand: The challenge to analyse empirically a very broad range of materials over longer stretches of time can be tackled using this technique for large-scale textual data. Thus, the point of departure is a theoretical problem, but I hope to demonstrate, how newer techniques for dealing with textual data help to solve a substantial empirical problem. The data basis for my paper will be a corpus of parliamentary debates of the German Bundestag, which covers all proceedings since 1949 and includes more than 230 million words. It is a resource that is being prepared in the PolMine project. The paper is also a use case how a political data science approach supports research on the policy process.