Measuring Path Dependency in Politics through Text Mining

This paper demonstrates novel methodology to measure path dependency in policy-making. The method takes large text corpuses of policy documents, plans, scenarios, roadmaps, preparatory text and the like, and analyzes the flow of concepts, phrases and blocks of text as later documents inherit bits of earlier documents. The analysis proceeds by looking at relative frequencies of words and phrases in documents through a measure called term frequency-inverse document frequency, and translates this into two networks: the document-document network of similarities and dissimilarities describes how conceptual usage is inherited between documents and the concept-concept network describes discourses or network areas of concepts that are used together and start to institutionalize in the political language.

The paper demonstrates the method through an analysis of energy policies in Finland, and how various technologies, policy tools, and conceptualizations rose and fell over a fifteen year period from 1997-2012, from right after joining the European Union, through the rise of climate change negotiation and the Kyoto treaty, and into a period when energy policy came to be incorporated with climate policy.