Combining Machine Learning and Econometrics to Examine the Historical Roots of Institutions and Cultures

Machine learning (ML) and associated computational advances have opened entirely new avenues for the processing and analysis of large data sets, especially those containing text. In this chapter, we show how ML can extend the scope of historical institutional and cultural analysis. We first provide an overview of some of the scattered existing literature using ML methods to study historical institutions and culture. We then use our own work on pre-nineteenth-century English caselaw and print culture to illustrate the possibilities and the challenges in using ML as a tool for systematic quantitative inquiry into the origins, change, and impact of institutions and culture. We highlight the power of ML for distilling core facts from large corpora and generating datasets amenable to analysis using conventional econometric analysis. We demonstrate how our work allowed us to explore the deep institutional roots of specific legal and cultural ideas, analyze the coevolution of ideas within caselaw and culture, examine the impact of caselaw on economic development both before and during the Industrial Revolution, and discern critical junctures in England’s legal and cultural development. Focusing on historical institutions and culture, the chapter illuminates the types of lessons that can be learned from the application of ML in new institutional economics. It also suggests a pathway that researchers applying ML to history can follow when trying to find a practical, implementable set of methods among the proliferation of new techniques that is usual when an area of research is in its infancy.