Identifying Delegation in Congressional Bills

Theories of delegation abound in political science, economics, and public administration, but there have been few empirical tests of these theories. One major roadblock is the lack of a comprehensive data set on acts of delegation. I take a necessary first step toward testing these theories by training a neural network to classify whether sections of bills are delegating public authority to the bureaucracy. Using a convolutional neural network architecture, I achieve excellent performance that vastly improves over the baseline.