Bureaucratic agencies create hundreds of regulations every year that have the force of law and outnumber statutes roughly ten to one. Most decisions about how those new regulations will be structured are made during the proposal development stage, which occurs after an action has been initiated but before it is published in the Federal Register for public comment. However, there is little research on this stage of regulatory development. This gap is particularly problematic since some scholars have argued that interest groups are most influential at lobbying the bureaucracy during this stage.

Both methodological difficulties and schisms in the theoretical literature have limited existing research on the proposal development stage. Agencies vary rather dramatically and idiosyncratically in how they develop proposals. This variation poses a challenge to cross-agency studies. Therefore, the majority of empirical studies limit their focus to a single agency or a handful of rules. Most large-N studies focus only on the later stages of rule development, after the political leadership has already signed off on major regulatory decisions.

These methodological difficulties are exacerbated by theoretical discordance. Interest group participation in government is a vibrant field in political science, spanning decades of research and multiple theoretical traditions. Two of the most dominant traditions, principal agent theory and policy subsystems theory, provide substantively different accounts of the roles that interest groups play in democratic governance.

Rather than attempting to prove or disprove either of the two theoretical approaches mentioned above, this paper seeks to improve the debate by exploring under which conditions each theory is likely to hold true. To achieve this end, the study relies on William T. Gormley’s (1986) well-known public administration typology, which uses policy salience and complexity to predict the types of issue networks that form around policy areas.

This paper extends Gormley’s typology to predict the applicability of principal agent and policy subsystems theories. Both theories generate a series of predictions about interest group participation—specifically, the identity of participants, their means of participation, and their subsequent influence on policy. This paper argues that these three dependent variables (identity, means, and influence) are shaped by the salience and complexity of a policy area. To support this claim, this paper relies on several different methods including simple statistical tests, hierarchical modeling, and graphical displays.

This study examines stakeholder participation in the development of 6,372 regulatory proposals initiated by 25 federal agencies between 2000 and 2005. It significantly contributes to existing empirical research by presenting an innovative semi-automated data collection method using recently-digitized federal publications as well as an original survey of federal rule writers.