A probabilistic distribution theory of bureaucratic behavior: Policy preferences as random variables

The relationship between Congress and the executive branch agencies is complex and uncertain. Congressional committees authorize and appropriate an immense array of federal programs, each having unique political and fiscal attributes. However, formal models of legislative-bureaucratic interaction tend to simplify the relationship to such an extent that the generic agencies and the generic legislatures interact in a manner that deviates substantially from empirical observation. I examine the current development of formal models in political science and public administration finding that there is a need for the inclusion of an uncertain and probabilistic element that characterizes the lack of precision human beings have about each others preferences and attitudes. Since greater interaction provides higher levels of familiarity, each single interaction can be treated as a sample point from some defined distribution of preference on a specific policy. This construct allows the use of standard statistical tools within the model of legislative-bureaucratic behavior. I find that agencies generally have very little direct ability to manipulate knowledge of congressional preferences to achieve desired program budget and policy levels. The effect of limited information is to severely restrict agency control of the appropriations process. This contradicts the agency dominance literature derived from formal models of legislative/bureaucratic behavior. The scope and range of federal programs in the United States assures us that each person is in some way affected by the administration of the relationships determining policy goals and budget levels. By modeling the fundamental characteristics of a complex set of interrelated behaviors and structures, it is possible to enhance our principle knowledge of how the legislative and bureaucratic branches of government interact under these particular circumstances.