Public agencies, devolution, and transaction costs

Thursday, March 12, 2015: 10:00 AM
Federica Iorio, M.A. , Public Policy and Administration, George Washington University, Washington, DC
Stuart Kasdin, Ph.D. , George Washington University, Washington, DC
This paper examines Congress’ choice in designing new programs as to whether to rely upon internal, central government provision or external, state-based administration. In the US government, with advice from the relevant federal agencies, the legislature creates the structure that defines the program.  The choice of program design, whether a grant to states or direct provision, comes from the Congress.  We examine what considerations go into the decision to devolve program administration to the states. The decision is based on transaction costs: direct provision of a good or service generates additional governance costs for an agency.

            These governance costs depend on the existing infrastructure of a government. As agencies grow, the complexity and diversity of the information content grows and becomes less efficient. Decentralization offers the opportunity to enhance the efficiency of the information flow and decision-making process. We will analyze whether Congress turns to devolution as a tool for overcoming organizational complexity and inefficiency as agencies’ size and complexity increase.

            The alternative to central provision of the good or service is to delegate program administration to the states using grants. There is a risk of opportunism by the states: they may opt to use the grant funds for other purposes, including purposes not supported by the Congress. We look at the cost of devolution based the level of goal congruence between the principal (federal government) and the agent (states), in terms of spending and ideology.

            We incorporated data from several databases including the Public Budget Database (Budget of the United States Government), the Catalog of Federal Domestic Assistance (CFDA), and the U.S. Census Bureau to analyze grant programs from 1986 to 2009.

            The analysis will rely on ordered logit regressions based on the degree of devolution. Grants to states vary in the level of strings that are attached to them.  Some grants impose matching requirements, while others establish maintenance of effort requirements to reduce opportunities for state substitution of grant funding. 

            We expect that as the governance cost and the risks of opportunism fall, programs will be decentralized and designed as block grants. When instead current governance costs are lower and the risk of grant diversion is high, programs will be administered as direct federal provision.