Federal redistribution programs are not implemented uniformly even when designed to be uniform. Building public support for redistribution at the local level is a prerequisite, which means the counterplan results in uniformity and the aff doesn’t

Delegation is part and parcel of politics because elected officials who write statutes and make policy rarely personally implement these directives on the ground. The central challenge in the design of government institutions is thus to provide street-level bureaucrats with sufficient authority to utilize their expertise while avoiding having this discretion be abused to pursue divergent policy goals. This dilemma remains at the heart of the study of bureaucratic politics. However, the dilemma presupposes a trade- off between the discretion thought necessary for bureaucratic effectiveness and democratic responsive- ness (e.g., Epstein and O'Halloran 1999, 6). In reality, there are ample arenas in which discretion and respon- siveness go hand-in-hand - with street-level bureau- crats exercising their authority in highly responsive ways. Across many policy areas, including Medicaid (Weissert 1994), occupational health and safety (Scholz 1986; Scholz, Twombly, and Headrick 1991), labor law (Schmidt 2002), Social Security disability insurance (Keiser 1999), and immigration (Lewis et al. 2013), local workers appear to use their discretion to align realized policy outcomes with the local community norms and values, even when their formal responsibility is to implement otherwise identical statutory language. Such behavior is a different sort of bureaucratic responsiveness, one that may conflict with the clear intent of the legislators and elected executives who crafted the relevant legislation. In these and many other settings, discretion potentially brings about too much responsiveness rather than too little of it from the point of view the laws' authors.

The present study examines bureaucratic adaptation in the context of the national food stamp program. Using administrative records from the federal Department of Agriculture and various state social service agencies, I document substantial local variation in food stamp enrollment rates and show that these differences correspond to local voter attitudes toward government economic policies. The existence of such differences in food stamp access is surprising because Congress took a number of steps to bring about national uniformity in the program in the late 1960s and early 1970s, when lawmakers centralized many key decisions over its design in Washington, DC. The second part of my anal- ysis, which takes advantage of state-level variation in the way the program is administered, offers an explana- tion why these centralization efforts did not ultimately produce their desired effect. Most US state governments retain administrative responsibility over food stamps. In these states, potential beneficiaries submit their applica- tions at local offices, but the caseworkers who staff these offices report directly to state-level agencies. In a dozen remaining states, by contrast, administrative respon- sibility for the program is devolved further to county governments, which are overseen by locally elected offi- cials. Surprisingly, I show that the degree of administrative centralization does not affect the level of observed bureaucratic responsiveness to the preferences of local voters, which is high in both institutional settings.

These findings contribute to an extensive literature on bureaucratic decision-making in the context of social welfare service provision (e.g., Keiser and Soss 1998; Schram et al. 2009; Weissert 1994). Much of this work, however, focuses on joint state-federal cash income support programs, which provide for substantial local discretion over eligibility, benefit levels, and sanctions enforcement (Haveman et al. 2015). Indeed, there is evi- dence that local discretion was built into these programs by design (Lieberman 1998). By contrast, food stamps are a nationally funded federal benefit with a single set of formula-based eligibility criteria, leaving little statutory room for local discretion.1 Indeed, as Currie (2003) notes, the food stamp program is designed to offset state and local variation by "providing a minimum federal safety net” (p. 206). Despite this objective, how I nevertheless find substantial variation in local implementation of this nominally federal program.

Documenting this variation and examining how affected by the design of political institutions also contributes to a broader understanding of public administration. First, the findings suggest that local political culture and norms play an important role in shaping the behavior of civil servants, highlighting an explanation for the intrastate variation in policy implementation that has been documented for a variety of programs. Although such variation is well-known in the literature, existing accounts generally focus on incentives of program administrators and their subordinates and organizational design as the key explanatory factors, overlooking the important role played by the local political context. Second, since the potentially troubling normative implications if otherwise similarly situated individuals receive disparate treatment from government officials due solely to place of residence, studying whether different institutional arrangements affect the uniformity of implementation is also valuable for identifying opportunities for socially desirable reform. Finally, accounting for local variation in policy implementation is important for correctly interpreting existing empirical research on democratic responsiveness and for understanding the precise mechanisms through which responsiveness is achieved. I return to these broader questions in the concluding section of the article.

After reviewing the relevant literature and providing an overview of the food stamp program, my analysis proceeds in two parts. First, I utilize administrative data from 36 states to document variation in food stamp participation at the county level. I combine these administrative records with a new dataset of county-level public opinion toward government redistribution, constructed using multilevel regression with poststratification (MRP), and show that differences in observed enrollment rates correspond to local voter attitudes: In counties where opposition to government redistribution is higher, food stamp enrollment is statistically and substantively lower. The analysis further shows that the strength of this relationship does not vary significantly between states with centralized and devolved program administration.