Search
Browse By Day
Browse By Time
Browse By Person
Browse By Policy Area
Browse By Session Type
Browse By Keyword
Browse Artificial Intelligence Presentations
Program Calendar
Sign In
Search Tips
State variation in SNAP (food stamp) policies has long been used for quasi-experimental research, yet we know surprisingly little about why states choose different policies in the first place. This paper explains state policy choices in federal programs through the lens of an office-seeking agent constrained by the state budget and administrative capacity. I study five major state options: Broad-Based Categorical Eligibility (BBCE), vehicle exemptions, in-person interview waivers, simplified reporting, and transitional benefits. Using best subset selection, a machine learning method for model selection, together with permutation tests, bootstrap resampling, and estimates restricted to pre-adoption years, I identify the most robust predictors of each option. Eligibility expansions such as BBCE respond to rising unemployment, but, as prior studies suggest, they also depend on political ideology. For options that ease administration, administrative capacity is the main driver: states with higher administrative costs and error rates are less likely to adopt the very options that would reduce their burden. Taken together, these findings suggest that state discretion allows federal programs to be tailored to local economic and political conditions, but this discretion may be limited for states whose administrative capacity is already strained.