According to decades of political behavior research, economic or labor-market risk should motivate support for social welfare policy, especially social insurance (e.g. Compton & Lipsmeyer 2019; Cusack, Iversen, & Rehm 2006; Rehm 2009). Public policy research, on the other hand, has focused on policy design and the impacts of different policy features on social and economic outcomes (e.g., Jacobs & Mettler 2018; Laenen 2018; Rosenthal 2019). Complex social programs transferring cash benefits to households, such as unemployment insurance, exhibit notable variation in benefit entitlements, duration of eligibility, and stringency of qualification criteria. We ask whether individuals hold distinct preferences over these multiple dimensions of social policy design. Do individuals have multidimensional policy preferences? Further, how does labor-market risk shape these preferences? Drawing from behavioral economics (e.g., Frederick, Loewenstein, and O’Donoghue 2002; Carvalho, Meier & Wang 2016), we argue that economic risk shapes individual time preferences by promoting present-biased decision making. Preferences for temporal policy dimensions, such as duration of eligibility, should therefore be more sensitive to economic insecurity. We test our expectations using a survey experiment in the nationally representative 2020 Cooperative Congressional Election Study in the US, where we manipulate labor market risk to determine multidimensional preferences for unemployment insurance.