Apart from conventional interest-based studies of regulatory policy, the structure of policy design itself is vital to explaining the long-run robustness of policy choices. Causal reasoning and goal setting act as a framework for assessing these choices. Linking policy instruments to the goals they seek to accomplish will be problematic, particularly if linkages are obscured by uncertainties, and inferences are complicated by non-linearities, feedbacks, and differences in the causal inferences drawn by experts and the public. Somewhere in the middle of this causal chain, policy-makers must choose the stages at which to insert both operational goals and policy instruments. Policy choice is rarely a stark one between one causal stage and the next, but instead involves constructing a portfolio of strategies and deciding how to assign relative weights to alternative causal stages without creating too many contradictions or overlaps. In following policy solutions over time, the sequencing of policy design is found to be important in establishing robust policies. Conflicting demands emerge to relate policies to both outcomes and causes, subject to a host of intervening considerations regarding the efficiency of policy choice. These tradeoffs are not static. Causal knowledge that had previously resided only in isolated expert communities will be diffused, albeit slowly and imperfectly, to policy-makers, the judiciary, and the public, leading to changes in perceptions of problems and hence solutions. Policy design that ignores more complicated expert causal

(cont.) Moving towards greater reliance on expert knowledge is clearly desirable, the key question this dissertation seeks to explore is the rate at which such a move can be made without threatening the legitimacy of a program. The three cases explored here, air pollution, antitrust, and climate change each presents a different set of actors, forum for regulation, and historical legacy. Each poses a dilemma for policy-makers over how to intervene credibly and effectively. The air pollution case and antitrust cases both offer over a century of experience with policy design over multiple jurisdictions. Lessons from history highlight the importance of: experimenting at early stages; tying goals, not instruments, to outcomes; placing instruments at many different causal stages; introducing expert understandings slowly; and anticipating long delays for adaptation or reform. Finally, the difficult case of climate change is presented and the lessons for causal reasoning and goal setting are applied in the hopes of identifying plausible alternative climate policies.