OPTIMIZATION IN PUBLIC POLICY - A RISK-BASED MULTI-PERIOD APPROACH

Abstract

In Public Policy, the objectives are multiple, competing and ambiguous. Trade-offs between objectives can be difficult to articulate. As such, it is more reasonable to adopt a risk-based approach – finding a course of action that has a high chance of achieving a basket of targets, taking uncertainty into account. We propose a novel optimization model to achieve this in the multi-period context, termed the Pipeline framework. Our model can tractably find such a policy if the uncertainty and decision variables are related in a manner we term pipeline dominance. It also lends the possibility of synthesis of analyses from earlier analytics stages at the most granular level. We utilize the model to re-examine Queueing Theory and illustrate it on the problems of bed capacity planning in healthcare, and manpower planning in public sector workforce management. While contextualized in Public Policy, the model may apply more widely to other multi-period problems.