This paper investigates how best to determine time-invariant policy rules in macroeconomic models with forward-looking constraints, where fully optimal policy is known to be time-inconsistent.  It proposes a new 'coefficient optimisation' approach that improves upon the timeless perspective...