Aggregate-Surplus Maximizing Monetary Policy

We study optimal monetary policy in a general heterogeneous-agent New Keynesian (HANK) economy. Instead of maximizing a social welfare function, the policymaker maximizes the joint surplus of all households, without taking a stance on how that surplus should be distributed. This is the policy that maximizes resources left over if winners from reform were to compensate losers. We derive the approximate optimal policy using linear-quadratic methods. The quadratic objective function nests the standard New Keynesian loss function and adds losses from incomplete markets and heterogeneous labor wedges. We show that optimal policy is the sum of time-dependent and state-contingent components. When labor wedges are homogeneous, monetary policy cannot improve household insurance: the time-dependent component is zero, and optimal policy coincides with the representative-agent New Keynesian (RANK) benchmark. Departures from RANK arise from heterogeneous labor wedges. In that case, the time-dependent component temporarily alleviates incomplete-market distortions but vanishes over time. The state-contingent component solves a dual-mandate problem, as in RANK, but with a higher weight on output-gap stabilization relative to inflation stabilization. In a calibrated model, heterogeneous labor wedges raise this weight by about 37 percent. We also find that a dual mandate with this higher weight achieves almost all of the aggregate-surplus gains from the full optimal policy and requires no compensatory transfers between winners and losers.