Expectations, stagnation and fiscal policy: a nonlinear analysis

This paper formalizes a nonlinear stochastic New Keynesian model with agent forecasts of future GDP and inflation modeled using adaptive learning (econometric models with parameters updated over time). The macroeconomic model has multiple steady states. The steady state targeted by policy-makers is locally but not globally stable under learning. The well-known ''unintended'' steady state, with mild deflation and interest rates at the zero-lower bound, is not stable under adaptive learning. A severe pessimistic expectations shock can trap the economy in a stagnation regime, underpinned by a very low-level steady state, with falling output and inflation turning to deflation. A large fiscal stimulus may be needed to avoid or emerge from stagnation, and the impacts of forward guidance, credit frictions, central bank credibility, and policy delay are studied. Our model encompasses a wide range of outcomes arising from pessimistic expectations shocks and can be used to interpret the impacts and policy options following negative demand and financial shocks such as those associated with the Great Recession.