A Nonlinear Dynamic Factor Model for Financial and Macroeconomic Data

We propose a nonlinear dynamic factor model to examine the impact of external shocks and internal forces on macroeconomic and financial data fluctuations. This model permits nonlinear dynamics, enables asymmetric, state-dependent, and size- dependent responses to shocks, and generates time-varying volatility and asymmetric tail risk behavior. We find evidence of nonlinear dynamics in a U.S. application on measuring the credit cycle. The nonlinear factor stimulates credit growth during booms but hinders recovery after crises, with shocks having longer, amplified effects during credit crunches. An extended model that separates first- and second-order effects of the factor reveals that credit cycles exhibit state dependence in which shocks that spur growth during modest credit conditions trigger sharp busts during excessive expansions,thereby predicting how credit conditions can flip from boom to crisis.