ABSTRACT
This paper investigates China’s macroeconomic fluctuations by using the agent-based New Keynesian (ABNK) model. The model features bounded rationality, heterogeneous expectations, and adaptive learning. The model is estimated by the Bayesian method combined with the differential evolution algorithm and is analyzed using the impulse response, forecast performance, and variance decomposition approach. The estimation results show that in the real economy, in addition to rational agents, there are also boundedly rational agents with adaptive learning expectations. The impulse responses of macroeconomic variables to shocks are more sensitive and last longer than those from the dynamic stochastic general equilibrium (DSGE) model. The ABNK model exhibits better out-of-sample forecast performance than both the vector auto-regression (VAR) and DSGE models.
Acknowledgments
We would like to thank Ali Kutan, two anonymous referees and seminar participants at the 2018 International Finance and Accounting Conference at Northwest A &F University for their useful discussions and comments, and Regal J. for editorial assistance.