Nowcasting Monthly Chinese GDP with Mixed Frequency Data: A Model Averaging Approach

Real-time nowcasting plays a crucial role in the fields of economics and finance by providing timely information, improving decision-making, supporting policy formulation, and assisting financial transactions. It holds significant importance in understanding and adapting to economic changes. The official release of chinese GDP data typically experiences a certain time delay and requires some time to acquire and publish. Nowcasting GDP can provide more timely economic indicator forecasts, bridging the information gap caused by the lag in data release. Conventional penalized least squares methods yield satisfactory nowcasting results when fitting GDP data, but they neglect the influence of residual autocorrelation. In this paper, we propose several nowcasting methods for monthly chinese GDP with mixed frequency data. We demonstrate that our proposed method outperform the conventional penalized methods in nowcasting chinese GDP. Furthermore, after removing residual autocorrelation, the JMA method has the smallest RMSE and achieves the best nowcasting performance.