China | Forecasting modeling for China's inflation

Summary

We are trying to assess a large group of forecasting models’ performance in predicting China’s inflation. Both linear and structural forecasting models are discussed, estimated and evaluated based on some typical criteria such as RMSE, MAE and Theil-U.

Key points

• Key points:
• The recent African Swine Flu outbreak in China, COVID-19 and global oil price dipping motivated us to assess a large group of forecasting models’ performance in predicting China’s inflation.
• Both linear and structural forecasting models are discussed, estimated and evaluated based on some typical criteria such as RMSE, MAE and Theil-U.
• Our results suggest that unlike widely found in the literature that complicated models are difficult to significantly beat the naïve random walk model, by adding monetary and economic indicators into the linear model, as well as the VAR-type models are found to significantly out-perform random walk model in China’s inflation forecasting case.
• Moreover, VECM, by including the error-correction term could also out-perform the reduced-form VAR to predict China’s inflation.
• These findings indicate that monetary and economic indicators indeed contain useful information for predicting inflation both in-sample and out-of-sample. In addition, adding structural forms, together with error-correction term in VECM, could improve models’ forecasting performance as well.

Geographies

Topics

Tags