Research Article
Nowcasting Vietnam’s RGDP Using a Kernel-Based Dimensional Reduction Method
@INPROCEEDINGS{10.1007/978-3-030-92942-8_10, author={Thanh Do Van}, title={Nowcasting Vietnam’s RGDP Using a Kernel-Based Dimensional Reduction Method}, proceedings={Nature of Computation and Communication. 7th EAI International Conference, ICTCC 2021, Virtual Event, October 28--29, 2021, Proceedings}, proceedings_a={ICTCC}, year={2022}, month={1}, keywords={Data mining Nowcasting Big data Dimensionality reduction Kernel tricks PCA}, doi={10.1007/978-3-030-92942-8_10} }- Thanh Do Van
Year: 2022
Nowcasting Vietnam’s RGDP Using a Kernel-Based Dimensional Reduction Method
ICTCC
Springer
DOI: 10.1007/978-3-030-92942-8_10
Abstract
The gross domestic product growth rate (RGDP for short) is one of the most important macroeconomic indicators often used for making economic policies and planning production and business development plans by government agencies and enterprise communities. Research to improve the forecast accuracy of this indicator has always been of interest to researchers. In Vietnam, this indicator is only released quarterly.
It is no longer appropriate to forecast the RGDP according to predictors at the same frequency as this indicator because, in the time interval between two quarters, there may be some political, socio-economic events occurring that have a substantial impact on many economic activities that cause the change of the RGDP in the current quarter and the next quarters. It is necessary to use another new forecast approach to overcome this limitation.
The purpose of this article is to build a model to nowcast the RGDP on a large dataset of predictors at higher frequencies than the quarterly frequency. Such a model is developed based on the dynamic factor model. Unlike previous studies, the factors in the built model are extracted from the input dataset by a variable dimension reduction method using kernel tricks and based on an RMSE-best model. The article also proposes a ragged-edge data handling method and reinforcement learning method, suitable for the regression method used to build the nowcasting model of the RGDP indicator.
